Precision Agriculture
Tamás, János
Created by XMLmind XSL-FO Converter.
Precision Agriculture
Tamás, János
TÁMOP-4.1.2.A/1-11/1-2011-0009
University of Debrecen, Service Sciences Methodology Centre
Debrecen, 2013.
Created by XMLmind XSL-FO Converter.
Tartalom
Tárgymutató ....................................................................................................................................... 1
PREFACE ........................................................................................................................................... ii
1. 1. AGRO-ECOLOGICAL DATA ACQUISITION ........................................................................ 3
2. 2. PROBLEMS AND SOLUTIONS OF FIELD SCALE ............................................................... 6
3. 3. DATA MANAGEMENT ........................................................................................................... 9
4. 4. SAMPLING STARTEGY ........................................................................................................ 16
5. 5. SPATIAL ANALYSIS TO EXPLORE SPATIAL HETEROGENITY ................................... 26
6. 6. BIOMASS EVALUATION BY AIRBORNE HYPERSPECTRAL IMAGE SPECTROSCOPY 39
7. 7. EARLY DETECETION IN PLANT PROTECTION .............................................................. 45
8. 8. THREE DIMENSIONAL DYNAMICAL SOIL WATER REGIME ...................................... 50
9. 9. SOIL CHARACTERIZATION AT LARGE SCALE .............................................................. 54
10. 10. ADVANCED IRRIGATION WATER SCHEMA DESIGN ................................................ 60
11. 11. WATER STRESS ................................................................................................................. 78
12. 12. RUNOFF ON AGRICULTURAL WATERSHED .............................................................. 86
13. 13. TRAFIC CONTROLL .......................................................................................................... 94
14. 14. APPLIED LAND CHANGE MODELLING ........................................................................ 99
15. 15. ANALYZE OF BIOMASS PRODUCTIVITY BY TIMESERIES .................................... 104
iii
Created by XMLmind XSL-FO Converter.
Az ábrák listája
4.1. 1. Figure Sugar beet yield t/ha ................................................................................................... 21
4.2. 2. Figure Sugar content (digestion%) ........................................................................................ 21
4.3. 3. Figure Calcium carbonate % (Scheibler) ............................................................................... 21
4.4. 4. Figure Nitrate-nitrogen mg/kg ............................................................................................... 22
4.5. 5. Figure AL-K content K2O mg/kg ........................................................................................... 22
4.6. 6. Figure11 pH and 0.05 mole EDTA+0.1 mole KCl soluble Cu .............................................. 23
4.7. 7. Figure12 pH and 0.05 mole EDTA+0.1 mole KCl soluble Cd .............................................. 23
5.1. 8. Figure Spatial distributions of residual data of the measured values and the surface calculated by
kriging. .............................................................................................................................................. 28
5.2. 9. Figure Average yield h-scatergram (Distance 10-1 m) ........................................................... 28
5.3. 10. Figure Semi-variogram and applied search direction/angle of the sugar beet yield samples 28
5.4. 11. Figure Sample taking areas of copper and pH of original sample numbers (left- middle) as well as
spatial distribution of copper with reduced sample points estimated by pH variogram (right) ........ 31
5.5. 12. Figure Variogram surface of original copper concentration (left) and pH (right) and cross
variogram surface of both (below) sampling sets ............................................................................. 31
5.6. 13. Figure Spatial mean of analysed data and visual methods of mapping ............................... 34
5.7. 14. Figure Eigenvectors and rotations angels of raw sugar (A), digested sugar (B), and yield (C)
where origo is the mean centre . (1) north direction ......................................................................... 34
6.1. 15. Figure Hyperspectral imaging systems can capture spectral, spatial, and radiometric information
in same time ...................................................................................................................................... 39
6.2. 16. Figure Spectral statistical curves of regional interest .......................................................... 40
6.3. 6. Table Spectral narrowband indexes of cereals ....................................................................... 41
6.4. 17. Figure NDVI 3D surface of winter wheat block .................................................................. 42
6.5. 18. Figure Spatial-Spectral cross-sections of winter wheat block ............................................. 42
7.1. 19. Figure Cercospora leaf spots -Cercospora beticola) ............................................................ 46
7.2. 20. Figure The first PCA of the homogeneous (1., 2.) and the heterogeneous (3.,4.) spatial distribution
parcels. .............................................................................................................................................. 47
7.3. 21. Figure Deterministic coefficients (R2) of Cercospora leaf spots (Cercospora beticola) (%) 47
7.4. 22. Figure Cercospora leaf spots map and the investigated parcels ........................................... 47
7.5. 23. Figure Cercospora leaf spots map and the investigated parcels (n=14). .............................. 48
8.1. 24. Figure: Left picture - Spatial distribution of soil plasticity, KA (apple trees illustrated as points on
the surface layer); Right picture - spatial distribution of soil density (soil penetration resistance *100kPa;
blanked sites means above detection level of the penetrometer). ..................................................... 51
8.2. 25. Figure The soil moisture tensions at 3 June and 28 July 2010. ............................................ 51
9.1. 26. Figure box-and-whisker plots of plasticity, pH and capillary rise of water ......................... 56
9.2. 27. Figure Spatial distribution of capillary rise for the grid based random point sampling method 57
9.3. 28. Figure Spatial distribution of pH for the grid based random point sampling method .......... 57
9.4. 29. Figure Spatial distribution of plasticity index for the grid based random point sampling method
........................................................................................................................................................... 57
9.5. 30. Figure Hyperspectral reflectance data most sensitive to Arany plasticity index .................. 58
10.1. 31. Figure Location of the experimental site ........................................................................... 60
10.2. 32. Figure Applied sensors to evaluate soil-plant environment ............................................... 60
10.3. 33. Figure Leaf Area scanner and field GPS ........................................................................... 61
10.4. 34. Figure Time domain reflectrometry (TDR) a sensitive water content (vol %) .................. 62
10.5. 35. Figure NDVI Time series .................................................................................................. 63
10.6. 36. Figure Major influential factors in determining crop coefficient (Kc) .............................. 64
10.7. 37. Figure Major influential time series component of PCA in determining crop coefficient (Kc)
........................................................................................................................................................... 66
10.8. 38. Figure Relationship between crop coefficient (Kc)and normalized difference vegetation index
(NDVI) values and tertiary equations for the 5 selected sites ........................................................... 66
10.9. 39. Figure Daily reference crop evapotranspiration (ETo) values modified by c constant factor and
the average ETo values in the 19 ha experimental area .................................................................... 67
10.10. 40. Figure Flowchart of CROPWAT FAO MODEL ............................................................. 70
10.11. 41. Figure The amount of effective rain with different methods ........................................... 71
10.12. 42. Figure ET0 changes in 2009 ............................................................................................ 71
10.13. 43. Figure . Air drought in 4 months in 2009 ........................................................................ 72
iv
Created by XMLmind XSL-FO Converter.
Precision Agriculture
10.14. 44. Figure Irrigation water demand of autumn ripening pear varieties on sand ..................... 72
10.15. 45. Figure Irrigation water demand of summer ripening pear varieties on sandy soil ........... 73
10.16. 46. Figure Data integration of precision water management ................................................. 74
11.1. 47. Figure Stem gage schematics (Source: Dynamax manual ................................................. 79
11.2. 48. Figure Transpiration of 1 m2 canopy and the climatic conditions ...................................... 80
11.3. 49. Figure Thermography image on apple trees (oC) .............................................................. 82
11.4. 50. Figure ................................................................................................................................. 82
12.1. 51. Figure Two typical digital elevation model of the water management .............................. 87
12.2. 52. Figure Relief pattern estimation based on contours of microrelief (Eastman,2010) .......... 88
12.3. 53. Figure Inserting a new point based on local min/max of 8 directional parabolas .............. 88
12.4. 54. Figure The improved sink filling process .......................................................................... 89
12.5. 55. Figure Boreholes data preparations with groundwater modelling tool and visual presentation of
soil properties in real 3D environment ............................................................................................. 89
12.6. 56. Figure NDVI map of Tedej farm ....................................................................................... 90
12.7. 57. FigureMaize before harvesting (left), canopy of maize (middle), NDVI of maize(right) .. 91
12.8. 58. Figure NDVI – LAI and LAI(g) correlation in maize ........................................................ 91
13.1. 60. Figure The number of class intervals in case of different correction sources. ................... 96
14.1. 61. Figure. Land Cover Gains and Losses ............................................................................... 99
14.2. 62. Figure Land Cover Net Changes ..................................................................................... 100
14.3. 63. Figure Land cover change map ........................................................................................ 101
14.4. 64. Figure Distance from 4-times frequency of excess water inundation .............................. 101
14.5. 65. Figure Transition potential map (sample) ........................................................................ 102
14.6. 66. Figure Soft prediction to land transforms ........................................................................ 102
15.1. 67. Figure NDVI time series of Nyírlugos in 2006 (January-December from to left by lines) 104
15.2. 68. FigureThe time series of NDVI data sources and linear trend line (2001-2006, n=72) ... 105
15.3. 69. Figure Time steps of biomass production periodic component and substances ............... 107
v
Created by XMLmind XSL-FO Converter.
A táblázatok listája
4.1. 1. Table Yield and quality of sugar beet ....................................................................................
4.2. 2. Table Soil analysis data .........................................................................................................
5.1. 4. Table The Pearson correlation values of the yield and the soil pH. .......................................
5.2. 5. Table Spatial weighted mean and standard weighed distance of sugar beet and soil data .....
7.1. 7. Table Spectral sampling of test size ......................................................................................
9.1. 8. Table Descriptive statistics of soil parameters .......................................................................
10.1. 9. Table Typical soil parameters of the sample plot ................................................................
10.2. 10. Table Irrigation schedule of pear orchard on sandy soil in 2002 .......................................
10.3. 11. Table Irrigation schedule of pear orchard on sandy soil in 2009 .......................................
11.1. 12. Table Canopy and leaf area of Jonagold and Granny Smith apple species ........................
13.1. 59. Figure Average standard deviation values in case of different corrections .......................
vi
Created by XMLmind XSL-FO Converter.
16
19
30
32
46
55
62
73
74
80
96
Tárgymutató
1
Created by XMLmind XSL-FO Converter.
PREFACE
During the last two decades there has been a very significant progress in the precision agriculture (PA). Its roots
and development go back to non military GNSS applications after the Persian Gulf War, and the agriculture
discovered the opportunities in the new technology. In one industry, where the effects of the environment mean
the highest risk on the quality and quantity of the yield, farmers all over the world desire continuously to get
more detailed information about the cultivated area and its every part.
The issue of the Information Society in the agriculture speciality is the so-called precision agriculture. The
precision agriculture is the most popular name of this farming form. At the same time, this system is also named
by some enhanced segments fist of all under the influence of English literature. The site specific production
more emphasizes the environmental claim and the sustainable economic form character, while the site specific
technology (SST) relates to the technological system utilizing properly the production area. The Spatial
Variable Technology (VRT) name enhances also technology aspects less considering data collection and spatial
decision supporting system (SDSS). The technology controlling by satellite, the Satellite Farming name makes
focus on the significance of global positioning system (GPS) and remote sensing, but point out to the
importance of the ground sensors and operational job computers even less.
The instruments to measure the parameters of cultivated area were limited previously. The local extremities
were covered by the data of mean samples, thus there were limited tools to dismiss this anomalies. GPS
instruments with high accuracy make real time steering possible. Nowadays the steering of agricultural machine
without people is easily attained with high level automata technology. The environmental expectations were
increased toward farmers in the last two decades because of known reasons. The headway of the Precision
Agriculture made a reply to this challenge. More compound sensors can probe the relationship of the soil,
nutrition, water and vegetation; and control the charging of the irrigation water nutrition and pesticide while
moving. The ground, airborne, and spaceborne remote sensing can help farmers and other experts to perceive
the environmental impacts of the agricultural production on the narrower and wider surrounding of the own
farms. The PA means a tool to modern farmers to provide quality food for the global food demand improving
environment.
The measurements controlled by GPS open up new opportunities to explore the agro-ecological potential of the
production area with very high spatial resolution (sub meter) and timescale. On the basis of measurements the
results of spatial decision support system (SDSS) making in the common GIS environment
On the basis of measurements the results of spatial decision support system (SDSS) making in the common GIS
environment can protect resources with high precision and it could help to manage in a sustainable way.
At the same time, this demands a new way of thinking – paradigm, where the function of the analysis and
evaluation with the environmental information technology is higher and higher beside traditional knowledge.
This sets new tasks to developers. The information standardization of the agricultural mechanical engineering is
unavoidable for different manufacturers. Nowadays, in the case of the telecommunication it can be considered
common that the field instrument comprises wireless technology. Prospectively the PA will spread in animal
husbandry and horticulture much more intensive than present. The acceleration of the developments in PA is
also expected in the area of the agricultural biomass production and the bioenergetics. To the development the
newer technology possibilities of internet (WEB2 technology) can open improvement opportunities for
engineers. The spreading of Web based mapping, mapping development (GoogleMap, GoogleEarth, Bingmap),
and mobile internet (smart phone) induces promising expansion of PA. It is interesting and notable that the
consultant systems of PA built to cheap smart phones appeared inundeveloped economic regions recently
passing over some middle technical steps. For today the predictions come true that the users’ imagination keep
the future development of PA within limits.
In this book the author summarizes theoretical and practical experiences in PA performed with colleagues
throughout more decades. The book is recommended to MSc students becoming agro-, bio- and environmental
engineers, farmers, and developers expecting that the book contribute to form collective spatial thinking and to
inspire readings to new idea.
The editor thanks the quoted authors and anonym colleagues without who this book would not have been
possible for mutual works.
ii
Created by XMLmind XSL-FO Converter.
1. fejezet - 1. AGRO-ECOLOGICAL
DATA ACQUISITION
Agricultural production is an activity of high economic and environmental risk, nowadays especially. Handling
of widening scientific knowledge and up-to-date technological appliances needs rethinking of foregoing theory
and practice; according to some researchers, total paradigm change is needed. The “sustainable development”
theory of Agenda 21 made up by the Rio Summit (1992) means long-term tasks for all branches of economy, so
as for agriculture. As a result, the change of land use planned from economic and environmental points of view
would realign the overall view of the Carpathian Basin for centuries in a territory where the ratio of agricultural
regions is one of the highest (72%) in Europe. Risks of decisions refer to any period or area can be decreased by
having more and more data and analysed information. Agricultural activity, as an open ecological system, has a
hardly elaborative data requirement even by the present information technological devices [1]. About half of the
input data interfering the results are changed during a fragment of the phenological cycle, that can not be
described exactly by such widely used ecological models as CERES[2] or CROPWAT [3] during particularly
extreme changes validated for Central-European model border conditions. These results in more data need in
space and time, which has properly been satisfied so far by the development of digital sensors, but the practice is
not able to keep up with the manipulation integration and interpretation of the collected data. Similar process
occurs on the field of data mining technology performed on Internet, where effective solution of information
production still has to be waited for. Seemingly, this causes contradiction, the decision maker uses smaller and
smaller part of the information available, but this data amount – in absolute sense – grows at a highly increased
rate in time. During many decisions the incomplete interpretation assumes the reliability of data or their interval
and value of uncertainty are not known. As a result, decision makers have no numerical risk values actualised
according to the information built up from the data, thus the farmer dares to undertake lower and lower risk
level in the smaller decision space.
How one can escape from the trap of this lack of data and data dumping that sometimes comes up at the same
time in destructive cases? The key element of the decision-making process is improvement of the interpretation
techniques that must be optimised during the whole data flow.
Optimisation of data mass, data quality and set of information effectively support the sustainable development at
the given decision level, where the decision maker knows the levels of uncertainty of the data, the error
spreading processes and the undertaken decision risk in space and time.
In decisions, the economic market information was dominant up to the mid 1990’s in the Hungarian
industrialised agriculture. The structure of ecological-environmental information data was determined by this
dominantly agrochemical point of view in Hungary. From amongst the ecological and economic environments
in spite of the renewing economic crisis the European Common Agricultural Policy is calculably stable, in
which the agro-ecological and rural development elements have been upgraded. The ecological environment
keeps on changing capriciously, in fact the rapid urbanization and the conversion of the environment causes
more often natural hazards (draught, floods, exceeds waters etc.) with significant damages. In the agro-decision
space the economic and classical crop production and animal husbandry questions mean smaller, while the
ecological tasks bigger role [4]. Determination of potential of natural resources, monitoring of the changes is an
agro-information task even nowadays in the ideal several-meter spatial resolution as well as in the several-hour
temporal resolution, whose mass distribution is a big challenge and high-tech task for establishment of an
information society.
Therefore amongst the interpretation techniques the possibilities and limitations of the agro-ecological digital
data acquisition modelling that is prevailing for sustainable development will be overseen.
In the document Agenda 21 a proposal was born to establish the conditions of Sustainable Agriculture and Rural
Development (SARD) that aims the establishment of food production and food safety as well as protection of
the resources at the same time. Previously, in Hungary after the world energy-crisis of the 1970’s several
research and governmental programmes were started to evaluate the volume and utilization of natural resources
used by agriculture [5, 6] . Results of the project called “Agro-ecological Potential of Agriculture on
Millennium” led by Láng et al. (1983) were systematised into a professional system by the Agro-Ecological
Integrated Information System (AIIR). AIIR is a professional system that manages and builds on settlement,
soil, meteorological, land use and crop production data [7].
3
Created by XMLmind XSL-FO Converter.
1. AGRO-ECOLOGICAL DATA
ACQUISITION
Its main advantage was that resources were evaluated by up-to-date, model methods. During the that time of
“industrial-like” production purposes, where the goal was the hospitalisation of productivity (y), the limited
possibility of sustainability of resources was pointed out and the questions of yield safety – risk as well as
production site conditions were discussed.
According to the model factors of the methodological researches led by Harnos [8] the stationary state of the
soil characteristics (x) in time (t) were assumed, where t is only one phenological cycle in evaluation of
sustainability, while x can change depending on time [x(t)]. A group of climatic factors can change rapidly and
randomly (ξ), while the other group applied in agricultural technology (u) containing all elements of production
(seeds, nutrients, plant protection, soil cultivation, irrigation and crop rotation etc.), can change in every
phenological cycle. The effect of the production site to the yield (ŋ) was described by the following equation:
ŋ = ŋ(x, ξ, u).
In the sustainability model it was interpreted that productivity of a given x production site is not decreased by
the u-type agricultural technology at t+1 time. That is, quantity of yield under the actual agro-technological
solution will not decrease in a further production cycle.
yact(t) ≥ y(t+1)
Productivity is expressed by E(ŋ(x(t), ξ, u)), where E is the deviation caused by the weather, whose value can be
characterised by the F(y) = P(ŋ(x, ξ)) distribution function, where F(y) = P(ŋ(x, ξ)) ≤ y means that yield of the
reference plant (y) will not be higher than the average yield expressed in t/ha. In the relation the P random
variable is 0 ≤ P ≤ 1. The approach above gave a possibility for the probability establishment of yield loss (pv)
as well. Hindrance of introduction of the theoretical model was that the analytical form of the ŋ(x, y, u) function
describing the connection amongst yield–production site–agricultural engineering and the x(t+1) = g(x(t), u(t))
function describing the state change of the production site were not exactly known.
In practice the studies were focused on: genetic and agricultural engineering development and stochastic change
of weather, where the county units were used as spatial references, under the actual technical conditions.
Agricultural technology is expressed by the change in time of ŋ. Accordingly the change of average yields in
time (y1, ŋ2) can be expressed by the ŋ(t, ξ) = y1(t) + ŋ2(t, ξ) potential function, where y1(t) describes the
stochastic of genetics and applied technology, while ŋ2 is that of the weather. During the almost two decades
passed since this studies the role and function of agriculture and its methods of research have been changed. The
need of data and the methodology of evaluation of the regional and local ecological information systems build
on large-scale field information systems are upgraded. Unfortunately, its introduction to practice delays because
of lack of intellectual and financial capitals supports.
1. Test questions:
1. How can you characterize agro-ecological potential of the optional farm?
2. What are the main disadvantages of industrialised agricultural production?
3. What type of factors should be take account of yield modelling?
2. References
1. Csáki Cs., Harnos, Zs., Rajkai, K., Vályi, I. [eds.] Hungarian agriculture: development potential and
environment. . J.K Parikh. Dordrecht, The Nederlands : Martinus Nijhoff Publishers, 1988. IIASA. old.: 253–
265. In: Sustainable development in agriculture.
2. Kovács, G.,J., Nagy, J. [eds.] Test runs of CERES-maize for yield and water use estimation. Soil, plant and
environment relationships. In Current plant and soil science in agriculture. J. Nagy. Debrecen, : University of
Debrecen, 1997. old.: 115–119.
3. Tamás, J., Nagy, J. [eds.] Evaluation of irrigation schedules by CROPWAT FAO model. Soil, plant and
environment relationships. Current plant and soil science in agriculture. . J. Nagy. Debrecen : University of
Debrecen, 1997. University of Debrecen. old.: 115–119.
4
Created by XMLmind XSL-FO Converter.
1. AGRO-ECOLOGICAL DATA
ACQUISITION
4. Swaminathian, M.S.From Stockholm to Rio de Janeiro. The road to sustainable agriculture. Madras : M. S.
Swaminathan Research Foundation, 1991. old.: 68.
5. Cascio, J., Woodside, G., Mitchell, P.ISO 14000 Guide, The New International Environmental
Standards. pp. New York : McGraw-Hill, 1996. old.: 217.
6. Láng, I., Csete, L., Harnos, Zs. Resources Adjustment and Farming Structures. European Review of
Agricultural Economics. 1988., 15. kötet, old.: 2–3.
7. Láng, I.Mezőgazdaság agroökológiai potenciálja az ezredfordulón. [Agro-ecological potential of agriculture
on millennium.]. Budapest: Tárcaközi Bizottság Jelentése, 1983.
8. Harnos, Zs.Az alkalmazkodó mezőgazdaság rendszere. [Systems of sustainable agriculture]. – Módszertani
kutatások. Budapest: Kertészeti és Élelmiszeripari Egyetem, 1991. old.: 91.
9. Buzás, I.A káliumellátás szerepe a sikeres cukorrépatermesztésben. Budapest: INDA 4231, 2001.
10. Poel, van der P.W., Schiweck, H., Schwartz, T.Sugar technology. Berlin: Dr. Albert Bartens KG, 1998.
11. Buzás, I. [ed.] Impact of soil and nutrient management on sugar beet yield and quality in Hungary. A. E.,
Buzás, I. Johnston. in: Balanced plant nutrition in sugar beet cropping systems for high yield and quality Proc.
of workshop of Beta R.I. : IMPHO, 2000. old.: 8.
12. L., Buzás I. - Kulcsár.Tápanyag-gazdálkodásunk és a cukorrépa-termesztés. 1999., 2. kötet, old.: 64-70.
5
Created by XMLmind XSL-FO Converter.
2. fejezet - 2. PROBLEMS AND
SOLUTIONS OF FIELD SCALE
1. Soils and land use
Evolution of soils is affected by geological, climatic, topographical and biological features as well as by the age
of soils. Analysing soil evolution factors, effects of human activity on soil cannot be neglected. This effect has
been particularly intensive during the last several hundred years. On the one hand, this human activity resulted
in the improvement of productivity of soils, while on the other hand, in certain regions, it accelerated soil
degradation processes. All of the above-mentioned soil factors exerted their effects together in the Carpathian
Basin and their interactions determined the form of appearance, the physical, chemical and biological features of
the certain soils. Processes in the soil represent opposite impact pairs that are in dynamic balance in space and
time [1]. These balance processes can shift to one or the other direction, strengthen, change periodically in time
or may have shorter periodic effects that can be temporary or permanent.
On the AGROTOPO 1:100 000 scale digital soil map of Hungary, Várallyay [50, 51] distinguished 3310
polygons based on 10 soil parameters. In case of 1:10 000 – 1:25 000 field-scale levels the characteristic values
of the heterogeneity of spatial patterns grow at a highly increased rate. With the increase of the spatial resolution
the standard deviation of attributive data increased as well. In this dimension the digital topographical data was
partly processed, which is the main data source for survey of soil resources and for performing an agroecological model. At the determination of soil characteristics on field, the representativity of parameters got for
a general purpose soil survey are interpreted by Webster [52] as how much it can explain from all the variances
of the data. He asserts that in case of soil-physical characteristics about half of them, while in case of some soilchemical characteristics less than one-tenth of all variances are revealed by the survey. It must be overcome by
the soil-characteristics estimation based on environmental correlation.
From amongst the soil characteristics the biological, chemical and physical parameters can be measured in
growing uncertainty order, respectively. Among the soil-physical characteristics, the water- and heat
management values of the same sample can be differentiated at several thousand-fold measured either in situ on
the field or in the laboratory.
Uncertainties of the soil-chemical measurements are grown in order of magnitudes by the preparation and the
extraction procedures prior to measurement, in comparison to the interval of value of analytical device
measurements. The up-to-date field sensors and remote sensing data significantly lessen the volume and the
deviation of sample preparation and of laboratory errors. While ten years ago several weeks were needed from
soil sample taking to the evaluation of the result, nowadays this can be lessened only to some hours. The
measurement limits for example in agro-chemistry in case of field devices decreased to ppm, while in case of
measurements in a laboratory to ppb level value.
Digital technology defined the numerically experimental fact, that the pedological changes in the Great Plain
regions of Hungary are significant even in vertically in sub-meter level and due to the intensive land use it is
much faster than it was estimated previously. This fosters new research directions. Applying simulation models
can help in determination of direction and volume of the processes. A group of wide-spread used models
contains such regional, for example erosion models like USLE and WEPP, while their other group deals with
the nutrient management of the given site, e.g. SOIL-SOIL-N or with water movement at point scale [24, 35].
Some of these models are for research, need a big volume of input data and provide a very detailed process
description, while the others are practice-orientated, robust models.
2. Water sources
After the Netherlands, the second largest area being threatened by floods and exceed waters can be found in
Hungary. Frequency of extreme values in the statistics of hydrology has grown trend-like during the past
decades[34]. The extreme changes in the water source together with the permanent contamination burden
increase the variability of the quality of waters as well [38, 39].
Runoff of exceed waters closed out from flood areas, the volume, period and frequency of inundation are
consequences of series of accidental hydrological events and phenomena [40]. Probability of floods in space and
6
Created by XMLmind XSL-FO Converter.
2. PROBLEMS AND SOLUTIONS
OF FIELD SCALE
time can be basically determined by two ways: on the one hand, as volume of frequency values of inundations,
on the other hand, as size of factor maps of exceed waters. Besides, combination of these two can occur, which
means the subjective correction of the first method by using the second one.
With the re-classification of the digital maps of such characteristics that take part in formation of exceed waters
and show relative spatial stability (for example aquifer hydraulic conductivity and maximum storage capacity of
the aquitard, convexity of the micro relief, critical possible depth of the underground water, land use), thematic
maps can be prepared. From their subsequent overlaying an exceed water risk map can be prepared [4].
There are several possibilities for spatial distribution and interpretation of the infiltration factor. The simplest
way would be making an isometric map if the co-ordinates of the sample taking sites were known. But this
solution might come up only if we would like to interpret a soil-genetically or soil-physically homogeneous
area. Then by choosing the proper interpolation technique a continuous surface can be modelled, that is suitable
for interpretation to “all points”. But the heterogeneous regions have several anisotropies.These anisotropies
have such impacts, that significantly change the regularities of spatial development of the infiltration factor.
These are mainly the factors that play role in formation of soils (for example the topography – i.e. the deep
areas have low water conductivity).
Therefore, if a map that pictures a detailed physical kind of soil is available whose profile analysis data (for
example the mechanical structure) are known – that are base of preparation of the map –, then through the
revealed relations thematic maps displaying the hydraulic conductivity of the bigger regions can be prepared.
These maps are true with the given input structure and boundary conditions. The spatial pattern of the change of
the accumulation conditions overpasses the speed of change of soil characteristics in time [45]. Mathematical
survey of time series analyses is indicated a fast developing field.
3. Climate
Material of knowledge of Hungary's climate potential is as revealed as that of soil sources. Climate potential for
country size and meso-regions were determined to study the maximum yield [2, 42, 48], which provides spatial
modelling possibilities. Based on their study, a part of the total radiation energy that reaches the plants (Q 0) is
photosynthetically active energy (Qp), whose certain part (ε) is used by the plants for biomass production (Y b)
Burgos [5] concludes the value of the maximum possible utilisation of radiation from the estimated value of the
energy reaches the Earth. Based on it, Varga-Haszonits et al. [49] calculated the value of ε to 22–23% for
Hungarian conditions. Relying upon the Campbell method [7] performed also by him, he compares the energy
of photons relate to the medium wave of the photosynthetically active radiation with the amount of energy
bound chemically in 1 mol material. Its value is about the same: 22%.
Climate exerts its effect – besides the energy conditions – through the fluctuation of water resources. During
modelling the Aridity Index that expresses the rate of potential evapotranspiration and precipitation during the
vegetation season, has an average value of 1.6–2.8 in case of winter wheat in the 1951–1991 period of time.
According to the time series analyses calculated by Harnos et al. [16], for the period of 1951–1983 there are 3–5
draughty years in every 15 years, which cause yield loss of higher than 5%. From amongst the three cases the
yield loss once is 10–15%, and once is over 15%.
Eight research stations were included in the above-mentioned survey. But the field scale forecasts are very
uncertain. The present destiny of field measuring stations of meteorological service is not able for supervised
teaching of satellite multispectral images. They cannot supply continuous real-time data for field-scale waterand heat management analyses but at the same time both the spatial and spectral resolutions improved in order
of size. The highest improvement is to be waited in the field of practical data supply of radar technology. In
2002 in research-level the hyperspectral measuring technology also appeared in Hungary [23].
4. Biological resources
7
Created by XMLmind XSL-FO Converter.
2. PROBLEMS AND SOLUTIONS
OF FIELD SCALE
Bio-diversity and homogenous monoculture of an agricultural crop seems to be an irresolvable contradiction.
The strongly fragmentised ecological islands poor in species increase the energy need of the whole agricultural
region. In Hungary the spatial mosaic-like structure of soils is confirmed by the land use databases originated
from the CORINE 1 : 50 000 and 1 : 10 000-scale air photos. But complete interpretation of data source will be
the future task of the agro-ecological modelling. According to Horváth [17], by a spatial extension of the soil–
plant–weather–pest system that can be considered homogeneous within a small plot a space specific complex
ecological model can be obtained [18, 25, 26]. The models of the ecological system are strongly connected to
and dependent from each other, however, each of them works by itself.
The ecological examination of spatial processes is very important in life-systems of most species. Population
ecology is a branch of ecology that studies the structure and dynamics of populations. Populations can be
defined at various spatial scales. Local populations can occupy very small habitat patches like a puddle. A set of
local populations connected by dispersing individuals is called a metapopulation. Populations can be considered
at a scale of regions, islands, continents or seas. Even the entire species can be viewed as a population [3].
Populations differ in their stability. Some of them are stable for thousands of years. Other populations persist
only because of continuous immigration from other areas [10].
Sharov [37] introduced the definition of physiological time via population ecological research to describe
biological maturity and time relationships. Spatial processes of ecological analysis are very important in lifesystems of most of the species. They may be so significantly modify system behaviour that local models would
be unable to predict population changes. Several methods are used for description of the spatial processes such
as random walk, diffusion model, dispersal mechanism, metapopulation analysis etc.
Random walk is simulated after several time steps until the distribution of organisms becomes close to the 1 or 2
dimensional normal distribution. The diffusion models can be applied to any initial distribution of organisms.
The combination of long- and short-distance dispersal mechanisms is known as stratified dispersal.
Metapopulation is a set of local populations connected by migrating individuals.
One of the main strengths of the geo-statistical analyses is the investigation of spatial variance and correlation.
5. Test questions:
1. Can you evaluate the impacts on soil evolution process?
2. Can you evaluate the impacts on soil degradation process?
3. What are the main reasons? Why the local spatial uncertainties of factors of hydrological cycle are relative
high?
4. What is the mean aridity index?
5. Why problem, if the biodiversity is getting to decrease on your farm?
6. How do you keep or increase biodiversities on your farm?
7. How do you mean the agricultural landscape degradation?
8
Created by XMLmind XSL-FO Converter.
3. fejezet - 3. DATA MANAGEMENT
With geo-statistics, the GIS analyst gains a wide range of tools to detect and describe expressions of spatial
dependency in a study area through sample data sets. (Very simply, spatial dependency refers to the extent to
which neighbouring points have similar attributes.) These tools contribute to an exploratory analysis of data by
helping to describe the nature of spatial dependency in the study area. These descriptions may then be used to
build predictive models for full surfaces. Any geo-statistical project begins, prior to sampling, with obtaining as
much knowledge as possible about the distribution characteristics of the phenomenon under study. In cases
where one does not have direct control over the production of sample data, the project begins by gathering
ancillary information about the study area, the sampling methods, and the sampling scheme. Next, if a geostatistical analysis is to be fruitful, it is necessary to examine the spatial arrangement of data samples visually
and produce summary statistics that reveal characteristics of the sample data distribution. Detecting and
interpreting special features, characteristics, or abnormalities of the data set are the first steps of exploratory data
analysis, the success of which will influence subsequent interpretations of geo-statistical measures of variability
and continuity. In addition to displaying a map of the sample locations with different palettes, one can analyse
histograms of the attributes and obtain a statistical summary of the data. With these results in hand, better
interpretations of spatial structure are likely as one begins geo-statistical analysis [13].
An exploration of the modeller (as called Spatial Dependence – IDRISI, Exploratory Spatial Data Analysis –
ArcMap) which provides tools for measuring spatial variability (or its complement, continuity) in sample data.
Model fitting is to build models of spatial variability with the assistance of mathematical fitting techniques.
The variogram surface is a representation of statistical space based on the variogram cloud. The variogram cloud
is the mapped outcome of a process that matches each sample data point with each and every other sample data
point and produces a variogram value for each resulting pair. Typically, uncovering spatial continuity is a
tedious process that entails significant manipulation of the sample data and the lag and distance parameters.
With the Spatial Dependence Modeller, it is possible to interactively change lag widths, the number of lags,
directions, and directional tolerances, use data transformations, and select among a large collection of modelling
methods for the statistical estimator. If the degree of spatial dependence decreases equally at the same rates for
all sample pair separation directions, the model design is isotropic. With model fitting, the continuity structures
suggested by the semi-variograms produced can be interpreted as well as any additional information we have
obtained. The parameters for the structure(s) will describe the mathematical curves that constitute a model
variogram. These parameters include the sill, range, and anisotropy ratio for each structure. When there is no
anisotropy, the anisotropy ratio is represented mathematically as a value of 1. The sill in model fitting is an
estimated semi-variance that marks where a mathematical plateau begins. The plateau represents the semivariance at which an increase in separation distance between pairs no longer has a corresponding increase in the
variability between them. Theoretically, the plateau infinitely continues showing no evidence of spatial
dependence between samples at this and subsequent distances. It is the semi-variance where the range is
reached.
The goal is to decide on a pattern of spatial variability for the original surface. To carry out this goal
successfully with limited information requires multiple views of the variability/continuity in the data set. This
will significantly increase the understanding and knowledge of the data set and the surface the set measures.
Finally, in the last section of the analysis, kriging, IDW and other spatial estimators can be used to test models
for the prediction and simulation of full surfaces. There are more types of kriging, which present different
results.
Ordinary kriging is known to be a Best Linear Unbiased Estimator, because it assumes that constant mean (μ)
unknown, simple kriging where μ is known and universal kriging where μ(s) is some deterministic (trend)
function. The final result of kriging will be to produce two images, a surface of kriged estimates and a surface of
estimated variances. Kriging estimates a new attribute for each location (pixel) on the basis of a local
neighbourhood. Cokriging is another useful geo-statistical tool that uses an additional sample data set to assist in
the prediction process. Cokriging assumes that the second data set is highly correlated with the primary data set
to be interpolated. Cokriging is useful, for example, when the cost of sampling is very high and other (cheaper
or available) sample data can instead [20].
The interested reader should consult Cressie [11], Isaaks and Srivastava [19] and later chapters for additional
explanations of kriging and general geo-statistical questions.
9
Created by XMLmind XSL-FO Converter.
3. DATA MANAGEMENT
1. Data conversion or comparability
In case of field-scale data collection and processing the agricultural application of machines having satellitebase GPS means technical breakthrough. The technological decision support system building around it is called
precision agriculture (PA). It has an almost decade-long past concerning from the 1992 – the year of the first
American conference. Standards for agro-environmental management systems have the same short past in
practice. Both areas, from the view of methodology and applicable techniques, were efficiently inspired by the
quite rapid and continuously traceable development in information technology. The more stringent quality
assurance and increasing agro-environmental protection requirements made the need of elaboration of complex
environmental indicators obvious for analysis of environmental management and for the loading capacity
analysis of the environment as a living system.
Similarly to the large-scale and efficient control functions of industrial closed systems, in open agricultural
environment system analysis of material and energy flow and life cycle assessment to be shown by us become
possible. The conditions of maintenance and effective operation of the environmental management systems are
the continuous and reproducible measuring, tracing, improvement and development that help to increase the
environmental effectiveness of production. For this purpose proper knowledge of the site and environment of
the agricultural production are essential.
For the common resolution of the above-mentioned topics, an alternative is offered by the simultaneous
introduction of the precision production system and the environmental management system. The precision
agriculture is about to increase the effectiveness of production with optimisation of raw materials’ (water, seeds,
chemicals etc.) making plant growing more economic.
Determining elements of precision agriculture are: high-precision, continuous site determination, geographical
information and remote sensing tools of analysis and the highly automated fieldwork [46].
This information system of precision agriculture can be the basis of establishment of the environmental
management systems. Its first step is the survey and the evaluation of the environmentally active factors of the
given production process and their effects. Knowing them the two most important fundamental principles, the
continuous measuring–monitoring and the in parallel made continuous improvement–development can be
realised.
For establishment of the environmental management systems the life cycle assessment provides help that
determines the materials and energies used during the production as well as quantity and quality of the released
potential toxic materials and wastes within the investigated production system. On the basis of these results the
environmental effects of production can be estimated. The ISO 14040-49 family of standards contains the
standards of life cycle assessment [9]. According to the examination standards, the life cycle (Hungarian
Standard ISO 14040, 1997) is the subsequent, connecting stages of the influence system of a product, from
purchasing the raw materials or from the formation of the natural resource to the reuse or waste disposal.
One of the problems of evaluation is the fact that effects of directly not measurable input materials must be
calculated to directly not comparable effects. In particular, this data integration process is often based on
techniques for the management of uncertainty.
Bayesian probability theory has been proven to be sensitive to inaccuracies in the input probabilities. The
Bayesian model often requires that events are independent of each other. This assumption is rarely true in real
life. Neural networks and Fuzzy sets have an alternative approaches to handle spatial (mapped) and measured
uncertainty. Keller [21] and Ultsch [47] give several examples of how to apply neural networks to
environmental problems.
Uncertainty in any data layer will propagate through an analysis and combine with other spatial and attributive
sources of error, including the uncertain relation of the data layer to the final decision set. In traditional GIS
analysis, uncertainty is not taken into account in the database. As a result, hard decisions are made with very
little concept of the risk involved in such decisions. Mays et. al. [31] demonstrates how simple it can be to work
with measurement error and its propagation in the decision rule. The task of the decision maker is to evaluate a
soft probability map and set an acceptable level of risk with which the decision maker is comfortable. By
knowing the quality of the data, the decision maker can view the decision risk occurring across an entire surface,
and make judgements and choices about that risk. Finally, any further analysis or simulation modelling of
impacts with such data increases the precision of those decisions as well.
10
Created by XMLmind XSL-FO Converter.
3. DATA MANAGEMENT
Results of the fuzzy analysis are the fuzzy layers, which are suitable for describing the decision on spatial
uncertainties. Fuzzy set theory [22, 53, 54] is a mathematical method to characterise and quantify uncertainty
and imprecision in data and functional relationships. Fuzziness represents situations where membership in sets
cannot be defined on a yes/no basis since the boundaries of sets are vague. The fuzzy sets are classifications of
data in which the boundary between classes is not distinct. The basics of fuzzy set theory have been presented
and discussed in numerous articles [6, 32, 43]. Fuzzy set membership functions describe the degree to which
data belongs to for each class (fuzzy set). These functions take on values between 0 and 1, and depict the grade
of membership (also known as the possibility) that a certain entity has in that class. For the description of these
continuous risk levels the fuzzy functions are more appropriate than the Boolean layers. Transient values
correspond to risk levels in accordance with the applied functions.
2. Data management
Data management of environmental models comparing to the business applications contains some special
problems. The amount of data to be processed is extremely large particularly due to the exponential increase of
spatial and spectral resolution values of air photos and space images. The data managing hardware and software
environments are heterogeneous and divided since data owners produce their data in different environments.
Standards of new generations of digital data collectors, for example mobile phones (MMS, GPRS) have not had
always-compatible data form. In case of precision agriculture, data formats of machine computers can only
partially standardised. Environmental data objects are often spatial-temporal and frequently uncertain. For
natural resources no standard object library was made that could follow behaviour of complex logical
connections, nested and joint objects.
In those situations, where one works with big volume or rapidly changing data, like in ecology, it seems
favourable to establish a divided database, where each database is to be installed to those users, who use those
most frequently. Successful application of divided database must meet the following requirements: Local
autonomy, namely each user place has authority over the database and the operational system. Thus all user
places are equal.
Conditions of continuous operation are given, even if different user places make different interventions in the
system at the same time. Physical and logical data independence must be assured. Physical fragmentation of the
logical field must not be perceptible for the users.
Combined application is not to cause data duplication. Relation database should make the data distribution
possible of those answers to the different places, but to give full value data as well.
False changes are to be deleted. Divided DBMS should run on different type of computers as well, so not to be
dependent on hardware. It must operate compatible with different operation systems since it may be supposed
that in each place of the organisation different operational systems had been installed. Due to the heterogeneous
network possibility mentioned in the introduction part, the network independence is a requirement as well. The
divided relation database must be independent from the probably different relation databases used in the user
places. It must be safe in data protection and loss of data points of view. In case of a divided relation database
risk of deterioration of the database is smaller than of central databases.
The most common used relation database is a logically planned, traditionally two-dimensional data management
system. Its disadvantage is that it describes the real world in tables. Records of these tables atomise the
connections and fragment them along some kind of logical organization. It is not able to reflect the embedment
of the objects. Formal differences of records in columns are not able to emerge. Logical normalisations and
physical planning could take a long time in complex relations. Time for reaching the data in big tables can
increase.
A part of data structures making spatial indexing possible, for example the KD-tree, the 4-tree, the R+ tree, the
BSP-tree are not easily matched into an object-oriented system since they cut the objects into pieces. It
contradicts the essence of object-oriented planning that uses complex units [27]. Since the R-tree, the Reactivetree, the KD2B-tree and the Sphere-tree data systems handle the objects as units; they can be used properly in
object-oriented modelling [33]. It makes keeping the relation systems possible without breaking the inner
connections. Most of the objects in the real world have 1: M-type relation system. This can be modelled in a
higher level of abstraction. Instead of attachment of simple attributes given by the relation database, it makes
input of complex attributes possible. The classification builds on the inner connection instead of the type of
entity. Origin of objects in the attributes turns traceable. The attribute and the geographical data are not divided.
11
Created by XMLmind XSL-FO Converter.
3. DATA MANAGEMENT
Application of object-oriented models in description and analysis of ecological relations is expected to bring
large progress.
3. Model creation
Prior to actual modelling certain preliminary works must be performed whose aims are to evaluate the resulting
situation on the basis of the available data. Based on the calculations made by the preliminary model the optimal
execution of the field exploration work, place and method of sample taking as well as planning of perception
data types and perceptual frequency must be determined.
Model creation begins with preparation of a conceptual model. In formation of the conceptual model that
fundamental principle becomes embodied that generally aim of the model is not to describe the process with
detailed mathematical methods, but to reproduce our idea about the problem to be solved, of course in a
properly controlled form. In case of the models available today the main question of model creation is not the
mathematical description anymore but choosing from amongst the processes to be taken into account. Choosing
the proper software the mathematical model in accordance with our decision is available.
Modelling helps to handle the complex problems, but determined by the modeller interactions that one cannot
follow in head or by simple calculation methods. In case of numerical resolution methods boundaries of the
resolution do not preclude any partial processes, therefore in case of uncertainty one can decide to take the
process into account or neglect it on the basis of the calculations made by just the method. The system-like
approach of modelling provides that one must take a clear stand on each of the parameters in the mathematical
description of the processes. Neglecting a parameter is possible only with changing in the structure of the model.
Most field ecologists are not good at abstraction. If they build a model they often try to incorporate every detail.
Most mathematicians are not good at interpretation of their models. Usually they think of “clean models and
dirty reality”. However, both abstraction and interpretation are necessary for successful modelling. Many system
properties are not represented in the model and some model properties cannot be found in real systems [37].
During the creation of a model the following points of view should be taken into account:
select the optimal level of complexity, but do not try to make a universal model, plan model development for
each time period and area, if possible, incorporate already existing simulation or stochastic-deterministic
models.
In case of ecological models three basic conditions must be cleared. These are as follows: type of the object
(flora, fauna, etc.) and the volume of the impact to the environment must be determined, the medium, in which
the processes take part must be determined in space and time, situation and response reactions of potential
impact receivers must be determined in space and time. Choosing the software is performed on the basis of
choosing conceptional model, and even in this case it would be practical to set up the conceptional model
independently from the software, if the available computer program is given.
In the second phase the possible simplifications, proper for the limitations of the software, must be decided on.
If the conceptional model defers significantly from the modelling possibilities of the available software, it is not
worth starting the detailed elaboration of the model. In case of the single model systems it is advantageous, if
the user have scope for action from that point of view how the given software is able to calibrate or validate the
environment.
First phase of elaboration of a model is the verification of the software. During verification we examine whether
the model gives a properly exact solution in case of the known, and mainly analytically resolvable tasks.
Parameters can be independent from time or changeable in time. By spatial extension they can be point-like,
linear or spatial-type ones.
During the calibration of the parameters the errors of the parameters are corrected. Aim of calibration is
generally not to decrease the deviations to minimum at the measuring point, but to display the character of the
process as precisely as the processes taken into account according to the conceptional model and the supposed
homogeneity during the parameter estimation make it possible.
Validation means the control of the calibrated model with use of such relevant cases that were not used during
calibration: for example, extreme phenomena or simulation of scenario or period left out from calibration. If the
results of validation are not satisfactory, the parameters must be modified or even the conceptional model too.
12
Created by XMLmind XSL-FO Converter.
3. DATA MANAGEMENT
Boundary conditions, characteristics of the medium and the object can be changed according to the chosen
scenario. These data are the basis of the evaluation, whose spatial visualization can be performed by
geographical information systems.
During post-control material and energy movement as well as the effect of intervention must be traced by
establishment of a monitoring network. Data received during the operation of the monitoring must be regularly
evaluated, generally by with the help of modelling as well, where the expectable trends can be foretold.
Estimation of the parameters of the model must be checked, and if it is not sufficient, the processes taken into
account – i.e. the conceptional part of modelling – must be reviewed.
4. Test questions:
1. What type agricultural problems can be solve by geostatistics?
2. What is the role of variogram when you try to describe spatial heterogeneity of the farm?
3. Do you compare different uncertainty treatment methods with?
4. Why important tool is the effective data management for PA?
5. What are the main advantages the modelling?
6. Please, enumerate the main steps of modelling?
5. REFERENCES
[1]
Andrén, O., Rajkai, K. & Kätterer, T. (1993): Water and temperature dynamics in a clay soil under
winter wheat: influence on straw decomposition and N immobilization. – Biol. Fertil. Soils 15: 1–8.
[2]
Antal, E. (1978): Éghajlati erőforrás és mezőgazdaság jelentősége. [Climatic potential and importance of
agriculture.] – Időjárás 76: 602–313. Aszály 1983. [Drought 1983.] KÉE, Budapest, pp. 9–43.
[3]
Berryman, A.A. (1981): Population systems: a general introduction. Plenum Press, New York.
[4]
Bíró, T. & Thyll, Sz. (1999): A belvíz-veszélyeztetettség térképezése. [Risk mapping of excess water.] –
Vízügyi Közlemények 81(4): 709–718.
[5]
Burgos, J.J. (1986): Equilibrium and extreme climatic conditions of world’s biomass and agrosystems.
Land use and agrosystem management under severe climatic conditions. – WMO Technical Note 148: 12–56.
[6]
Burrough, A.P. & McDonnel, A.R. (1998): Principles of Geographical Information Systems. Oxford
University Press, Oxford, 327 pp.
[7]
Cambell, I.M. (1977): Energy and atmosphere. A physical–chemical approach. John Wiley and Sons
Ltd, London.
[9]
Csáki, Cs., Harnos, Zs. & Láng, I. (1984): Agricultural development and ecological potential: the case
of Hungary. Kieler Wissenschaftsverlag Vauk
[10]
Clark, L.R., Geier P.W., Hughes, R.D. & Morris, R.F. (1967): The ecology of insect populations.
Methuen, London.
[11]
Cressie, N.A.C. (1993): Statistics for spatial data. Wiley and Sons Publisher, New York, 887 pp.
[13]
Eastman, J.R. (1999): Idrisi32 tutorial. Worcester, MA USA, 291 pp.
[14]
Environmental Management Standard. McGraw-Hill, New York, pp. 3–81.
[15]
[16]
Harnos, Zs. (1993): Időjárás és időjárás-termés összefüggéseinek idősoros elemzése. [Time
series analysis of climate-yield relationships.] – In: Baráth, Cs., Győrffy, B., Harnos, Zs.: Aszály 1983. [Drought
1983.] OTKA kiadvány, pp. 9–43.
13
Created by XMLmind XSL-FO Converter.
3. DATA MANAGEMENT
[17]
Horváth, L. (2002): Térbeli inhomogenitások kezelésének módszertani problémái a precíziós
növénytermesztésben. [Spatial inhomogeneity problems in precision farming.] – In: Proc. Agrárinformatika
2002, Debrecen, pp. 335–343.
[18]
Hufnagel, L. & Gaál, M. (2002): Többváltozós állapotsíkrendszerek alkalmazása valós és szimulált
adatsorok kezelésében. [Multivariate state-planes in service of analysis of field- and simulated data.] – In: Proc.
Agrárinformatika 2002, Debrecen, pp. 315–325.
[19]
Isaaks, H.E. & Srivastava, M.R. (1989): An introduction to applied geostatistics. Oxford University
Press, Oxford, pp. 3–553.
[20]
Johnston, K., ver Hoef, M.J., Krivoruchko, K. & Lucas, N. (2001): Using ArcGIS Geostatistical
Analyst. ESRI, Redland, USA, 287 pp.
[21]
Keller, H.B. (1995): Neural nets in environmental applications. – In: Avouris, N.M.& Page, B. (eds.):
Environmental Informatics – Methodology and Applications of Environmental Information Processing. Kluwer
Academic Publisher, Norwell, Mass., pp. 142–165.
[22]
Klir, G.J. & Yuan, B. (1995): Fuzzy Sets and Fuzzy Logic. Prentice Hall, New Yersey, 150 pp.
[23]
Kardeván, P., Vekerdy, Z., Róth, L., Sommer, S., Kemper, T., Jordán, Gy., Tamás, J., Pechmann, I.,
Kovács, E., Hargitai, H. & László, F. (2003): Outline of scientific aims and data processing status of the first
Hungarian hyperspectral data acquisition flight campaign, HYESNS 2002 HUNGARY. – In: Proc. 3rd EARSeL
Workshop on Imaging Spectroscopy, 13–16 May 2003, German Remote Sensing Data Centre,
Oberpfaffenhofen, Germany (in press).
[24]
Kovács G.J., Ritchie J.T. & Németh, T. (1995): Testing simulation models for assessment of crop
production and nitrate leaching in Hungary. – In: Agricultural Systems, Vol. 49., No 4. Elsevier Sci. Ltd.,
Oxford, England, pp. 385–397.
[26]
Ladányi, M., Horváth, L., Gaál, M. & Hufnagel, L. (2003): An agro-ecological simulation model
system. – Applied Ecology and Environmental Research 1(1–2): 47–74.
[27]
Laurini, R. & Thomson, D. (1992): Fundamentals of spatial information systems. Academic Press,
London, 673 pp.
[29]
Láng, I. (2002): Agrártermelés és globális környezetvédelem. [Agricultural production and global
environmental protection.] Mezőgazda, Budapest, 211 pp.
[31]
Mays, M.D., Holzhey, C.S., Bogardi, I. & Bardossy, A. (1995): Managing risk and variability with
fuzzy soil interpretations. – In: Robert, P.C., Rust, R.H. & Larson, W.E. (eds.): Proceedings of site-specific
management for agricultural systems. American Society of Agronomy, Madison, WI, USA, pp. 187–200.
[32]
Moore, I.D. (1996): Hydrological Modelling and GIS. – In: Goodchild, M.F., Parks, B.O. & Johston,
D.C. (eds.): GIS and Environmental Modelling. Progress and Research Issues.
[33]
Oostrerom, P. (1993): Reactive data structures for geographical information systems, spatial
information systems. Oxfords University Press, 198 pp.
[34]
290.
Pálfai, I. (1994): Az Alföld belvíz-veszélyeztetettségi térképe. – Vízügyi Közlemények 76(3): 278–
[35]
25 pp.
Racskó, P. (1991): Some aspects of crop modelling. Workshop on Agricultural modelling, Visegrád,
[37]
Sharov, A.A. (1992): Life-system approach: a system paradigm in population ecology. – Oikos 63:
485–494.
[38]
Somlyódi, L. & Straten, G. van (eds.) (1986): Modelling and Mapping shallow lake eutrophication –
with application to Lake Balaton. Springer Verlag, London.
[39]
Somlyódi, L. (2000): Magyarország vízgazdálkodási stratégiai kérdései. MTA Budapest.
14
Created by XMLmind XSL-FO Converter.
3. DATA MANAGEMENT
[40]
Somogyi, S. (ed.) (2000): A XIX századi folyószabályozások és ármentesítések földrajzi és ökológiai
hatásai. [Geographical and Ecological Impacts of River Regulation in XIX century.] MTA Geographical
Science of Hungarian Academy, Budapest, 302 pp.
[41]
Stefanovits, P., Filep, Gy. & Füleky, Gy. (1999): Talajtan. [Soil Science.] Mezőgazda, Budapest, pp.
15–470.
[42]
Szász, G. (1985): A klímapotenciál fogalma és alkalmazása a mezőgazdasági termelésben. [Definition
of Climapotential and Adaptation in Agricultural Production.] – In: A klímapotenciál és az agrometeorológiai
információk népgazdasági hasznosítása. Magyar Meteorológiai Társaság, Budapest.
[43]
Tamás, J., Csillag, J. & Murányi, A. (1998): Risk mapping of potential heavy metal pollution
modelling combined effects of pH, clay mineral and organic matter. – In: Filep, Gy. (ed.): Soil pollution. Soil,
water and environment relationships, Debrecen, pp. 225–239.
[44]
[45]
Tamás, J. & Bíró, T. (2001): Vízkészlet modellezés. [Modelling of water resources.]
University of Debrecen, 201 pp.
[46]
Tamás, J. (2001): Precíziós mezőgazdaság. Mezőgazdasági Szaktudás Kiadó, Budapest, 164 pp.
[47]
Ultsch, A. (1995): Einsatzmöglichkeiten von neurolanen Netzen im Umweltbereich. – In Page, B. &
Hilty, M.L. (eds.): Umwetinformatik- Infortikmethoden für Umweltforschung. Oldenbourg, Munich, Vienna.
[48]
Varga-Haszonits, Z. (1981): A gazdasági növények terméshozamának éghajlati potenciálja. [Climatic
potential of yield of the crops.] – MTA Közlemények 14(2–4): 253–270.
[49]
Varga-Haszonits, Z., Varga, Z., Lantos, Zs., Vámos, O. & Schmidt, R. (2000): Magyarország éghajlati
erőforrásainak agroklimatológiai elemzése. [Agroclimatic Analysis of Hungarian Climatic Potential.]
Mosonmagyaróvár, 219 pp.
[50]
Várallyay, Gy. (1985): Magyarország 1:100 000 méretarányú agrotopográfiai térképe. – Agrokémia és
Talajtan 34: 234–248.
[51]
Várallyay, Gy. (1993): Soil data-bases for sustainable land use: Hungarian case study. Soil resilience
and sustainable land use. – In: Greenland, D.J. & Szabolcs, I. (eds.): CAB International, pp. 469–495.
[52]
Webster, R. (1977): Quantitative and numerical methods in soil classification and survey. Clarendon
Press, Oxford.
[53]
Zadeh, L. (1965): Information and Control 8: 338–353.
[54]
Zimmerman, H.J. (1985): Fuzzy set theory and its application. Martinus Nijhoff, Dordrecht, 363 pp.
15
Created by XMLmind XSL-FO Converter.
4. fejezet - 4. SAMPLING STARTEGY
The sugar beet is very sensitive for the soil characteristics of the cultivated land. This demand is supplied by
strictly and highly industrialised cultivation technology [10]. The technology has to be applied in a site-specific
way, obviously. Thus, it is important to use an effective and representative soil and plant sampling method [11].
This study was carried out at a sugar beet farm having moderately heavy clay soil at the Small Plain in Hungary.
In addition to the classic soil and plant analysis, the necessary and the toxic microelement contents were also
examined. The applied 20 x 20 m high density grid sampling technique supported the geo-statistical analysis of
the soil – plant relationship.
A 2 ha trial plot on chernozem soil was marked out in the neighbourhood of city Komárom. Soil samples, 0-30
cm deep, and plant samples at the intersection points of a 20 x 20 m square grid, were taken. Sugar beet served
as a marker plant as sugar beet reacts very sensitively to changes in soil characteristics.
Sugar beet root yield and pollution, sugar content (digestion%), potassium, sodium and alpha-amino-nitrogen
contents were evaluated (Table. 1). Sugar yield/ha, the white sugar content (Hc%) were calculated from above
data [12]
Hc%=digestion% - [0,343(K+Na)+(0,094N)+0,2],
where
K=K content of the sugar beet, mmol/100g sugar beet,
Na=Na content of the sugar beet, mmol/100g sugar beet,
N=alpha-amino-N content of the sugar beet, mmol/100g sugar beet.
Hc% was used to calculate the white sugar yield/ha and the thick juice purity quotient % (Q%):
Q%=99,36 – 1,427(K’+Na’+N’)/digestion%,
where
K’=K content of the sugar beet, mmol/kg sugar beet,
Na’=Na content of the sugar beet, mmol/kg sugar beet,
N’=alpha-amino-N content of the sugar beet, mmol/kg sugar beet.
In soil samples the following parameters were studied: pH(KCl), SP (= saturation percentage), calcium
carbonate (Scheibler) (further on: CaCO3 %, humus, total salt, 1 mol KCl soluble nitrate+nitrite-nitrogen
(further on: nitrate-nitrogen) and Mg, AL-soluble phosphorus, K and Na, 0.05 mole EDTA+0.1 mole KCl
soluble Zn, Cu, Mn, Fe, Mo, Al, B, Cd, Cr, Co and Pb contents (Table 2).
Plants were analysed at the Sugar Factory Petőháza and soil samples at the Laboratory of the Plant and Soil
Protection Service of County Vas, Hungary.
4.1. táblázat - 1. Table Yield and quality of sugar beet
Plant
S.beet
S.beet
Nett
Thick
White
Sugar
White
test
yield
pollution
yield
juice
sugar
yield
sugar yield
t/ha
%
t/ha
Q%
%
t/ha
t/ha
I/1
72,6
7,13
67,42
89,00
12,62
10,36
8,51
I/2
63
7,44
58,31
91,87
13,73
9,26
8,00
I/3
81,2
5,74
76,54
90,60
12,49
11,48
9,56
16
Created by XMLmind XSL-FO Converter.
4. SAMPLING STARTEGY
I/4
75,5
7,19
70,07
92,94
14,64
11,74
10,26
I/5
96,8
1,09
95,75
88,80
12,49
14,82
11,96
I/6
82
7,62
75,75
89,40
11,54
10,74
8,74
I/7
25,1
11,37
22,25
91,09
13,23
3,48
2,94
I/8
91,2
8,45
83,50
91,02
12,79
12,66
10,68
I/9
119
2,74
115,74
88,07
11,14
16,27
12,90
I/10
115,9
10,29
103,98
87,56
10,61
14,18
11,03
II/1
65,4
7,84
60,27
91,34
13,59
9,59
8,19
II/2
49
6,00
46,06
92,95
15,35
8,05
7,07
II/3
72,1
6,76
67,23
92,74
14,67
11,31
9,86
II/4
83,1
0,00
84,87
91,59
14,16
14,18
12,02
II/5
96,3
4,85
91,63
90,09
12,79
14,24
11,72
II/6
94,6
5,21
89,67
91,45
13,65
14,37
12,24
II/7
57,8
8,27
53,02
92,47
14,09
8,63
7,47
II/8
100
5,23
94,77
92,84
14,82
16,10
14,04
II/9
90,6
7,08
84,18
92,41
14,00
13,66
11,78
II/10
103,1
7,04
95,84
90,79
12,94
14,82
12,40
III/1
62,1
5,16
58,90
90,71
13,39
9,40
7,88
III/2
90,4
7,20
83,89
92,28
14,81
14,30
12,42
III/3
96,8
4,33
92,61
90,82
13,63
14,93
12,63
III/4
101
7,34
93,59
90,13
13,15
14,79
12,31
III/5
71,3
6,54
66,64
90,97
13,50
10,65
9,00
III/6
100,1
7,58
92,51
91,42
13,53
14,67
12,51
III/7
95,2
6,01
89,47
91,23
13,49
14,19
12,07
III/8
108
7,72
99,67
90,16
13,19
15,69
13,15
III/9
130,1
26,03
96,24
91,93
14,06
15,68
13,53
III/10
84,1
5,85
79,18
92,01
14,00
12,90
11,08
17
Created by XMLmind XSL-FO Converter.
4. SAMPLING STARTEGY
IV/1
82,6
4,02
79,28
90,66
12,73
12,08
10,09
IV/2
89,6
0,00
93,49
92,82
14,78
15,81
13,81
IV/3
101,9
5,08
96,73
93,42
15,33
16,80
14,82
IV/4
100,1
6,01
94,08
92,58
14,67
15,83
13,80
IV/5
85,6
4,86
81,44
91,70
13,99
13,38
11,39
IV/6
95,3
6,42
89,18
92,11
14,35
14,97
12,79
IV/7
87,4
5,70
82,42
91,22
13,03
12,73
10,74
IV/8
77,9
6,78
72,62
89,75
12,11
10,78
8,80
IV/9
93,8
7,43
86,83
92,09
13,41
13,63
11,64
IV/10
100,2
6,50
93,69
92,13
14,19
15,54
13,29
V/1
37
7,83
34,10
91,48
13,26
5,38
4,52
V/2
81,3
0,00
81,44
91,94
14,57
13,84
11,86
V/3
82,1
5,10
77,91
94,24
15,99
13,97
12,46
V/4
80,1
6,16
75,17
93,62
15,91
13,49
11,96
V/5
105,8
27,38
76,83
90,25
12,88
11,89
9,90
V/6
99,8
6,03
93,79
91,97
14,04
15,26
13,17
V/7
82,4
6,40
77,13
90,74
13,50
12,43
10,41
V/8
86,3
6,88
80,36
90,68
12,64
12,23
10,16
V/9
72,9
0,00
81,83
88,89
12,72
12,85
10,41
V/10
82,8
6,50
77,42
92,62
14,49
13,01
11,22
VI/1
75,7
3,04
73,40
91,78
13,35
11,42
9,80
VI/2
89,1
6,07
83,69
91,42
12,97
12,70
10,85
VI/3
59,9
6,25
56,15
93,31
15,21
9,72
8,54
VI/4
65,5
6,94
60,96
93,38
14,51
10,11
8,85
VI/5
60,3
7,20
55,96
92,50
13,53
8,80
7,57
VI/6
70,3
1,16
69,48
92,07
14,41
11,69
10,01
VI/7
76,8
0,00
82,52
92,56
14,66
13,95
12,10
18
Created by XMLmind XSL-FO Converter.
4. SAMPLING STARTEGY
VI/8
92,2
5,83
86,83
91,88
13,90
14,20
12,07
VI/9
88,9
6,74
82,91
92,10
14,65
14,14
12,15
VI/10
99,3
6,84
92,51
92,32
14,33
15,35
13,25
Mean
84,61
6,44
79,33
91,48
13,70
12,75
10,84
Minim.
25,10
0,00
22,25
87,56
10,61
3,48
2,94
Max.
130,10
27,38
115,74
94,24
15,99
16,80
14,82
Data are represented in diagrams 2-3 dimensions. For example: Fig. 1 represents yield evolution and Fig.2
changes in sugar content within the plot.
4.2. táblázat - 2. Table Soil analysis data
Soil test
1
2
3
4
code
I/1
I/2
I/3
I/4
pH (KCl)
7,48
7,44 7,35 7,36
SP
%
37
39
Total salts
%
0,00
0,00 0,00 0,00
CaCO3
%
11,0
7,0
Humus
%
2,07
2,04 1,93 2,21
NO3-N
mg/kg
6,4
8,1
P2O5
mg/kg
272
319 231 196
K2O
mg/kg
553
319 295 296
Mg
mg/kg
113
151 90
Zn
mg/kg
3,25
1,47 1,68 1,38
Cu
mg/kg
5,71
4,76 5,00 4,17
Mn
mg/kg
10,9
17,3 30,1 35,1
B
mg/kg
0,26
0,38 0,29 0,31
Soil test
31
32
code
IV/1
IV/2 IV/3 IV/4
pH (KCl)
7,34
7,30 7,38 7,45
19
Created by XMLmind XSL-FO Converter.
34
5,0
6,4
33
37
4,0
4,1
97
34
4. SAMPLING STARTEGY
SP
%
34
33
33
Total salts
m/m%
0,00
0,00 0,00 0,00
CaCO3
m/m%
5,0
3,0
Humus
m/m%
2,70
2,16 2,30 2,22
NO3-N
mg/kg
2,7
3,2
P2O5
mg/kg
220
225 250 221
K2O
mg/kg
216
177 264 225
Mg
mg/kg
89
100 95
Zn
mg/kg
1,25
1,65 1,51 1,33
Cu
mg/kg
4,16
4,71 3,72 2,38
Mn
mg/kg
20,4
43,2 24,5 11,4
B
mg/kg
0,24
0,41 0,36 0,39
Soil test
41
42
code
V/1
V/2 V/3 V/4
pH (KCl)
7,65
7,57 7,51 7,39
6,0
3,7
43
10,0
3,4
122
44
SP
%
29
30
Total salts
m/m%
0,00
0,00 0,00 0,00
CaCO3
m/m%
17,0
17,0 3,0
Humus
m/m%
1,36
1,42 2,20 2,26
NO3-N
mg/kg
2,3
2,7
P2O5
mg/kg
176
196 189 192
K2O
mg/kg
178
202 228 184
Mg
mg/kg
100
91
Zn
mg/kg
0,87
1,10 1,22 1,42
Cu
mg/kg
3,14
3,29 3,50 3,44
Mn
mg/kg
10,1
7,7
20
Created by XMLmind XSL-FO Converter.
33
39
4,2
32
4,0
3,0
125 109
14,7 28,9
4. SAMPLING STARTEGY
B
mg/kg
0,20
4.1. ábra - 1. Figure Sugar beet yield t/ha
4.2. ábra - 2. Figure Sugar content (digestion%)
Similar large differences were observed in soil analysis results. Fig. 3 represents changes in CaCO3 %. Fig. 4
those of nitrate-nitrogen and Fig. 5 those of AL-K content within the plot [9]. Similar variability is found in all
the measured and calculated parameters.
4.3. ábra - 3. Figure Calcium carbonate % (Scheibler)
21
Created by XMLmind XSL-FO Converter.
0,23 0,29 0,26
4. SAMPLING STARTEGY
4.4. ábra - 4. Figure Nitrate-nitrogen mg/kg
4.5. ábra - 5. Figure AL-K content K2O mg/kg
22
Created by XMLmind XSL-FO Converter.
4. SAMPLING STARTEGY
As theoretically the measured differences could also be attributed simple failures in plant production
technologies correlation was studied between soil analysis results and yield indices. A simple linear regression
is used here to show tendencies of correlation of the numerous correlations the effect of soil N and K contents
on yield quantity and quality are especially important. As shown in Fig. 6 there is a convincing negative
correlation between nitrate-N content and sugar content while practically no correlation (R2=0,013) was found
between the nitrate-N content and yield quantity . In the same time, an increase in soil nitrate content spoils the
thick juice purity quotient to a great extent which results in reducing the reception price of sugar beet.
This soil well supplied with K, an increase in soil AL-K content did not increase sugar beet yield, its effect,
however, on white sugar content is still noticeable
The results show clearly the correlation between pH (or calcium carbonate) and availability of microelements
and heavy metals. For example: Fig. 6. shows the 0.05 mole EDTA+0.1 mole KCl soluble Cu and Fig. 7. that of
Cd as affected by pH.
4.6. ábra - 6. Figure11 pH and 0.05 mole EDTA+0.1 mole KCl soluble Cu
4.7. ábra - 7. Figure12 pH and 0.05 mole EDTA+0.1 mole KCl soluble Cd
23
Created by XMLmind XSL-FO Converter.
4. SAMPLING STARTEGY
It’s generally known that in contrast to most microelements, Mo-availability is improved by increasing pH.
There is no positive correlation was found between pH and 0.05 mole EDTA+0.1 mole KCl soluble Mo-content.
The effect of K on increasing sugar content is seen clearly even in this soil well supplied with K. An increase in
pH and calcium carbonate content, respectively, decreases clearly the microelement and heavy metal content
extracted by 0.05 mole EDTA+0.1 mole KCl solution. However, no positive correlation was found between pH
and the extracted Mo content. It would be worth while to test if this general solvent can be used to characterise
Mo availability in a soil fairly supplied with Mo.
1. Test questions:
1. Why important are parameters of soil N and K contents on yield quantity and quality?
2. Why important are parameters of soil micro elements on yield quantity and quality?
2. References
1. Csáki Cs., Harnos, Zs., Rajkai, K., Vályi, I. [ed.]Hungarian agriculture: development potential and
environment. . J.K Parikh. Dordrecht, The Nederlands : Martinus Nijhoff Publishers, 1988. IIASA. pp. 253–265.
In: Sustainable development in agriculture.
2. Kovács, G.,J., Nagy, J.Test runs of CERES-maize for yield and water use estimation. Soil, plant and
environment relationships. In Current plant and soil science in agriculture . [ed.] J. Nagy. Debrecen, : University
of Debrecen, 1997. pp. 115–119.
3. Tamás, J., Nagy, J. [ed.] Evaluation of irrigation schedules by CROPWAT FAO model. Soil, plant and
environment relationships. Current plant and soil science in agriculture. . J. Nagy. Debrecen : University of
Debrecen, 1997. University of Debrecen. pp. 115–119.
4. Swaminathian, M.S.From Stockholm to Rio de Janeiro. The road to sustainable agriculture. Madras : M. S.
Swaminathan Research Foundation, 1991. p. 68.
5. Cascio, J., Woodside, G., Mitchell, P.ISO 14000 Guide, The New International Enviromental Standars. , , pp.
. New York : McGraw-Hill, 1996. p. 217.
6. Láng, I., Csete, L., Harnos, Zs. Resources Adjustment and Farming Structures. European Review of
Agricultural Economics. 1988, Vol. 15., pp. 2–3.
24
Created by XMLmind XSL-FO Converter.
4. SAMPLING STARTEGY
7. Láng, I.Mezőgazdaság agroökológiai potenciálja az ezredfordulón. [Agro-ecological potential of agriculture
on millennium.]. Budapest : Tárcaközi Bizottság Jelentése, 1983.
8. Harnos, Zs.Az alkalmazkodó mezőgazdaság rendszere. [Systems of sustainable agriculture]. – Módszertani
kutatások . Budapest : Kertészeti és Élelmiszeripari Egyetem, 1991. p. 91.
9. Buzás, I.A káliumellátás szerepe a sikeres cukorrépatermesztésben. . Budapest : INDA 4231, 2001.
10. Poel, van der P.W., Schiweck, H., Schwartz, T.Sugar technology. Berlin : Dr. Albert Bartens KG,, 1998.
11. Buzás, I. [ed.] Impact of soil and nutrient management on sugar beet yield and quality in Hungary. A. E.,
Buzás, I. Johnston. in: Balanced plant nutrition in sugar beet cropping systems for high yield and quality Proc.
of workshop of Beta R.I. : IMPHO, 2000. p. 8.
12. L., Buzás I. - Kulcsár.Tápanyag-gazdálkodásunk és a cukorrépa-termesztés. 1999, Vol. 2., pp. 64-70.
25
Created by XMLmind XSL-FO Converter.
5. fejezet - 5. SPATIAL ANALYSIS TO
EXPLORE SPATIAL HETEROGENITY
Spatial agrochemical data are sensitive to scales and scales of measurement. In case study introduces an
implementation of complex spatial statistical analysis with GIS highly contributes to the description of the
spatial correlation of certain parameters and the spatial directions and extents of their effects. Analogous to the
traditional statistical parameters were introduced to spatial analyses such as mean centre, weighted mean centre,
standard distance and standard ellipse of deviation were calculated based on some representative attribute data
sets of points. In addition to the analysis of those parameters considerable from industrial point of view and the
mathematical interpolations, the optimal sampling number and the sampling points were determined. The data
were integrated to a geographical information system. The developments in Geographic Information Systems
allow local and regional variations in crop performance to be more readily explored through co-mapping with
data for soil and other factors. This offered the possibility of refining the use of both factory records and crop
models.
Quality parameters of sugar beet change according to production sites. Decreasing of variability is exceptionally
important from the point of view of the sugar industry. Accurate spatial determination of the site and fluctuation
of this variability is considered to be the base of this work. Standardisation of field sampling and comparability
of studies were subjects of several studies [11]. These studies assumed that the number of repetitions is steady in
space, but the randomly made sampling technique can keep the sampling errors under an acceptable level.
However in this manner the sampling sites and their spatial locations can not be evaluated / retrieved in the
future. The differences in plant states can be evaluated with accurate co-ordinates on the basis of DGPS and
other real-time corrections. The precision agricultural technologies building on this technical possibility perform
rate application input material output [27]. The technology in its present state provides more information for the
producer than before. However the digital soil map determined by such way does not give direct information on
the relationship between the measurement data and the true data. Conditions of data production, for example the
form of estimation or its spatial uncertainty, are not known. Particularly handling of randomly occurred extreme
data could cause problems. A yield peak can happen in the sensor of a harvester due to actual yield increase
because of local weeding or errors in fertiliser spreading. The secondary data processing based on these data ,
where at first on the basis of the sample taking points a yield surface is made, then according to the spreading
width of the cultivating machines are classified to zones considered to be homogeneous, can make both types of
error.
Namely they at first faulty evaluate the yield in space due to local error then because of generalisation this error
is spread globally to the homogenised zone. We have also no information on the direction and volume of the
error propagation happened meanwhile. Some researchers offer the automatic ignorance of the extreme data,
whose volume and influence can also difficult to be assessed. The traditional statistic indexes: average,
deviation, t, f, chi test; do not take into consideration the locations of the samples compared to each other. In our
examinations spatial variability examinations were performed on sugar beet test plant, the application conditions
of the estimation methods were evaluated and such statistical analogues that can spatially be interpreted were
sought, which meet the traditional statistical indexes.
The examinations were made in a sample area located in north-western Hungary with sugar beet test plant.
According to the traditional sample taking procedure N=60 samples were taken in regular 20 x 20 m grid, where
besides the plant micro- and macro elements, the sugar industrial quality parameters the agro-chemical
examination of soils were performed as well. The spatial relations of the sample values are shown mainly on the
basis of the average yield. Data were pictured digitally in ±1 m spatial accuracy, the database was supplemented
with a 1:10,000 scale topographical map. The integrated geographical information system was made less than
8.2 version of ArcMap. For statistical examinations Geo-statistical Analyst (ESRI), Genstat 6.0 and Variowin
2.2 softwares were used. On the basis of the point database five interpolation methods were compared to each
other. Cross-correlation calculation was made with 31 examination parameters for the Parson-values. For its
spatial analogy an application for cokriging was shown. Description of this method is available for example in
the works of Isaaks, E. H.; Srivastava (1989)[8]; Cressie (1990)[3]; Wackernagel (1995)[31].
In case of the statistical analogues of the traditional statistical indexes solutions applied for description of space
in geometry were utilised. Monmonier (1993) [19]calls attention to the fact that in this case the co-ordinate
system needs to be carefully structured data source so that it orients to proper direction for GIS project, it
26
Created by XMLmind XSL-FO Converter.
5. SPATIAL ANALYSIS TO
EXPLORE SPATIAL
HETEROGENITY
situates with proper arbitrary origin, and it uses suitable measurements unit and interpretation scale. With a coordinate system defined, the mean centre can be calculating as follows (1.equation).
where xmc, ymc = are co-ordinates of mean centre, xi, yi are co-ordinates of point i, and n is the number of points.
The weighted mean centre of distribution can be found by multiplying the x and y co-ordinates and mean of
each point by weights (measured values) assigned to them (2. equation).
where xwmc, ywmc defines the weighted mean centre and wi is the weight at i measured point
Standard distance is the spatial analogy of standard deviation in descriptive statistics. While standard deviation
indicates how observations deviate from mean, standard distance indicates how points in a distribution deviate
from mean centre. Attribute values can be used as weights when calculating their mean centre and also possible
to weight when calculating the weighted standard distance as following (3. equation):
xmc,ymc = are co-ordinates of mean centre, xi, yi are co-ordinates of point i, and n is the number of points and fi is
the weight for xi, yi point.
The traditional statistical measurement is not able to take into consideration the spatial locations of number = N
sample taking points. Spatial relations between the x-value characteristic (i.e. sugar content and signed as z(x))
of the plant defined by x1, y1 co-ordinates in the given z measuring point and described with given h distance
vector and the same characteristic of a plant defined x2, y2 co-ordinates being in the point of z(x+h) can't be
evaluated.
The work hypothesis of exact field trials is that providing the repetition number of the trial treatment plots
situated randomly in space the treatment data of each plot differentiate from the others in space and the results
exceeding a given certainty limit value depends on significantly on the treatment. Basis of the work hypothesis
of the geo-statistical examinations is the so-called Tobler law that says the characteristics of the objects (sample
taking points) located nearer to each other are more similar than of the point located farther in space. The spatial
variance of the closer sample taking points is smaller and their auto-correlation is bigger than of the farther
points. This γ (h variability) grows with distance, while the auto-correlation decreases and in case of a given
range the variability stabilises at given sill value and the correlation approaches to 0 at this same place (
Figure8.).. The variogram is sensitive to local and global outliers. Local outlier means an attribute value of the
sampling point exceeds its neighbourhood, while the global outlier exceeds the means of the examined area.
Each observed z(x) can contribute to several estimates of γ (h). So one exceptionally large or small measured
value z(x) will tend to swell γ' (h) whenever it is compared with other values. The result is to inflate the average.
But the effect is not general. The point that is closer to the edge of the examination area contributes less, than
the one is closer to the centre. If the sample points along a regular line would be taken into consideration, the
initial and end points act only one time in each lag, while the others inside the line more times.
If the points are scattered in space the effect of the extreme date can lesser be estimated. In the database of the
average yields distributed near normally a similar case can be found in the NW space, if the biggest and smallest
values are close to each other in space. Thus their effect succeeds less than of the high yield values located in
the centre of the examination space. At the same time the surface estimation made by ordinary kriging underand overestimated these spaces.
27
Created by XMLmind XSL-FO Converter.
5. SPATIAL ANALYSIS TO
EXPLORE SPATIAL
HETEROGENITY
5.1. ábra - 8. Figure Spatial distributions of residual data of the measured values and
the surface calculated by kriging.
Exceptional contributions to the semi-variance can be identified by drawing an h-scattergram surface or scatter
diagram for each lag, h. An h-scatter diagram is a graph in which the z(x) are plotted against the z(x+h) with
which they are compared in computing γ' (h). The closer the points lie to the diagonal line with gradient 1, the
stronger is. In the figure in the upper part the variogram fitted onto the value pairs can be seen, while in the
lower part the average yield h-scattergram surface and the seeking directions can be seen.
5.2. ábra - 9. Figure Average yield h-scatergram (Distance 10-1 m)
5.3. ábra - 10. Figure Semi-variogram and applied search direction/angle of the sugar
beet yield samples
28
Created by XMLmind XSL-FO Converter.
5. SPATIAL ANALYSIS TO
EXPLORE SPATIAL
HETEROGENITY
On the figure above can be observed that the yield samples showed the smallest variance in N-S direction. The
fitted model function is γ ' (h) = 296.4 * spherical (17.3). The experimental sample taking distance, which was
10 m originally, can be grown to near twofold but keeping the variance level.
The variogram function was used during Ordinary Kriging interpolation. Effects of other interpolators to the
same yield data were analysed as well, which were the follows: Global Polynomial (GP), Local Polynomial
Interpolation (LP) are quick deterministic interpolators that are smooth (inexact), Inverse Distance Weighted
(IDW), Radial Basis Functions (RBF) methods are a quick deterministic interpolator that is exact. In case of
each of all methods the cross-validation was performed as well residual values were calculated to all (N=60)
measurement sites, their average and deviation to each method (Table 3.).
3. Table Residual Results of Different interpolation methods
Measured
yield
LP Error
RBF Error
IDW Error
GP Error
Ordinary
kriging
t/ha-1
Mean
84.16
-0.93
-0.10
0.17
0.07
0.08
Standard deviation
17.83
16.54
14.88
15.39
17.04
14.97
A residual is the difference between the Z value of a point in a data file and the interpolated Z value at the same
XY location on a grid surface. Residual values are reported as either positive or negative values. If the Z value
in the data file is greater than the Z value derived from the grid surface, the residual value is positive.
29
Created by XMLmind XSL-FO Converter.
5. SPATIAL ANALYSIS TO
EXPLORE SPATIAL
HETEROGENITY
A negative residual value indicates that the Z value from the data file is less than the interpolated Z value at that
point. On the basis of the residual maps and their statistical values it can be ascertained, that the GP can be used
successfully in that case if the surface changes slowly and gradually.
The data located in the edges can play an important role in surface formation. Local Polynomial Interpolation
(LP) is more flexible than the global polynomial method, but there are more parameter decisions, but it is less
flexible and more automatic than kriging. Inverse Distance Weighted (IDW) can be a good way to take a first
look at an interpolated surface.
Radial Basis Functions (RBF) provides prediction surfaces that are comparable to the exact form of kriging,
making it less flexible and more automatic than kriging.
Kriging is a moderately quick interpolator that in this case was exact because it was not involved measurement
error model in variogram. Kriging assumes the data come from stationary stochastic process and used ordinary
kriging methods assume normally distributed data.
Not only spatial distribution of yield must be determined in practice but its correlation to other parameters and
soil examination values needed for sugar industry. The Pearson correlation values were calculated for
31parameters. From the correlation matrix received in such way the values -1 ≤-0,5 and a 0,5≥1 were kept (4.
Table).
5.1. táblázat - 4. Table The Pearson correlation values of the yield and the soil pH.
Mean of yield
pH (KCl)
t ha-1
net sugar beet yield t ha-1
0.97
Carbonic chalk m/m%
0.83
raw sugar yield
0.91
Humus m/m%
-0.75
0.87
Cu mg/kg
-0.69
thick juice purity quotient %, 0.96
(: Q%, )
Mn mg/kg
-0.74
Fe mg/kg
-0.72
Al mg/kg
-0.74
Cd mg/kg
-0.78
Co mg/kg
-0.72
Pb mg/kg
-0.76
t ha-1
white sugar yield
Parameters of the table were calculated as follows:
-raw sugar yield: net sugar beet yield t/ha x digestion %/100
-white sugar yield t/ha, : net yield sugar beet t/ha x white sugar content of the sugar beet %/100,
where
white sugar content of the sugar beet %= digestion % - [0,343(K+Na)+(0,094N)+0,2]
K = K content of the sugar beet, mmol/100g sugar beet
30
Created by XMLmind XSL-FO Converter.
5. SPATIAL ANALYSIS TO
EXPLORE SPATIAL
HETEROGENITY
Na = Na content of the sugar beet, mmol/100g sugar beet
N = alpha-amino-N content of the sugar beet, mmol/100g sugar beet.
-thick juice purity quotient Q%, :=99,36 – 1,427(K’+Na’+N’)/digestion%,
where
K’=K content of the sugar beet, mmol/kg sugar beet
Na’=Na content of the sugar beet, mmol/kg sugar beet
N’=alpha-amino-N content of the sugar beet, mmol/kg sugar beet.
The statistical examination above does not take the places and spatial values of the sample into consideration.
As analogy in this case the spatial effect of cokriging was analysed. Cokriging jointly analyses the spatial
variance of the two variables. On its basis spatial variance with less number of samples can be calculated on the
basis of the variogram defined to bigger sample number.
Cokriging uses statistical models that allow a variety of map outputs. Cokriging assumes the data come from a
stationary process, and used methods assume normally distributed data.
From among the results maps of copper and pH are shown here. ( Figure3.).
5.4. ábra - 11. Figure Sample taking areas of copper and pH of original sample numbers
(left- middle) as well as spatial distribution of copper with reduced sample points
estimated by pH variogram (right)
Copper and pH show a negative spatial correlation, i.e. increasing several decimal pH levels already detects less
micro element content and distribution. In both cases the cokriging functions make decrease of number of
samples possible.
5.5. ábra - 12. Figure Variogram surface of original copper concentration (left) and pH
(right) and cross variogram surface of both (below) sampling sets
31
Created by XMLmind XSL-FO Converter.
5. SPATIAL ANALYSIS TO
EXPLORE SPATIAL
HETEROGENITY
In the model the number of the regular sample grid were randomly decreased to 70% (signed with white cross).
pH and Cu r=-0,69 showed medium values, which were signed by the opposite meaning spatial pattern of the
above figure.
On the basis of the co-variogram the spatial pattern of Cu was re-estimated by ordinary kriging in the 30%
reduced database, where the standard error was 12% between the original and reduced number of sample
estimation. Since the Fe, Mn, Al, Cd, Co, Pb showed a closer correlation with the pH, at the same error level
further decrease in the number of samples and in case of same number of samples further decrease in
uncertainty can be reached. In practice the parameter that can be measured cheaper can decrease the
measurement costs of several spatially well correlating elements.
On the basis of the equations the spatial weighted mean and standard distance values were calculated for the
different parameters (table). For the examination area the geometrical spatial mean is x(35), y(55) and the
standard distance is 33,41( Table 3.).
5.2. táblázat - 5. Table Spatial weighted mean and standard weighed distance of sugar
beet and soil data
Measuring point
Spatial
mean
(X)
Spatial
(Y)
35
55
mean Standard Weighted
distance mean
33,41
sum of
sum of
sum of
weighted (x)
weighted (y)
weight
Average yield tha-1
34.7785 57.5187
33.0611 175607
290429
5049.3
pH(KCL)*
35.1054 54.9917
33.456
15648.6
24513.1
445.76
saturation percentage*
33.8719 55.5688
33.4
70860
116250
2092
Total salt (m/m%),*
20
0
1.4
0.7
0.07
10
32
Created by XMLmind XSL-FO Converter.
5. SPATIAL ANALYSIS TO
EXPLORE SPATIAL
HETEROGENITY
CaCO3(m/m%)*
39.4189 50.8959
33.9
16280
21020
413
Humus (m/m%),*
34.1157 55.8579
33.016
4268.9
6989.5
125.13
Contamination of the sugar 33.9164 58.4595
beet (m/m%),
32.89
12990
22390
383
Digestion %r
34.8487 57.4433
33,096
165880
273430
4760
Net sugar beet yield t ha-1
35.4976 54.4539
33.2421 29250
44870
824
Raw sugar yield t ha-1
35.2549 57.1895
32.9375 26970
43750
765
White sugar yield t ha-1
35.3988 56.6718
32.9546 23080
36950
652
Q%
35.351
33.295
52754.2
963.88
54.7311
34074.1
K-content of the sugar beet 34.7532 55.7278
mmol/kg sugar beet
33.5336 92996
149122
2675.9
Na content of the sugar 33.9698 56.7187
beet mmol/kg sugar beet
33.3717 21119
35262
621.7
alpha-amino N content of 35.1002 54.9235
the sugar beet, mmol/kg
sugar beet
33.396
301420
5488
NO3-N+NO2-N, 1 mol KCl 30.0307 62.2761
soluble, mg/kg soil
33.8225 9790
20302
326
P2O5 soluble, mg/kg soil
34.3553 53.0497
33.2863 428410
661530
12470
K2O soluble, mg/kg soil
33.9419 56.2672
33.6628 532480
882720
15688
Na soluble, mg/kg soil
34.6702 57.9553
33.2343 162950
272390
4700
Mg 1 mol KCl soluble, 34.3305 56.2369
mg/kg soil
33.33
226890
371670
6609
Zn*
33.9291 54.4593
33.8873 2739.1
4396.5
80.73
C u*
32.8236 53.634
33.3894 7055.1
11528.1
214.94
Mn*
30.099
59.5723
30.1001 62916
124524
2090.3
Fe*
33.2593 55.8289
32.4939 18276
30678
549.5
Mo*
34.0116 40.5814
30.7535 58.5
69.8
1.72
Al*
31.6297 58.3228
31.5015 9995
18430
316
B*
32.2859 56.3163
32.764
1071.7
19.03
Cd*
32.0963 55.6118
32.7386 103.35
179.07
3.22
192630
614.4
33
Created by XMLmind XSL-FO Converter.
5. SPATIAL ANALYSIS TO
EXPLORE SPATIAL
HETEROGENITY
Cr*
34.1549 55.1698
33.6255 89.52
144.6
2.621
Co*
28.6805 59.2803
29.5924 263
543.6
9.17
Pb*
31.4789 56.6745
32.4218 2937.3
5288.3
93.31
Remarks: 3-7 rows – soil; 22-32 rows -0,05mole EDTA+0,1 mole KCl soluble, mg/kg soil*
5.6. ábra - 13. Figure Spatial mean of analysed data and visual methods of mapping
5.7. ábra - 14. Figure Eigenvectors and rotations angels of raw sugar (A), digested sugar
(B), and yield (C) where origo is the mean centre . (1) north direction
34
Created by XMLmind XSL-FO Converter.
5. SPATIAL ANALYSIS TO
EXPLORE SPATIAL
HETEROGENITY
The origin of the above left figure is in the geometrical centre of the examination area. The co-ordinates of the
parameters are displayed correctly according to the serial number of the figure. Distribution of the density
values shows the relative homogeneity of the examination area. Excluding the pH, lime and Mo content the
spatial mean values of the examined characteristics moved to the N-W direction. To the spatial average its
dispersion value can be added, as the radius value of the circle drawn into the origin of the spatial average. They
can be jointly mapped in co-ordinate spaces.
Similarly to our results, Lee and Wong (2001)[16] declare based on population data that a weighed mean centre
provides a better description of the central tendency than a mean centre when points or locations have different
frequencies or occurrences of the phenomena studied. In this case, standard distance is usually used as the radius
to draw a circle around the mean centre to give the extent of spatial spread of the point distribution it is based
on. This can provide visual comparison of extent of spread among different types of measured data.
In this example a 20 X 20 m sample grid 31 soil- and plant parameters were examined in case of sugar beet
sample plant. For analysing the spatial distribution of yield data 3 exact and 2 smooth interpolators were tested.
The spherical model used in case of ordinary kriging interpolation provided good result. The about 17 m griddistance still estimates the spatial variance well. Approach of the extreme values at the edges of the examination
areas provided higher spatial uncertainty. The pH distribution of the soil in case of a several decimal deviation
already showed negative correlation to the examined microelements. Cokriging significantly improved the
estimation of the less examined microelement, if spatial correlation made it possible.
The spatial weighted mean and spatial weighted deviation values were established for characterisation of the
examination space. They are available for visual mapping, thus can show the location of the parameters coordinate-correctly in the examination space. This interpretation technique considers the spatial position of each
point to another individually (distance and direction), and the value of the plant and soil parameters. Mapping
the sample area in GIS environment, the coordinates of the spatially weighted mean centre values of the
measured plant and soil parameters correlated to the mean centre values showed a northwest direction. As a new
visual analysis, the spatially weighted mean centre values of the parameters as eigenvectors were projected to
the mean centre values as origin. To characterize the production yield, the raw and digested sugar contents of
the sample area, and the absolute rotation angles of the generated vectors were determined, which indicate
numerically the inhomogeneity of the area. The generated spatial analogues are applicable to characterise
visually and quantitatively the spatial positions of sampling points and the measured parameters in a quick way.
However, their disadvantage is that they do not provide information on the clustering and direction of the spatial
correlation similarly to the original statistical parameters.
To mitigate this adverse effect, the spatial correlation methods were introduced. Based on variogram calculation,
the yield surface showed homogeneity in North – South direction. Considering the results, decrease of sampling
distance to 17 m can be suggested. The direction of the variability of yield could be modelled with a direction
variogram based on analysis of the variogram surface.
35
Created by XMLmind XSL-FO Converter.
5. SPATIAL ANALYSIS TO
EXPLORE SPATIAL
HETEROGENITY
In the study, developed methodological processes are presented for the analysis of spatial relationship between
measured production and soil parameters. 5 spatial evaluation methods for yield surface were compared. On the
basis of the analysed methods, it can be stated that different methods (LP, RBF) should be used, when the
reasons for locally extreme yields (mean: outliner data) are in focus than in case when the yield surface of the
whole area is estimated (IDW, GP).
Using adequate parameters the kriging method is applicable for both functions. Similarly to the results of an
ordinary Pearson correlation analysis, spatial correlation analysis was shown using soil pH and Cu concentration
data. The results of cross variogram analysis and the North – South direction of the variogram surface showed
negative correlation. Based on simulation calculations, decrease of 30% in sampling points resulted in increase
of 12% in error for the total sample number considering Cu concentration. The method provides a tool to
decrease the cost of sampling and sample analyses of spatially correlating features, and to increase the reliability
of spatial estimation using a better sampling strategy with the same sample number.
1. Test questions:
1. How do you calculate the mean centre of sampling points?
2. How do you calculate the weighted mean centre of sampling points?
3. How do you calculate the weighted standard distance of sampling points?
4. When do you apply deterministic interpolators?
5. When do you apply smooth interpolators?
6. When do you use cokriging?
7. If you are mapping the different spatial parameters of table 5th, this will be useful PA information and why?
2. REFERENCES:
1. Berzsenyi Z. Dang Q.Lap.: (2000) Különböző tenyészidejű kukorica(Zea mays L.) hibridek
növekedésanalízise Hunt.Parsons modellel és többváltozós módszerekkel. Növénytermelés. 49:623.640.
2. Cohran, W. G.: 1977. Sampling Techniques. 3rd edition. John Wiley and Sons, New York
3. Cressie, N. A. C. Statistics for Spatial Data; John Wiley and Sons, New York, 1991; 900.
4. Csathó, P.-Árendás T.-Németh T.: 1998. New, environmentally friendly fertiliser advisory system, based on
the data set of the Hungarian long-term field trials set up between 1960 and 1995. Communications in Soil
Science and Plant Analysis. 29. 2161-2174
5. Elek É.-Kádár I.: 1980. Állókultúrák és szántóföldi növények mintavételi módszere. MÉM NAK. Budapest.
1-55.
6. Fisher, R. A.: 1925. Statistical Methods for Research Workers. Oliver and Boyd, Edinburgh.
7. Huzsvai L.-Nagy J.: 1995. Kísérletek optimalizálása a földművelési, növénytermesztési kutatások
tervezésében. Növénytermelés, 44:5-6. 483-491.
8. Isaaks, E. H.; Srivastava, R. M. An Introduction to Applied Geostatistics, Oxford University Press, New
York, 1989; 561.
9. Journel, A.G. Huijbregts, C.: 1978. Mining Geostatistics, Academic Press, 1.600.
10.
Kádár I.: 1998. A szennyezett talajok vizsgálatáról. Kármentesítési kézikönyv 2. Környezetvédelmi
Minisztérium. Budapest. 1-151.
11.
Kádár I.Kiss E.: 2000.A cukorrépa (Beta vulgaris L.) műtrágyázása karbonátos vályog csernozjom
talajon. Növénytermelés. 49:677-690.
36
Created by XMLmind XSL-FO Converter.
5. SPATIAL ANALYSIS TO
EXPLORE SPATIAL
HETEROGENITY
12.
Kádár, I.; Kiss, E. Mineral fertilisation of sugar beet (Beta vulgaris L.) on calcareous loamy chernozem
soil, Növénytermelés, 2000; 49.677-690.
13.
Krige, D. G.: 1966. Twodimensional weighted moving average trend surfaces for oreevaluation.
Journal of the South African Institute of Mining and Metallurgy. 66:13.38.
14.
Kulcsár L.: 1999. A cukorrépa N-felvételének vizsgálata különböző termőhelyeken. Agrokémia és
Talajtan. 48: 543-560.
15.
Lee, J., Wong, S, W, Statistical analysis with ArcView GIS. John Wiley and Sons, New York, 2001; 1191.
16.
Lee, J.Wong, S. W.: 2001. Statistical analysis with ArcView GIS. John Wiley and Sons, New York, 1191.
17.
Levine, N.Kim, K.E.Nitz, L. H.: 1995. Spatial analysis of Honolulu motor vechiclevehicle crashes I.
Spatial patterns. Accident Analysis and Prevention, 27(5). 675-685.
18.
Matheron, G.: 1965. La Theorie des Variables Regionalisees et ses Applications. Masson, Paris.
19.
Monmonier, M. Mapping out. Chicago: University of Chicago Press. 1993
20.
Nagy J.: 1996. A növényszám és a talajművelés kölcsönhatása a kukoricatermesztésben.
Növénytermelés. 45:5-6 543-552.
21.
Németh T.Jolánkai M.: 2002. A precíziós növénytermesztés elemei. in Nagy J. (szerk.): 2002. EU
konform mezőgazdaság és élelmiszerbiztonság. Debreceni Egyetem, 12-21.
22.
Oliver, M.Badr, L.: 1995. Determining the spatial scale of variation in soil random concentration.
Mathematical Geology, 27:893.922.
23.
Ruzsányi L.: 1981. A műtrágyázás és az öntözés hatása a cukorrépa termésére és a gyökér beltartalmi
értékére. Növénytermelés. 30: 363-369.
24.
Sváb J.:1979. Többváltozós módszerek a biometriában. Mezőgazdasági Kiadó, Budapest.
25.
Szabó J.Bakos J.Dobos A.Cservenák R.Pásztor L.Pográny K.: 2002. Üzemi szintű agrár geoinformációs rendszer a mezőgazdasági szaktanácsadás támogatására. in Nagy J. (szerk.): 2002. EU
konform mezőgazdaság és élelmiszerbiztonság. Debreceni Egyetem, 22-31.
26.
Tamás J.:2001. Precíziós Mezőgazdaság, Mezőgazdasági Szaktudás Kiadó, Budapest, 1-160.
27.
Tamás, J, Preciziós Mezőgazdaság, Mezőgazdasági Szaktudás Kiadó, Budapest,2001; 1-160.
28.
Thapar, N.Wong, D.Lee, J.: 1999. The changing geography of population centroids in the United States
between 1970 and 1990. The Geographical Bulletin, 41:45-56.
29.
Tukey, J. W.:1977. Exploratory Data Analysis. Addison.Wesley, Reading. MA
30.
Vajdai I. (szerk.):1984. A cukorrépa termesztése. Mezőgazdasági Kiadó. Budapest.
31.
Wackernagel, H. Multivariate Geostatistics, Springer, Berlin, 1995; 251.
32.
Webster, R..-McBratney, A. B.: 1987. Mapping soil fertility at Broom’s Barn by simple kriging.
Journal of the Science of Food and Agriculture, 38. 97-115.
33.
Webster, R.-Oliver, M. A.: 1990. Statistical methods in soil and land resource survey. Oxford
University Press, Oxford.
34.
Wong, D. W.: 1999. Geostatistics as measures of spatial segregation. Urban Geography, 20(7). 635647.
37
Created by XMLmind XSL-FO Converter.
5. SPATIAL ANALYSIS TO
EXPLORE SPATIAL
HETEROGENITY
35.
Youden, W. J.Mehlich, A.: 1937. Selection of efficient methods for soil sampling. Contributions of the
Boyce Thompson Institute for Plant Research, 9:5970.
38
Created by XMLmind XSL-FO Converter.
6. fejezet - 6. BIOMASS EVALUATION
BY AIRBORNE HYPERSPECTRAL
IMAGE SPECTROSCOPY
Effective physical measurements and spatial estimation of agricultural biomass production have not available till
now, because data uncertainty of traditional field sampling was too high at large scale farm level. Nowadays,
precision farming technology introduced the real time and high scale yield mapping methods, where can be
evaluate spatial pattern of the total biomass [1]
Remote sensing spectral data acquisition is very effective way to analyse time series of agricultural plots and to
receive further information about different qualitative and quantitative plant parameters.
Conventional commercial spectrometers or spectrophotometers are usually able to measure optical spectrum in
one measuring spot at a time. Airborne hyperspectral imagery provides the potential for more accurate and
detailed information extraction than possible with any other type of remotely sensed data. The “hyper” in
hyperspectral refers to the large number (>80) of measured wavelength bands. Field and laboratory
spectrometers usually measure reflectance at many narrow, closely spaced wavelength bands, so that the
resulting spectra appear to be continuous curves. When a spectrometer is used in an imaging sensor, the
resulting images record a reflectance and intensity spectrum for each pixel in the image.
6.1. ábra - 15. Figure Hyperspectral imaging systems can capture spectral, spatial, and
radiometric information in same time
The first Hungarian aerial hyperspectral imaging programme took place within the framework of the 2002
HYSENS project. Based on early results of flight campaign were analyzed water quality and urban vegetation.
Widely used Vegetation Indices (VIs) are combinations of surface reflectance at two or more wavelengths
designed to highlight a particular property of vegetation. Hyperspectral narrowband indexes are more
sophisticated measures of general quantity than the traditional satellite broadband indexes.
Many of these indices are currently unknown in agricultural practice or under-used. The main objective of this
paper is to examine the potential of AISA DUAL airborne hyperspectral sensor data to create a narrowband
vegetation indexes distribution map of cereal agricultural fields.
39
Created by XMLmind XSL-FO Converter.
6. BIOMASS EVALUATION BY
AIRBORNE HYPERSPECTRAL
IMAGE SPECTROSCOPY
In this study the images were taken by an AISA DUAL airborne hyperspectral imaging spectrometer of
Hungarian University of Debrecen and FVM GKI. This service has operated from 2007 in Hungary. AISA is a
dual sensor system, which provides seamless hyperspectral data in the full range from 400 to 2500nm.
The transmissive imaging spectrograph is nearly independent of the polarization in the incoming light, and
provides high diffraction efficiency and uniform spectral resolution of 10 nm over the full SWIR range. All the
optics (fore optics and imaging spectrograph) and the detector assembly are temperature-stabilized. Applied
AISA systems are push broom imaging sensors, consisting of a hyperspectral and high-performance GPS/INS
sensor and a data acquisition unit housed in a rugged PC [2]. AISA sensors employ SPECIM’s high quality
transmissive imaging spectrographs which feature sub-pixel smile and keystone distortions, and very low
polarization dependency.
A real-time fibre optic down welling irradiance sensor (FODIS) can be integrated into the sensors to monitor
the illumination conditions. Auxiliary components include a mount to connect the sensor to the GPS/INS unit,
and regulated power supply. C-MIGITS III and Oxford RT3100 are a standard GPS/INS option in AISA
systems. It is a high-performance, integrated 3-axial inertial navigation sensor for monitoring the aircraft
position and attitude. The systems include CaliGeo control and operation software, which allows data
acquisition settings to be tailored for individual flight mission requirements. Calibrating imaging spectroscopy
data to surface reflectance is an integral part of the data analysis process, and is vital if accurate results are to be
obtained [3]. The identification and mapping of materials and material properties is best accomplished by
deriving the fundamental properties of the surface, its reflectance, while removing the interfering effects of
atmospheric absorption and scattering, the solar spectrum, and instrumental biases[4]. The objectives of
calibrating remote sensing data are to remove the effects of the atmosphere (scattering and absorption) and to
convert from radiance values received at the sensor to reflectance values of the land surface. At the time of data
acquisition, it is very important to characterize the calibration site with a field spectrometer [5]. We measured it
with our field also used special reference materials. The best conditions in which to make field measurements
are clear skies, near solar noon, at temperatures. A small calibration artefact could distort an absorption feature,
causing a misidentification [6]. An accurate calibration shows the fundamental properties of surface materials,
and is key to ulinking remotely sensed surface properties with laboratory data.
Analyzing vegetation using remotely sensed data requires knowledge of the structure and function of vegetation
and its reflectance properties. The absorption and reflection of solar radiation is the result of many interactions
with different plant materials, which varies considerably by wavelength. Water, pigments, nutrients, and carbon
are each expressed in the reflected optical spectrum from 400 nm to 2500 nm, with often overlapping, but
spectrally distinct, reflectance behaviours [7]. The reflected optical spectrums are also changing in different
phenological phases and plant morphology
The leaf area index LAI is the green leaf area per unit ground area, which represents the total amount of green
vegetation present in the canopy. The LAI is an important property of vegetation, and has the strongest effect on
overall canopy reflectance. The MLA is the average of the differences between the angle of each leaf in a
canopy and horizontal and depends on plant genetics, morphology and actual physiological status of species.
Test sites of the remote sensing was close to Siofok city (UL Geo 18º 0’ 0.26” E; 46º 54’ 3.14” N; spectral
interval 398-973 nm, average bandwidth 10 nm).
First step was data exploitation, where spectral statistics of region of interest site was calculated. On this figure
we can observe the real spectral differences in green (550nm), red (650nm) and infrared (780 nm) channels.
6.2. ábra - 16. Figure Spectral statistical curves of regional interest
40
Created by XMLmind XSL-FO Converter.
6. BIOMASS EVALUATION BY
AIRBORNE HYPERSPECTRAL
IMAGE SPECTROSCOPY
The most parameters of vegetation indexes are very sensitive in these intervals. The calculated vegetation
indexes were summarized in table 1.
6.3. ábra - 6. Table Spectral narrowband indexes of cereals
The Modified Red Edge Simple Ratio (mSR 705) index is a modification of the traditional broadband SR index.
It differs from the standard SR because it uses bands in the red edge and incorporates a correction for leaf
specular reflection.
Applications include precision agriculture, forest monitoring, and vegetation stress detection. The value of this
index ranges from 0 to 30. The common range for green vegetation is 2 to 8 [8]. Actual numbers of Siofok site
were: x =11,14, max. = 18,7; min. = 1,55.
The Red Edge Normalized Difference Vegetation Index (NDVI 705) is intended for use with very high spectral
resolution reflectance data. Applications include precision agriculture (an information- and technology-based
agricultural management system to identify, analyze, and manage site-soil spatial and temporal variability),
forest monitoring, and vegetation stress detection[9].
41
Created by XMLmind XSL-FO Converter.
6. BIOMASS EVALUATION BY
AIRBORNE HYPERSPECTRAL
IMAGE SPECTROSCOPY
This VI differs from the NDVI by using bands along the red edge, instead of the main absorption and
reflectance peaks. The NDVI 705 capitalizes on the sensitivity of the vegetation red edge to small changes in
canopy foliage content, gap fraction, and senescence. The value of this index ranges from -1 to 1. The common
range for green vegetation is 0.2 to 0.9 [10]. Actual numbers of Siofok site were: x =0,83, max. = 0.87; min. =
0,17.
Every different vegetation index can be visualized in 3D continuous surface. This surface also can be use to
make different directional cross profile, which very effective tool to find spatial anomalies.
6.4. ábra - 17. Figure NDVI 3D surface of winter wheat block
6.5. ábra - 18. Figure Spatial-Spectral cross-sections of winter wheat block
The Vogelmann Red Edge Index (VOG1-2) is a narrowband reflectance measurement that is sensitive to the
combined effects of foliage chlorophyll concentration, canopy leaf area, and water content. Applications include
vegetation phenology (growth) studies, precision agriculture, and vegetation productivity modelling [11]: The
value of this index ranges from 0 to 20. The common range for green vegetation is 4 to 8. Actual numbers of
Siofok site were: x=2,59, max. = 2,84; min. = 1,33.
The Photochemical Reflectance Index (PRI) is a reflectance measurement that is sensitive to changes in carotene
pigments (particularly xanthophylls pigments) in live foliage [12]. Actual numbers of Siofok site were: x=0.016,
max. = 0,07; min. = -0,05.
Carotene pigments are indicative of photosynthetic light use efficiency, or the rate of carbon dioxide uptake by
foliage per unit energy absorbed. As such, it is used in studies of vegetation productivity and stress.
42
Created by XMLmind XSL-FO Converter.
6. BIOMASS EVALUATION BY
AIRBORNE HYPERSPECTRAL
IMAGE SPECTROSCOPY
Applications include vegetation health in forests, and agricultural crops prior to senescence. The value of this
index ranges from -1 to 1. The common range for green vegetation is -0.2 to 0.2.
Stress-related pigments include carotenes and anthocyanins, which are present in higher concentrations in
weakened vegetation. Carotenes function in light absorption processes in plants, as well as in protecting plants
from the harmful effects of high light conditions. The Carotene Reflectance Index 1-2 (CRI1-2) is a reflectance
measurement that is sensitive to carotene pigments in plant foliage. Higher CRI1 values mean greater carotene
concentration relative to chlorophyll [13]. The value of this index ranges from 0 to more than 15. The common
range for green vegetation is 1 to 11. CR2 provides better results in areas of high carotene concentration. Actual
numbers of Siofok site were: =6,99, max. = 8.7; min. = 4.58.
Anthocyanins are water-soluble pigments abundant in newly forming leaves and leaves undergoing senescence.
The Anthocyanin Reflectance Index 1 (ARI1) is a reflectance measurement that is sensitive to anthocyanins in
plant foliage. Increases in ARI1 indicate canopy changes in foliage via new growth or death[14]. The ARI2 is a
modification of the ARI1 which detects higher concentrations of anthocyanins in vegetation. The value of these
indexes ranges from 0 to more than 0.2. The common range for green vegetation is 0.001 to 0.1. Actual numbers
of Siofok site were: x=0.00012, max. = 0,0009; min. = 0,000086.
Water content is an important quantity of vegetation because higher water content indicates healthier vegetation
that is likely to grow faster and be more fire-resistant. The Water Band Index (WBI) is a reflectance
measurement that is sensitive to changes in canopy water status. As the water content of vegetation canopies
increases, the strength of the absorption around 970 nm increases relative to that of 900 nm. The common range
for green vegetation is 0.8 to 1.2. Applications include canopy stress analysis, productivity prediction and
modelling, fire hazard condition analysis, cropland management, and studies of ecosystem physiology. WBI is
defined by the following equation. Actual numbers of Siofok site were: x=1,22, max. = 1,59; min. = 0,87.
1. Test questions:
1. Can you give a definitions about airborne hyperspectral image spectroscopy?
2. Do you describe an AISA DUAL system?
3. How do you calculate Modified Red Edge Simple Ratio (mSR 705) index, Red Edge Normalized Difference
Vegetation Index and can you determine some important application fields?
4. How do you calculate Vogelmann Red Edge Index, Reflectance Index and can you determine some
important application fields?
5. Question: How do you calculate Anthocyanin Reflectance Index, Water Band Index and can you determine
some important application fields?
2. References
[1] Németh, T., Neményi, M., Harnos, Zs. (2007). A precíziós mezőgazdaság módszertana. Szegedi Egyetemi
Kiadó.p.238
[2] http://www.specim.fi/
[3] Green, R.O., Eastwood, M.L., Sarture, C.M., Chrien, T.G., Aronsson, M., Chippendale, B. J., Faust, J. A.,
Pavri, B. E., Chovit, C. J., Solis, M., Olah, M. R., &Williams, O. (1998).Imaging spectroscopy and the Airborne
Visible/Infrared Imaging Spectrometer (AVIRIS).Remote Sensing of Environment65: 227-248
[4] Clark, R.N., King, T.V.V,, Klejwa, M., Swayze, G., &Vergo, N. (1990). High Spectral Resolution
Reflectance Spectroscopy of Minerals, J. Geophys Res.95, 12653-12680.
[5] Clark, R.N., Swayze, G.A., Livo, K.E., Kokaly, R.F., King, T.V.V., Dalton, J.B., Vance, J.S., Rockwell,
B.W., Hoefen, T., &McDougal, R. R., (2002). Surface Reflectance Calibration of Terrestrial Imaging
Spectroscopy Data: a Tutorial Using AVIRIS in Proceedings of the 10th Airborne Earth Science Workshop, JPL
Publication 2:1.
43
Created by XMLmind XSL-FO Converter.
6. BIOMASS EVALUATION BY
AIRBORNE HYPERSPECTRAL
IMAGE SPECTROSCOPY
[6] Conel, J.E., Green, R.O., Vane, G., Bruegge, C.J., &Alley, R.E. (1987).AIS-2 radiometry and a comparison
of methods for the recovery of ground reflectance. In: Proceedings of the Third Airborne Imaging Spectrometer
Data Analysis Workshop (G. Vane, Ed.), JPL Publ. 87-30, Jet Propulsion Laboratory, Pasadena, CA, pp. 18-47
[7] Asner, G. P. (1998). Biophysical and Biochemical Sources of Variability in Canopy Reflectance, Remote
Sensing of Environment, 64:234-253
[8] Datt, B., (1999). A New Reflectance Index for Remote Sensing of Chlorophyll Content in Higher Plants:
Tests Using Eucalyptus Leaves. Journal of Plant Physiology 154:30-36
[9] Gitelson, A.A. Merzlyak, M.N. (1994). Spectral Reflectance Changes Associated with Autumn Senescence
of Aesculus Hippocastanum L. and Acer Platanoides L. Leaves. Spectral Features and Relation to Chlorophyll
Estimation.Journal of Plant Physiology 143:286-292
[10] Sims, D.A,. Gamon, J.A. (2002). Relationships Between Leaf Pigment Content and Spectral Reflectance
Across a Wide Range of Species, Leaf Structures and Developmental Stages. Remote Sensing of Environment
81:337-354
[11] Vogelmann, J.E., Rock, B.N., &Moss, D.M. (1993).Red Edge Spectral Measurements from Sugar Maple
Leaves.International Journal of Remote Sensing 14:1563-1575
[12] Champagne, C., Pattey, E., Bannari, A., &Stratchan, I.B. (2001).Mapping Crop Water Status: Issues of
Scale in the Detection of Crop Water Stress Using Hyperspectral Indices. In: Proceedings of the 8th
International Symposium on Physical Measurements and Signatures in Remote Sensing, Aussois, France. pp.7984
[13] Gitelson, A.A., Zur, Y., Chivkunova, O. B., &Merzlyak, M. N. (2002).Assessing Carotenoid Content in
Plant Leaves with Reflectance Spectroscopy.Photochemistry and Photobiology 75:272-281
[14] Gitelson, A.A., Merzlyak, M.N., &Chivkunova, O. B. (2001). Optical Properties and Nondestructive
Estimation of Anthocyanin Content in Plant Leaves.Photochemistry and Photobiology 71:38-45
[15] Gamon, J.A., Penuelas, J., Field, C.B. (1992).A Narrow-Waveband Spectral Index That Tracks Diurnal
Changes in Photosynthetic Efficiency.Remote Sensing of Environment 41:35-44
44
Created by XMLmind XSL-FO Converter.
7. fejezet - 7. EARLY DETECETION IN
PLANT PROTECTION
The core of the modern environmentally friendly precision plant protection technology is the exact mapping of
the infected area. The delineation of the infection makes the plant protection more effective and at the same time
lowers the quantity of the harmful and costly pesticides. The paper presents the allocation of the disease with
remote sensing techniques. The examinations of the spatial variability of sugar beet leaf area were performed in
the frame of a large scale field pest control experiment in the Tedej model area.
The effects of different pest control treatments on leaf area - applied on the model area - were studied
Cercospora sp. leaf destruction by processing DAIS 7915 images. The new devices of the precision agriculture
enable to use new, environmentally sound methods which allow exact dosage of pesticides to cut down the
amount of harmful chemicals getting into the food chain [6]. Remote sensing technologies provide an important
tool to aid site-specific management of crops. Remote sensing has the potential to provide real-time analysis of
the attributes of a growing crop that can assist in making timely management decisions that affect the outcome
of the current crop. Laudien et al. (2003)[3] analysed sugar beet disease using hyperspectral remote sensing
measurements, separating the health and the disease plants.This study presents the aims and methods of image
analysis of hyperspectral data for evaluating the spatial distribution of sugar beet diseases.
The investigated area is situated in the eastern part of the Hungarian Great Plain, next to a small rural town
called Hajdúnánás. In the model area selected for Cercospora examination in large parcel scale shroud
protection experiments, rhizomania tolerant species (Triplex) were used, with same treatment in each parcel.
The examination included the analysis of effects of chemicals’ treatment time, treatment number, and different
fungicides. Also we collected reference picture from red and infrared channels by TETRACAM hand camera
(Tetracam Inc.Chatsworth California). Tetracam “off the shelf” products include several camera assemblies
utilizing CMOS image sensors for both visible and near-infrared photography. Tetracam camera is personnel
routinely handle both small, well-defined design tasks and full project management assignments. Tetracam
engineers commercialized the low-cost multi-spectral sensor for agricultural remote sensing. The technical
specifications of TETRACAM camera :
Images carry Red, Green and NIR data;Color LCD viewfinder; Review of stored images in three modes; Raw
false color; Monochrome IPVI; Colorized IPVI; Image storage on Compact Flash card; USB interface for
connection to host computer; External GPS input; External 6VDC input; Interchangable C-mount lens system;
CNC machined aluminum housing; 5.5 x 3.1 x 2.0 inches (not including lens).
When take a photo, the cells in plant leaves are very effective scatterers of light because of the high contrast in
the index of refraction between the water-rich cell contents and the intercellular air spaces. Vegetation is very
dark in the visible (400-700 nm) because of the high absorption of pigments which occur in leaves (chlorophyll,
protochlorophyll, xanthophyll, etc.). There is a slight increase in reflectivity around 550 nm (visible green)
because the pigments are least absorptive there. In the spectral range 700-1300 nm plants are very bright
because this is a spectral no-man's land between the electronic transitions which provide absorption in the
visible and molecular vibrations which absorb in longer wavelengths. There is no strong absorption in this
spectral range, but the plant scatters strongly as mentioned above. From 1300 nm to about 2500 nm vegetation is
relatively dark, primarily because of the absorption by leaf water. Cellulose, lignin, and other plant materials
also absorb in this spectral range. Red chanel is used to indicate a band covering all or part of the portion of the
visible spectrum perceived as red by the human eye (600-700 nm). Examples LandsatTM 3 band. Near Infrared
chanel is used to indicate a band covering all or part of the near-infrared portion of the spectrum (800-1100 nm
or a subset of these wavelengths). Examples: Landsat TM4. Nearly all of the commonly used vegetation indices
are only concerned with red-near-infrared space. The camera captures green, red and near-infrared bands. The
band / image color relationship for the NIR_R_G image is.
An important part of the entire camera and software system is the ability to calibrate the software supplied with
the camera, both image processing softwares (PixelWrench's Index plugin and Briv32). Calibration consists of
taking an image of the Teflon calibration tag under the same conditions as the images under study. This image is
used to teach the application software what the spectral balance of that day's sunlight is. Have to place the
calibration tag on the ground or hold it level with the ground and photograph it. It need not fill the entire frame
and it must not be overexposed. Make sure to avoid a direct reflection of the sun. The calibration picture should
45
Created by XMLmind XSL-FO Converter.
7. EARLY DETECETION IN
PLANT PROTECTION
be taken under the same conditions and as near in time to the study images as possible. Directional anisotropy in
brightness temperature and NDVI are calculated to be less than 7 to 12% respectively for zenith view angles less
than 30 deg. but range up to 22 to 40% for zenith view angles of 60 deg. Analysis of variation in local standard
deviations with spatial resolution shows a maximum peak corresponding to crop row spacing with rapid fall-off
at larger scales.Tetracam use the ADC formats and stores images in a "losslessly compressed" form. This
method reduces the storage requirements somewhat without causing any image degradation. The resulting file
must be decompressed, formatted and presented to the application screen using the software supplied with the
camera. All (or only the selected images) are written to a DCA file on disk. Field worker can then re-open the
DCA file at any time to retrieve images for additional processing.
Satellite observations of agricultural and other plant canopies in the thermal and near IR regime have generally
been at spatial scales of tens to hundreds of meters. Advances in sensor technology will extend our capabilities
for IR measurements from space to yield improved spatial resolutions. The airborne hyperspectral image
spectrometry also a novell method of the weed survey, which was applied on field of Hajduság loess platau.
This spectraly detailed datasourse was more effective to distinquis weed species than earlier used a neaer-field
broad-band remote sensining methods[7]. The results of thematic maps were important tool to estimate Shannon
biodiversity index in ecological forecast.
In our case study,for the determination of the Cercospora leaf spots (Cercospora beticola), bonitation were
applied in a way, that vegetation were examined in each parcel 250 by 25m, than an average was
calculated.(Figure 19).
7.1. ábra - 19. Figure Cercospora leaf spots -Cercospora beticola)
DAIS-7915 hyperspectral image was applied for image analysing (Table 1.) that was used evaluating in other
environmental protection aspects [4].
7.1. táblázat - 7. Table Spectral sampling of test size
Image type
Image Data
Image size
Ground
resolution
DAIS - 7915
19/08/2002
3990m*8735m 5m
Bands number Bandwidth (μm)
79
1.-32.: 0,40-1,00
33.-40.: 1,50-1,80
41.-72.: 2,00 – 2,50 )
73.: 3,00-5,00
46
Created by XMLmind XSL-FO Converter.
7. EARLY DETECETION IN
PLANT PROTECTION
74.-79.: 8,00-12,60
To calculate the homogeneity of the parcels principal component analysis was carried out and the first
component (giving the highest variability) V01 was analysed with histogram matching. By the majority of the
parcels normal distribution could be observed, in case of 5. and 8. parcel two different population were
recognizable with different means and standard deviations, also with a recognizable spatial distribution. The tprobe based on the first principal component justified, that the 5. and the 8. parcels are significantly different.
7.2. ábra - 20. Figure The first PCA of the homogeneous (1., 2.) and the heterogeneous
(3.,4.) spatial distribution parcels.
Between the observed Cercospora infection and the reflectance and radiance values regression was calculated.
By the trendline matching in any case the linear regression represented the highest R square. The results are
shown on Figure 21.
7.3. ábra - 21. Figure Deterministic coefficients (R2) of Cercospora leaf spots
(Cercospora beticola) (%)
From the reflectance values of the DAIS sensor the LANDSAT red B3 NIR B4 channel NDVI was calculated
and was compared with infection rate. In the vegetation analysis the plant-stress analysis method the so called
red edge position (REP) and the Chlorophyll Absorptions Integral (CAI) are the most commonly used [5]. For
the determination of the Cercospora infection I this case the REP method [3] was used. The green reflectance
maximum of the B4 channel (551 nm) from the first derivative of the hyperspectral image curve, the first peak
of the NIR REP value from the 780 nm B17 channel was calculated. After the determination of the REP
regression was calculated with the values of the infection (Figure 22.)
7.4. ábra - 22. Figure Cercospora leaf spots map and the investigated parcels
47
Created by XMLmind XSL-FO Converter.
7. EARLY DETECETION IN
PLANT PROTECTION
The R Square (R2 = 0,663) was higher, than in the case of NDVI, dislocation of the red edge position could be
observed towards the lower values with the growth of the infection. Cercospora map was made by image
processing from the B37 band of georectified hyperspectral data. This map shows the spatial distribution of the
disease inside the parcel (Figure 23).
7.5. ábra - 23. Figure Cercospora leaf spots map and the investigated parcels (n=14).
The spatial pattern of each treatment was examined by principal component analysis using the values
representing the variability of vegetation as the principal component. The results show different inside some of
the parcels that support the Cercospora leaf spots map. Based on the regression analysis it could be concluded
that each NIR channel is suitable for the estimation of leaf area changes, although MIR and TIR range is less
sensible to vegetation changes. In the course of calculations NDVI, REP values were obtained from preprocessing and compared to values of leaf contamination/infestation. The highest regression rates could be
estimated from the iteration model (R2 =0,731, p<0,05) calculated from the 1,668 μm range that can be used
making a detailed Cercospora leaf spots map.
1. Test questions:
1. What type plant disease can modify the normal spectral reflectance curve and why?
2. Which spectral VIR and NIR/SWIR intervals are very sensitive for plant stress conditions?
3. Why effective tool PCA is to compress data?
2. References
[1] Gyula, S. - Béla, U. - László V. - Balázs K.: 2007. Application of by-products of bioetanol production in
feeding, environmental and feeding safety concerns of utilization. Cereal Research Communications, Vol. 35,
No. 2 pp 1065-1068
[2] Lakner, Z. - Kóbor, K .- Pozsonyi, F. - Pándi, F: 1993. The Possibilities and Chances of a Hungarian
Bioethanol Program. Acta Agronomica Hungarica, 42. 3-4. pp.424-428.
48
Created by XMLmind XSL-FO Converter.
7. EARLY DETECETION IN
PLANT PROTECTION
[3] Laudien, R. - Bareth, G. - Doluschitz, R.: 2003. Analysis of hyperspectral field data for detection of sugar
beet diseases. EFITA 2003 Conference, Debrecen, pp.375-381.
[4] Nagy, A. - Tamás, J. - Burai, P.: 2007. Application of advanced technologies for the detection of pollution
migration In Proc. VI. Alps-Adria Scientific Workshop, HAS Obervellach, Austria, pp. 805-809.
[5] Smith, K.L. - Steven, M.D. - Colls, J.J.: 2004. Use of hyperspectral derivative ratios in the red-edge region to
identify plant stress responses to gas leaks. Remote Sensing of Environment, 92, pp.207-217.
[6] Tóth T, - Németh T, - Bidló A, - Dér F, - Fekete M, - Fábián T, - Gaál Z, - Heil B, Hermann T, - Horváth E,
Kovács G, - Makó A, - Máté F, - Mészáros K, - Patocskai Z, - Speiser F, - Szűcs István, - Tóth G, - Várallyay
Gy, - Vass J, - Vinogradov Sz.: 2006. The Optimal Strategy to Improve Food Chain Element CyclesDevelopment of An Internet Based Soil Bonitation System Powered by a Gis of 1:10 000 Soil Type Maps.
Cereal Research Communications 34: (1) pp. 841-844 .
[7]Sárkány, E., Lehoczky, É, TAMÁS, J. : 2008.Spreading , ecology and damages caused by the common
milkweed (Asclepias syriaca L.) in Hungary.VII. Alps-Adria Scientific Workshop.Stara Lesna, Slovakia, 2008.
DOI: 10.1556/CRC.36.2008.Suppl.1
49
Created by XMLmind XSL-FO Converter.
8. fejezet - 8. THREE DIMENSIONAL
DYNAMICAL SOIL WATER REGIME
Most of the water regime assessments are one dimensional whether it is horizontal or vertical. This study aims
to survey the dynamic change of soil heterogeneity and water regime in 3 dimensional scales in orchards. The
main goal is to establish such a precision decision support system, with which the water management properties
of soil can be meliorated, and the irrigation management can be rationalized.
The limited water resources and the increasing frequency of extreme hydrological events (floods, water-logging,
over-moistening and drought) due to the high territorial and temporal variability of atmospheric precipitation;
the heterogeneous (micro) relief; and the unfavourable physical/hydro physical characteristics of soils are
pressing to improve agricultural water use efficiency and necessitates an efficient control of soil moisture
regime in the Carpathian Basin [2, 3,4,5].Hungary has favourable agro ecological potential for pomaceous fruit
production. Nowadays one of the novel and widespread achievement of pear production is the intensive pear
plantations with high tree density [6, 7, 8]. Orchards are relatively not highly water consumer comparison with
cereal species. However, to ensure optimal water capacity values calculating breeding season it is the most
important risk factor. It cannot imagine the fruit production without modern irrigation system.
Due to the heterogeneous terrain surface special attention has to be paid to places with different location in order
to examine all of the different soil varieties. The coordinates of the sampling points were collected by GPS.
Systematic sampling strategy was carried out based on the number of the rows and apple trees to collect as much
information as possible with possibly the least number of samples. The soil samples were collected from the
surface and 40cm, 70 cm depth. The upper limit of soil plasticity (K A) – according to Arany – was measured in
order to determine the spatial distribution of the physical characteristics of the examined site. Using the samples
with the original soil structure, maximal (pF=0) and minimal (pF=2) water holding capacities were determined
to evaluate water bearing properties of the examined soils. Soil density was also measured on field. 3T System
penetrometer was used to measure the soil penetration resistance at each centimetre for 60 cm depth and
expressed in kPa. It also measures the soil moisture content at each 1-cm-thick layer. Cone angle of the probe
which penetrates to the soil and collects the soil density data was 60˚.
The matrix potentials of soil were measured by analogous tensiometers, from 1. June 2010 to 31. August 2010.
The measured pressure value was converted to water height in cm so as the water content of soils can easily be
determined if the pF curves of soils are known. The tensiometers can successfully be used on light sandy soils to
automate the monitoring of the water regime of soils. Gauges were set at 6 sampling points in 40 and 70 cm
depth. The soil moisture tension values were always measured in the morning at the same time. The pF curves
of the soil were measured in 40 and 70 cm depth.
Based on soil plasticity, sites with different characteristics could be distinguished in every layer (surface, 40 cm
and 70 cm). The spatial variability of soil plasticity, thus the physical features of the soil appeared differently in
each layer (Figure 24.). Since the maximal saturation percentage (K A=30) measured at the sampling point with
the lowest altitude, it was caused by the micro and meso relief. Significant differences among physical
characteristic of the three soil layers can not be found.
Soil density of the soil varies between 1.51 – 1.57 t m-3.The measured pF curves are typical for sandy soils. The
water management properties of soils can be determined by this pF curves. The available water capacity was
8.61 V/V% at 40 cm depth and 9.24 V/V% at 70 cm depth. The amount of water, that a soil doesn’t hold against
gravitational forces and could be drained, was 23 V/V% at 40 cm layer and 18 V/V% at 70 cm layer.
Concerning these data, the examined sandy soil with low capillarity loses the great amount of its water content
even in the case at low tensions. So the water retention of this soil is slight, and have small amount of available
water regime, which means that this soil can only satisfy the water content of the apple orchard for a short term
dry period.
Despite the fact, that there is a sandy soil at the examined site, surplus water occurred at the whole vegetation
period in 2010, due to the frequent, intensive and large amount of precipitation. The presence of surplus water
was even supported by the compacted layer at 40-60 cm depth. Even in the 20-30 cm soil layer the soil density
reached and exceeded the 3MPa soil penetration resistance value, which is the threshold for the high soil
density, according to Birkás (2002) measurements [1]. The mean penetration resistance values of deeper soil
layers were clearly exceeded this threshold. The high soil density changed dramatically the water holding
50
Created by XMLmind XSL-FO Converter.
8. THREE DIMENSIONAL
DYNAMICAL SOIL WATER
REGIME
capacity, infiltration intensity and water saturation properties of the sandy soil. At the Eastern part of the
examined site, sandstone layer with extreme high soil density was found at 30-40 cm depth. At this layer
penetration resistance values exceeded the upper limit of the measurement range (10000 kPa) of the
penetrometer, therefore it was not possible to measure further soil layers (Figure 24.).
8.1. ábra - 24. Figure: Left picture - Spatial distribution of soil plasticity, KA (apple
trees illustrated as points on the surface layer); Right picture - spatial distribution of
soil density (soil penetration resistance *100kPa; blanked sites means above detection
level of the penetrometer).
The soil water tensions measured by tensiometers were varied between pF 0 and 2.5 due to the extreme
precipitation circumstances in 2010. Based on both of the soil moisture tension values and pF curves the rate of
easily drainable gravitation pore volume was considerable. Tensiometers at 40 cm depth resulted fast and
significant respond to precipitation; the tension was dropped markedly. While in 70 cm depth, rainfall had slight
effect on soil moisture tension, total water contentwas only measured at concerned dates, after long lasting
heavy rainfalls, e.g. at the end of July (Figure 25).
This phenomenon also suspects the presence of compacted layer at 40-70 cm depth.To determine the accurate
amount of the drainable gravitation water regime total and field water capacity was also measured and described
in 3 dimensions. In accordance with the results the amount of drainable water regime was about 20.6 V/V% at
40 cm depth and 18.6 V/V% at 70 cm mainly.
8.2. ábra - 25. Figure The soil moisture tensions at 3 June and 28 July 2010.
51
Created by XMLmind XSL-FO Converter.
8. THREE DIMENSIONAL
DYNAMICAL SOIL WATER
REGIME
The total drainable water regime is 920 m3 ha-1 from the upper 40cm soil layer and 1460 m3 ha-1 from the upper
70 cm soil layer. The harmful surplus water can be infiltrated by the loosening of the compacted soil layer in the
50-70 cm depth or led off by vertical drainage. Therefore one, solo knife coulter is suggested to use at 80-90cm
depth. The narrow loosening width does not injure considerably the pomaceous tree root zone, but the surplus
water can be infiltrated to deeper soil layers. Thus the harmful effects (fruit crack, falls, and tree necrosis) can
be prevented.
The examinations were carried out at an intensive apple orchard in Debrecen-Pallag in 2010. The examination
site is the part of the Experimental Pomology Site and Study-Farm of the University of Debrecen, Centre for
Agricultural and Applied Economic Sciences. Upper limit of plasticity according to Arany, maximum and
minimum water holding capacity, pF, matrix potential of soils, soil density and soil penetration resistance were
measured in two soil layers to obtain appropriate information on the physical and water management properties
of the soil.
1. Test questions:
1. Do you know why important these pF values: 0; 2; 4,2?
2. What kind of relationships exist among differential soil porosity- plasticity- available water content?
3. Do you describe an optimal soil physical status on sandy-clay soil?
2. References
[1] Birkás, M.: 2002. Környezetkímélő és energiatakarékos talajművelés. Akaprint Nyomdaipari Kft.
[2] Pálfai, I.: 2000. The role and significant of water in the Hungarian Plain. (In Hungarian) Nagyalföldi
Alapítvány. Békéscsaba.
[3] Somlyódy, L.: 2000. Strategy of Hungarian water management (In Hungarian). MTA Vízgazdálkodási
Tudományos Kutatócsoportja, Budapest. 370.
[4] Várallyay, Gy.: 2002. The role of soil and soil management in drought mitigation . In: Proc. Int. Conf. On
Drought Mitigation and Prevention of Land Desertification, Bled, Slovenia, April 21-25 2002. ICID-CIIC. (CD)
[5] Várallyay, Gy.: 2007. Soil resilience (Is soil a renewable natural resource?). Cereal Research
Communications35. 1277–1280.
52
Created by XMLmind XSL-FO Converter.
8. THREE DIMENSIONAL
DYNAMICAL SOIL WATER
REGIME
[6] Takács, F.: 2009. Almaalanyok értékelése két művelési rendszerben a nyírségi termesztő körzetben. PhDthesis Corvinus Egyetem, Budapest.
[7] Soltész, M., Szabó, T.: 1998. Alma. 119-155. p. In: Soltész, M. (Szerk.): Gyümölcsfajta-ismeret és használat. Mezőgazdasági Kiadó, Budapest. 513 p.
[8] Hrotkó, K.: 1999. A gyümölcsfajták alanyai. 407-506. p. In: Hrotkó K. (Szerk.): Gyümölcsfaiskola.
Mezőgazda Kiadó, Budapest. 550 p.
53
Created by XMLmind XSL-FO Converter.
9. fejezet - 9. SOIL
CHARACTERIZATION AT LARGE
SCALE
Soil, as one of the most important natural resources, is affected by several soil tillage practices, which are often
applied in rates following certain soil parameters. However, where soils are heterogeneous in space and time
soil information data sampled at densities that reflect the scale of spatial variability is required for the
development of soil maps and decision support strategies. Soil has been surveyed on a national basis in many
countries, for a broad variety of planning purposes. However, soil maps specific for (map updating and
upgrading of pedotransfer functions) do not form adequate sources of information for planning and evaluating
land use and for many other soil-related activities at regional scale. Nowadays, the sustainable agriculture, an
innovative, integrated and internationally standardized approach aiming to increase the efficiency of resource
use and to reduce the uncertainty of decisions, is globally recognized as a potentially viable means of meeting
the future demands on balancing agricultural productivity, economic stability, resource utilization, degradation,
and environmental impacts [1]. Soil resource management is one aspect of sustainable agriculture that is needed
to overcome limitations in economical sources, while maintaining or enhancing environmental quality [5]. Use
of precision farming technologies requires better understanding of soil variability in physical, hydraulic and
chemical properties [2]. As part of the decision-making process, the choice of a suitable sampling strategy is
crucial to the interpretation of the results. However, a widenumberofdifferentguidelines on soil sampling is
recommended, at international level. The evaluation of different sampling approaches requires well
characterised tools stable in time.
Measurements of soil physical and hydraulic properties are time-consuming andexpensive. Inaddition, a large
number of measurements are necessary to quantify their space-time variability. Reliable measurement of these
properties is confounded by the extreme spatial heterogeneity and inherent nonlinearity of soil characteristics.
Therefore, it is desirable to develop simplified methods to characterize soil media properties over large areas
[3].
Remote sensing data are also useful in helping to define management units. Remote sensing offers the potential
for identifying fine scale spatial patterns in soil properties across a field, and optimizing soil sampling strategies
to quantify those patterns.
Gomez et al. (2006) as summarized in research in environmental monitoring, modelling and precision
agriculture need good quality and inexpensive soil data[3]. Hence we need the development of more time- and
cost-efficient methodologies for soil analysis. Visible and near infrared reflectance (VIS–NIR) spectroscopy is
aphysicalnon-destructive, rapid, reproducible method that provides inexpensive prediction of soil physical,
chemical and biological properties according to their reflect ancein the wave length range from 400 to 2500 nm.
In this chapter will be introduce more test field, one of themis located at Debrecen-Pallag (lat. 47°29’N,
long.21°39’E) region East-Hungary. The soil type in the area is acidic sandy soil with thin interstratified layers
of colloid and sesquioxide accumulation, poor inclay, nutrient and humus Nagyt al.,2006). The use of different
soil sampling strategies and techniques may affect the results obtained from the analysis of samples collected in
the same area. For the investigation, 57 surface soil samples were taken, in order evaluate different sampling
methods, the effectiveness of classical, grid-basedrandom and method based on hyperspectral remotely sensed
data classification at dedicated wave length (Kovács et al. 2009).
The data set of the first sampling was used the classical methods. At sampling sites, atotalof
11soilcoresweretakentoadepthof20cm at5m intervals on a grid measuring 100 m×100 m and with the centre
point of the gridatthesamplelocation.Thesamplesweretakenfrom theupper0-20cm layer of the soil and according
tothe Hungarian sampling standard.
Thesecondsamplingstrategyincludedrandom grid-basedpointsampling withgridof100m×100m,andadditional
samples were taken by using increased sampling density, for the validation calculations.
The
evaluation
of
the
potential
of
hyperspectral
reflectance
data
representativesamplenumberbasedonthe samedatasetasforgrid-based attributes’ analyses.
54
Created by XMLmind XSL-FO Converter.
in
defining
9. SOIL CHARACTERIZATION
AT LARGE SCALE
Hyperspectral remotely sensed data were processed by using ENVI software.The objective of the statistical
analysis wastoidentifythespectral regions contributing to predict the chosensoilproperties,andtodetermine under
what uncertainty the reflectance spectra could be used for the prediction. Multiple linear regression for the
specification of the relationship betweenaresponsevariable(Y)andasetofdependentvariables(X) were
carriedout.Forthebestcorrelatedband selection, multiple stepwise regression at 5% significant level was used,
while, to identify the most informative wavelength range, the Principal Component Analysis was applied.
Allsoilsampleswereair-driedatambienttemperature.Soilsamples
weresievedtoremovestonesand
plantdebris,andmixedthoroughlyto obtainarepresentativesample.After dryingtheywerepassedthrougha2 mm
mesh sieve. For the characterisation of the soil the following soil parameters were determined: (1) The
mechanical composition wasdeterminedbythe plasticityindexaccordingtoArany,whichquantifies the amount of
water in cm3 added to 100 g air-dry soil sample to obtain a yarn (upper limit of plasticity).(2)The pH of each
subsample were measured potentialmetrically by first preparing suspensions of each mine waste sample in
deionised water (1 : 2.5 w/w), allowing each to equilibrate for 24 h at room temperature before measurement,
and (3) The samples used for the capillary rise ability test were tubular samples with thin walls (5 mm) and thick
walls (10 mm) and were all 1350 mm long. The capillary rise height was measured 5 hours intervals by visual
observation.
As summarized in Bocchi et al. (2000), the first step in soil characterization is to record the main properties and
then seek plausible explanations for their distributions in the light of the statistical analysis [2]. The analytical
data were processed using a classical statistical approach to test the normality of data distribution.
The development of computational resources coupled with the development of geostatistical prediction methods
have allowed the possibility of mapping soil by considering different kinds of secondary information. In order to
provide an improved prediction map, the soil variables were interpolated using the ordinary kriging method, a
univariate interpolation method based on a weighting scheme widely used in soil science to estimate the
unknown primary variable at unsampled location as a linear combination of neighbouring observations
(Webster and Oliver,2001). Taking into consideration different sampling sizes 11, 22 and 33 samples randomly
selected and using ordinary kriging interpolation, the spatial distributions of the soil physical
properties from the different methods were evaluated.
Debrecenareaischaracterizedby sandsoiltype. Comparison between themeasured and predicted soilproperties
demonstratesimilarpatternsand
valuerangesforthesoilproperties
examined.Theupperlimitofplasticity
accordingtoArany(KA)varies between25-28,whichdescribessoiltextureandthusindicatesthatthesoil in question is
sand [7].The percentage of clay in the samplesrangesfrom25%to30%, correspondingtothesandtexture.In
addition,thecapillaryriseabilitytestofvalueshownslightvariability,as
thevalueofthecapillary
waterrangedbetween20and40(20).Considering pH,thestudyareaistypicallyslightlyacidic,however,thereareoutliers
in thehighlyacidicregion(pH=4.11)despitethattheareaconsideredisonly 11 ha, and the soil tillage
practiceappliedisuniform.
9.1. táblázat - 8. Table Descriptive statistics of soil parameters
Sample
size
Mean
Median
S.D.
Variance Kurtosis
Skewness Min
Max
pH
5,34
5,35
0,71
0,511
4,43
1,18
4,11
7,12
plasticityi
ndex
28,72
29
1,10
1,218
0,69
0,65
27
31
capillary
rise
33,8
33
2,35
5,563
-0,78
0,599
31
38
5,38
5,3
0,63
0,4
5,19
2,08
4,87
7,07
(A)
(B)
pH
11
11
55
Created by XMLmind XSL-FO Converter.
9. SOIL CHARACTERIZATION
AT LARGE SCALE
plasticityi
ndex
27,6
27
1,85
3,45
2,22
1,56
26
32
capillary
rise
31,6
31,5
4,37
19,11
0,646
0,24
24
40
pH
5,37
5,3
0,57
0,32
2,64
1,42
4,63
7,07
plasticityi
ndex
27,6
27
1,52
2,33
1,92
1,29
26
32
capillary
rise
30,2
30,5
4,51
20,41
0,70
-0,22
20
40
pH
5,36
5,3
0,49
0,24
3,26
1,48
4,63
7,07
plasticityi
ndex
27,4
27
1,36
1,87
2,8
1,41
26
32
capillary
rise
30,7
31
4,11
16,94
0,76
-0,38
20
40
(B)
(B)
22
33
Upper table shows descriptive statistics of the soil parameters. The variability of variance for all sampling
approaches
changed
very
slightly
thevariabilityinstandarddeviationresultingfrom
thesmall-scale
heterogeneity.Asthesamplesizewas increased,thedecreasein degreeof variance and standard deviation were
directly proportional. The raw data follow neither normal nor lognormal distribution confirmed by
Kolmogorov–Smirnov test. The most heavily skewed parameter was calculated for pH for random data set with
sample number 22, having a skewness of 2.08 and kurtosis of 5.19; while, the capillary rise for average
sampling showed the least difference fromnormal distribution for skewness.
In accordancewiththeinformationdemonstratedby thedescriptive statistics, box-and-whisker plotdescribing
median,lowerand
upperoutliers,andlowerandupperquartiles.The
averagesamplingstrategyresultsindatashowingalmost lognormal distribution hiding hot spots at the site, while in
case
of
point
samples,extremevaluescanalsobe
detected.Forthelatterstrategy,boxandwhiskerplotalsoconfirms,thattheoptimalsamplingnumberis11,
sincetakingintoconsiderationeven
22points,thevariancedoesnotshow
significantdifferenceforanyinvestigatedparameter,thus,variability
for
attributes can be well-estimated in case of a representative data set N=11.
9.1. ábra - 26. Figure box-and-whisker plots of plasticity, pH and capillary rise of water
56
Created by XMLmind XSL-FO Converter.
9. SOIL CHARACTERIZATION
AT LARGE SCALE
9.2. ábra - 27. Figure Spatial distribution of capillary rise for the grid based random
point sampling method
9.3. ábra - 28. Figure Spatial distribution of pH for the grid based random point
sampling method
9.4. ábra - 29. Figure Spatial distribution of plasticity index for the grid based random
point sampling method
57
Created by XMLmind XSL-FO Converter.
9. SOIL CHARACTERIZATION
AT LARGE SCALE
Distribution
mapsalsodemonstrate
thatincreasingsamplesizeresults
in
increasinglydetailedinformationontheinvestigatedparameters,inspace, however, even for 11 measured data,
informative and representative maps can be generated revealing the main spatial trends.
Sincereflectanceisaproperty,whichis
derivedfrom
theinherentspectral
behaviourof
eachsoilcomponent,whenhyperspectralremotelysensed reflectancedataareanalysedandtheirrelationtosoilcondition
is investigated, itisusefultoidentifytheparametertowhichthetechniqueis mostlysensitive.
Consideringthree
parameters,onlyrelationshipbetween
plasticityindexandreflectancemeasuredwasstatisticallyproved,though
thedetermination
coefficientislow,only0.30,eveninthiscase.Forthe others, significant relation for any bands could not be proved.
Iterative series generated by regression calculation correlated best with band B116 of
wavelength908.78nm,inthenear-infraredrange(Fig.7). Subsequently, Principal Component Analysis (PCA) also
confirmed very weak correlation for the other soil physical properties and individual bands.
9.5. ábra - 30. Figure Hyperspectral reflectance data most sensitive to Arany plasticity
index
However, more detailed data set being more variable would provide better results even for capillary rise, as
reflectance is expected to be in relation to soil moisture. Considering pH, sensitivity of reflectance has not been
studied, yet, though sensitivity to calcite mostly abundant in soils having neutral pH may be informative, as
well. Cohen et al. (2005) also suggested that NIR spectroscopy could be used as a rapid analytical tool for soil
quality assessment and soil management. Furthermore, low costs of sample evaluation would allow high spatial
and temporal resolution for routine monitoring across large areas, which may greatly reduce management
uncertainty.
58
Created by XMLmind XSL-FO Converter.
9. SOIL CHARACTERIZATION
AT LARGE SCALE
In addition), the maps from kriged estimates showed that combination of geostatistical techniques and digital
data from aerial photograph were accurate enough to improve the identification of soil management zones,
which is the first step for site- specific management. However, they examined several soil parameters in relation
to the methods simple linear regression and regression equations for ordinary kriging estimates, e.g. pH, organic
matter, sand, clay, silt, phosphorus, potassium and found poor statistically proved relations to the reflectance
data. The ordinary kriging, exhibited the highest R2 since it does not take into account the secondary
information and only uses the primary soil variable, and compare the simple linear regression, where higher
predictions errors were obtained.
Among the investigated samplings strategies, average and random grid- based, approaches that use to
geostatistical techniques such as kriging and box-and-whisker plot showed smoothing effect on hot spots at the
classic average sampling strategy.
Considering representativity, N=11 samples were proved statistically representative for the study area the spatial
distribution maps also demonstrate that increasing sample size results in increasingly detailed information on the
investigated parameters, in space, however, even for 11 measured data, informative and representative maps can
be generated revealing the main spatial trends. Though several results suggest the applicability of hyperspectral
remote sensing in spatial analysis of soil condition, in this case, where variability was small, relationship
between any sensitive band and any considered soil physical or chemical data could not be proved.
1. Test questions:
1. How impact sampling density the uncertainty level of spatial map and cost/benefit ratio?
2. Do you agree that higher sampling number in every case result more information, if not how define the
optimal interval?
2. REFERENCES
[1]Birkás M., Dexter, A. R., Kalmár T., Bottlik L. 2006. Soil qualiy-soil condition- production stability, Cereal
Res Communications, 34, 1, 135-138.
[2]Bocchi, S., Castrignan`o, A., Fornaro, F., Maggiore, T. 2000. Application of factorial kriging for mapping
soil variation at fieldscale. Eur. J. Agron. 13, 295–308.
[3]Chang , D. H., Islam, S., 2000. Estimation Of Soil Physical Properties Using Remote Sensing And Artificial
Neural Network. Remote Sens. Environ. 74, 534–544.
[4]Cohen, M.J., Prenger, J.P., DeBusk, W.F. 2005. Visible-Near infrared reflectance spectroscopy for rapid,
non-destructive assessment of wetland soil quality. Journal of Environmental Quality 34, 1422–1434.
[5]Corwina, T.D.L., Lescha, S.M., Osterb, J.D., Kaffkac, S.R. 2006. Monitoring management-induced spatio–
temporal changes in soil quality through soil sampling directed by apparent electrical conductivity. Geoderma
131, 369-387.
[6]Filep Gy. 1995. Talajtani vizsgálatok. Egyemi segédlet (In Hungarian), 1-156.
[7]Füleky Gy., Vicze M., 2007. Soil and archaeological evidences of the periods of the tell development of
Százhalombatta-Földvár. Atti Soc. tosc. Sci. nat., Mem., Serie A, 112.
[8]Gomez, C., Rossel, V. R. A., McBratney A.B. 2008. Soil organic carbon prediction by hyperspectral remote
sensing and field vis-NIR spectroscopy: An Australian case study.Geoderma 146, 403–411.
59
Created by XMLmind XSL-FO Converter.
10. fejezet - 10. ADVANCED
IRRIGATION WATER SCHEMA
DESIGN
The climatic extremities are highly influence the productivity of agriculture. Sometimes water surplus causes
problems, and decreases the yield, and, on the other hand, drought is at least such dangerous, when plants cannot
get enough water. In similar cases, these problems appear in the same year, sometimes in the same area in
Central-Europe [12]. One of the basic conditions of plant production is water, so it is indispensable the deep
knowledge of water management of the given production site.
GPS based precision water management technologies are still not spread widely because of the data shortage
[6]. The major problem of these technologies is that they not have any information about timely dynamic data
exchange. The recent models can not eliminate this problem, so they often calculate with high error. Using
satellite or airborne images, we can get that information that supply with the absent data. By the usage of these
data, origins from the above mentioned sources, we can give more accurate approach for the appropriate water
management practice.
The Global Positioning System (GPS) - Remote Sensing (RS) - Water Management Model (WMM) model
combined with CROPWAT irrigation model can be a suitable basic for precision irrigation for a given site,
completed with other data that characterize the production site. This water management strategy can
significantly contribute to the sustainable water management, because it services enough information for
farmers about their plots. In this study, the authors present a solution for getting timely dynamic data from
satellite images, which is being build into an integrated GPS-RS-WMM model, to contribute to the prevention
and elimination of such problems that arise from the extreme water regime conditions, and create the possibility
of the development of the sustainable water management practice and precision irrigation.
In this study a 19 hectare sugar beet field was investigated in the Hajdúsági loess ridge, Northeast-Hungary
(Figure 31). The soil type on experimental site is chernozem.
10.1. ábra - 31. Figure Location of the experimental site
This study focuses on the water management improvement of this area, altogether with the observation of
climatic extremities. In the first part of the research, we collected data about the plant agro-technology.
10.2. ábra - 32. Figure Applied sensors to evaluate soil-plant environment
60
Created by XMLmind XSL-FO Converter.
10. ADVANCED IRRIGATION
WATER SCHEMA DESIGN
Leaf area of sugar beets was measured by ADC AM 100 typed leaf are scanner, continuously tracking during
the whole phenological phase, to get information about the transpiration surface, which is important to the
determination of the amount of daily transpiration. In line with this, the water demand in the area was mapped,
so we sampled the GPS registered points in every 0,2 m layers till 1,5 m depth to get acquainted with the actual
water content of the soil.
10.3. ábra - 33. Figure Leaf Area scanner and field GPS
61
Created by XMLmind XSL-FO Converter.
10. ADVANCED IRRIGATION
WATER SCHEMA DESIGN
Near the surface, we applied TDR method [8], and those samples, which were collected from the deeper layers,
were analyzed in laboratory.
10.4. ábra - 34. Figure Time domain reflectrometry (TDR) a sensitive water content (vol
%)
Sample points were selected by geostatistical evaluation, which was based on 1:10000 scale digital elevation
TIN model and digital soil maps [5, 2]. All collected data was prepared in ArcGIS environment. The major
parameters of the experimental field are shown in Table 1.
10.1. táblázat - 9. Table Typical soil parameters of the sample plot
Soil horizon
pH
Flexibility index
Cum. salt%
AL soluble K2O AL soluble P2O5
ppm
ppm
Aa
6,7-7,3
38-39
0,03-0,04
202-214
A1
7,6-8,3
39-41
0,04-0,05
B
7,8-8,4
37-39
0,03-0,04
C1
8,4-8,8
36-40
0,05-0,06
C2
8,6-8,9
33-36
0,06-0,07
27-29
9 Landsat images were possessed into database (Image source: Institute of Geodesy, Cartography and Remote
Sensing, 2005), which cover the research area and the full phenological period. Landsat products are distributed
via FTP electronic transfer only. The Landsat 7 satellite is part of NASA's Earth Observing System (EOS). The
EOS Data and Information System component (EOSDIS) provides a structure for data management and user
services for products derived from launched EOS satellite instruments, future missions and relevant NASA
Earth science data for the foreseeable future. Within the EOSDIS framework, the Distributed Active Archive
Centres (DAACs) are responsible for providing data and information services to support the customer
community. These centres are responsible for data archival, product development and distribution along with
user support. They are distinguished from one another by their data subject area. The LP DAAC is responsible
for land processes data of which Landsat 7 is a part. The DAACs are ulinked by the Warehouse Inventory
Search Tool (WIST) web portal which allows users to submit cross-discipline data (e.g. MODIS, ASTER)
queries using spatial and temporal criteria, examine search results for relevancy using built-in tools, and submit
orders via the EOS ClearingHOuse (ECHO) to the appropriate data providers. Landsat archive searching and
62
Created by XMLmind XSL-FO Converter.
10. ADVANCED IRRIGATION
WATER SCHEMA DESIGN
downloading is now performed through the EarthExplorer and Global Visualization Viewer (GloVis) - two web
portals developed by the EROS to replace the dated Global Land Information System (GLIS). EarthExplorer
allows searches from Macs, PCs and Unix computers across multiple USGS maintained data sets. These data
sets include Landsat 7, Landsats 1 - 5, AVHRR and aerial photography. GloVis simplifies the scene selection
process via efficient retrospective examination of all acquisitions for a given WRS location. Two different
product generation systems are used by EDC; the NASA-built Level 1 Product Generation System (LPGS) and
the EDC-procured National Landsat Archive Production System (NLAPS). An EarthExplorer search is
performed by identifying the Landsat holding and then specifying a location via world map or place/address
name. Additional criteria may include acceptable cloud cover, date range, and data type. A search commences
and a results page is presented that lists all scenes meeting the search criteria. A set of ulinks appears with each
scene that allows one to examine the browse, download the scene, examine the scene's Earth footprint, and view
the scene's metadata. Of particular metadata interest to users of ETM+ data are the cloud cover, data quality
scores and the gain states for the individual bands, which are needed to convert the scaled digital numbers to
radiance units. The GloVis portal provides a rapid way of examining the entire acquisition history for a specific
WRS location. Once a collection type is selected from a pull-down menu, a user either enters a longitudelatitude coordinate or clicks on a world map to zero-in on the desired land area. A 3 by 3 Landsat browse grid
appears with compass keys that allow scene shifting navigation to the WRS of interest. A pull-down Map Layer
menu allows for the overlay of land features such as major cities, rivers, roads, railways, and country
boundaries. Search limits (e.g. cloud cover, date range) can be set using the Tools pull-down menu. Once a
scene is selected and added to the order box it can be downloaded or ordered. A downloadable message is
splashed on the browse if the scene is currently online. Otherwise, an on-demand order must be submitted. The
Landsat 7 Science Data User's Handbook is a living document prepared by the Landsat Project Science Office at
NASA's Goddard Space Flight Centre in Greenbelt, Maryland (NASA, 2010).
The time-series were calculated from these images. In our research, Normalized Difference Vegetation Index
(NDVI) was also calculated from the Landsat images. NDVI can be used as an indicator of relative biomass and
greenness [1, 3]. It is an index that provides a standardized method of comparing vegetation greenness between
satellite images. The formula to calculate NDVI is: NDVI = (near infrared band - red band) / (near infrared band
+ red band). The greener colours indicate higher biomass and browner colours the bare soil on next figure.
10.5. ábra - 35. Figure NDVI Time series
63
Created by XMLmind XSL-FO Converter.
10. ADVANCED IRRIGATION
WATER SCHEMA DESIGN
CROPWAT is a decision support system developed by the Land and Water Development Division of FAO [9].
Its main functions are to calculate reference evapotranspiration, crop water requirements and crop irrigation
requirements; to develop irrigation schedules under various management conditions and scheme water supply;
to evaluate rainfed production and drought effects and efficiency of irrigation practices. The results of
Hungarian CROPWAT model testing was published by Tamás & Nagy (1996), and more soil physical
comparative model evaluation were made by Várallyay & Rajkai (1989)[10, 11]. In this software, reference crop
evapotranspiration (ETo) can be calculated by the utilization of Penman-Monteith equation. For the equation the
following variables are required: minimum-maximum temperature, air humidity, wind speed and daily sunshine.
The data used were collected from the weather station of Debrecen (Lat. 45°N, 21°E).
Using the calculation of water balance regime, evapotranspiration can be determined indirectly with the
following equation (FAO series No. 56.):
Dr, i = Dr, i-1 - (P - RO)i - Ii - CRi + ETc, i + DPi
where
Dr, i root zone depletion at the end of day i [mm],
Dr, i-1 water content in the root zone at the end of the previous day, i-1 [mm],
Pi precipitation on day i [mm],
ROi runoff from the soil surface on day i [mm],
Ii net irrigation depth on day i that infiltrates the soil [mm],
CRi capillary rise from the groundwater table on day i [mm],
ETc, i crop evapotranspiration on day i [mm],
DPi water loss out of the root zone by deep percolation on day i [mm].
The crop evapotranspiration, ET c, is calculated by multiplying the reference crop evapotranspiration, ET o, by a
crop coefficient, Kc:
ETc = Kc*ETo
where
ETc crop evapotranspiration [mm d-1],
Kc crop coefficient [dimensionless],
ETo reference crop evapotranspiration [mm d-1].
Crop coefficient is computed for four different crop growing stages, which are the following: initial stage, crop
development stage, mid-season, late season.
Determination of Kc is very difficult because of the integrated results of the several influential factors (as
illustrated by Figure 36). CROPWAT model estimates Kc for a given place with large error by 3 point. The
value of the crop coefficient is between 0,3-1,2 depending on crop varieties, so its average estimation error can
reach the 200-300 %. Since the value of the actual crop water requirement based on model sensitivity analyses
depends on the actual value of Kc significantly, the error propagation influences the reliability of the whole
model. One of the aims of our searches is the rise of the accuracy of this parameter using remote sensing data
source.
10.6. ábra - 36. Figure Major influential factors in determining crop coefficient (Kc)
64
Created by XMLmind XSL-FO Converter.
10. ADVANCED IRRIGATION
WATER SCHEMA DESIGN
Considering runoff direction, the amount of effective rain was calculated by D8 runoff algorithm (published by
Moore et al. (1993)) from the digital elevation model [7]. Statistical and image analyses were performed by
different software packages. We carried out principal components analysis by SPSS 12., and we made spatial
statistic evaluation by clustering by IDRISI Kilimanjaro and ENVI 4.2.
Time series were prepared from the Landsat images, from which principal components analysis was run off by
the usage of IDRISI GIS software.
Principal Components Analysis (PCA) is also known as Empirical Orthogonal Function (EOF) Analysis. It is a
very powerful technique for the analysis of variability over space and time. The images in a time series highly
correlate with one another from one phenological moment of time to the next phase. PCA transforms the series
into a set of image components that are orthogonal (i.e., independent of each other) in both time and space. They
are also ordered in terms of the amount of variance that they explain from the series. In theory, one can produce
as many components as there are images in the original series. However, in practice, almost all the variance can
be explained by only a small number of components, with the remainder expressing noise and high frequency
variations. The easiest way to understand PCA is to think about the time series of values for a single pixel across
time as a vector. If you imagine that each date represents a dimension then the series can be completely
described by a single point in that space. The image is made up of many pixels, so we will in fact have a space
occupied by many vectors. The correlation between any pair of vectors is inversely proportional to the angle
between them (in fact, the cosine of that angle is equal to the correlation coefficient). The first component is the
average vector (i.e., a vector that is as close as possible to the entire collection of vectors). It is known as an
eigenvector (meaning characteristic vector) and its length is know as the eigenvalue, which expresses the
amount of variance it explains. The cosine of the angle between this eigenvector and each pixel vector indicates
its loading on the component – i.e., the pixel vector’s correlation with the eigenvector. After the first component
has been calculated, its effects are removed from the pixel vector field. The new vectors thus express the
residuals after removing the effects of the first component. Then the process is repeated to extract the second
each other. Ultimately it is possible to extract as many components as there are pixel vectors. Note that if one
were to calculate PCA this way, the components would be graphs (an expression of the vector recast back onto a
time dimension) and the loadings would be images. Efficient computation of PCA is actually done with matrix
algebra and starts with a matrix of intercorrelations. While the correlations could be between the pixels as
illustrated here, it is in fact more efficient with geographical data to start with the correlations between the
images. PCA has many uses, but in image time series analysis, it is primarily an exploration tool. It is
remarkably effective in organizing the underlying sources of variability in the data. However, components aren’t
always pure. If at any level in the analysis there are two or more sources of variability that have roughly equal
weight, then PCA will tend to produce mixed components. The usual solution to this is known as rotation of the
axes.
65
Created by XMLmind XSL-FO Converter.
10. ADVANCED IRRIGATION
WATER SCHEMA DESIGN
Principal components analysis showed the variances for the test area. As a result of this analysis we found that
the first principal components image (Figure 37) is the most suitable for further analysis, so clustering was made
from this image and 5 parts of the sample plot was selected, which considered homogeneous.
10.7. ábra - 37. Figure Major influential time series component of PCA in determining
crop coefficient (Kc)
We created Region of Interest (ROI) that covered the 5 sites. A single pixel site has 9 average NDVI values,
which were calculated from the 9 time steps. Regression equation was fitted for these average values,
determining the parameters, and strength of the correlation. We could appreciate the NDVI values which were
estimated between 2 measuring dates with the help of the regression equation. We found strong correlation
between NDVI and Leaf Area Index (LAI), and LAI is in strong correlation with Kc. Daily values of the tertiary
equation had to be suited for the Kc values. Since K c=1,2 is equal with the maximum of the tertiary equation, we
have NDVI value enumerated to Kc value by proportional matching (Figure 38).
10.8. ábra - 38. Figure Relationship between crop coefficient (Kc)and normalized
difference vegetation index (NDVI) values and tertiary equations for the 5 selected sites
66
Created by XMLmind XSL-FO Converter.
10. ADVANCED IRRIGATION
WATER SCHEMA DESIGN
The model was fed with the actual water supply data using the soil-cartograms, and with the help of the field
measurements considering rooting depth which is increasing in time. To calculate sugar beet crop water
requirements was taken into account five homogenous plot size actual water regime based on rainfed water
condition. The amount of total rain was modified by the value of the runoff and we calculated the maximum rain
infiltration rate. Dry matter production and also the length of the individual growing stages were measured on
site. Crops factors (Kc) and a field response factor to estimate yield reduction due to drought stress (K y) were
determined, and rooting depth measured. Kc and Ky factors have to be given for each growing stages. We had the
ETo values for the whole sample plot, origin from an A-type evaporation pan of the weather station, but these
values were needed to be multiplied by a c constant factor to get the correct data, which are now suitable for the
characterization of free water surface evaporation. The modified ETo values are illustrated by Figure 39.
10.9. ábra - 39. Figure Daily reference crop evapotranspiration (ETo) values modified
by c constant factor and the average ETo values in the 19 ha experimental area
We examined the difference between ET o values estimated with the Penman-Monteith equation by CROPWAT
model and calculated ETo values based on the A-type pan evaporation measured on the weather station. We
found that estimated ETo values have exceeded calculated values with 7-12%. We also compared the difference
of crop water requirement (CWR) values between the 5 sites and the CWR values estimated by CROPWAT. We
determined that CROPWAT model is overestimates the CWR in the initial stage and in the mid-season, but it
correlates well in the crop development stage.
Since the value of the actual crop water requirement based on model sensitivity analyses depends on the actual
value of Kc significantly, the error propagation influences the reliability of the whole model. Calculated K c,
which is computed form NDVI calculated form Landsat TM time series images, showed more accurate value
67
Created by XMLmind XSL-FO Converter.
10. ADVANCED IRRIGATION
WATER SCHEMA DESIGN
compared to Kc, which origins from CROPWAT irrigation model. Using calculated Kc from NDVI, we can
eliminate the 200-300% estimation error, which is caused by CROPWAT calculation.
Strong correlation was found between NDVI and LAI, and LAI is in strong correlation with Kc. Analyses from
satellite images are appropriate for the calculation of reference crop evapotranspiration (ETo).
ETo values estimated with the Penman-Monteith equation by CROPWAT modelhave exceeded calculated ET o
values based on the A-type pan evaporation measured on the weather station with 7-12%. The crop water
requirement values estimated by CROPWAT model have also exceeded the calculated values that based on
measured data.By theresults it is suggested to use the measured data in practice.
GIS-RS-WMM integration is a good solution for the discovering of the water regime of an area and for the
achievement of the proper water management practice by the available information. Using this method, farmers
can realize a cost effective water management practice. Drip irrigation spread in the orchards is due to the 9095% of water use efficiency [27]. The drip irrigation is good delivery system because it allows doing the
agricultural procedures between the rows in the orchards and providing the water and nutrient used by
fertigation to reach the high density root zone. The wing lines running from the head line are set up fixed on the
stay system of on the soil in the tree rows and their emitters output 1-5 1/hour. The distances between emitters
on the hose are chosen in accordance with planting of the trees in the rows [21]. This type of irrigation has no
influence on the air humidity during hot days. The area of the test site is 2,74 ha, with 1660 planting density
tree/ha possible spaces of pear trees. The physical characteristic of the soil is sandy and the pear orchard is not
irrigated. The irrigation modelling was set by CROPWAT 8.0 based on the climatic, crop and soil data inputs of
the last 10 years. Special attention was taken on 2009, because of the long term drought. Climatic data and rain
fall data was obtained from the meteorological station next to the pear orchard. One of the key roles of irrigation
is the proper determination of evapotranspiration. Using the calculation of water balance regime
evapotranspiration can be determined indirectly with the following equation, based on FAO 56 [13]:
where
ETC j crop evapotranspiration on day i [mm],
Dr j root zone depletion at the end of day i [mm],
Dr j j water content in the root zone at the end of the previous day, i-1 [mm],
Pj precipitation on day i [mm],
ROj runoff from the soil surface on day i [mm],
Ij net irrigation depth on day i that infiltrates the soil [mm],
CRj capillary rise from the groundwater table on day i [mm],
DPj water loss out of the root zone by deep percolation on day i [mm].
Crop and fruit evapotranspiration can also be calculated from climatic data and by integrating directly the crop
or fruit resistance, albedo and air resistance factors in the Penman-Monteith approach, i.e., ET 0. As there is a
considerable lack of information for different crops or fruits the Penman-Monteith method is used for the
estimation of the standard reference crop or fruit to determine its evapotranspiration rate. From the original
Penman-Monteith equation and the equations of the aerodynamic and surface resistance, the FAO 56 PenmanMonteith method to estimate ET0 is expressed as
where:
ET0 reference evapotranspiration [mm day-1],
68
Created by XMLmind XSL-FO Converter.
10. ADVANCED IRRIGATION
WATER SCHEMA DESIGN
Rn net radiation at the crop surface [MJ m-2 day-1],
G soil heat flux density [MJ m-2 day-1],
T mean daily air temperature at 2 m height [°C],
u2 wind speed at 2 m height [m s-1],
es saturation vapour pressure [kPa],
ea actual vapour pressure [kPa],
es - ea saturation vapour pressure deficit [kPa],
D slope vapour pressure curve [kPa °C'1],
ypsychometric constant [kPa °C"1]
Differences in leaf anatomy, stomata characteristics, aerodynamic properties and even albedo cause the crop or
fruit evapotranspiration to differ from the reference crop or fruit evapotranspiration under the same climatic
conditions. Due to variations in the crop, and in fruits as well, characteristics throughout its growing season, Kc
for a given crop or fruit changes from sowing till harvest. In the crop coefficient approach the crop or fruit
evapotranspiration, ETC, is calculated by multiplying the reference crop or fruit evapotranspiration, ETQ, by a
crop coefficient, Kc:
ETC crop evapotranspiration [mm d-1],
Kc crop coefficient [dimensionless],
ET0 reference crop evapotranspiration [mm d -1].
Crop coefficient is computed for the following crop growing stages:
initial stage,
crop development stage,
mid-season,
late season.
The value of the crop coefficient is between 0.3-1.2 depending on crop varieties, so its average estimation error
can reach the 200-300%. Based on the FAO 56 and FAO 24 irrigation and water papers the Kc values of a pear
orchard without ground cover are between 0.4 in initial stage, 0.9 in mid- and 0.65 in late season [16, 17 13].
Girona et al. (2004) is measured 0.85 Kc of pears in lysimeters, although, in the case of an another lysimeter
study in Portugal, the mean Kc of pear (for the mid-season stage) was 0.5, below the tabled Kc for pear orchards
with no ground cover (0.95) or with active ground cover (1.2), confirming the need for adjusting published
values for local conditions [18,15]. In our studies, crop coefficients were set based on the recommendation of
Irrigation and Drainage FAO 56 paper, without ground cover crop (weed free cultivated).
Since the value of the actual crop water requirement based on model sensitivity analyses depends on the actual
value of Kc significantly, the error propagation influences the reliability of the whole model. One of the aims of
our further researches is the rise of the accuracy of this parameter using remote sensing data source.
Besides evapotranspiration, the amount of precipitation is also determined the amount or irrigation water.
However, the efficiency of the total precipitation is never 100%, therefore the effective rain should be
calculated. All rainfall options refer to the calculation of the effective rainfall based on the actual rainfall data.
The options are:
69
Created by XMLmind XSL-FO Converter.
10. ADVANCED IRRIGATION
WATER SCHEMA DESIGN
Fixed percentage: The fixed percentage is to be given by the user to account for the losses due to runoff and
deep percolation. Effective rainfall is a fixed percentage of actual rainfall.
10.10. ábra - 40. Figure Flowchart of CROPWAT FAO MODEL
Dependable rainfall (FAO/AGLW formula): Based on an analysis carried out for different arid and sub-humid
climates, an empirical formula was developed in the Water Service of FAO to estimate dependable rainfall, the
combined effect of dependable rainfall (80% probability of exceedance) and estimated losses due to Runoff
(RO) and Deep Percolation (DP). This formula may be used for design purposes where 80% probability of
exceedance is required. Calculation according to (monthly step):
Peff= 0.6 * P - 10 for Pmonth <= 70 mm;
Peff = 0.8 * P - 24 for Pmonth > 70 mm
Empirical formula: Same formula as for Dependable rainfall but with the possibility to change the parameters,
which may be determined from an analysis of local climatic records (monthly step):
Peff = a * Pmonth - b for Pmonth<= Z mm
Peff = C * Pmonth - d for Pmonth> Z mm
values for a, b, c, d and z arc correlation coefficients.
USDA Soil Conservation Service: Formula developed by USCS, where effective rainfall can be calculated
according to (monthly step):
Peff=Pmonth*(125-0.2*Pmonth)/125 for Pmonth<=250mm
Peff = 125 + 0.1 * Pmonth for Pmonth> 250 mm
70
Created by XMLmind XSL-FO Converter.
10. ADVANCED IRRIGATION
WATER SCHEMA DESIGN
Rooting depth of the pear trees were 1,2 m, the crop height was 3 m. Soil data was derived from field
measurements, water depletion factor was set to 0,45 for pear orchard, based on the recommendation of
Irrigation and Drainage FAO 56 paper.
The hydrology and climatic parameters were different from the mean values in 2009. The amount of the
precipitation was less than the average, but its distribution was differed in time from the average values, which
was very disadvantageous for pomaceous orchards. At the beginning of 2009 the precipitation had usual,
adequate amount, similarly to a regular precipitation status at the end of winter.
However, there was a severe drought in April and May. This deficiency was partly compensated in June, but
there were also less precipitation than average causing severe drought in July and August.
10.11. ábra - 41. Figure The amount of effective rain with different methods
There are several factors which influence unfavourably the efficiency of the precipitation, such as the most
important runoff, deep percolation, lateral flow, evaporation. These related factors can be characterized with
different dynamic both in spatially and in time, therefore several approximate methods are used to calculate the
effective rain. In Újfehértó, the amount of the precipitation and effective rain were the followings in 2009 year:
472.9 mm total rain; 413.5 mm effective rain calculated with USDA method; 213.5 mm calculated with FAO
method; 336.1 mm calculated with Empirical method.
Since the examination site is flat and there were no measurements for interception, deep percolation and runoff,
the FAO/AGLW formula was designed for arid climate, empirical method is for Mediterranean climate and
USDA method is suitable for semi- humid climate, USDA method was used to calculate the amount of the
effective rain in this research.The determination of evapotranspiration is one of the most uncertain out of all
agro hydrological factors.
There are several method is known to calculate the potential evapotranspiration. In this research PenmanMonteith method was used (Figure 42.).
10.12. ábra - 42. Figure ET0 changes in 2009
71
Created by XMLmind XSL-FO Converter.
10. ADVANCED IRRIGATION
WATER SCHEMA DESIGN
The relative humidity was below the critical 65% for 4 months, causing areal drought for pear orchard. Due to
the low water capacity of sandy soil the water content decreased far below field water capacity, which often
caused air drought between 11-15 o’clock in April, May, July and August, because of the low relative humidity
(Figure 43).
10.13. ábra - 43. Figure . Air drought in 4 months in 2009
The irrigation water demands of both summer and autumn ripening pear varieties were calculated on sandy soil
in Újfehértó. Based on the results of 2009, the CROPWAT model obviously shows, that a large amount of water
should be irrigated, especially in summer in the case of both variety (Figure 45., 6).
10.14. ábra - 44. Figure Irrigation water demand of autumn ripening pear varieties on
sand
72
Created by XMLmind XSL-FO Converter.
10. ADVANCED IRRIGATION
WATER SCHEMA DESIGN
10.15. ábra - 45. Figure Irrigation water demand of summer ripening pear varieties on
sandy soil
Since autumn ripening pear varieties have greater water demand, the irrigation schedule was determined for this
variety on sandy soil. Based on the irrigation schedules of the last 10 years the mean amount of the total gross
irrigation is between 230-270 mm, within 3 irrigation interval (Table /), although in 2009, due to heavy drought,
the total gross irrigation was 355.4 mm/year on sandy soil calculating with 45% total available water depletion
in 5 irrigation interval (Table2.).
10.2. táblázat - 10. Table Irrigation schedule of pear orchard on sandy soil in 2002
Date
Day
Stage
DepI
Net Irr
Gr. Irr
Flow
Tree How
%
Mm
mm
1/s/ha
1/tree/h
1 Jul
78
Dev
47
69.6
73.26316
0.108712
1.29
26 Jul
103
Mid
46
68.6
72.21053
0.334308
3.97
2 Sep
141
End
45
68
71.57895
0.218016
2.59
6 Oct
End
End
6
The Flow represents the continuous water discharge needed to satisfy pear orchard irrigation requirements over
the irrigation interval period. It is expressed in litre per second per hectare and calculated converting the Gross
irrigation depth into a permanent supply. Gross irrigation represents the water depth (expressed in mm) applied
to the field. Since the Irrigation efficiency is usually lower than 100%, only a fraction of the Gross irrigation
73
Created by XMLmind XSL-FO Converter.
10. ADVANCED IRRIGATION
WATER SCHEMA DESIGN
depth, that is, the Net irrigation depth, effectively reaches crop root zone. Although the drip irrigation can
provide minimal water lost, its efficiency often exceeds 90-95%. Therefore the calculated total gross irrigation
value was calculated with 95% efficiency [27]. Tree flow is expressed in litre per hour per hectare for 12 hours’
irrigation period a day. Area demand of a pear tree was 16 m 2 based on the row spacing (8m) and plant-to-plant
distance (2m).
10.3. táblázat - 11. Table Irrigation schedule of pear orchard on sandy soil in 2009
Date
Day
Stage
Depl
Net Irr
Gr. Irr
Flow
Tree
flow
flfloflow
%
Mm
mm
1/s/ha
l/tree/h
26 May
42
Dev
45
61.1
64.3
0.18
2.11
12 Jul
89
Mid
47
70.3
74.0
0.18
2.16
28 Jul
105
Mid
46
69.6
73.3
0.53
6.29
17. Aug
125
Mid
46
69.3
72.9
0.42
5.02
11 Sep
150
End
45
67.3
70.8
0.33
3.89
6 Oct
End
End
30
In 2002 and 2009 the amount of available water was moderate and very low. Considering 45% depletion, the
water can only be supplied by irrigation to prevent pear orchard from yield deficiency. The water should be
utilized in 3 (2002) or 5 (2009) irrigation periods in the dates given in Table 1 and table 2. The sizing of the
irrigation system was set to the maximum 0.551/s/ha, which is 6.31/tree/h based on the irrigation schedule of
2009, as the most drought year of the last ten years. This amount should be utilized by drip emitters or
microsprinklers. 6.31/tree/h can be carried out with a drip emitter having 16 mm wing lines diameter, 4 1/h
water flow at 3 atm pressure. The distance between the drip emitters is 2 m, in accordance with the planting of
trees. The 9 wing lines should be set up fix on the soil surface in each row with 300 m length. To support the
adequate air humidity for the pear orchard, 2 1/tree/h can be irrigated with microsprinklers, for example with a
jet pulse emitters, set up below crown cloud. The increased air humidity can provide more effective yield safe,
and decrease the evapotranspiration as well, to some extent. End of process a special GIS model was the
interface to change and integrate different data source and optimise water distribution in space and time.
10.16. ábra - 46. Figure Data integration of precision water management
74
Created by XMLmind XSL-FO Converter.
10. ADVANCED IRRIGATION
WATER SCHEMA DESIGN
The presented real time field data acquisition and integration with remote sensing spectral data is a very
effective tool to increase resilience to weather extremities.
1. Test questions:
1. How can be used Time Series Analysis to isolate homogeneous patches of agricultural blocks?
2. How do you calculate evapotranspiration?
3. Can you characterize CROPWAT model?
2. References
1. Boone, R. B., K. A. Galvin, et al. 2000. Generalizing El Nino effects upon Maasai livestock using
hierarchical clusters of vegetation patterns. Photogrammetric Engineering & Remote Sensing 66(6). pp. 737744.
2. Cressie N. A.C. 1993 Statistics for spatial data. John Wiley & Sons, Inc., 900 p.
3. Chen, D., W. Brutsaert 1998. Satellite-sensed distribution and spatial patterns of vegetation parameters over a
tallgrass prairie. Journal of the Atmospheric Sciences 55(7): pp. 1225-1238.
4. FAO series No. 56., http://www.fao.org/docrep/X0490E/X0490E00.htm
5. Isaaks E. H., Srivastava R. M., 1989. Applied geostatistics. Oxford University Press, Inc., New York, pp 561.
6. Marques da Silva J. R., Alexandre C., 2003. The spatial variability of irrigated corn yield in relation to field
topography. In J. Stafford, A. Werner (eds.): Precision agriculture. Wageningen Academic Publishers, The
Netherlands. pp. 385-391.
7. Moore I., Grayson R., Ladson A., 1993. Digital terrain modelling: a review of hydrological,
geomorphological and biological applications. In K. Beven, I. Moore (eds). Terrain analysis and distributed
modelling in hydrology. Jonh Wiley & Sons, Chichester, pp. 7-30.
8. Rajkai K., Rydén B. E., 1992. Measuring areal soil moisture distribution by the TDR method. Geoderma, 52.
pp. 73-85.
75
Created by XMLmind XSL-FO Converter.
10. ADVANCED IRRIGATION
WATER SCHEMA DESIGN
9. Smith M., 1992. CROPWAT a computer program for irrigation planning and management. FAO series 46.
Rome, pp. 1-51.
10.
Tamás J., Nagy J., 1996. Evaluation of irrigation schedules by CROPWAT FAO model for maize
species in Hungary. In Book of Abstracts, 4th ASA-congress. Veldhoven - Wageningen, The Netherlands,
Vol. 1. pp. 122-124.
11.
Várallyay GY., Rajkai K., 1989. Model for the estimation of water (and solute) transport from the
groundwater to overlying soil horizons. Agrokémia és Talajtan, 38. pp. 641-656.
12.
Várallyay GY., 2005. Klímaváltozások lehetséges talajtani hatásai a Kisalföldön. “Agro-21” Füzetek,
Klímaváltozás – hatások – válaszok. 43. pp. 11-23.
13.
Allen, R.G., Pereira, L.S., Raes, D. & Smith, M. (1998): Crop évapotranspiration: Guidelines for
computing crop requirements. Irrigation and Drainage Paper No. 56, FAO, Rome, Italy, 300.
14.
Bosnjak, D., Gvozdenopvic, D. & Moldovan, S. (1997): Water requirements of pear William grown on
loamy sand soil. Acta Hort. 449:133-138
15.
:Conceiçâo, N., Paço, T.A., Silva, A.L. & Ferreira, M.I. (2008)Crop coefficients for a pear orchard
(Pyrus communis L.) obtained using Eddy covariance. Acta Hort. 792: 187-192.
16.
Doorenbos, J. & Kassam, A.H. (1979): Yield response to water. Irrigation and Drainage- Paper n. 33.
FAO, Rome, Italy, 193 pp.
17.
Doorenbos, J. & Pruitt, W.O. (1977): Guidelines for Predicting Crop Water Requirements. Food and
Agriculture Organisation of the United Nations. FAO Irrigation and Drainage Paper, 24: 143.
18.
Girona, J., Marsai, J., Mata, M. &del Campo, J. (2004): Pear crop coefficients obtained in a large
weighing lysimeter. Acta Hort. 664: 277-281.
19.
Hrotkó K. (1999): A gyümölcsfajták alanyai. 407-506. p. In: Hrotkó K. (Szerk.): Gyümölcsfaiskola.
Mezőgazda Kiadó, Buda¬pest.
20.
Nagy, A., Tamás, J., Fórián, T., Nyéki, J., Soltész, M., Szabó, Z. (2010) Irrigation modeling in a pear
orchard. . Int. J. Hort. Sci., 16(3): 75-79.
21.
73.
Nemeskéri, E. (2009): Irrigatioin of pear (A review) International Horticultural Science, 15 (1-2): 65-
22.
Pálfai, I. (Ed.) (2000): The role and significant of water in the Hungarian Plain. (In Hungarian)
Nagyalföldi Alapítvány. Békés¬csaba
23.
Papp, J. (2000): A körteültetvények vízgazdálkodása és öntözése. 249-255. (In: Göndör J-né (szerk.):
Körte) Mezőgazda Kiadó, Budapest
24.
Soltész, M., Szabó, T. (1998): Alma. 119-155. p. In: Soltész, M. (Szerk.): Gyümölcsfajta-ismeret és használat. Mezőgazdasági Kiadó, Budapest. 513 p.
25.
Somlyódy, L. (2000): Strategy of Hungarian water management (In Hungarian). MTA Vízgazdálkodási
Tudományos Kutatócsoportja, Budapest. 370. p.
26.
Takács, F. (2009): Almaalanyok értékelése két művelési rend¬szerben a nyírségi termesztő körzetben.
PhD-thesis Corvinus Egyetem, Budapest
27.
Tóth, A. (1995): Az esőszerű és mikroöntözés gyakorlata. KITE RT. Nádudvar.
28.
Várallyay Gy. (1989): Soil water problems in Hungary. Agrokémia és Talajtan, 38: 577-595.
29.
Várallyay Gy. (2002): The role of soil and soil management in drought mitigation . In: Proc. Int. Conf.
On Drought Mitigation and Prevention of Land Desertification, Bled, Slovenia, April 21-25 2002. ICIDCIIC. (CD)
76
Created by XMLmind XSL-FO Converter.
10. ADVANCED IRRIGATION
WATER SCHEMA DESIGN
30.
Várallyay Gy. (2005): Klímaváltozások lehetséges talajtani hatásai a Kisalföldön. ‘Agro-21” Füzetek,
43: 11-23.
77
Created by XMLmind XSL-FO Converter.
11. fejezet - 11. WATER STRESS
Apple production is important agro-economical sector in Europe. The fruit growing is very sensitive for drought
conditions. Already short water scarcity period can made irreversible damage in fruit quality. This is the reason
that permanently has to optimize the available water content in root zone. In Hungary about 100.000 hectares of
orchards can be found, from which apple is cultivated on one of the largest areas. The total area of apple and
pear orchards is more than 45000 ha. Apple orchards cover about 60% of the total pomiculture in Hungary,
although in the last period the production was reduced [2]. The production of marketable horticulture products is
difficult without quality horticulture practice, which in many cases is the primary condition of appropriate
management and irrigation systems
Hungary has favourable agro ecological potential for apple fruit production [30, 21], although intensive apple
orchards need irrigation to avoid plants from water stress and increase the apple yield security and quality. The
data of the Central Statistics Office show that 28% of the apple and pear orchards can be irrigated, but only 21%
is irrigated. Since horticulture is a water demanding sector, high quality fruit-production is difficult without
proper irrigation.Furthermore in some horticultural farms there is no irrigation applied, or its techniques is
improper. There are several experiments going on around the world to develop methods of irrigation, which
draw different technology combination for the water and energy saving micro-irrigation. One of the biggest
professional challenges of the following years is to develop the water resource management for apple and pear
trees. For this the water norm of the trees has to be identified in the different phenological stages, the irrigation
turns, and technology and the transpiration surface.
In orchard plant’s leaf area must determine not only in absolute terms, but compared with the growing area (T)
also. The rate of two values called leaf area index (LAI), which is leaf area (m2) per 1m2 soil surface. The LAI is
the most suitable index in the cultivation practice for the plant mass [17].
(m2/m2)
LAI – as biophysical status – is in a close relationship with the amount of biomass, with the photosynthesis and
transpiration scale [14]. The size and number of leaf determine the leaf area (LA). The number of leaf produced
is reduced by water stress, nutrient deficiencies and is under hormonal control. The leaf area index change
species, stage of development, methods of cultivation, density of crop. Wagenmakers (1989) found that leaf area
per tree of apples and pears decreased linearly with planting density [19].Verheij (1972) showed that with apple
the leaf area per tree declined with increasing planting density even with unpruned trees, and was accompanied
by a relative suppression of lateral growth in the lower parts of the trees [18].
The leaf area determines the area of transpiration. Transpiration is taking place through the stomas of the lower
epidermis [6]. The stoma is a pore, found in the leaf and stem epidermis that is used for gas exchange. The pore
is bordered by a pair of specialized parenchyma cells known as guard cells which are responsible for regulating
the size of the opening and are the closest thing a plant has to a muscle. [5] In the case of different apple
varieties, the number of stomas can be different: 200-450 pores/m2[10,16]. The apple varieties growing faster
the density of stomas is higher, while smaller varieties have less stoma density [3]. Cowart (1935) established
that less stoma can be found in the leaves of lower position. The extent of transpiration depends on several
physical factors as well as the opening or the closing of the stomas [23]. One of the most significant factors
influencing transpiration is the wind [13]. Beukes (1984) established a negative correlation between
transpiration and wind velocity [4]. The transpiration rate is also determined by the leaf temperature and the
available moisture content of the soil. Transpiration is increasing if more water is available in the soil: in case of
optimal water supply apple trees transpirated more water and the conductance of the stomas increased [4,10, 8].
During the day the water potential of apple and pear trees is decreasing with the increasing transpiration: It was
observed the lowest water potential in noontime. Landsberg et al. (1975) and Fernandez et al. (1997) examined
apple and pear trees under droughty and control conditions [12, 8]. They concluded that the closing of the
stomas is an effective physiological control that reduces the rate of transpiration.
In this study the transpiration properties and response to heat stress of Granny Smith and Jonagold apple species
were measured. The examinations were set at an orchard near to Debrecen without irrigation. The investigation
period was from 5th of July to 5th of September 2011, with 5 day pause, caused by technical problems. The
78
Created by XMLmind XSL-FO Converter.
11. WATER STRESS
chosen trees were two years old, without fruits, in 20l container, placed in the mixture of mulch matrix, wood
chips and sawdust and grown at the same climatic conditions. Since different rootstocks affect differently the
water management of apple trees, species with the same rootstock (MM106) were chosen. One of the most
important goals was to determine the leaf area of the trees which is a basic data for transpiration modelling.
Another aim was to measure the transpiration rate of the apple trees. The transpiration measurements were
carried out by sap flow meter. After determining the transpiration and transpiration surface of the examined
trees, it is possible to define the transpiration rate for one square meter. On 30th September 2011 the leaves were
collected from each tree. During the leaf sampling the canopy was divided into two parts. The leaves from the
lower branches were separately collected from the leaves of the upper branches.
Scanning of the leaves was carried out in the laboratory of the Institute of Water- and Environmental
Management of the Centre for Agricultural and Applied Economic Sciences of the University of Debrecen. The
leaf areas were measured by Area Meter 100 (AM 100) leaf area scanner developed by the Analytical
Development Company. After scanning, the data were stored in the memory of the AM 100, then they were
downloaded to a computer. Nevertheless the digital display of the device ensures a permanent monitoring of the
measurement data. By means of the scanner, not only the surface area of the leaves with different sizes and
colours, but also of the leaves damaged by pests could be determined. After each scanning the length and width
of the leaf, the average and total area and the number, the time and the date of the scanning were displayed and
stored.
The transpiration was measured by Sap flow meter based on Stem Heat Balance theory. Sensors were set on the
trunk of the apple trees. The Dynagage sap flow sensor consists of a flexible heater, a thermopile to measure
radial heat loss, and differential thermocouple pairs to measure the axial temperature differences q u-qd. All of
these sensors and heater are mounted on a cork substrate and housed inside white, reflective foam, thermal
insulating collar.
Once the sensor is installed on the stem surface, both the sensor and the stem sections above and below the
sensor are completely covered by a heat insulator to minimize thermal perturbations caused by the ambient
environment. Power is supplied continuously to the heater from a regulated DC power source. The Dynamax
loggers have a power down mode so that power is saved at night and the stem is preserved from overheating.
During the power down mode and at the transitions to power on, the sap flow is not computed to maintain the
accumulated flow accurately during this unbalanced transition. Figure 47 shows a stem section and the possible
components of heat flux, assuming no heat storage
11.1. ábra - 47. Figure Stem gage schematics (Source: Dynamax manual
The heater surrounds the stem under test and is powered by a DC supply with a fixed amount of heat, Qh. Qh is
the equivalent to the power input to the stem from the heater, Pin. Qr is the radial heat conducted through the
79
Created by XMLmind XSL-FO Converter.
11. WATER STRESS
gage to the ambient. Qv, the vertical, or axial heat conduction through the stem has two components, Qu and
Qd. By measuring Pin, Qu, Qd, and Qr, the remainder, Qf can be calculated. Qf is the heat convection carried by
the sap. After dividing by the specific heat of water and the sap temperature increase, the heat flux is converted
directly to mass flow rate. Flow rate is expressed as g moisture/hour.
The apple tree leaves collected at the end of the final phenological state were scanned and the total transpiration
and the average leaf areas were determined (Table 1.). Concerning Granny Smith species 144 leaves from lower
part of canopy and 244 from upper part were collected. In the case of Jonagold 136 leaves from lower part of
canopy and 183 from upper part were collected. Comparing the leaf size within and transpiration surfaces, it can
be stated, that in the case of both species the photosynthetically active canopy area was larger in the upper part
of the canopy. This was simply caused by the structure of tree for better utilization of the light. Leaf area was
also larger (significance level p<0.05) in the upper part of the canopy to achieve larger photosynthetic surface.
Naturally the transpiration is also higher in this region.
11.1. táblázat - 12. Table Canopy and leaf area of Jonagold and Granny Smith apple
species
Leaf area (cm2)
Granny Smith
Jonagold
Upper part of canopy
∑
5413.5
4723.7
1 leaf (mean)
22.19
25.81
Lower part of canopy
∑
2187.9
2542.3
1 leaf (mean)
15.19
18.69
Whole canopy
∑
7601.4
7265.9
1 leaf (mean)
19.59
22.77
Comparing the measured area parameters between species, there were some differences is found. Though only a
slight total canopy differences was found (4.44% differences between species) the average leaf area was
significantly (significance level p<0.05) larger in the case of Jonagold species. That means Jonagold species
with less leaf number reached almost the same canopy area, than Granny Smith, since there is 18% differences
between the species concerning number of the leaves, which probably due to species properties.
For the characterization of the water use and stress of the fruit trees, the transpiration of them was investigated,
concerning the climatic parameters, such as temperature and rainfall. In order to compare the transpiration
properties of apple trees, the measured sap flow (g/day) was adjusted to one square meter of canopy.
On the base of values of transpiration only a small difference can be found between species. Concerning the
whole transpiration period, there was an only 4874 g difference in transpiration, which is less than 5 l. Therefore
it can be declared, that significant difference in transpiration can not be found between the examined species.
11.2. ábra - 48. Figure Transpiration of 1 m2 canopy and the climatic conditions
80
Created by XMLmind XSL-FO Converter.
11. WATER STRESS
During the investigation time there were wet and cool as well as dry and extremely hot periods. In the case of
dry and extreme hot periods (9th – 13th of July and 24th – 29th August) the transpiration is decreased, due to
the lack of water in the soil and the huge water potential. The decreased transpiration means that stomata closure
occurred thus, according to several studies the photosynthesis is also blocked and reach its minimum [15]. Not
only the transpiration is blocked but also the CO2 uptake, therefore reactive hydroxyl radicals are produced
which are harmful for the chloroplasts and cell membranes, which eventually causes the water stress symptoms
in plants.
Concerning the wet periods there was sufficient water supply for the apple trees. While the temperature
increased the transpiration increased as well, furthermore, positive correlation (r=0.77) was found between the
maximum temperature and transpiration.
In the case of both species the photosynthetically active canopy area and leaf area was larger in the upper part of
the canopy for better utilization of the light. Leaves of Jonagold tree are significantly larger than Granny Smith
which probably due to species properties. In the case of dry and extreme hot periods the transpiration is
decreased, water stress occurred. Since there is a positive correlation was found between the maximum
temperature and transpiration properties Sap Flow meter can be suitable for not only transpiration monitoring,
but also for irrigation scheduling and water stress assessment.
Next study is focusing on canopy temperature as a good indicator of plant water status. The infrared
thermography is considered for the identification of plant water stress and is also used as a tool for irrigation
scheduling method [31]. If plant water stress increases, transpiration decreases and plant temperature may
exceed air temperature. On the other hand, non-stressed plants will have canopy temperatures less than air
temperature, particularly when vapour pressure deficit (VPD) is not greater than 4 kPa [28]. The crop water
stress index (CWSI) relates canopy–air temperature difference to net radiation, wind speed and vapour pressure
deficit [23]. However, a surrogate measure is calculable from the temperatures of the canopy and reference leaf
surfaces corresponding to fully transpiring and non-transpiring canopies [24,27]. Thus, by monitoring plant
canopy temperature and the temperatures of wet and dry leaves, it is possible to estimate the underlying plant
water stress status and therefore, intelligently control the related irrigation process.
The examinations were carried out at a micro-irrigated intensive apple orchard in Debrecen-Pallag, University
of Debrecen. The HEXIUM mobile infra red camera with microblometer sensor and 384x288 pixel resolution
was used for the water status detection of the apple trees. There is an IR camera included. With 50 frames/sec a
real-time imaging is possible. The images can be seen on the colour LCD. With the 16 bits/pixel image you get
an image that can be coloured later as you like with a post-processing algorithm. The camera sensitivity is 0,05
°C, and its measurement interval is between. -20 and 120 °C, but the optimal temperature is between -25 and 60
°C for proper working. A key procedure for the evaluation of crop water stress from plant canopy temperature
81
Created by XMLmind XSL-FO Converter.
11. WATER STRESS
was to calculate CWSI based on the data collected from IR thermography systems. The CWSI described by
(Jones et al., 2002) is of the following generic form [26]:
where Td and Tw represent the reference temperatures for dry (nontranspiring) and wet (fully transpiring) leaf
surfaces respectively. Tc is the temperature of the transpiring surface, i.e., the actual measured temperature of
all sunlit leaves to represent the sunlit portion of the canopy. Although alternative methods for estimating
reference temperatures may be found [20], reference leaves, which are artificially treated real leaves with known
conductance to water vapour, can be physically embedded in the scene [24] and so the reference temperatures T d
and Tw can be estimated from the leaf temperature distribution. In this study the reference temperatures T d and
Tw can be estimated from the leaf temperature and air distribution. Before the CWSI calculation masking was
made in order to eliminate the background ad keep only the apple trees on the infra red image. The image
processing was made based on Hunyadi et al (2010), in IR Player, Surfer9 and the CWSI analysis were carried
out in Idrisi Tajga software environment [22].
Thermography survey was made on 25th of August, 2011. Since heat warming was announced at that day by the
Hungarian Meteorological Services it was an appropriate opportunity for surveying the effect of extreme high
temperature on the apple tree canopy. The survey was made between 04:30 and 16:00, thermography images
were taken in every 30th minutes.
Once the temperatures Td, Tw and Tc are identified using the pixel-by-pixel temperature data associated with the
IR image.
11.3. ábra - 49. Figure Thermography image on apple trees (oC)
Parallel to the thermography investigation, the air temperature was also measured by analogous thermometers in
shade. The air temperature reached its minimum at dawn. Based on thermography image, the temperature
parameters of apple yield, leaf, trunks were investigated.
11.4. ábra - 50. Figure
82
Created by XMLmind XSL-FO Converter.
11. WATER STRESS
The result suggests that parallel to the air temperature, the temperature of the examined apple tree parameters
also increased. Furthermore, the leaf temperature exceeded the air temperature, which shows water deficiency
and the inadequate transpiration. Based on CWSI image, those parts of the canopy were easily eliminated, where
the increased water stress occurred. This due to water deficiency, caused by the lack of precipitation and
irrigation, stomata closure can occur thus, according to several studies [29] the photosynthesis is blocked and
reach its minimum. Not only the transpiration is blocked but also the CO2 uptake, therefore reactive hydroxyl
radicals are produced which are harmful for the chloroplasts and cell membranes, which eventually causes the
water stress symptoms in plants.
Based on CWSI image, it can be stated, that irrigation was urgently needed for the orchard. This result is also
confirmed by soil water monitoring survey, which was carried out by tensiometers. According these results,
irrigation should have been started on 10th of August. Therefore, based on the CWSI the irrigation was started
the following day.
In this study the leaf temperature exceeded the air temperature, which shows water deficiency and the
inadequate transpiration. Based on CWSI image, those parts of the canopy were easily eliminated, where the
increased water stress occurred. It can be stated, that irrigation was urgently needed for the orchard. This result
is also confirmed by soil water monitoring survey, which was carried out by tensiometers.
1. Test questions:
1. What is the operational theory of sap flow system?
2. What it means crop water stress index and how can be applied in irrigated orchard?
2. References
1. Allayne V.-Larsen F. E.-Higgins S. S. (1989): Water relations of container-grown, virus tested and common
apple (Malus domestica Borkh) rootstocks. Scientia Horticulturae38: 117-129.
2. Gonda, I. – Apáti, F. (2011): Almatermesztésünk helyzete és jövőbeni kilátásai. [The currently status and
future prospects of our apple production.] In: Tamás, J. (eds.): Almaültetvények vízkészletgazdálkodása.
[Water resource management of apple orchards.] Debreceni Egyetem, AGTC, Kutatási és Fejlesztési Intézet,
Kecskeméti Főiskola, Kertészeti Főiskolai Kar, Debrecen. 13-25.
3. Beakbane B., Majumdar P. K. (1975):A relationship between stomatal density and growth potential in apple
rootstocks, J. Hortic. Sci.50: 285–289.
4. Beukes D. J. (1984): Transpiration of apple trees as related to different meteorological, plant and soil factors.
J. Hortic. Sci.59: 151-159.
5. Boldizsár A. (2007): Evaporation and microclimate studies in the Balaton lake reeds. PhD thesis. Keszthely.
27.
6. Boyer J. S. (1985): Water transport. Annual Review of Plant Physiology. Vol. 36: 473-516
7. Cowart F. F. (1935): Apple leaf structure as related to position of leaf upon the shoot and to type of growth.
Proc. Am. Soc. Hortic. Sci.33: 145–148.
8. Fernandez R. T.-Perry R. L.-Flore J. A. (1997): Drought response of young apple tree rootstocks. II. Gas
exchange, chlorophyll fluorescence, water relations, and leaf abscisic acid. J. Amer. Soc. Hort. Sci.122: 841848.
83
Created by XMLmind XSL-FO Converter.
11. WATER STRESS
9. Goode H. G.-Higgs K. H. (1973): Water potential-water content relationships in apple leaves. J. Exp. Bot.30:
965.
10.
Gower S.T., Kucharik C. J., Normann J. M. (1999): Direct and indirect estimation of leaf area index,
fapar, and net primary production of terrestrial ecosystms. Remote Sensing of Environmental. 70: 29-51.
11.
Jackson J. E. (2003): Biology of apples and pears. Cambridge: Cambridge University.
12.
Landsberg J. J.-Beadle C. L.-Biscoe P. D.-Butler D. R.-Davidson B.-Incoll L. D.- James G. B.-Jarvis P.
G.-Martin P. J.-Neilson R. E.-Powell D. B. B.-Slack E. M..-Thorpe M.R.-Turner N. C.-Warrit B.-Watts W.
R. (1975): Diurnal energy, water and CO2 exchanges in an apple (Malus pumila) orchard. J. Applied Ecol.
12: 659-684.
13.
Mansfiled T. A.-McAinh M. R. (1995): Hormones as regulators of water balance. In: Plant Hormones.
(Ed.): Davies P. J. Dordrecht : Kluwer Academic Press. 598-616.
14.
Nemani R. R.-Running S. W. (1989): Testing a theoretical climate-soil-leaf area hydrologic
equilibrium of forests using satellite data and ecosystem simulation. Agricultural and Forest Meteorology
Volume 44, (3-4), 245-260.
15.
Pethő M. (1996): Mezőgazdasági növények élettana, Akadémia Kiadó, Budapest
16.
Slack E. M. (1974):Studies of stomatal distribution on the leaves of four apple varieties. Journal of
Horticultural Science 49: 95-103.
17.
Szász G. (1988): Agro meteorology, Mezőgazdasági Kiadó, Budapest, 462.
18.
Verheij E. W. M. (1972):Competition in apple, as influenced by Alar sprays, fruiting, pruning and tree
spacing. PhD thesis. Wageningen.
19.
Wagenmakers P. S. (1989): High-density planting system trial with pear.4. International Symposium
on Research and Development on Orchard and Plantation Systems, Dronten (Netherlands), 29 Aug - 2 Sep
1988
20.
Guilioni, L. – Jones, H.G. – Leinonen, I. – Lhomme, J.P.: 2008. One the relationships between stomatal
resistance and leaf temperatures in thermography. Journal of Agricultural and Forest Meteorology 148,
1908–1912.
21.
Hrotkó, K.: 1999. A gyümölcsfajták alanyai. 407-506. p. In: Hrotkó K. (Ed.): Gyümölcsfaiskola.
Mezőgazda Kiadó, Budapest. 550 p.
22.
Hunyadi, G. – Tamás, J. – Kosárkó, M.:2010. Coherences between the inside and surface temperature
in sewage sludge based compost prisms [In Proc: ECSM 2010. 2nd European Conference on Sludge
Management.] Budapest, Hungary.1-8.
23.
Jackson, R.D. – Idso, S.B. – Reginato, R.J. – Pinter, P.J.: 1981. Canopy temperature as a cropwaterstress indicator. Journal of Water Resources Research 17: 1133–1138.
24.
Jones, H.G.: 1999a. Use of infrared thermometry for estimation of stomatal conductance as a possible
aid to irrigation scheduling. Journal of Agricultural and Forest Meteorology 95: 139–149.
25.
Jones, H.G.: 1999b. Use of thermography for quantitative studies of spatial and temporal variation of
stomatal conductance over leaf surfaces. Journal of Plant Cell and Environment 22 (9): 1043–1055.
26.
Jones, H.G. – Stoll, M. – Santos, T. – de Sousa, C. – Chaves, M.M. – Grant, O.M.: 2002. Use of
infrared thermography for monitoring stomatal closure in the field: application to grapevine. Journal of
Experimental Botany 53 (378), 2249–2260.
27.
Moller, M. – Alchanatis, V. – Cohen, Y. – Meron, M. – Tsipris, J. – Osrovsky, V.: 2007. Use of
thermal and visible imagery for estimating crop water status of irrigated grapevine. J. Exp. Bot. 58: 827–838.
28.
Olivo, N. – Girona, J. – Marsal, J.: 2009. Seasonal sensitivity of stem water potential to vapour
pressure deficit in grapevine. Irrigation Science 27: 175–182.
84
Created by XMLmind XSL-FO Converter.
11. WATER STRESS
29.
Pethő M.: 1996. Mezőgazdasági növények élettana, Akadémia Kiadó, Budapest
30.
Soltész, M. – Szabó, T.: 1998. Alma. 119-155. p. In: Soltész, M. (ed.): Gyümölcsfajta-ismeret és használat. Mezőgazdasági Kiadó, Budapest. 513 p.
31.
Wang, D. – Gartung, J.: 2010. Infrared canopy temperature of early ripening peach trees under
postharvest defecit irrigation. Agricultural Water Management, 97: 1787-1794.
85
Created by XMLmind XSL-FO Converter.
12. fejezet - 12. RUNOFF ON
AGRICULTURAL WATERSHED
Accurate micro relief is a key factor of farm runoff condition, but as model information, it is time consuming to
determine and contain high spatial uncertainty. Digital elevation models are mainly successful at describing
macro relief, but their usage as micro relief modelling in water management has several weaknesses. Like other
digital models (grid, raster) TIN is used to describe steep mountain, hilly areas, which have a little or no human
impact on them. However, agricultural land use coverstypically flat areas, where slope is below 12 %. There are
only a few examples of DEM usage (terracing, rice paddy field) in the micro relief analysis of extreme flat
lands.
The largest part of Hungarian Great Plan is suitable site to test this modelling problem. Pálfai (1994) was set up
an exceed water risk assessment scale, by analyzing the water logging maps between 1961 and 1980. From the
43860 km2 of Hungary’s Plan land, 24800 km2 are exposed to the danger of exceed water, where the risk of
drought is also high [16].
On intact, plain flat areas, natural, continuous slope catchment areas that ensure water flow are rare. After a
longer rainy period, the water which not evaporates or infiltrates to the soil accumulates and creating flood,
especially in deeper areas, with heavier soil texture, and no runoff [22]. To avoid this, on flat lands, besides
enhancing the effectivity of precipitation, the storage of water, and the usage of drainage patters are also
required [17]. The process of water accumulation starts with the presence of water on the land, continues with
runoff in surface water receptors, or canals, and ends as water leaves the small watershed [21].
The characteristics of this process are determined by precipitation, as a key factor, as well as the relief, soil
conditions and crop canopy. Moreover, natural or artificial field objects that constrain or enhance runoff
(accelerate or decelerate water flow) are also important factors of the process. The role of relief must be
emphasized. This factor determines how much active energy has the water to move on the field, and also a
major factor in the determination the soil’s water characteristics. Dobos et al., (1997, 2000) was developed and
tested an effective method to reduce the errors of hilly area’s water accumulation in Hungary, by the usage of
DEM, together with satellite images [12, 13]. On flat lands, water usually has not sufficient energy to move, or
the runoff is very slow. Slope degrees are too low; the energy they provide is for the moving water to overflow
natural or artificial field objects. So on flat areas, accumulation processes are only take place after melioration,
which results controlled water accumulation. Among others, Blaskó stated that in Hungary, the Small and the
Great Plain are characterized with the above-mentioned processes, which, together with high SAR content of
ground waters, increase the risk of salinization, at given location and time.
The presented theoretical model of exceed water accumulation determines the process with four correlated
phases. In the first phase, dipping starts as a result of precipitation or snow melt. Next, in deeper parts of the
field, dip water accumulates, resulting about 10 to 200 m2 large exceed water areas, as deep as 10 to 40
centimetres. In the third phase, if the amount of water reaches the threshold of exceed water, it moves to the
deeper areas, and result is a network of water logging areas. The last phase of accumulation is the water
movement in the canals. The hydraulic description of this type of movement is simple. Supposing that the theory
of continue occurs, the amount of runoff can be calculated from hydraulic factors. The model can be modified
with leaching and evaporation processes. Relief has a double effect on exceed water. Micro relief determines the
formation, and macro relief determines the runoff of exceed water [18]. Hectic micro relief enhances the
formation of waterlogged areas, but block runoff. In steeper catchments accumulation is faster, in declivous
areas field, water storage also reduces runoff. Among soil characteristics, its infiltration rate and hydro
conductivity are the most important. These factors are mainly determined by soil texture.
Kertész et al. (1997) pointed out that the analysis of flood period’s satellite images and the usage of different
spectral bands enhance the spatial isolation of not only water bodies, but also water logging areas [14].
Moreover, their method increases the accuracy of isolation, and reduces its spatial uncertainty. The researches
of Bíró and Tamás (2002) on the Bihar plain (East Hungary) revealed that accumulation, as process in time can
be much more precise, if it corrected with factors that create and determine exceed water vulnerability [11]. The
basis of the process is that the relative water logging frequency map is modified with maps of soil characteristics
(hydraulic conductivity, field capacity, depth of first water storage layer), critical (5% probability) ground water
level, surface convex, and land use categories. (SPSS) In this method, when the layers are transformed to one,
the method. The values of s weight is calculated by spatial multiply regression between the relative frequency
86
Created by XMLmind XSL-FO Converter.
12. RUNOFF ON
AGRICULTURAL WATERSHED
map, and the layers representing the modifying factors. By the reclassification of the maps created this way,
several types of vulnerability maps can be formed, having different number of categories, and so serving
different needs. Crucial part of this model is the digital micro relief data, with the required accuracy. Subsurface
processes can be determined with special 3D sub models. In SURFER, ESRI, IDRISI software environment, the
grid, contour, TIN, raster and DEM (digital elevation model) models of Tedej Ltd.’ 1500 hectare large plain flat
area was created. The sources of the models were analogous paper-based 1:10000 scale maps, 1 m spatial
resolution digital ortophotos, Landsat multispectral satellite images, TRIMBLE based DGPS data, and +- 2 mm
accuracy field nivelling surveys. Comparative analysis was done, and suggestions were made to make the
models more accurate from a water management point of view.
Grid, raster, contour DEM is known as it hardly tracks and follows critical spatial variability, but on the other
hand, it’s easily computerized and wide spread. The detailed geostatistical analysis of grid and raster was
published earlier [19, 20], so in the following section a hardly known TIN micro relief model development is
introduced. TIN has a good spatial accuracy and reality, but its GIS transformation to other models often results
information loss. Triangulated Irregular Networks (TIN) are the most commonly used structure for modelling
continuous surfaces using a vector (point, polyline) data model. With triangulation, data points with known
attribute values (e.g. elevation) are used as the vertices (i.e. corner points) of a generated set of triangles. The
result is a triangular tessellation of the entire area that falls within the outer boundary of the data points (known
as the convex hull). The Delaunay triangulation process is most commonly used in TIN modelling. A Delaunay
triangulation is defined by three criteria: 1) a circle passing through the three points of any triangle (i.e., its
circumcircle) does not contain any other data point in its interior, 2) no triangles overlap, and 3) there are no
gaps in the triangulated surface. A natural result of the Delaunay triangulation process is that the minimum
angle in any triangle is maximized fat triangles. The number of triangles (Nt) that make up a Delaunay TIN is Nt
= 2(N-1)-Nh, and the number of edges (Ne) is Ne = 3(N-1)-Nh, where N is the number of data points, and Nh is
the number of points in the convex hull. Normally, little data preparation is necessary when point data is used to
create a TIN. If iso-line data are used as input to TIN generation, both the non-constrained and constrained
options are available and a better TIN result can be expected. This process ensures that triangle edges do not
cross iso-lines, and the resulting TIN model is consistent with the original iso-line data. Not all triangles will
necessarily meet the Delaunay criteria when the constrained triangulation is used. On a micro relief level, main
delineations are irrigation-drainage canals and plough faults. In several locations, based on the drainage
delineation, depressions were located, which must be removed from the model, to make the runoff processing
successful.
Although a pit-free input DEM will speed processing, its not a requirement. Depressions (pits) in the input
Digital Elevation Model (DEM) data are automatically detected and filled before the watersheds are delineated.
If micro relief was analyzed on the basis of iso-line, DEM showed flat area on locations, where great water
logging areas often figure. Figure 51 presents more virtually flat areas.
12.1. ábra - 51. Figure Two typical digital elevation model of the water management
Experiments with actual terrain pointed out that the true surface is probably not flat, but rather rises above the
TIN facets. The edges of the TIN facets that lie below the true surface in this case are examples of what are
called “tunnel edges”. Tunnel edge is any triangle edge that lies below the true surface. An optimal TIN will be
generated if iso-lines are used as the original input for TIN generation, the constrained triangulation is used, and
B/T edges are removed.
New points to solve task, termed critical points, are created at the midpoints of the B/T edges (Figure 51). The
areas around the critical points are then re-triangulated.
87
Created by XMLmind XSL-FO Converter.
12. RUNOFF ON
AGRICULTURAL WATERSHED
12.2. ábra - 52. Figure Relief pattern estimation based on contours of microrelief
(Eastman,2010)
When a B/T edge is shared by two triangles, four new triangles result. When a B/T edge is part of the TIN
boundary, and thus used by only one triangle, two new triangles result. Once the critical points have been placed
and the triangulation has been adjusted, the next step is to assign appropriate attribute values (e.g., elevations) to
these new points. In this case, the recommended method for determining the attribute of a critical point uses a
parabolic shape. The parabola, as a second-order non-linear polynomial method, was chosen because it
combines computational simplicity and a B/T Edges Critical Points Re-Triangulation.
On the surveyed site it was assumed that the contours describe a micro depression and that the deepest point
beyond the lowest contour has a somewhat rounded negative peak. Given this, a parabolic surface (U-shaped
bowl) could be image fitted to the down of the depression. The particular parabolic surface we choose would
depend on the shape of the nearby terrain. If slopes were gentle leading up to the lowest contour, then a surface
with gently sloping sides and a wide sink was chosen. But if slopes were gently, a surface with more vertical
sides and a narrower deep was chosen. Once a particular surface was chosen, all critical points on the tunnel
edges at the negative peak of the sink could be projected onto the parabolic surface. Each of them could then be
assigned the elevation of the surface at their location. The actual implementation of the interpolation differs
from the general logic described above, in the fact that two-dimensional parabolas are used rather than parabolic
surfaces. Up to eight parabolas, corresponding to eight directions, are fit through each critical point location. An
attribute for the critical point is derived for each parabola, and the final attribute value assigned to the point is
their average.
12.3. ábra - 53. Figure Inserting a new point based on local min/max of 8 directional
parabolas
A parabola is defined by the following equation: (X-a)2 = 2p(Y-b) , where the point (a,b) defines the centre (top
or bottom) point of the parabola and the parameter p defines the deepness of the shape. When p is positive, the
parabola is U-shaped. When p is negative, the parabola is inverted. The larger the absolute value of p, the wider
the parabola. Figure shows calculated several parabolas and their equations.
88
Created by XMLmind XSL-FO Converter.
12. RUNOFF ON
AGRICULTURAL WATERSHED
To translate the general parabolic equation to the critical point attribute interpolation problem, we re-label the
axes of Figure from X,Y to S,H where S represents distance from the origin (o) and H represents the attribute
value (e.g. elevation) from the origin (o). The origin is defined by the location and attribute of the original point.
In the example of the hilltop, a sharp peak would be modelled by the linear method in contrast to the rounded
peak of the parabolic method. The implementation of this cointrained TIN method presents on next figure.
12.4. ábra - 54. Figure The improved sink filling process
Left picture shows the isolated vertex points of contour lines, on the middle picture presented the improved TIN
model and on the right picture where blue patch indicated a real runoff process originated from Landsat satellite
image analysis. This independent remote sensing data shows that new model running correctly.
Runoff process is also affected by water infiltration into deep root zone. Full physical-chemical soil analysis of
samples gained from 2-2,5 m deep drillings was done. The selection and location of samples were based on the
relief, and their DGPS coordinates were recorded using the borehole method from hydrogeology as analogy, a 3
D TIN models of the sample site was created. Experience gained from surface TIN models was considered when
a complex 3 D water management environment was built (.
12.5. ábra - 55. Figure Boreholes data preparations with groundwater modelling tool
and visual presentation of soil properties in real 3D environment
89
Created by XMLmind XSL-FO Converter.
12. RUNOFF ON
AGRICULTURAL WATERSHED
Also important factor of runoff process modelling is to know the soil covering spatial pattern. In agricultural
condition this means, to get actual information from vegetation map. The inexpensive method of the high
resolution vegetation mapping apply remote sensed data source. One of the most frequently methods to analyse
vegetation is LAI (Leaf Area Index), what can be measured from Vegetation Index (VI) based on remote
sensing data. Leaf area determines the evapotranspiration and the yield too. Leaf Area Index (LAI) is a key
biophysical variable influencing land surface photosynthesis, transpiration, and energy balance [6].
The electromagnetically reflected data are can be use to determine the soil moisture, so can be direct evaluated
the soil moisture and evapotranspiration. Hyper spectral technology is improved in the spectral resolution of
images - accomplished by increasing the number of spectral bands and decreasing the bandwidth of each band.
In this study, aerial hyper and multispectral images were compared, as well as multispectral digital camera
images with the background data about the test farm site. Hyper spectral records were obtained by a 80channeled aerial spectrometer (Digital Airborne Imaging Spectrometer /DAIS 7915/), where intensive crop
growing was carried out. The detailed GIS database was collected about the test sites. The additional air and
ground images with a TETRACAM ADC wide band multispectral camera were taken in blue, green and near
infrared bands. There have been strong correlations between a red and near-infrared transmittance ratio and LAI.
Chlorophyll absorbs high red energy while plant foliage has relatively low transmittance (reflectance) of red
energy. Plant cell walls, in particular lignin, because scattering of near-infrared energy resulting in high nearinfrared transmittance and reflectance [9].
TETRACAM ADC was developed for agricultural purposes, and it can be used in environmental or plant
protection [2]. LAI was measured by Montgomery method [4] from 6 different morphology of maize’s. We
separated the green leaves from dry leaves to evaluate the photosynthetically active leaf area. Soil moisture,
infra and contact temperature were measured to calibrate the records.
Hyperspectral records were used to compare the measured data in soil or vegetation with bands data. First, raw
remote sensing data was transformed for proper mapping. Second, principal component analysis, unmixing and
likelihood classification were used to analyse prepared data. NDVI was calculated to determine the dispersion of
biomass within a table (Figure56.).
12.6. ábra - 56. Figure NDVI map of Tedej farm
Greener colour indicates higher biomass production where lower runoff value should be expected. The ground
validation of airborne data is necessary to determine accurate relationship between reflectance and vegetation.
90
Created by XMLmind XSL-FO Converter.
12. RUNOFF ON
AGRICULTURAL WATERSHED
Records were taken by TETRACAM ADC multispectral camera in different periods. We measured canopy,
NDVI (Normalized Difference Vegetation Index) and SAVI (Soil Adjusted Vegetation Index). NDVI was
compared with the LAI, and it proved strong regression (r=0.82). Green leaf area (LAIg) was also evaluated in
the test side. There was stronger correlation between LAIg and NDVI (r=0.92) than between leaf area index and
NDVI (Figure 57.). SAVI was computed from red and near infrared reflectance and adjustment soil factor (L).
12.7. ábra - 57. FigureMaize before harvesting (left), canopy of maize (middle), NDVI of
maize(right)
12.8. ábra - 58. Figure NDVI – LAI and LAI(g) correlation in maize
These results offered new opportunities mainly isolating exceed water areas, likely to have no runoff, and in
exploring dynamic spatial relationship between them. The method enables the spatial evaluation of depression
or micro relief locations, where spatial sampling density is low, or completely missing. The 3 D TIN model
presented in this study can be a new approach in soil modelling, as it offers a powerful tool in visualising soil
data sources, having only attribution subsurface data. By this method, they can be used in virtual 3 D
applications. It is highly effective in evaluating processes dynamic in time and space, like soils’ water
management. By providing reliable, fast, and relatively cheap land cover data, remote sensing is an important
data resource. It can be fitted to detect biomass of vegetation. High spectral data give several spectral bands and
high resolution data to choose the right relationship between the bands and the object of research. There is
strong relationship between reflectance and vegetation types. We need to choose the suitable bands to make
more effective method. We have proven strong correlations between LAI and NDVI from multispectral data.
High ground resolution in the multispectral datacube gives special spatial information for high scale runoff
modelling.
91
Created by XMLmind XSL-FO Converter.
12. RUNOFF ON
AGRICULTURAL WATERSHED
1. Test questions:
1. Can you typify different digital elevation models?
2. Why important process is pit removing?
3. How the deep percolation, the vegetation and precipitation intensity impacted on runoff process ?
2. REFERENCES
1. Kruzsilin, A. C. (1958): Az öntözéses termelés biológiai vonatkozásai. – Mezőgazdasági Kiadó. Budapest
2. Nagy S., P. Reisinger, J. Tamás (2004): Möglichkeiten der Anwendung von multispektralen. Journal of Plant
Diseases and Protection, Stuttgart
3. Pechmann I. – Tamás J. – Kardeván P. – Vekerdy Z. – Róth L. – Burai P. (2003): Hiperspektrális technológia
alkalmazhatósága a mezőgazdasági talajvédelemben (EU Konform Mezőgazdaság és Élelmiszerbiztonság,
Gödöllő, Hungary)
4. Petrasovits I. (1988): Az agrohidrológia főbb kérdései. Akadémia Kiadó, Budapest
5. Pontailler J.Y., Hymus G.J., Drake, B.G. (2003): Estimation of leaf area index using ground-based remote
sensed NDVI measurements: validation and comparison with two indirect tecniques.
6. Running, S.W., Nemani, R. R., Peterson, D.L., et al. (1989): Mapping regional forest forest
evapotranspiration and photosyntesis by coupling satellite data with ecosystem simulation. Ecology 70: p
1090-1101.
7. Sheets K.R., Hendrickx, M.H. (1995): Nonivasive Soil Water Content Measurment Using Electromagnetic
Introduction. Water Resources Research, 31, p 2041-2409
8. Tamás, J. (2001): Precíziós mezőgazdaság. Mezőgazdasági Szaktudás Kiadó, Budapest
9. Turner, D.P., Cohen, W.B., Kennedy, R.E., Fassnacht, K.S., Briggs, J.M. (1999): Relationships between Leaf
Area Index and Landsat TM Spectral Vegetation Indices across three temperate zone sites: Remote Sensing
Environment, 70, p 52-68.
10.
Blaskó L., Czimbalmos R., Tamás J., 2003. Evaluation of a long-term experiment on a salt affected soil
with structural B-horizon(solonetz) by means of GIS methods In: Proc. VIIIth.Timisoara's Academic days,
Scientifical Papers Faculty of Agriculture. Editura Mirton Timisoara. 369-375.
11.
Bíró T., Tamás J. 2002. Evaluation of probability of surplus water based on small-scale digital model
In: Proc. Preventing and Fighting Hydrological Disaster, International Association of Hydrological Sciences,
Timesoara, IAHS/AISH 173-177.
12.
Dobos E., Baumgardner, M.F., Micheli, E., 1999. The use of potential drainage density for the
delineation of erosional and depositional surfaces. In Proc. 10th International Soil Conservation Conference.
West Lafayette, Indiana, USA
13.
Dobos E., Micheli E. M.F. Baumgardner, Larry Biehl, Todd Helt, 2000. Use of combined digital
elevation model and satellite radiometric data for regional soil mapping. Geoderma. 97:367-391.
14.
Kertész, Á. - Márkus, B. - Tózsa, I. 1997.Land Use Change Analysis by GIS. In Land Use and Soil
Management, Edited by Filep, Gy., Debrecen, pp. 265-283
15.
Márkus B. 1999 .Error Modelling in GIS Environment, in GIS for Environmental Monitoring, ed. H.P.Bahr - Th. Vögtle, E. Schweizerbart'sche Verslagsbuchhandlung, Stuttgart,. pp. 190 - 200.
16.
Pálfay I.1994.: Az Alföld belvíz-veszélyeztetettsége. Vízügyi Közlemények LXXVI. Évf. 3. füzet.
92
Created by XMLmind XSL-FO Converter.
12. RUNOFF ON
AGRICULTURAL WATERSHED
17.
Somlyódy L. (szerk.) 2000.: A hazai Vízgazdálkodás stratégiai kérdései. MTA Vízgazdálkodási
Kutatócsoportja, Budapest.
18.
Tamás J., Juhász Cs. 1997. Simulation of hidrological limiting conditions based on integrated computer
models In: Nagy, J.(eds.) Soil, Plant and Environment Relationships, Current plant and soil science in
agriculture. Hungary173-186.
19.
Tamás J. 1999. Analysis of uncertaintinity in the design of sampling strategy In: Analele Universitatii
din Oradea, Fascicula Agricultura si Horticultura 5. 7-14.
20.
Tamás J. 2001. Risk of Inland Waters In: Proc. International conference on water and nature
conservation in the Danube-Tisza river, Debrecen. 19-21.
21.
Thyll Sz. Bíró T.1999.: A belvízveszélyeztetettség térképezése. Vízügyi Közlemények LXXXI. Évf. 4.
709-718. p
22.
Várallyay Gy. 1993. Soil data-bases for sustainable land use: Hungarian case study. Soil Resilience and
Sustainable Land Use. (Eds. Greenland D. J., and Szabolcs I.) CAB International, pp. 469-495.
93
Created by XMLmind XSL-FO Converter.
13. fejezet - 13. TRAFIC CONTROLL
Precision – or site specific farming is getting more and more popular in central and east, Europe and appeared in
West Asia or China and South Africa as well. The main reason of this spreading is the potential of efficiency
increase may be achieved by decreased input utilization and improving quality. Beside economic advantages the
precisely controlled and positioned chemical use and fuel saving has positive effect on the environment as
reducing its load.
According to BLACKMORE (1999) precision farming is not a technology rather is to be considered as a
management process [4]. He mentioned some examples which represent that precision farming may be realised
without applying new technique or elements. However, precision farming in case of intensive arable farming is
based on high-tech equipments require significant investment. Similar opinion is stated by GYŐRFFY (2000)
and supported by NEMÉNYI et al. (1998) as well according to whom precision farming means the
complementation or rather the further development of modern agricultural machines [6,8]. The fundamental of
site specific technology is satellite system based positioning.
Positioning accuracy of the applied GPS system may be defined by the demand of the given field operation but
cost of accuracy upgrade of GPS receivers may be a limiting factor. 15-20 cm pass-to-pass accuracy with
manual or assisted steering is accepted in case of fertilizer spreading or spraying but autopilot with 2.5 cm
accuracy is needed for seeding or within-row cultivation.
The cost of upgrade from 15-20 cm (EGNOS) to 6-10 cm (OmniStar) pass-to pass accuracy is approximately 1
500 EUR, while to 2.5 cm pass-to-pass and year-to-year level (RTK) is about 4 600 EUR. With respect to its
cost or rather the accuracy demand of the field operations, the question raises whether to what extent can this
positioning accuracy be realized in the level of steering of agricultural machinery.
Accuracy and availability of GPS positioning using different correction signals were investigated by many
authors e.g. [12, 13, 9, 5]. TUCHBAND (2008) analyzed the absolute and relative positioning accuracy using
code and phase measurements applying precise point positioning (PPP) method [14]. The effects of different
models (ionosphere, clock errors, and orbit models) on positioning errors were investigated as well. These
investigations are very important in order to determine the factors are influencing positioning accuracy or rather
their degree of impact.
Operation of GPS receivers in agriculture is however significantly different. BAIO (2012) compared the
navigation accuracy of manual steering and auto-guidance in case of sugar cane harvester [3]. The author found
that steering accuracy increased applying both in day and night whereas operational field efficiency was the
same. HUDSON et al. (2007) made a very comprehensive study in the frame of which the accuracy (mean error
and RMS error) and precision (standard deviation of error) of three tractor-guidance methods – foam-marker,
light-bar, and assisted-steering systems (Trimble EZ-Steer) – were investigated [7]. Two different travelling
speeds were applied (5.6 – and 11.5 km/h) during a total 504 field passes of parallel swathing operation.
The engineers found that steering using foam marker was significantly less accurate comparing to using lightbar or assisted steering. It was stated as well that field speed caused no significant difference in mean error and
RMS error. However, a significant interaction was found between guidance method and field speed for both
mean and RMS errors. According to the authors, both indices increased at higher speed in case of foam marker
guidance and assisted steering whereas decreased when using light bar guidance. ADAMCHUK et al. (2008)
carried out some series of examinations investigating the positioning accuracy of GPS receivers [2].
They used a test cart carrying the GPS receivers running in an I-beam track and measured the steering accuracy
of autopilot systems in field trials as well. For these latter measurements visual sensor-, or rather potentiometerbased systems were built and applied. ABIDINE et al. (2004) carried out similar investigation but from practical
approach [1].
The authors set up a split-plot trial in the frame of which field operations were done using an RTK autopilot
controlled tractor. Steering accuracy was evaluated by measuring plant damage caused by the given field
operation.
According to the authors no significant plant damage occurred even at 7 miles per hour (mph) forward speed
and cultivator disk spacing of 2 inches from the plant line. SZLÁMA (2011) carried out a comprehensive
investigation with respect to achievable positioning accuracy using RTK system for agricultural purpose [11].
94
Created by XMLmind XSL-FO Converter.
13. TRAFIC CONTROLL
He used John Deer’s single RTK base station and repeater network in Hungary and investigated the positioning
accuracy within different distances from the base stations. This so-called “RTK network” is in fact a group of
single RTK base stations which was established in order to provide 2 cm accurate RTK correction signal for
precision farming applications.
The author analyzed the correlation between the distance from the base station and positioning accuracy or
rather signal strength. It was stated that 2 cm accuracy could have been realized only if the distance of the base
station not more than 5.61 km. Unfavourable was shading effects of natural and artificial.
The present investigation was not to identify the static or dynamic accuracy of the given GPS receiver. The
primary goal was to investigate the achievable steering accuracy under real field circumstances using different
correction signals with hydraulic autopilot system. objects (e.g. relief or building) were experienced as well. The
results of this experiment are very remarkable from a farming point of view. However, it should be mentioned
that measurements were applied are stationary type what is still differ from agricultural practice where
kinematic positioning is needed.
The authors investigated the steering accuracy of a NH T 6030 tractor installed with Trimble Autopilot
hydraulic robot pilot system controlled by a Trimble FMX display (FMX 1). The tractor was built in 2008 and
was used for 2 769 operation hours till the beginning of the tests. All function of the display was unlocked by a
Master Unlock Code was provided by Trimble Agriculture Europe Gmbh. With this, the receiver could have
been set to use different source of correction and was able to receive the signal of GLONASS satellites as well.
The firmware version of the FMX display was 5.11.52214 and 5.15 in case of the Navigation Controller, which
is responsible for tilt compensation as well. The test tractor was installed with another FMX display (FMX 2)
(firmware: 5.11.52214) using a Trimble Ag252 GPS external GPS receiver (firmware: 3.7) for recording
position information as well. Test swats in North – South and East – West directions were marked out ensuring
200 m long straight sections after the turns. Autopilot was used following the same test AB lines in 10
repetitions in North – South, South – North, East- West and West – East directions. Repetitions were done in
each directions using different GPS or rather correction signals such as no correction, EGNOS, OmniStar VBS,
OmniStar HP, RTK from own base station or rather network RTK.
The base station (Trimble AgGPS 450 RTK Base station with in-built radio) as well as the test field is owned by
Helianthus Ltd. farming in town Adony, 60 km South of Budapest, capital of Hungary. The radio of the base
station operated at 450.0000 MHz with 12.5 kHz channel spacing and 2 W output power. The base station was
not GNSS unlocked, thus was not able to receive GLONASS satellite signal. The GPS coordinates of the test
field are: N 47.103214, E 18.86393. The distance from the RTK base station was approximately 600 m. The
base station was installed at the top of a silo tower, approximately 20-23 m above the ground level.
When using RTK network, correction was obtained from the base station located in Budapest utilizing a Trimble
AG3000 GPRS modem with Telenor SIM card. The test area was free of significant slope and plants, its surface
was flat. The plot was opened from every directions, there was no natural neither artificial objects may have
disturb the reception of GPS or rather correction signal. The soil was not freshly cultivated thus it was naturally
compacted, the wheels of the tractor sunk into the soil only 5-6 cm deep.
The average size of soil particles was in a range of 5-15 cm. They were chopped to a given degree during the
pre-test runs thus test runs were done in a quite similar soil surface. The tractor was driven in 9th gear in a fixed
engine revolution of 1 520 1/min ensuring 6.4 km/h travelling speed. FMX 1 controlled the Autopilot system
and recorded a so-called “Track 3d” file consisting information about the test runs such as actual position, date
and time of the given position record, source and age of correction, HDOP, height, heading and a value called
“Offline” among others.
The value offline shows the distance from the actual swath in centimetre. Same data file was recorded by FMX
2 using RTK correction from own base station all the time. The goal of this double measurement was to avoid or
decrease the measurement error caused by the inaccuracy of the GPS signal.
Using the same receiver for both Autopilot control and position mapping the error of positioning affects both
processes and thus steering- and positioning accuracies cannot be separated. Even as application of different
GPS signals causes different impact on the measurements. As position mapping was done using RTK
positioning the effect of its inaccuracy can be considered identical during the experiment. The Autopilot system
was calibrated according to the factory standard.
95
Created by XMLmind XSL-FO Converter.
13. TRAFIC CONTROLL
Data collection started only after the tractor’s hydraulic oil reached the normal operation temperature and the
given correction signal had converged entirely. State of convergence was followed using Trimble’s AgRemote
application. Measurements took place between 11 – 17 August, 2011. Data processing was carried out using
ArcGIS 10 GIS software and Microsoft Excel 2010 application. In the course of processing we used WGS84
geographical coordinate system, so the differences between the measurements can be determined in decimal
degree. To convert decimal degree to meter unit we have used Spatial Analyst / Distance operator / Raster
operation in ArcGIS 10. Our aim was to reach the mm precision in data processing, so the map resolution was
0.0000001 decimal degree.
Accordingly, one pixel meant 0.0008 m (0.8 mm) on the resulted raster map. The standard deviations and
averages were calculated and the diagrams were prepared in spreadsheet as visualization. Data recorded by the
FMX Display were downloaded and separated. Thus, it could be seen those turns, which were different to the
replications. Based on the distribution diagram was created the data set recorded using GNSS correction was
chosen as reference as its standard deviation was the smallest.
Deviations of waypoints using each correction in each direction comparing to the midline of GNSS
measurement were determined in ArcGIS 10 software environment. Data were analysed following a different
approach too. The statistical parsing of the offline distance values recorded by the FMX display has been carried
out as well.
The results of the ArcGIS based analysis shown that the highest steering accuracy was achieved using RTK
correction from GNSS network and own RTK base station. In case of own base RTK, GNSS network and the
XP-HP corrections the standard deviation values were lower in East – West and West – East directions than in
North – South and South – North directions.
Without correction is occurred reverse (Table 1). In case of OmniStar VBS, the standard deviation values of EW direction were lower, while the highest values were detected in the case of without correction and EGNOS
correction at the same direction.
13.1. táblázat - 59. Figure Average standard deviation values in case of different
corrections
Correction source
East-West
West-East
North-South South-North
GNSS
4.088
5.240
9.560
9.458
RTK
5.804
5.128
8.697
10.672
XP-HP
7.671
6.838
7.951
8.673
EGNOS
17.123
11.387
10.788
16.072
VBS
9.269
13.667
9.236
15.899
No CORR
53.687
44.805
18.953
27.847
Distance values defined in ArcGIS in pixel level were compared with the offline distance data of the FMX
display. Direct statistical coherence was not found. However, the standard deviation of GNSS reference data and
offline distance data set were similar, but the values shown higher difference in N-S and S-N directions than in
E-W and W-E directions. Furthermore, data were classified into intervals of tenth of a millimetre. The more
accurate steering was shown the less class intervals. According to this investigation the best performances were
achieved using GNSS and base station RTK or rather OmniStar HPcorrections.
13.1. ábra - 60. Figure The number of class intervals in case of different correction
sources.
96
Created by XMLmind XSL-FO Converter.
13. TRAFIC CONTROLL
The results of the present study justified, that the different correction sources are greatly influence the accuracy
of any autopilot system. The performance of GNSS RTK correction signal surpassed even the own base RTK
signal what is very remarkable, and attributable most possible to the fact that GNSS RTK is a more
sophisticated technology comparing to the single base RTK. The results of the OmniStar signals noteworthy as
well.
The OmniStar XP/HP’s performance is clearly comparable with single base RTK’s one whereas VBS signal was
more accurate and consequent than EGNOS. Certain agricultural operations require different precision levels. In
the case of fertilization the decimetre accuracy is accepted, but in case of seeding a steering inaccuracy less than
2.5 cm is expected.
Choose and adjust to a given operation of the appropriate correction source is necessary for precision
agricultural practice. Using inadequate correction signal could saddle surplus costs on the farmer, which could
reduce the profitability of crop production.
1. Test questions:
1. Why and when is the RTK correction of signals important for PA farmers?
2. When do you choose the EGNOS/WAAS or OmniStar or Ground Virtual GNSS system?
2. References
1. Abidine, A.Z. – Heidman B.C. – Upadhyaya, S.K. – Hills, D.J.(2004): Autoguidance system operated at high
speed causes almost no tomato damage. California Agriculture. 58 (1): 44-47.
2. Adamchuk, V.I. – Stombaugh, T.S. – Price, R.R.(2008): GNSS-based auto-guidance in agriculture. SiteSpecific Management Guidelines SSMG-46. International Plant Nutrition Institute. Norcross, Georgia. 08/08
3. Baio, F.H.R.(2012): Evaluation of an auto-guidance system operating on a sugar cane harvester. Precision
Agriculture. 13 (1): 141-147.
4. Blackmore, S.(1999): Developing the principles of precision farming. In: Proceedings of Agrotech 99.
Barretos Institute of Technology. Barretos, Brazil. 15-19 November 1999. 133-250.
5. Eissfeller, B. – Dötterböck, D. – Junker, S. – Stöber, C.(2011): Online GNSS data processing – status and
future development. 53rd Photogrammetric Week. Stuttgart, Germany.
6. Győrffy, B.(2000): A biogazdálkodástól a precíziós mezőgazdaságig. [From the organic farming to the
precision agriculture.] Agrofórum. 11 (2): 1-4. (in Hungarian)
97
Created by XMLmind XSL-FO Converter.
13. TRAFIC CONTROLL
7. Hudson, G. – Shofner, R. – Wardlow, G. – Johnson, D.(2007): Evaluation of three tractorguidance methods
for parallel swathing at two field speeds. The Student Journal of the Dale Bumpers College of Agricultural,
Food and Life Sciences Discovery. 8: 61-66.
8. Neményi, M. – Pecze, Zs. – Petróczki, F.(1998): Agrárműszaki feladatok a térinformatikai adatbázis
felvételénél illetve annak bővítésénél. [Agricultural engineering tasks in connection with establishment or
enlargement of geo-referred data base.] VII Térinformatika a felsőoktatásban szimpózium elıadások
összefoglalói. Budapest. 94-98.
9. Noack, P.O. – Muhr, T.(2008): Integrated controls for agricultural applications – GNSS enabling a new level
in precision farming. In: 1st International Conference on Machine Control & Guidance. Zurich. 24-26. June
2008. 1-8.
10.
Riczu, P., Mesterházi, P., A., Forián, T., Fehér, G., Bíró, J., Tamás, J.(2012) Evaluation of dfferent
GPS signal corrections to improve field accurancy of autopilot system. International Scientific Conference on
Sustainable Development and Ecological Footprint. West-Hungarian University, Sopron, 1-6.p.
11.
Szláma, Zs.(2011): RTK műholdas irányítórendszer pontosságának elemzése, a pontosság változása a
környezeti hatások és jelismétlő alkalmazásának függvényében. [Precision analysis of RTKSatellite Control
System, Changes in precision under the influence of physical – geographical features and depending on the
use of repeater stations.] Thesis. 63 p.
12.
Tamás, J. – Lénárt Cs. (2003): Terepi térinformatika és a GPS gyakorlati alkalmazása. [The practical
application of field GIS and GPS.] DE-ATC, Debrecen. 199 p. (in Hungarian)
13.
Tiberius, C. – Verbree, E.(2004): GNSS positioning accuracy and availability within Location Based
Services: The advantages of combined GPS-Galileo positioning. In: 2nd ESA Workshop on Satellite
Navigation User Equipment Technologies, NAVITEC '2004. ESTEC Noordwijk, December 2004, 1-12.
14.
Tuchband, T.(2008): Nagy pontosságú abszolút GPS helymeghatározási technika pontossági
vizsgálata. [Accuracy analysis of high accuracy GPS positioning technique] In: Doktori kutatások a BME
Építımérnöki Karán. 12 p.
98
Created by XMLmind XSL-FO Converter.
14. fejezet - 14. APPLIED LAND
CHANGE MODELLING
Land-Use modelling is a rapidly growing scientific field because land-use changing is one of the most important
ways that humans influence can cause for agro-environment. For the sustainable utilization of the land
ecosystems, it is essential to know the natural characteristics, extent and location, quality, productivity,
suitability and limitations of various land uses [4].
The analysis of the spatial extent and temporal change of land-use categories using remotely sensed data is of
critical importance to agricultural sciences [6]. The rational land-use and proper soil management are important
elements of sustainable (agricultural) development, having special importance both in the national economy and
in environmental protection [7]. Space-born remote sensing has a good potential for change detection and good
data availability and is, therefore, well suited for the monitoring of land-use change over a time period. IDRISI
LCM (Land Change Modeller, Clark lab.) is organized around a set of five major task areas:
• Analysing past landcover change,
• Modelling the potential for land transitions,
• Predicting the course of change into the future,
• Assessing its implications for biodiversity,
• Evaluating planning interventions for maintaining ecological sustainability [2].
Pontius et al. (2001) give the most complete peer-reviewed description and application of GEOMOD currently
in IDRISI [1]. GEOMOD is a grid-based land-use and land-cover change model, which simulates the spatial
pattern of land change forwards or backwards in time. The minimum input requirements are: the beginning time,
the ending time, an image of the beginning time for two land cover types that must be denoted by 1 and 2, and
an estimate of the number of cells of each of the two categories at the ending time.
Most users also include either a suitability map that has already been created or driver maps that GEOMOD uses
to create its own suitability map. Some users have a map of the true landscape at the ending time, which can be
used for validation as described below. GEOMOD’s most important output is a map of the simulated landscape
of developed versus non-developed cells at the ending time. If any of the maps have a mask, then the modeller
should make certain that all maps have the exact same mask. GEOMOD has been designed such that it can take
maximum advantage of data that can vary highly in availability, completeness, precision, currency, and
accuracy. For example, GEOMOD requires only one beginning land-use map for calibration, while some
algorithms for other popular models require maps from four times for calibration [5].
The Change Analysis process provides a rapid quantitative assessment of change by graphing gains and losses
by landcover category. A second option, net change, shows the result of taking the earlier landcover areas,
adding the gains and then subtracting the losses. The third option is to examine the contributions to changes
experienced by a single landcover.
Gains and losses (Figure 61) are based on change of values between 2000 and 2006 land cover pixels and
exclude pixels with no change. Values are shown in hectares some of the gains and/or losses are offset by pixel
misclassification and generalization of cluster groups. The biggest gains are in the Pastures, Coniferous forest
and Arable lands.
14.1. ábra - 61. Figure. Land Cover Gains and Losses
99
Created by XMLmind XSL-FO Converter.
14. APPLIED LAND CHANGE
MODELLING
The Land Cover Net Changes graphs (Figure 62) show the total net changes in hectares for the 26 land cover
classes. The graph shows the Natural Grassland and the Broad-leaved Forest areas have declined more than
25.000 hectares and the Coniferous Forest and the Pastures areas have increased more than 23.000 hectares.
14.2. ábra - 62. Figure Land Cover Net Changes
According to the land cover change map the changes are based on a pixel value change analysis. Detail land
cover changes are shown on the legend. Pixels with no change have been assigned a no-data value and are not
shown on the legend. The most noticeable changes are in the Natural Grassland, Arable Lands, Complex
Cultivation Patterns and Broad-leaves Forest which is shown in sample size (Figure 63). The next step is to
create a group transitions into a set of sub-models and to explore the potential power of explanatory variables.
The Transition Potentials module of IDRISI software allows one to group transitions into a set of sub-models
and to explore the potential power of explanatory variables. Variables can be added to the model either as static
or dynamic components. Static variables express aspects of basic suitability for the transition under
consideration, and are unchanging over time. Dynamic variables are time-dependent drivers such as proximity
100
Created by XMLmind XSL-FO Converter.
14. APPLIED LAND CHANGE
MODELLING
to existing development or infrastructure and are recalculated over time during the course of a prediction. Once
model variables have been selected, each transition is modelled using either Logistic Regression or IDRISI’s
extensively enhanced Multi-Layer Perceptron (MLP) neural network. After a detailed assessment of empirical
modelling tools (such as Weights-of-Evidence, Empirical Probabilities, Empirical Likelihoods, etc.), it was
found that these two approaches offer the strongest capabilities, particularly the MLP. The MLP neural network
has been extensively enhanced to offer an automatic mode that requires no user intervention.
The result in either case is a transition potential map for each transition – an expression of time-specific
potential for change. Static variables express aspects of basic suitability for the transition under consideration,
and are unchanging over time.
According to the results of running LCM model the number of transitions was high so it was reduced to just 4
transitions (ignore transitions less than 15.000 ha). In order to predict changes, we will need (at any moment in
time) to be able to create a map of the potential of land to go through each of these transitions so called
transition potential maps
14.3. ábra - 63. Figure Land cover change map
The evidence likelihood transformation is a very effective means of incorporating categorical variables into the
analysis. It was created by determining the relative frequency with which different land cover categories
occurred within the areas that transitioned from 2000 to 2006. The numbers thus express the likelihood of
finding the land cover at the pixel in question if this were an area that would transition. According to the high
Cramer’s V values the broad-leaved forest, the complex cultivation patterns and the rice fields have a strong
association with all categories. The driver map was selected for the LCM model. It was used for creation the
distance map from 4-times frequency of excess water inundation (Figure 64).
14.4. ábra - 64. Figure Distance from 4-times frequency of excess water inundation
Each transition was modelled using Multi-layer Perceptron (MLP) neural network which is a dynamic process.
The MLP constructs a network of neurons between explanatory variables and output classes (the transitions and
persistence classes), and a web of connections between the neurons that are applied as a set of (initially random)
weights.
101
Created by XMLmind XSL-FO Converter.
14. APPLIED LAND CHANGE
MODELLING
These weights structure the multivariate function. When the MLP completes its training, it is up to us to decide
whether it has done well enough and whether it should re-train either with the same parameters, but a different
random sample, or with new parameters. As launched by LCM IDRISI, the MLP starts training on the samples
it has been provided of pixels that have and have not experienced the transitions being modelled.
At this point, the MLP is operating in automatic mode whereby it makes its own decisions about the parameters
to be used and how they should be changed to better model the data. After training has been completed, we need
to classify to complete the process of transition potential modelling. The result in either case is a transition
potential map for each transition (Figure 65).
14.5. ábra - 65. Figure Transition potential map (sample)
If we want to determine the potential of land transform (Figure 66), we may consider the distance to excess
water hazarded areas.
14.6. ábra - 66. Figure Soft prediction to land transforms
One of the major components of land cover change prediction is modelling transition potential. According to the
LCM IDRISI the primary tool for transition potential modelling is the MLP Neural Network. In LCM, land
cover change prediction utilizes two land cover maps from two different dates (2000 and 2006) to predict what
the land cover will be in the future or what land cover need to transform in the future. The particular transitions
of interest were specified for the sub-model and specified a variable which drive the type of transition taking
place. As a consequence the resulted maps and the integrated database can be utilized in numerous land related
activities (e.g. land-use and agricultural planning).
1. Test questions:
1. Could you enumerate 5 steps of IDRISI Land Change Modelling?
2. What dos Transition Potentials mean and what is the role of Multi-Layer Perceptron?
2. References
1. Pontius Jr, R G, J Cornell and C Hall. 2001. Modelling the spatial pattern of land-use change with Geomod2:
application and validation for Costa Rica. Agriculture, Ecosystems & Environment 85(1-3) p.191-203.
102
Created by XMLmind XSL-FO Converter.
14. APPLIED LAND CHANGE
MODELLING
2. Eastman, R(2009) IDRISI Taiga, Guide to GIS and Image Processing, Clark Labs Clark University,
Worcester, MA, USA
3. Büttner, G. – Bíró, M. – Maucha, G. – Petrik, O.: 2001. Land cover mapping at scale 1:50.000 in Hungary:
Lessons learnt from the European CORINE programme, in: A decade of Trans-European Remote Sensing
Cooperation, Buchroithner, M.F. (Ed.), Balkema. 25-31.
4. Rejmur Rahman Md. – Saha, S. K.: 2009. Spatial dynamics of cropland and cropping pattern change analysis
using Landsat TM and IRS P6 LISS III satellite images with GIS. Geo-spatial Information Science, 12 (2):
123-134.
5. Silva, E. and K. Clarke. 2002. Calibration of the SLEUTH urban growth model for Lisbon and Porto,
Portugal. Computers, Environment and Urban Systems 26. pp. 525-552.
6. Tamás, J.: 2003. Problems and solutions of field scale agro-ecological data acquisition and data
interpretations in agroinformatical domain. Applied Ecology and Environmental Research, International
Scientific Journal, 1 (1-2). pp. 143-159.
7. Várallyay, Gy.: 2006. Soil degradation processes and extreme soil moisture regime as environmental
problems in the Carpathian Basin. Agrokémia és Talajtan, 55 (1-2): 9-18.
103
Created by XMLmind XSL-FO Converter.
15. fejezet - 15. ANALYZE OF
BIOMASS PRODUCTIVITY BY
TIMESERIES
According to Csete and Láng sustainable competitiveness is accomplished through sustainable farming methods,
whose key motives are sustainable production, adaptability, quality to any extent and favourable investments
levels, consequently this kind of competitiveness is altogether different from any old practice.
The U.S. Department of Agriculture (USDA) and the National Aeronautics and Space Administration (NASA)
signed a Memorandum of Understanding (MOU) to strengthen future collaboration. In support of this
collaboration, NASA and the USDA Foreign Agricultural Service (FAS) jointly funded a new project to
assimilate NASA's Moderate Resolution Imaging Spectroradiometer (MODIS) data and products into an
existing decision support system (DSS) operated by the Production Estimates and Crop Assessment Division. It
produces objective, timely and regular assessments of global agricultural production outlook and the conditions
affecting food security. In monitoring crop conditions for a specific region, remotely sensed vegetation index
data are used to track the evolution of the growing season compared to reference long-term mean conditions [3].
A global normalized difference vegetation index (NDVI) is produced from MODIS data, and is referred to as
the "continuity index" similar to the existing archive of NOAA-AVHRR derived NDVI. SPOT Vegetation
NDVI data and MODIS NDVI data from the Terra and Aqua platforms represent improvements in the ability to
monitor land photosynthetic capacity. The croplands are highly variable both temporally and spatially.
Croplands vary from year to year due to events such as drought and fallow periods, and they vastly differ across
the globe in accordance with characteristics such as cropping intensity and field size [1]. To describe this
temporally heterogeneity a global NDVI time-series database, with a spatial resolution of 250 meters has been
assembled using a 16-day compositing period, allowing for regular comparisons of growing season dynamics
[2].
Basic data source is made after dual average by MODIS Terra remote sensing data. This MODIS NDVI dataset
is reprojected and mosaiced to suit the Nyírlúgos region. There was maps made in Nyírlugos which is after
mean of territorial matrix of 10x10 km with 250 meter rescission. These data are monthly frequency data of 16
day period which had respect for in 6 years duration. The analysed areas are in black square in next Figure.
15.1. ábra - 67. Figure NDVI time series of Nyírlugos in 2006 (January-December from
to left by lines)
104
Created by XMLmind XSL-FO Converter.
15. ANALYZE OF BIOMASS
PRODUCTIVITY BY
TIMESERIES
The progress of biomass process is a climate dependent periodicity in time so it is manifest to analyze the
observation of time series by time steps methods.
Forecast, operation, simulation are solvable this way. Basic data source consists of dual averaging NDVI values
counted by MODIS Terra remote sensing data. There was made maps in Nyírlugos which is originated from
means of territorial matrix of 10x10 km with 250 meter resolution. These data are monthly frequency covering
16 day periods which had respected for in 6 years duration (Figure 68.).
15.2. ábra - 68. FigureThe time series of NDVI data sources and linear trend line (20012006, n=72)
105
Created by XMLmind XSL-FO Converter.
15. ANALYZE OF BIOMASS
PRODUCTIVITY BY
TIMESERIES
The mean of presented time series of NDVI data sources in region of Nyírlugos is 0.536; standard deviation:
0.136; maximum: 0.759; while minimum: 0.316.
Discrete time steps in time describe phenomenon of biomass growth. The components of time series model were
determined in four steps. The applied theoretical model were described by Kontur et al., (1993), used for
another phenomenon (groundwater hydrograph) which was continuous in time and space. This adopted time
series analysis based on the next formula:
Where Ti trend, Pi periodic, Ai autoregressive and Vi contingent component. In addition data sources in were
terminated by equation 1. First of all the trend was detached from the series for term 2001-2006. Due to this
process the long term effect was eliminated from the data source. This means the effect of serial of wet and
drought years resulted of biomass production changing was filtered out.
The detachment of linear trend component:
i= 1, 2, …., N (time steps)
With proper replacement the element of parameter vectors d0 and d1, independent variable i=1, 2,…N,
dependent variable Y1, Y2,… YN and errors y1, y2,…yN can be obtained the series without trend.
where
d0 = constant of trend function
d1 = coefficient of trend function
N = number of element
i = ith element
Y = mean of original time series
106
Created by XMLmind XSL-FO Converter.
15. ANALYZE OF BIOMASS
PRODUCTIVITY BY
TIMESERIES
Calculated result of Equation 2. is the following: Ti=0.5113+0.0007i. The Pi periodic component of time steps
of annual biomass intensity is exists in temperate climate. As next step the periodic component was detached by
using the model. In that case when the periodic time of periodic component is known, the determination of the
period amplitude has to be done only. We suppose there is a 12 months periodic in biomass change. The
calculation of one time period is the following:
where:
a0 constant of period
a and b coefficient of function
The relevant period time is r = 12 months. The value of i can be counted in month. The a0, a and b parameters
were determined. In this case y1 time step without trend with Pi periodic component was converged (Eqv. 5.):
The determination “a” and “b” (Eqv. 6.) is needed to count Pi.
. equ
The a*cos(2π*i/12) periodic component value was 0,000131776, while b*sin(2 π i*i/12) periodic component
value was 2,80205E-05 between 2001-2006. Pi periodic component and substances are demonstrated by Figure
69.
15.3. ábra - 69. Figure Time steps of biomass production periodic component and
substances
After the isolation of the periodic component from the data source the autoregressive and random components
are remained. During the interpretation of the prepared model the autoregressive component expressed the
connecting effect of biomass periods (year by year) which are written in plant production practice as year effect
while random component determines the uncertainty of model. Then analyzes was continued with determination
of pi random component supposing the trend and periodic component effects are not exist anymore. So there is
only autocorrelation between time series values. The one step autoregressive sub-model is the following:
107
Created by XMLmind XSL-FO Converter.
15. ANALYZE OF BIOMASS
PRODUCTIVITY BY
TIMESERIES
where t refers time step and Vi is autoregressive component, the c1 is constant of the above function. c1’ equal to
autocorrelation factor, which is rj (Eqv.8):
Value of c1’ is 0,274262498 while mean of Vi value is 0,04 the minimum 0,08 the maximum 0,11. When taking
into consideration the all regularity these values mean the precision of the determination of biomass production.
The random component depends on the climate and cultivation condition mainly. But only that part of it which
is not affected by the periodicity and the trend. Higher random factor can be occur anywhere (e.g. extreme year)
but it is more likely on irregular increasing interval of biomass-(e.g. indeterminate sow conditions).
It was introduced a math process first time in Hungarian biomass research which is made by dissociation of 6
years time series from earth satellite multispectral remote sensing data in region of Nyírlugos. This model
consists of linear trend, periodic, autoregressive and random components which were successfully transformed
forward. Result of trend analyse filtered out long term effect. This means that the effects of biomass increase
and decrease like wet or drought years were detached. With detaching periodic component sinus changing was
separated from data source. Autoregressive analyses quantify statistical rules which resulted from dependence of
sequential years. Random component demonstrated the uncertainty of model. This is the factor which makes
biomass production characteristics undetermined in an added point or in investigational space. This means by
taking into consideration the all well-known influential factor divergent value can be appeared according to the
expected value. After adjusting model component successfully backward transformation was carried out which
procedure proved the applicability clearly. This confirms the fact that significant regularity can be assumed in
biomass growth in such a dynamic model was set in this study.
1. Test questions:
1. Can you describe equation of time series analysis?
2. What is the mean of periodic component?
3. What is the mean of autoregressive component?
4. Can you explain the role of Earth Observation for PA farmers?
2. REFERENCES
1. Anyamba, A., J. Eastman R., Tucker, C. J. (1998). Warm Enso Event of 1997/98: NDVI Precursors and
Drought Pattern Prediction for Southern Africa. Greenbelt, NASA pp. 1¬40. Csete, L., Láng, I.(2005) A
fenntartható agrárgazdaság és vidékfejlesztés. Magyarország az ezredfordulón.MTA Társadalomkutató
Központ. Marosi-Print Kiadó. Budapest. 1¬313. Kontur I., Koris K., Winter J. (1993). Hidrológiai
számítások. Akadémiai Kiadó, Budapest. pp. 143-184.
2. Tamás J., Németh T. (szerk.) (2005): Agrárkörnyezetvédelmi indikátorok elmélete és gyakorlati
alkalmazásai. Debreceni Egyetem, Debrecen, p. 138
3. Tucker, C. J., Vanpraet, C. L., Sharman, M. J , van Ittersum, G. (1985) Satellite remote sensing of total
herbaceous biomass production in the Senegalese Sahel: 1980–1984,” Remote Sens. Environ., vol. 17, pp.
233–249,
108
Created by XMLmind XSL-FO Converter.
15. ANALYZE OF BIOMASS
PRODUCTIVITY BY
TIMESERIES
109
Created by XMLmind XSL-FO Converter.