ANALYSIS OF THE SPATIAL AND TEMPORAL EVOLUTION OF
VEGETATION COVER IN THE SPANISH CENTRAL PYRENEES:
THE ROLE OF HUMAN MANAGEMENT.
Sergio M. Vicente-Serrano*, Teodoro Lasanta** and Alfredo Romo***
*Departamento de Geografía y Ordenación del Territorio. Universidad de Zaragoza. C/
Pedro Cerbuna 12. Ciudad Universitaria. 50009. e-mail: svicen@posta.unizar.es
**Instituto Pirenaico de Ecología (CSIC). Campus de Aula Dei. Apdo. 202. 50015Zaragoza.
***Laboratorio de Teledetección. Departamento de Física Aplicada I. Facultad de Ciencias.
Universidad de Valladolid. 47071. Valladolid. España.
ABSTRACT:
Vegetation cover increment has been identified at global scales using satellite
images and vegetation indices. This fact is usually explained by global climatic change
processes as CO2 and temperature growth. Nevertheless, although these causes can be
important, the role of socio-economic transformations must be considered in some spaces,
since in several areas of Northern Hemisphere an important change in management
practices has been detected. The rural depopulation and land abandonment have reactivated
the natural vegetal regeneration processes. This work analyses the vegetation evolution in
the central Spanish Pyrenees from 1982 to 2000. The analysis has been taken using
calibrated-NDVI temporal series from NOAA-AVHRR images. A positive and significant
trend in NDVI data has been identified from 1982 to 2000 coinciding with temperature
increment in the study area. However, the spatial differences in magnitude and sign of
NDVI trends are significant. The role of land management changes in the twenty-century is
considered as hypothesis of the spatial differences in NDVI trends. The role of land-cover
and human land-uses on this process has been analysed. The highest increment of NDVI is
detected in lands affected by abandonment and human extensification. The importance of
management changes in vegetation growth is discussed, and we indicate that although
climate has a great importance in vegetal evolution, the land-management changes can not
be neglected in our study area.
KEY WORDS:
Vegetation evolution, revegetation processes, vegetal trends, abandonment, extensification,
NDVI, NOAA-AVHRR, Pyrenees, Spain.
INTRODUCTION
Vegetation cover changes play an important role in the development of
environmental processes (van Wijgaarden, 1991) due to the strong relationships between
the biosphere and parameters such as atmospheric CO2 content (Braswell and others, 1997;
Zeng and others, 1999), the great influence of vegetation on the hydrological cycle at
different spatial scales (Changnon and Demissie, 1996; Aber and others, 1995; Kergoat,
1998; Hutges and others, 1998); on sediment transport (Kok and others, 1995; García-Ruiz
and others, 1995); and on landscape structure and diversity (Kammerbauer and Ardon,
1999; Olson and others, 2000).
Numerous studies (Riebsame and others, 1994; Lucht and others, 2002; Myneni and
others, 1998; Kawabata and others, 2001) point out the recent increment of vegetation
cover in different world ecosystems, adducing that the principal cause is the rise in
temperatures and/or precipitation. Peñuelas and Filellas (2001) show significant changes in
vegetation cycles (flowering, fructification, vegetative period, etc), and Fitter and Fitter
(2002) indicate that temperature increment implies more favourable conditions in cold
regions, with an advance of flowering phases and a longer vegetative period.
Nevertheless, with the influence of global climatic change, the regional or local
conditions must be considered in the processes of vegetation increase. The vegetation
evolution depends on topographical conditions (Florinsky and Kuryakova, 1996). The
vegetation succession phases (Vicente-Serrano and others, 2002), soil type (Farrar and
others, 1994) and the human management and land use (Hester and others, 1996; Stohlgren
and others, 1998) also play major roles. The greatest growth in vegetal biomass has been
detected in medium and high latitudes in the Northern Hemisphere (Kawabata and others,
2001; Lucht and others, 2002; Shabanov and others, 2002; Slayback and others, 2003).
These papers have studied regions where the main effect is expected to be climatic and
where the human activities have little influence on landscape and vegetation cover.
Moreover, the vegetation increment coincides with a temperature growth that facilitates
longer vegetative periods, high evapo-traspiration rates and significant vegetation growth.
In this paper a region with dominant human influence is studied to point out that human
factors cannot be neglected in many other areas because management changes,
abandonment and extensification can influence the vegetation increment.
European Mediterranean mountain areas are part of territories in which important
land use changes have taken place as a consequence of depopulation and the abandonment
of traditional economic activities (García-Ruiz and Lasanta, 1990; Bazin and Roux, 1992;
Lasanta, 1990). For centuries, the entire territory was exploited for a wide range of
purposes. Nevertheless, throughout the twentieth century, only the most fertile areas (valley
bottoms) have been occupied by intensive land uses whereas the slopes have been
abandoned, and an intense revegetation process has thus been reactivated (García-Ruiz,
1990; MacDonald and others, 2000; Lasanta and others, 2000). The Mediterranean
mountain system constitutes an appropriate space to study if traditional land use location
and abandonment processes affect the vegetal biomass increment.
The objectives of this work are two. Firstly checking whether the increment of
vegetal biomass, identified by other works on a global scale, is also manifested in a
representative Mediterranean mountain area, and if the climatic changes can have some
influence in this process. Secondly, to determine the role of human land management in the
spatial and temporal changes in vegetation cover. The analysis has been achieved using a
multi-temporal series of NDVI obtained from satellite images (NOAA-AVHRR).
STUDY AREA
The study area is located in the high basin of the Aragón river, tributary of the Ebro
river, in the Spanish central Pyrenees (Figure 1). The surface of the basin is 1772 km2, with
a wide range of altitudes, from 2886 m in the Collarada peak to less than 500 m in the Yesa
reservoir.
Relief and lithology are placed in parallel bands with a NW-SE direction (SolerSampere and Puigdefábregas, 1972). In the northern part (Axial Pyrenees), the lithology is
Palaeozoic (limestones, schists and clays). The most elevated peaks of the basin are located
in the sierras interiores (limestones and sandtones), with elevations over 2500 m. The next
lithological band corresponds to Eocene flysch areas, with ridged relieves, moderate slopes
(between 20 and 50 %) and altitudes between 800 and 2200 m (García-Ruiz and
Puigdefábregas, 1982). The southernmost sector corresponds to the inner depression, with
eocene marls, forming a wide valley dominated by quaternary deposits (terraces and
glacis), altitude range from 500 to 900 m and soft slopes (less than 20%).
The basin has a transition climate that oscillates between Atlantic (North and East)
and Mediterranean influences (South and West). In the Inner depression the mean annual
precipitation is 800 mm. This value is overcome in the rest of the basin; above 1500 m the
precipitation is higher than 1500 mm. The annual variability is very high, and the rainy
season extends from October to June. Mean temperature is 12 ºC in the Inner depression.
During the cold season (November to April) the isotherm of 0 ºC is located at 1.549 m over
sea level (García-Ruiz and others, 1985).
The traditional management has been maintained for centuries, based on an integral
use of land resources and guaranteeing the population's food in a self-supplying economy
with very scarce exchanges with the exterior and a very high sheep census. This fact
implied important transformations in natural spaces. Several forest areas were cut down,
and cultivation areas and livestock pastures established. The forests located in low slopes
(Quercus faginea, Pinus sylvestris and Fagus sylvatica) were deforested and the lands
occupied for cereal cultivation (Lasanta, 1989). The forests located at high altitude were cut
down too, and summer livestock pastures were created (Montserrat, 1992; García-Ruiz and
Valero, 1998; Ferrer, 1988). The land use in the landscape patches was discriminated on the
basis of potential productivity and environmental limitations, since subsistence farming and
population's food depended on the balance between land exploitation and natural resource
conservation (Puigdefábregas and Fillat, 1986).
During the twentieth century, the Pyrenees underwent important economic
transformations, with a high depopulation and extensification of large areas (García-Ruiz,
1988). This process was induced by the development in communications and the difficulty
to compete with flat areas in the Ebro valley (of easier access, near decision centres, and
with great soil fertility, easy mechanisation and with the creation of irrigated lands). At
present, only the flat areas in the bottom valleys with good accessibility are cultivated with
tractors, while all slopes have been abandoned. The abandoned fields cover 480.5 km2
(27,3% of the studied area); and constitute the dominant landscape in the flysch areas
(Lasanta, 1988). On the other hand, the sheep and cattle census dropped pronouncedly with
the transhumance system crisis (migration of sheep to barrens and fallows of the Ebro
valley during the cold season). During the twentieth century the basin lost more than 80%
of the sheep, and the livestock has exercised minimal grazing pressure on large areas,
especially in the low slopes (Vicente-Serrano, 2001). Sheep and cattle were concentrated
exclusively on cultivated lands and in summer pastures. García-Ruiz and Lasanta (1993)
point out that cultivated lands contribute between 66% and 79% of livestock feeding, and
the summer pastures contribute between 18 and 27%. The low slopes (forests, shrub and
abandoned fields) contribute only between 3 and 7%.
METHODOLOGY
Using Remote Sensing for vegetation monitoring
Remote Sensing has been widely used for monitoring vegetation dynamic since
allows analyse large areas with a high temporal frequency (Gutman, 1991). The high
temporal frequency and the availability of long time series of NOAA-AVHRR images
make this data very useful for monitoring vegetation changes (Gutman and others, 1995).
Utility of remote sensing for vegetation monitoring is based on the response of
vegetation cover to radiation in visible and near-infrared regions of electromagnetic
spectrum. Visible radiation is mainly absorbed by vegetation in photosynthesis processes
while near infrared radiation is principally reflected, owed to the internal structure of leaves
(Knipling, 1970). High vegetation activity is characterised by low reflectivity of solar
visible radiation and high reflectivity in the near infrared region of the spectrum.
Different indices have been developed for monitoring and measuring vegetal status
using spectral data (Bannari and others, 1995). Nevertheless, the mostly used is the
Normalized Difference Vegetation Index (NDVI, Tucker 1979) that is calculated using the
expression [NDVI = (Infrared - Red)/(Infrared + Red)].
There are some shortcomings in NDVI use for monitoring vegetation status. The
relationship among vegetation parameters (leaf area index, vegetation cover, green-biomass
and NDVI) are often non-linear (Gillies and others, 1997; Choudhury and others, 1994),
since the NDVI signal saturates before the maximum biomass is reached (Carlson and
others, 1990). Moreover, background soil properties such as surface soil color affect NDVI,
introducing some errors (Huete, 1988).
Nevertheless, despite these shortcomings, numerous authors have pointed out the
close relationship between NDVI and several ecological parameters. The NDVI measures
the fractional absorbed photosynthetically active radiation (FPAR, Myneni and others,
1995), and exhibits a strong relationship with vegetation parameters just as green leaf area
index (Baret and Guyot, 1991; Carlson and Ripley, 1997), green biomass (Tucker and
others, 1983; Gutman, 1991; Cihlar and others, 1991; Diallo and others, 1991; Wylie and
other, 2002) or fractional vegetation cover (Gillies and others, 1997, Duncan and others
1993).
In fact, the NDVI extracted from remote sensing is an excellent tool for monitoring
vegetation status and its temporal dynamic. However, the creation of NDVI temporal series
is problematic due to problems related to the non-uniformity of satellite time-series that can
restrict the satellite use for temporal analysis of vegetation cover (Gutman and others, 1995;
Goward and others, 1991). Satellite changes, orbit drift and sensor degradation of NOAA
satellites cause temporal inhomogeneities in AVHRR data (Price, 1991). Satellite orbit drift
results in a systematic change of illumination and local time of observation that produces
artificial temporal trends in the data (Kogan and Zhu, 2001), whereas satellite changes
cause breaks in temporal curves. These shortcomings involve that adequate tools are
needed to calibrate the NDVI for subsequent temporal analysis (Staylor, 1990).
Nowadays different NDVI global data series of contrasted calibration reliability are
available from AVHRR data (PAL and GIMMS NDVI, Justice and Townshend, 1994;
James and Kalluri, 1994) that have been widely used in the ecosystem monitoring (Pelkey
and others, 2000; De Fries and others, 2000; Salinas-Zavala and others, 2002). PAL-NDVI
database (available in http: //daac.gsfc.nasa.gov) has monthly NDVI data from 1981 to
2000, and is very useful tool to determine trends in vegetation cover by means of satellite
observations. The calibration of this series has been meticulous, and it has been taken a lot
of effort to develop post-launch calibration coefficients, tested in areas without vegetation
cover where a high NDVI temporal stability is assumed (NASA, 2003). Moreover, the
homogenization of the series has been checked with good results (Smith and others, 1997;
Kaufmann and others, 2000). Kaufmann and others (2000) analyse whether the changes in
solar zenith angle (SZA), due to orbital drift and sensor changes, affect significantly the
temporal homogeneity of NDVI series. These authors show that NDVI data series
calibrated using the PAL post-launch calibration coefficients (Rao and Chen, 1995 and
1999) are not affected by SZA changes. In fact, they highlight that NDVI is not statistically
significant related to SZA except for biomes with relatively low leaf area. Kawabata and
others (2001), Pelkey and others (2000), Shabanov and others (2002), among other authors,
have used the PAL-NDVI series for analysis of vegetation trends at continental and global
scales; and in this paper the PAL-NDVI series have been used to analyse the temporal
evolution of vegetation in the Aragon river basin from 1982 to 2000.
The spatial resolution of PAL-NDVI database (8 km of grid cell size) is insufficient
to determine spatial differences in vegetation evolution. It is necessary to use higher spatial
resolution databases to determine the spatial differences in vegetation trends. In order to
solve this problem, we have used another NDVI temporal database. The new NDVI
database has been created in Spain by the LATUV (Remote Sensing Laboratory of
Valladolid University) at Local Area Coverage resolution (LAC, 1 km2 of grid cell size).
The LATUV has one reception antenna of AVHRR images since 1993. The images are
received daily and atmospherically and geometrically corrected (Illera and others, 1996a).
After the clouds are eliminated (Delgado, 1991). Residual atmospheric errors in the
correction process are reduced with the creation of monthly composites using the maximum
value composite method (MVC; Holben, 1986). The calibration process is contrasted, and
standard published post-launch calibration coefficients are used (Rao and Chen, 1995 and
1999). These coefficients have also been used by the NASA to calibrate the PAL-NDVI
database, guaranteeing the temporal homogeneity of the NDVI series (Kaufmann and
others, 2000). The LATUV-NDVI series has been extensively applied in the Iberian
peninsula with numerous purposes: identification of drought areas (González-Alonso and
others, 1995, 2000 and 2001), yield productivity and natural vegetation monitoring
(Vázquez and others, 2001; Illera and others, 2000), or forest fire danger (Illera and others,
1996b; Calle and others, 2000).
The principal deficiency of this database is that it only contains information from
1993 to 2000. The temporal coverage of both databases is different, for this reason, the
PAL-NDVI database has only been used to analyse whether in the whole of the study area
there is a significant increment of NDVI as well as the role of climatic variables in the
process. The LATUV-NDVI database has been used to determine whether between 19932000 significant spatial differences are identified in vegetation trends, and the magnitude of
the changes.
Finally, a problem can be formulated in NDVI data related to the topography of the
study area. The area analysed has a complex topography and viewing geometry. For this
reason, non-lambertian reflectance of plant canopies and topography would affect
reflectance of red and near infrared of NOAA-AVHRR satellites. Topographical effects are
not considered in radiometric correction of LATUV-NDVI series, nevertheless the band
ratio used in the calculation of NDVI is useful for reducing variations caused by surface
topography (Holben and Justice, 1981). Although the band ratio does not eliminate totally
the problems caused by topography and view-Sun geometry, in low spatial resolution
images (In the case of LAC-AVHRR) the problems are not very serious. Burgess and
others (1995) analysed the topographic effects on AVHRR-NDVI considering mountainous
spaces. They highlighted that in a complex topographic area errors caused by topography
using 1 km NDVI data are approximately 3 %, and they suggest that, for operational
correction of AVHRR NDVI data, it is reasonable to ignore high-order topographic effects
such as sky occlusion and adjacent hill illumination.
Both NDVI series (PAL and LATUV) have a monthly temporal resolution that
records the annual vegetal cycles, very contrasted in the study area because the high
thermal differences between the hot and cold seasons. The objective of this work is to
analyse if trends in vegetation are being recorded in the Pyrenees, and the monthly
temporal resolution can difficult the recognition of trends because the seasonal NDVI
oscillation. To solve this problem we have used only annual data, using the value of annual
integrated area under the monthly NDVI values as measure of vegetation productivity
during the year. Numerous authors have proven that integrated annual curves of NDVI
(NDVI) are highly related to vegetal biomass productivity during the year (Tucker and
others, 1981 and 1983; Tucker and Sellers, 1986; Prince and Tucker, 1986; Prince, 1991).
This method summarises the annual NDVI information and facilitates the identification of
temporal trends, because only one data per year of vegetal productivity is analysed. The
final PAL-NDVI database used in this work contains 13 annual temporal series that reflect
the temporal vegetation dynamics in cells of 8 x 8 km with 19 temporal records everyone.
The average of the 13 series was used to analyse the vegetation trend in the whole of the
study area from 1982 to 2000. LATUV-NDVI database contains 1772 spatial series of 8
temporal records. Every series reflects the dynamic of NDVI annual integrated values in
cells of 1 km2 from 1993 to 2000.
Statistical analysis
Trends in NDVI were identified by means of non-parametric correlation coefficients
(-Spearman) using annual NDVI values and one series of time in years (i.e. in PAL-NDVI
series 1982 was considered as year 1 and 2000 as 19). A non-parametric coefficient was
selected because it is more robust than parametric coefficients and does not make it
necessary to assume the normality of the data series (Lanzante, 1996). The values of 
indicate if there are significant trends in vegetation evolution. Positive and significant
values indicate an increase of the vegetal biomass and negative values indicate a regressive
trend. Trends were considered significant when p<0.1.
An important shortcoming of this method is that non-parametric coefficient only
shows the existence of significant-trends in NDVI series, but not the magnitude of change.
A little but sustained change can result a higher coefficient than a higher but abrupt change.
To determine the areas that have underwent the highest changes in vegetation biomass we
have used regression analysis between the series of time (independent variable) and the
annual NDVI temporal series (dependent variable). The results are several models (i.e. one
for each spatial series of 1 x 1 km of grid size in LATUV-NDVI data series) with the form
y = mx + b. The slope of the models (m) indicates really the evolution of annual NDVI
values, and we can consider that higher slope values coincide with high biomass growth.
Therefore, firstly we have determine the areas that have suffered a positive trend in the
growth of vegetation using non-parametric correlation and secondly we have analysed the
magnitude of change using linear regression.
We analysed whether the recent climate evolution has some influence on the
temporal evolution of NDVI using the annual data of mean temperature and precipitation
by means of the averages of 12 weather stations located within the Aragon river basin.
Trend analysis was applied to climatic data to determine if trend of the same sign of
vegetation is detected, and, after, the relationship between NDVI, temperature and
precipitation was analysed.
Results of the recent vegetal evolution (1993-2000) were related spatially to land
cover and actual land uses for determining the role of human management in the process.
Two digital maps (land-cover and land-use) were generated (scale 1:100.000) using
interpretation of aerial photographs (1990) and fieldwork to identify mainly the abandoned
fields and the areas used by livestock food (Lasanta and Errea, 2001). We used aerial
photographs previous to 1990 to determine with more accuracy the abandoned fields (In
1990 several areas were affected by vegetation colonization, which did not allow to detect
the old fields limits). The abandoned field areas previous to 1957 were considered as landuse of abandoned fields independently of the actual land cover. We did not use the digital
standard land-cover maps (i.e. CORINE) because the categories are not well adapted for
characteristics of land-cover in the study area and it does not reflect the actual land-use of
the territory. Between 1990 and 2000 there are small changes in land management within
the study area and a high stability of land cover, and principally, land-use has been
dominant (Lasanta and others, 2003).
In the land cover map (Figure 2) 10 categories were included: cultivation, pastures,
shrub cover, leafy forests, coniferous forests, mixed forests, bare rock and alpine pastures,
badlands, water and urban areas, but in the following analysis the last three categories were
not considered because their coverage is small. The land use map includes: summer
pastures, pastures used during the rest of the year, yielding forest, abandoned fields,
cultivated areas, low vegetated areas and bare rock. The land-use map reflects the current
use of the territory, but also considers the past land use. We have taken special care in the
map of agricultural spatial evolution (mainly abandonment of fields) and shepherding areas,
since they constitute the principal signal of management change, which should have
important consequences in vegetation cover dynamic.
We analysed the percentage of surface covered by the different land-uses and landcovers affected by significant trends to identify the land-use and land-cover categories in
that vegetation growth is more important. We solved the problem of the different spatial
resolution of land cover and land use maps (1:100,000), and the temporal series obtained
using satellite images, by converting the digital maps into grids of the same size as NOAAAVHRR images. This included assigning to each cell the code of the dominant vegetation
cover or land use in the cell.
Two variance analysis were used to determine whether there are significant
differences in the magnitude of change (slope of the regression models) between the
different land covers and land-uses and to determine the influence of the human
management on the evolution of the vegetal biomass in the study area. Bonferroni contrast
was used to identify the categories between there are differences statistically significant.
RESULTS
Temporal evolution of NDVI from 1982 to 2000 using PAL-NDVI series
The monthly evolution of the NDVI in the study area between 1982 and 2000 is
shown in figure 3, where the annual cycles of vegetation are recorded. There are important
inter-annual differences in maximum and minimum annual peaks. There is a clear
increment in the final years of the 90’s decade, with annual maximum NDVI values higher
than in the 80’s decade. Thus, a temporal increase of monthly NDVI values can be
detected, mainly in the summer months.
Figure 4 shows the temporal evolution of the annual integrated values of NDVI
between 1982 and 2000 in the whole of the study area. There is a positive trend and
statistically significant ( = 0.78, p < 0.01). Thus, we can confirm that a growth in vegetal
biomass is identified in the study area in 1980 and 1990 decades. Nevertheless, the growth
is not continuous, and in some years (such as 1986, 1992 or 1995) a decrease of NDVI
values are detected in relation to the general trend.
Relationships between NDVI evolution and climatic variables (precipitation
and temperatures)
Figure 5 shows the temporal evolution of mean annual temperature and total annual
precipitation in the study area. There is an important increase of mean annual temperature
between 1981 to 2000, from averages about 11 ºC in the beginning of the 1980-decade to
values around 12 ºC en the latter years of the 1990-decade. Trend is positive and significant
in temperature evolution ( = 0.51, p < 0.05), although it is not in precipitation ( = 0.16, p
= 0.50). Thus, only 2 precipitation weather stations of the 12 considered have significant
and positive trends (p < 0.05). While in temperature analysis 6 weather stations have
significant and positive trends.
The evolution of NDVI annual values can be related to these climatic variables,
because the temperature increase can make longer annual vegetative cycles that determine a
higher vegetal productivity. In figure 6 the relationship between mean annual temperature
and average NDVI is shown. There is a high positive relationship between both
parameters (Figure 6). The correlation is positive (r = 0.57) and statistically significant (p <
0.01). Nevertheless, the correlation between annual precipitation and NDVI is nonsignificant (r = -0.078, p = 0.38). Therefore, the annual NDVI is conditioned mainly by
mean temperature values, and the influence of precipitation has less importance. The close
relationship between temperature and NDVI would explain the increment of vegetation
productivity in this area associated to a high increase of temperatures between 1982 and
2000, which implies longer vegetative cycles, more assimilation of atmospherical CO2,
higher evapo-transpiration rates, and softer climatic conditions in a mountain environment
where the environmental constrains usually restrict the vegetation growth.
Spatial distribution of temporal vegetal evolution
Although a general increment of NDVI has been recorded in the study area
between 1982 and 2000 using the average of the PAL 8 km2 series, the spatial differences
in vegetal evolution are very contrasted. These differences have been detected even at PALNDVI resolution. In figure 7 the spatial distribution of slope in analysis of vegetation
evolution is shown. Slope values are positive in all the pixels that represent the vegetation
of the study area. Although the spatial resolution is coarse, we can distinguish that some
areas have undergone a higher vegetation increase recording a non-uniform spatial
increment.
We can not compare these results with other environmental and land-use variables
due to the coarse resolution, and for analysing the factors that can determine the spatial
differentiation of trends we have used the LATUV-NDVI series at 1 km2 of grid cell size.
Comparison of PAL (8 km) and LATUV NDVI (1 km) series for monitoring vegetation
dynamics.
Firstly, we have checked if both (LATUV and PAL) NDVI series are comparable
for analysing trends in the study area. Figure 8 shows the temporal evolution of monthly
NDVI-PAL and LATUV series between 1993 and 2000 using the average of every cell of
both grids. The evolution of curves is temporally similar, and the correlation coefficient
between the series is 0.74 (p < 0.01). The atmospheric correction and calibration process is
similar, and although higher values are recorded in PAL-NDVI data-series in general, the
comparability has not problems.
Spatial differences in NDVI temporal evolution (1993-2000)
Figure 9 shows the areas in which the trend in vegetation increment is statistically
significant. Trend is positive and significant in 414 km2 that represents the 24 % of the
study area. Negative and significant trends are no detected. There are clear spatial patterns
of trend in NDVI. Areas with significant trends are located mainly on spaces with
elevations from 1000 to 1600 m, on high slopes where the main management process is the
agricultural abandonment during the twenty century. These spaces coincide with flysch
lithological area, where agricultural fields were concentrated during traditional
management (until middle twenty century). These areas have been abandoned and natural
vegetal regeneration is the dominant process at present (Lasanta and others, 2000).
Figure 10 shows the spatial distribution of slope values from regression analysis, where
annual NDVI in every cell was the dependent variable and series of time (in years from
1993 to 2000) the independent variable. The picture 1 shows the original values whereas
picture 2 enhances the general spatial patterns using a low pass filter (3 x 3). It is shown
that higher slopes are concentrated in areas with positive and significant trends, in spaces
covered by shrub-lands, mixed and coniferous forest that settle areas of abandoned fields.
Influence of land-uses and land-covers in the spatial differentiation of trends in NDVI.
Table 1 shows the percentage of the different land-covers affected by significant and
non-significant trends in NDVI. The percentage oscillates between 28.7 % in coniferous
forests and 11 % in the bare rock and summer pastures land-cover. Coniferous forests,
mixed forests and shrub-land areas show an important percentage of their surface affected
by significant trends in NDVI values from 1993 to 2000. Therefore, the spatial
distribution of vegetation trends has clear patterns determined by the differences in
vegetation covers. Trend evolution is not independent of the land-covers in the study area
and the present vegetation of the basin affects the NDVI evolution.
More significant are differences in NDVI trends related to present land-uses
(Table 2). The percentage of study area affected by significant trends oscillates between
10.4 % in areas of low vegetation cover and 30.8 % in abandoned fields. Nowadays, areas
with high extensification as abandoned fields and pastures where livestock makes use in
non-summer months represents the highest percentages of area affected by positive and
significant trends. For this reason, the present management practices determine
significantly the rates of vegetation annual growth in the study area, coinciding the landuses of low utilisation with areas of more positive trends. We must highlight that
abandoned fields and pastures in non-summer months represent the 52.8 % of the study
area, and between both land-use categories group the 69.1 % of areas with positive and
significant trends.
Differences in the magnitude of change in NDVI related to land-uses and land-covers
Table 3 shows the results of analysis of variance where it is indicated that
significant differences are recorded in slope of temporal regressions related to different
land-covers.
Figure 11 shows means and confidence intervals of slope values for the different
land-covers. Cultivations, mixed forest and shrub-land areas have the highest mean values
of slopes (0.043, 0.045 and 0.046 respectively). Pastures, leafy forest, and specially areas of
bare rock with alpine pastures have the lowest slopes. This fact indicates that in these areas
the magnitude of vegetation biomass increment has been lower.
Table 4 indicates that significant differences are recorded between slope values for
the different present land-uses.
Figure 11 shows the means and confidence intervals of slope values in relation to
different land-uses. The highest mean values belong to cultivated areas and abandoned
fields (mean = 0.51 in both land-uses), whereas summer pastures, and areas of bare rock
with alpine pastures indicate low slope values and indeed, a slower NDVI increment.
Pastures in non-summer seasons also show high values of slope, whereas yielding forest
and low vegetated areas present a higher variability caused by the low number of cases in
both land-covers (1.78 and 4.33 % of the study area, respectively).
We must highlight that abandoned lands and present cultivations have similar mean
slope values. Nevertheless we have proved in trend analysis that the proportion of
cultivations affected by significant and positive trends is low (16.1 %) in contrast with the
proportion of abandoned fields affected by significant and positive trends (30.8 %). This
fact indicates that although growth in both land uses is similar (equal mean slopes), the
change has been more sustained along the time in abandoned fields, and it is reflected in the
high percentage of these spaces affected by positive trends. Nevertheless, in cultivations the
change has not been progressive and only in the latter years of the 1990 decade a significant
change has been recorded, probably caused by the high temperatures and precipitation
recorded during these years. This fact indicates that abandoned fields are incrementing their
vegetation biomass progressively, more independently of climate conditions, than other
land-uses.
DISCUSSION AND CONCLUSIONS
A positive trend in the temporal evolution of vegetal biomass (using the annual
integrated NDVI: NDVI) has been shown in the Spanish central Pyrenees using nonparametric test and linear regression. Both methods have been widely used in the analysis
of temporal series for determining the signification and magnitude of changes.
The result of the work does not seem to be consequence of inhomogeneities or
artificial trends in the series, which have been previously calibrated. The results obtained in
this work coincide with the evolution observed at global or continental scales by other
authors. Kawabata and others (2001) observe positive and significant trends of NDVI
between 1981 and 1990 in large areas of the world. They found a high relationship between
temperature and NDVI annually (r2 = 0.3) and seasonally (r2 = 0.7) and concluded that
thermal global increment could be the principal cause of positive evolution of vegetal
biomass. Shabanov and others (2000) show the highest increases of NDVI in high latitudes
of north hemisphere, mainly in spring, confirming the climatic cause of trends. Myneni and
others (1997) obtain similar results analysing the 1981-1991 period. They detect a seasonal
amplitude increment of the NDVI curves, around 10%, and an increase of the vegetative
period (12 +/- 4 days). Slayback and others (2003) find significant positive trends in
averaged NDVI for latitude bands above 35ºN, with trends in general higher in the 1990s
than in the 1980s. Lucht and others (2002) obtain similar results, identifying a strong
vegetal increment from 1991, coinciding with the temperature increment at global scale
(Houghton and others, 2001).
In studies of smaller territories, other authors obtain similar trends, but in the
explanation of causes of vegetal evolution introduce the human role. Fuller (1998) analysed
7 years of NDVI composites from 1987 to 1993 in Senegal, and he observed spatial
differences in the temporal trends of NDVI depending on agricultural and pasture
management. Pelkey and others (2000) indicate in Tanzania that the creation of protected
natural areas increases the growth of the vegetation cover and vegetal biomass with respect
to areas with more intensive use.
In the central Spanish Pyrenees we have shown that, in general, there is a significant
and positive trend that can be related to the increment of annual mean temperature. The
close relationship between temperature and NDVI seems confirm it. The results obtained
in the present work coincide with the results obtained globally in areas with important
climatic and environmental limitations, confirming that not only high latitudes are being
affected by vegetation increment. Climatic change processes are also affecting mountainous
areas, where temperature is the main restricted factor for vegetation growth.
Nevertheless, in contrast to high northern latitudes, where there is a higher spatial
homogeneity in vegetation cover and trends (Lucht and others, 2002; Slayback and others,
2003; Myneni and others, 1997) due to vegetation homogeneity (coniferous forests) and
low human management, the growth vegetation rhythm in the mountain area analysed in
this paper is very contrasted in relation to present land-covers and land management
practices.
We must highlight that the results obtained could be affected by the number of landuses and land-cover selected, because this mountainous area is complex and the spatial
diversity is very high. Nevertheless, at the NOAA images spatial resolution (1 km2) the
general land-covers and land-uses are represented in the land-cover and land-uses maps
used. The classes assignment to 1 km2, using the category more represented (in surface) in
the pixel, could introduce some errors, however it is necessary to guarantee the spatial
comparison between NDVI data set and the categorical information. Moreover the results
are spatially consistent and clear NDVI patterns are recorded, which coincide with the
spatial distribution of land-uses and land-covers.
Abandoned fields present the greatest NDVI increment, followed by shrub cover
areas and forests. The smallest increments correspond to areas with scarce or null
vegetation, including the badland areas, alpine pastures and summer pastures. These results
point out the important role of human management in the process of vegetal biomass
evolution. Areas intensively used in the past and nowadays abandoned or used extensively
with low cattle pasture, have the greatest increment of vegetal cover with the most
significant positive trends in NDVI. The management practices and land-covers in the
study area have not experienced any changes between 1990 and 2000 (Lasanta and others,
2000 and 2003). For this reason the changes in NDVI between 1993 and 2000 are mainly
caused by vegetation cover growth or leaf area increment in the different land-cover
categories (shrub-lands, pastures and forests).
Other works (Lasanta and others, 2000; Vicente-Serrano and others, 2002) carried
out in the central Pyrenees considering a longer period (the second half of twenty century),
conclude that in low and medium slopes the revegetation process has been more intense
than in other areas. Agricultural abandonment and shepherding decrement have produce a
gradual recovery of vegetal cover condition previous to intensive uses. Abandoned fields
are covered by herbaceous species (Brachipodium pinnatum, Carex flacca, Galium
lucidum, Sanguisorba minor,...) immediately after abandonment. 3-5 years later the first
shrubs appear (Genista scorpius and Rosa sp). Fields are completely covered with shrubs 25
or 30 years after abandonment. Later on, Crataegus monogyna, Prunus spinosa and
Juniperus communis appear, and 60 years from the time of abandonment the first trees
appear (Pinus sylvestris) (Montserrat, 1990; Molinillo and others, 1997). Trends in vegetal
biomass growth in abandoned field landscapes are very evident.
Forest spaces are also suffering a general biomass increment, although a lower one
with respect to the low slopes of abandoned fields. At present, the forests are scarcely used
(De la Riva, 1997) and there is an advance in the maturity stages. The forest areas with
fewer NDVI increments coincide with woodlands conserved in the traditional farming
system (beech forests and some oak woods). This is true because their localisation was
unfavourable to the installation of cultivation fields and there was a conservation interest
because the community extracted firewood, wood, and livestock pastures in these areas
(Puigdefábregas, 1981). Other forests (coniferous and mixed forest) advance more quickly
in their vegetative stages because they are succession woods that arise after the
abandonment process, i.e., the first decades of the twentieth century (Lasanta, 1988).
Summer pastures present a low vegetal increment. This evolution can be explained
by the greater amount of cattle shepherding in summertime more intensive than in other
spaces of the study area (Vicente-Serrano, 2001). However, the extreme environmental
conditions will limit the succession process and vegetation recovery.
In summary, this paper has proven that in a representative area of the Mediterranean
mountain regions, there is a significant increment of vegetal activity, mostly explained by
temperature growth. However, the intense management transformations during twenty
century can not be neglected, since the temporal evolution is spatially heterogeneous and
the highest vegetation growth is recorded in abandoned areas.
ACKNOWLEDGEMENTS
Data used by the authors in this study include data produced through funding from
the Earth Observing System Pathfinder Program of NASA's Mission to Planet Earth in
cooperation with the National Oceanic and Atmospheric Administration. The data were
provided by the Earth Observing System Data and Information System (EOSDIS),
Distributed Active Archive Center at Goddard Space Flight Center which archives,
manages and distributes this data set. The authors wish to acknowledge financial support
from the following projects: “La recuperación del espacio agrícola como estrategia de
gestión integrada del territorio en área de montaña: El ejemplo de los Altos Valles del
Aragón y del Gállego” (P049/2000), financed by the DGA; “Caracterización espaciotemporal de las sequías en el valle medio del Ebro e identificación de sus impactos”
(BSO2002-02743); “La identificación de fuentes de sedimento y áreas generadoras de
escorrentía en relación con los cambios de uso del suelo” (REN/2000-1709-C04/01/GLO),
and “Efectos erosivos del fuego a lo largo de un gradiente climático. Aportaciones para la
gestión de áreas quemadas” (REN/2002-00133), financed by the CICYT. We would like to
thank to the three anonymous reviewers for their helpful comments.
LITERATURE CITED
Aber, J.D., S.V. Ollinger, C.A. Federer, P.B. Reich, M.L. Goulden, D.W. Kicklighter, J.M.
Melillo, and R.G. Lathrop. 1995. Predicting the effects of climate change on water
yield and forest production in the northeastern United States. Climate Research 5:
207-222.
Bannari, A., D. Morin, F. Bonn and A.R. Huete. 1995. A review of vegetation indices.
Remote Sensing Reviews 13: 95-120.
Baret, F. and G. Guyot. 1991. Potential and limits of vegetation indices for LAI and APAR
assessment. Remote Sensing of Environment 35: 161-173.
Bazin, G., and B. Roux. 1992. Les zones de montagne et défavorisées méditerranéennes
françaises. Situation et évolution récente. Etudes économiques. París.
Braswell, B.M., D.S. Schimel, and E. Linder. 1997. The response of global terrestrial
ecosystems to interannual temperature variability. Science 278: 870-872.
Burgess, D.W., P. Lewis, and J.P.AL. Muller. 1995. Topographic effects in AVHRR NDVI
data. Remote Sensing of Environment 54: 223-232.
Calle, A., J.L. Casanova, A. Romo and F. González-Alonso. 2000. An operational system
for the forest fire risk assessment by means of NOAA images. Pages 363-368 in J.L.
Casanova (ed.) Remote Sensing in the 21st Century: economic and environmental
applications. Balkema. Rotterdam.
Carlson, T.N., and D.A. Ripley. 1997. On the relation between NDVI, fractional vegetation
cover, and leaf area index. Remote Sensing of Environment 62: 241-252.
Carlson, T.N., E.M. Perry, and T.J. Schumugge. 1990. Remote estimation of soil moisture
availability and fractional vegetation cover for agricultural fields. Agricultural and
Forest Meteorology 52: 45-69.
Changnon, S.A., and M. Demissie. 1996. Detection of changes in streamflow and floods
resulting from climate fluctuations and land use-drainage changes. Climatic Change
32: 411-421.
Choudhury, B.J., N.U. Ahmed, S.B. Idso, R.J. Reginato, and C.S.T. Daughtru. 1994.
Relation between evaporation coefficients and vegetation indices studied by model
simulations. Remote Sensing of Environment 50: 1-17.
Cihlar, J., L. St-Laurent, and J.A. Dyer. 1991. Relation between the normalized difference
vegetation index and ecological variables. Remote Sensing of Environment 35: 279298.
De la Riva, J. 1997. Los montes de la Jacetania. Caracterización física y explotación
forestal. Publicaciones del Consejo de Protección de la Naturaleza en Aragón. Serie
Investigación. Zaragoza, 358 pp.
Delgado, J. 1991. Clasificación y análisis de nubes mediante imágenes Meteosat. Ph. thesis.
Universidad de Valladolid. Unpublished.
De Fries, R.S., M.C. Hansen,. and J.R.G. Townshend. 2000. Global continuous fields of
vegetation characteristics: a linear mixture model applied to multiyear 8 km
AVHRR data. International Journal of Remote Sensing 21: 1389-1414.
Diallo, O., A. Diouf, N.P. Hanan, A. Ndiaye, and Y. Prévost. 1991. AVHRR monitoring of
savana primary production in Senegal, West Africa: 1987-1988. International
Journal of Remote Sensing 12: 1259-1279.
Duncan, J., D. Stow, J. Franklin and A. Hope. 1993. Assessing the relationship between
spectral vegetation indices and shrub cover in the Jordana Basin, New Mexico.
International Journal of Remote Sensing 14: 3395-3416.
Farrar, T.J., S.E. Nicholson, and A.R. Lare. 1994. The influence of soil type on the
relationships between NDVI, rainfall and soil moisture in semiarid Botswana II.
NDVI response to soil moisture. Remote Sensing of Environment 50: 121-133.
Ferrer, C. 1988. Los recursos pascícolas del Pirineo aragonés. Pages 23-65 in Actas de la
XXVIII Reunión Científica de la Sociedad Española para el Estudio de los Pastos.
SEEP. Jaca.
Fitter, A.H., and R.S.R. Fitter. 2002. Rapid Changes in Flowering Time in British Plants.
Science 296: 1689-1691.
Florinsky, I.V., and G.A. Kuryakova. 1996. Influence of topography on some vegetation
cover properties. Catena 27: 123-141.
Fuller, D.O. 1998. Trends in NDVI time series and their relation to rangeland and crop
production in Senegal, 1987-1993. International Journal of Remote Sensing 19:
2013-2018.
García-Ruiz, J.M. 1988. La evolución reciente de la agricultura de montaña y sus efectos
sobre la dinámica del paisaje. Revista de Estudios Agrosociales 146: 7-37.
García-Ruiz, J.M. 1990. Geoecología de las áreas de montaña. Geoforma Ediciones.
Logroño. 337 pp.
García-Ruiz, J.M., and T. Lasanta. 1990. Land-use changes in the Spanish Pyrenees.
Mountain Research and Development 10: 267-279.
García-Ruiz, J.M., and T. Lasanta. 1993. Land-use conflicts as result of land-use change in
the Central Spanish Pyrenees: a review. Mountain Research and Development 13:
295-304.
García-Ruiz, J.M., and J. Puigdefábregas. 1982. Formas de erosión en el flysch eoceno
surpirenaico. Cuadernos de Investigación Geográfica 8: 85-124.
García-Ruiz, J.M., and B. Valero. 1998. Historical geomorphic processes and human
activities in the Central Spanish Pyrenees. Mountain Research and Development 18:
309-320.
García-Ruiz, J.M., J. Puigdefábregas, and J. Creus. 1985. Los recursos hídricos
superficiales del Alto Aragón. Colección de Estudios Altoaragoneses. Huesca. 224
pp.
García-Ruiz, J.M., T. Lasanta, L. Ortigosa, P. Ruis Flaño, C. Martí, and C. González. 1995.
Sediment yield under different land-uses in the Spanish Pyrenees. Mountain
Research and Development 15: 229-240.
Gillies, R.R., T.N. Carlson, J. Cui, W.P. Kustas and K.S. Humes. 1997, A verification of
the triangle method for obtaining surface soil water content and energy fluxes from
remote measurements of the Normalized Difference Vegetation Index (NDVI) and
surface radiant temperature. International Journal of Remote Sensing 18: 31453166.
González-Alonso, F., J.M. Cuevas, J. Casanova, A. Calle and P. Illera. 1995. Drought
monitoring in Spain using satellite remote sensing. Pages 88-90 in P. Askne (ed.)
Sensors and environmental applications of remote sensing. Balkema. Rotterdam.
González-Alonso, F., A. Vázquez, J.M. Cuevas, A. Calle and J.L. Casanova. 2000.
Towards an operational monitoring of drought conditions in Spain using satellite
images. Pages 403-405 in J.L. Casanova (ed.) Remote Sensing in the 21st Century:
economic and environmental applications. Balkema. Rotterdam.
González-Alonso, F., A. Calle, A. Vázquez, J.L. Casanova, J.M. Cuevas and A. Romo.
2001. Seguimiento de la sequía en España, en el año 2000, mediante técnicas de
teledetección espacial. Pages 83-85 in Rosell, J.I. and Martínez-Casasnovas (eds.)
Teledetección, medioambiente y cambio global. Universidad de Lleida. Lleida.
Goward, S.N., B. Markham, D.G. Dye, W. Dulaney, and J. Yang. 1991. Normalized
difference vegetation index measurements from the advanced very high resolution
radiometer. Remote Sensing of Environment 35: 257-277.
Gutman, G. 1991. Vegetation indices from AVHRR: an update and future prospects.
Remote Sensing of Environment 35: 121-136.
Gutman, G., D. Tarpley, A. Ignatov, and S. Olson. 1995. The enhanced NOAA global land
dataset from the advanced very high resolution radiometer. Bulletin of the American
Meteorological Society 76: 1141-1156.
Hester, A.J., D.R. Miller and W. Towers. 1996. Landscape scale vegetation change in the
Cairngorms, Scotland, 1946-1988: implications for land management. Biological
Conservation 77: 41-51.
Holben, B.N. 1986. Characteristics of maximum-value composite images from temporal
AVHRR data. International Journal of Remote Sensing 7: 1417-1434.
Holben, B.N. and C.O. Justice. 1981. An examination of spectral band rationing to reduce
the topographic effect on remotely sensed data. International Journal of Remote
Sensing 2: 115-133.
Houghton, J.T., Y. Ding, D. Giggs, M. Noguet, P. Van del Linden, X. Dai, A. Maskell, and
C.A. Johnson. 2001. Climate Change 2001: The scientific Basis. Cambridge
University Press. Cambridge.
Huete, A.R. 1988. A soil adjusted vegetation index (SAVI). Remote Sensing of
Environment 25: 295-309.
Hutges, R.W.A., P. Kabat, S.W. Running, W.J. Shuttleworth, and others. 1998. Biospheric
Aspects of the Hydrological Cycle. Journal of Hydrology 212-213: 1-21.
Illera, P., J. Delgado and A. Calle. 1996a. A navigation algorithm for satellite images.
International Journal of Remote Sensing 15: 577-588.
Illera, P., A. Fernández and J. Delgado. 1996b. Temporal evolution of the NDVI as an
indicator of forest fire danger. International Journal of Remote Sensing. 17: 10931105.
Illera, P., J. Delgado, A. Fernández-Unzueta and A.A. Fernández Manso. 2000. Integration
of NOAA-AVHRR and meteorological data in a GIS-application for vegetation
monitoring in Castilla y León, Spain. Pages 47-54 in J.L. Casanova (ed.) Remote
Sensing in the 21st Century: economic and environmental applications. Balkema.
Rotterdam.
James, M.E. and S.N.V. Kalluri. 1994. The Pathfinder AVHRR land data set: an improved
coarse resolution data set for terrestrial monitoring. International Journal of Remote
sensing 15: 3347-3363.
Justice, C.O. and J.R. Townshend. 1994. Data sets for global remote sensing: lessons learnt.
International Journal of Remote Sensing 15: 3621-3639.
Kammerbauer, J., and C. Ardon. 1999. Land use dynamics and landscape change pattern in
a typical watershed in the hillside region of Central Honduras. Agriculture,
Ecosystems and Environment 75: 1-2
Kaufmann, R.K., L. Zhou, N.V. Shabanov, R.B. Myneni and C.J. Tucker. 2000. Effect of
orbital drift and sensor changes on the time series of AVHRR vegetation index data.
IEEE Transactions in Geoscience and Remote Sensing 38: 2584-2597.
Kawabata, A., K. Ichii, and Y. Yamaguchi. 2001. Global monitoring of interannual changes
in vegetation activities using NDVI and its relationships to temperature and
precipitation. International Journal of Remote Sensing 22: 1377-1382.
Kergoat, L. 1998. A model for hydrological equilibrium of leaf area index on a global
scale. Journal of Hydrology 212-213: 268-286.
Knipling, E.B. 1970. Physical and physiological basis for the reflectance of visible and near
infrared radiation from vegetation. Remote Sensing of Environment 1: 155-159.
Kogan, F.N., and X. Zhu. 2001. Evolution of long-term errors in NDVI time series: 19851999. Advances in Space Research 28: 149-153
Kok, K., B.M.W. Clavaux, W.M. Heerebout, and K. Bronsveld. 1995. Land degradation
and land cover change detection using low-resolution satellite images and the
CORINE database: a case study in Spain. ITC Journal 1995-3: 217 –227.
Lanzante, J.R. 1996. Resistant, robust and non-parametric techniques for the analysis of
climate data: theory and examples including applications to historical radiosonde
station data. International Journal of Climatology 16: 1197-1226.
Lasanta, T. 1988. The process of desertion of cultivated areas in the Central Spanish
Pyrenees. Pirineos 132: 15-36.
Lasanta, T. 1989. Evolución reciente de la agricultura de montaña. Geoforma Edición.
Logroño. 220 pp.
Lasanta, T. 1990. Tendances actuelles de l’organisation spatiale des montagnes espagnoles.
Annales de Géographie 551: 51-71.
Lasanta, T., and M.P. Errea. 2001. Usos agrarios del suelo. Pages 43-46 in El medio físico
y su peligrosidad en un sector del Pirineo Central. Ministerio de Ciencia y
Tecnología and IGME. Zaragoza.
Lasanta, T., S.M. Vicente Serrano, and J.M. Cuadrat. 2000. Marginación productiva y
recuperación de la cubierta vegetal en el Pirineo: un caso de estudio en el valle de
Borau. Boletín de la AGE 29: 5-28.
Lasanta, T., S.M. Vicente Serrano, and J.M. Cuadrat. 2003. Mountain mediterranean
landscape evolution caused by the abandonment of traditional primary activities: a
study of the Spanish central Pyrenees. Submitted to Applied Geography
Lucht, W., I.C. Prentice, R.B. Myneni, S. Sitch, P. Friedlingstein, W. Cramer, P. Bousquet,
W. Buermann, and B. Smith. 2002. Climatic control of the high-latitude vegetation
greening trend and Pinatubo effect. Science 296: 1687-1689.
MacDonald, D., J.R. Crabtree, G. Wiesinger, T. Dax, N. Stamou, P. Fleury, J. Gutiérrez
Lazpita, and A. Gibom. 2000. Agricultural abandonment in mountain areas of
Europe:
Environmental
consequences
and
policy
response.
Journal
of
Environmental Management 59: 47-69.
Molinillo, M., T. Lasanta, and J.M. García-Ruiz. 1997. Managing mountainous degraded
landscapes after farmland abandonment in the Central Spanish Pyrenees.
Environmental Management 21: 587-598.
Montserrat, G. 1990. Estudio de la colonización vegetal en campos abandonados del valle
de Aísa (Jaca, Huesca). Informe del Proyecto LUCDEME: Erosión y colonización
vegetal en campos abandonados, Jaca. 77 pp.
Montserrat, J. 1992. Evolución glaciar y postglaciar del clima y la vegetación en la
vertiente sur del Pirineo: Estudio palinológico. Instituto Pirenaico de Ecología. Jaca
147 pp.
Myneni, R.B., F.G. Hall, P.J. Sellers and A.L. Marshak. 1995. The interpretation of spectral
vegetation indexes. IEEE Transactions on Geoscience and Remote Sensing. 33:
481-486.
Myneni, R.B., C.D. Keeling, C.J. Tucker, G. Asrar, and R.R. Nemani. 1997. Increase plant
growth in the northern high latitudes from 1981-1991. Nature 386: 698-702.
Myneni, R.B., C.J. Tucker, G. Asrar, and C.D. Keeling. 1998. Interannual variations in
satellite-sensed vegetation index data from 1981 to 1991. Journal of Geophysical
Research 103: 6145-6160.
NASA. 2003. PAL Daily Tile Readme.
http://daac.gsfc.nasa.gov/campaign_docs/land_bio/glbdst_data.html
Olsson E.G.A., G. Austrheim, and S.N. Grenne. 2000. Landscape change patterns in
mountains, land use and environmental diversity, Mid-Norway 1960-1993.
Landscape Ecology 15: 155-177.
Pelkey, N.W., C.J. Stoner, and T.M. Caro. 2000. Vegetation in Tanzania: assessing long
term trends and effects of protection using satellite imaginery. Biological
Conservation 94: 297-309.
Peñuelas, J., and I. Filellas. 2001. Responses to a Warming World. Science 294: 793-795.
Price, J.C. 1991. Timing of NOAA afternoon passes. International Journal of Remote
Sensing 12: 193-198.
Prince, S.D. 1991. Satellite remote sensing of primary production: comparison of results for
sahelian grasslands 1981-1988. International Journal of Remote Sensing 12: 13011311.
Prince, S.D. and C.J. Tucker. 1986. Satellite remote sensing of rangelands in Botswana II:
NOAA-AVHRR and herbaceous vegetation. International Journal of Remote
Sensing 7: 1555-1570.
Puigdefábregas, J. 1981. El pinar altoaragonés de Pinus sylvestris. Estructura y producción
primaria neta. Ph thesis. University of Navarra. 757 pp. Unpublished.
Puigdefábregas, J., and F. Fillat. 1986. Ecological adaptation of traditional land-uses in the
Spanish Pyrenees. Mountain Research and Development 6: 63-72.
Rao, C.R.N., and J. Chen. 1995. Inter-satellite calibration linkages for the visible and nearinfrared channels of the Advanced Very High Resolution Radiometer on the
NOAA-7, -9 and -11 spacecraft. International Journal of Remote Sensing 16: 19311942.
Rao, C.R.N., and J. Chen. 1999. Revised post-launch calibration of the visible and nearinfrared channels of the Advanced Very High Resolution Radiometer on the
NOAA-14 spacecraft. International Journal of Remote Sensing 20: 3485-3491.
Riebsame, W.E., W.B. Meyer, and B.L. Turner. 1994. Modelling land use and cover as part
of global environmental change. Climatic Change 28: 1-10.
Salinas-Zavala, C.A., A.V. Douglas, and H.F. Díaz. 2002. Interannual variability of NDVI
in Northwest Mexico. Associated climatic mechanisms and ecological implications.
Remote Sensing of Environment 82: 417-430.
Shabanov, N.V., L. Zhou, Y. Knyazikhin, R.B. Myneni and C.J. Tucker. 2002. Analysis of
interannual changes in northern vegetation activity observed in AVHRR data from
1981 to 1994. IEEE Transactions on Geoscience and Remote Rensing 40: 115-130.
Slayback, D.A., J.E. Pinzon, S.O. Los and C.J. Tucker. 2003. Northern hemisphere
photosynthetic trends 1982-99. Global Change Biology 9: 1-15.
Smith, P.M., S.N.V. Kalluri, S.D. Prince and R.S. DeFries. 1997. The NOAA/NASA
pathfinder AVHRR 8-km land data set. Photogrammetric Engineering and Remote
Sensing 63: 12-31.
Soler-Sampere, M., and C. Puigdefábregas. 1972. Esquema litológico del Alto Aragón
Occidental. Pirineos 106: 5-15.
Staylor, W.F. 1990. Degradation rates of the AVHRR visible channel from the NOAA-6, 7, and –9 spacecraft. Journal of Atmospheric and Oceanic Technology 7: 411-423.
Stohlgren, T.J., T.N. Chase, R.A. Pielke, T.G.F. Kittel, and J.S. Baron. 1998. Evidence that
local land use practices influence regional climate, vegetation and stream flow
patterns in adjacent natural areas. Global Change Biology 4: 495-504.
Tucker, C.J. 1979. Red and photographic infrared linear combinations for monitoring
vegetation. Remote Sensing of Environment 8: 127-150.
Tucker, C.J. and P. Sellers. 1986. Satellite remote Sensing of primary production.
International Journal of Remote Sensing 7: 1395-1416.
Tucker, C.J., B.N. Holben, J.H. Elgin and J.E. McMurtrey. 1981. Remote Sensing of total
dry matter accumulation in winter wheat. Remote Sensing of Environment 11: 171189.
Tucker, C.J., C. Vanpraet, E. Boerwinkel and A. Gaston. 1983. Satellite remote sensing of
total dry matter production in the senegalese sahel Remote Sensing of Environment.
13: 461-474.
Van Wijngaarden, W. 1991. The green cover of the Earth: a dynamic resource in a
changing environment. ITC Journal 1991-3: 113-121.
Vázquez, A., F. González-Alonso, A. Calle, J.M. Cuevas, J.L. Casanova. 2001. Producción
asegurada de cereales y anomalías en el índice de vegetación en España peninsular:
1995-1998. Pages 37-40 in Rosell, J.I. and Martínez-Casasnovas J.A. (eds.)
Teledetección, medioambiente y cambio global. Universidad de Lleida. Lleida.
Vicente-Serrano, S.M. 2001. El papel reciente de la ganadería extensiva de montaña en la
dinámica del paisaje y en el desarrollo sostenible. El ejemplo del Valle de Borau.
Publicaciones del Consejo de Protección de la Naturaleza de Aragón. Zaragoza. 181
pp.
Vicente-Serrano, S.M., T. Lasanta, and J.M. Cuadrat. 2002. Diferencias espaciales en el
proceso de revegetación: influencia de los factores ambientales y de la gestión en el
Pirineo central. Frankfurter Geowissenschaftliche Arbeiten – Serie D: Physische
Geographie. In press.
Wylie, B.K., D.J. Meyer, L.L. Tieszen, and S. Mannel. 2002. Satellite mapping of surface
biophysical parameters at the biome scale over the North American grassland. A
case study. Remote Sensing of Environment 79: 266-278.
Zeng, J., D. Neelin, K.M. Lan, and C.J. Tucker. 1999. Enhancement of interdecadal climate
variability in the Sahel by vegetation interaction. Science 286: 1537-1540.
CAPTIONS FOR ALL TABLES
Table 1- Surface percentage in different land-covers affected by significant and non-significant trends in
NDVI evolution (1993-2000).
Table 2- Percentage of surface in different land-uses affected by significant and non-significant trends in the
NDVI evolution (1993-2000).
Table 3- ANOVA analysis. Independent variable is the slope of regression between NDVI values and time
series in years. Categorical factor is land-cover.
Table 4- ANOVA analysis. Independent variable is the slope of regression between NDVI values and
temporal series of years. Categorical factor is land-use.
Land-covers
Cultivations
Non-significant Significant Total
87.9
12.1
100
78
22.1
100
Leafy forest
80.4
19.6
100
Coniferous forest
71.3
28.7
100
Mixed forest
74
26
100
Shrub-lands
76.2
23.8
100
89
11
100
Pastures
Bare rock and summer pastures
Table 1- Surface percentage in different land covers affected by significant and non-significant trends in
NDVI evolution (1993-2000).
Land uses
Non-significant Significant Total
Summer pastures
84.26
15.74
100
Pastures used during the rest of the year
75.32
24.68
100
Yielding forest
80.15
19.85
100
Abandoned fields
69.20
30.80
100
Cultivated areas,
83.80
16.20
100
Low vegetated areas
89.56
10.44
100
Bare rock and summer pastures
89.08
10.92
100
Table 2- Percentage of surface in different land-uses affected by significant and non-significant trends in the
NDVI evolution (1993-2000).
Source
Between groups
Sum of Squares Df
0.076
Mean Square F-Ratio P-Value
6
0.012
Within groups
1.38 1765
0.0008
Total (Corr.)
1.46 1771
15.89
<.001
Table 3- ANOVA analysis. Independent variable is the slope of regression between NDVI
values and time series in years. Categorical factor is land-cover.
Source
Sum of Squares Df
Mean Square F-Ratio P-Value
Between groups
0.129
6
0.02
Within groups
0.923 1765
0.00052
Total (Corr.)
1.052 1771
26.87
<0.01
Table 4- ANOVA analysis. Independent variable is the slope of regression between NDVI values and
temporal series of years. Categorical factor is land-use.
CAPTIONS FOR ALL FIGURES
Figure 1: Study area
Figure 2- Land cover classes in the study area (1:100.000)
Figure 3- Monthly evolution of NDVI in the study area from NDVI-PAL data serie (1981-2000). Series
indicates the average of the pixels in the study area.
Figure 4- Annual integrated NDVI evolution from 1982 to 2000 in the whole study area (average of the
partial spatial series).
Figure 5- Annual mean temperature and total annual precipitation evolution (1982-2000) in the high Aragon
river basin.
Figure 6- Relationship between annual integrated NDVI and annual mean temperature (1982-2000)
Figure 7- Spatial differences of slope in regression analysis, using annual integrated PAL-NDVI series from
1982 to 2000 as dependent variable, and time series as dependent variable.
Figure 8 – Relationship between PAL and LATUV NDVI temporal data series. Both series are made using
the average of every pixels in the study area.
Figure 9- Spatial distribution of significant trends in NDVI data at 1 km2 grid cell size. LATUV-NDVI
series (1993-2000).
Figure 10- Spatial distribution of slopes from regression analysis, where annual integrated NDVI is the
dependent variable and time series independent variable (1993-2000).
Figure 11- Means and confidence intervals of slope in function of the different land covers. 1. Cultivations, 2.
Pastures, 3. Leafy forest, 4: Coniferous forest, 5. Mixed forest, 6. Shrublands, 7. Bare rock and alpine
pastures.
Figure 12- Means and confidence intervals of slope in function of land-uses: 1- Summer pastures, 2-Pastures
used during the rest of the year, 3-Yielding forest, 4-Abandoned fields, 5-Cultivated areas, 6-Low vegetated
areas, 7-bare rock and summer pastures.
Figure 1: Study area
Land cover:
Cultivations
Pastures
Shrub cover
Leafy forest
Coniferous forest
Mixed forest
Bare rock
Badlands
Water (reservoirs)
Urban areas
Figure 2- Land cover classes in the study area (1:100.000)
0
.
8
0
.
7
0
.
6
NDVI
0
.
5
0
.
4
0
.
3
0
.
2
0
.
1
1
9
8
1
1
9
8
2
1
9
8
3
1
9
8
4
1
9
8
5
1
9
8
6
1
9
8
7
1
9
8
8
1
9
8
9
1
9
9
0
1
9
9
1
1
9
9
2
1
9
9
3
1
9
9
4
1
9
9
5
1
9
9
6
1
9
9
7
1
9
9
8
1
9
9
9
2
0
0
0
Y
E
A
R
S
Figure 3- Monthly evolution of NDVI in the study area from NDVI-PAL data serie (1981-2000). Series
indicates the average of the pixels in the study area.
7
.
5
7
.
0
 NDVI
6
.
5
6
.
0
5
.
5
5
.
0
1
9
8
2
1
9
8
4
1
9
8
6
1
9
8
8
1
9
9
0
1
9
9
2
1
9
9
4
1
9
9
6
1
9
9
8
2
0
0
0
Y
E
A
R
S
Figure 4- Annual integrated NDVI evolution from 1982 to 2000 in the whole study area (average of the
partial spatial series).
1
3
TEMPERAATNURUEA(LºCM)EAN
1
2
1
1
1
0
1
9
8
2
1
9
8
4
1
9
8
6
1
9
8
8
1
9
9
0
1
9
9
2
1
9
9
4
1
9
9
6
1
9
9
8
2
0
0
0
1
9
9
4
1
9
9
6
1
9
9
8
2
0
0
0
Y
E
A
R
S
1
3
0
0
1
2
0
0
1
1
0
0
PRECIPTAION(m) ANUAL
1
0
0
0
9
0
0
8
0
0
7
0
0
6
0
0
1
9
8
2
1
9
8
4
1
9
8
6
1
9
8
8
1
9
9
0
1
9
9
2
Y
E
A
R
S
Figure 5- Annual mean temperature and total annual precipitation evolution (1982-2000) in the high Aragon
river basin.
1
3
.
0
1
2
.
5
1
2
.
0
TEMPERAMTUEARNE(AºNC)UAL
1
1
.
5
1
1
.
0
1
0
.
5
1
0
.
0
5
.
0
5
.
5
6
.
0
6
.
5
7
.
0
7
.
5


N
D
V
I
Figure 6- Relationship between annual integrated NDVI and annual mean temperature
(1982-2000)
Figure 7- Spatial differences of slope in regression analysis, using annual integrated PALNDVI series from 1982 to 2000 as dependent variable, and time series as dependent
variable.
0
.
8
0
.
8
N
D
V
I
(
P
A
L
)
0
.
7
0
.
7
0
.
6
0
.
5
0
.
5
NDVI(LATUV)
NDVI
0
.
6
0
.
4
0
.
3
0
.
4
N
D
V
I
(
L
A
T
U
V
)
0
.
2
1
9
9
3 1
9
9
4 1
9
9
5 1
9
9
6 1
9
9
7 1
9
9
8 1
9
9
9 2
0
0
0
Y
E
A
R
S
0
.
3
0
.
2
0
.
20
.
30
.
40
.
50
.
60
.
70
.
8
N
D
V
I
(
P
A
L
)
Figure 8 – Relationship between PAL and LATUV NDVI temporal data series. Both series
are made using the average of every pixels in the study area.
Figure 9- Spatial distribution of significant trends in NDVI data at 1 km2 grid cell size. LATUV-NDVI
series (1993-2000).
Figure 10- Spatial distribution of slopes from regression analysis, where annual
integrated NDVI is the dependent variable and time series independent variable (19932000).
0.05
SLOPE
0.04
0.03
0.02
1
2
3
4
5
6
7
LAND COVERS
Figure 11- Means and confidence intervals of slope in function of the different land
covers. 1. Cultivations, 2. Pastures, 3. Leafy forest, 4: Coniferous forest, 5. Mixed forest,
6. Shrublands, 7. Bare rock and alpine pastures.
0.06
SLOPE
0.05
0.04
0.03
0.02
1
2
3
4
5
6
7
LAND USES
Figure 12- Means and confidence intervals of slope in function of land-uses: 1- Summer pastures, 2-Pastures
used during the rest of the year, 3-Yielding forest, 4-Abandoned fields, 5-Cultivated areas, 6-Low vegetated
areas, 7-bare rock and summer pastures.