AN ABSTRACT OF THE THESIS OF Michael C. Roberts for the degree of Doctor of Philosophy in Crop and Soil Science Title: presented on July 16, 1991 . Field Sampling and Mapping Strategies for Balancing Nitrogen to Variable Soil Water Across Landscapes. Redacted for Privacy Abstract approved:_ F.E. Bolton Farmers know that wheat yields vary across landscapes in accordance with soil depth and water storage. This study was conducted to (i) quantify the correlation between yield and water storage as affected by landscape position, (ii) develop soil, water, and yield sampling strategies acceptable to farmers, and (iii) map soil management units. Two study areas were selected within the Columbia Plateau winter wheat region. Soil nitrogen, available water, and grain yield patterns were estimated by the use of reconnaissance transects across landscapes. Grain yields and crop *** available water were positively correlated (r2 = 0.96 ). Data suggest that gently rolling fields (slopes < 12%) can be sampled on a square 158 meter grid (variance estimate of 0.5 Mg ha-1). Sharply rolling fields with more variation down slope than across slope (anisotrophy) can be sampled with a rectangular grid; transects should traverse slopes at 70 to 158 meter spacing (variance est. 0.5 Mg ha-1), and be sampled at 40 meter intervals (variance est. 0.25 Mg ha 1 ). This arrangement will ensure that each landscape position on the transect is sampled. Landscape units mapped from aerial photographs and digital terrain models (tlree dimensional surface models of field topography) were analyzed with geographic information systems (GIS) to generate soil management maps that correlate yield with variable soil water storage across landscape positions. Using soil management maps to vary nitrogen application adjusted to available water levels across landscapes will optimize grain protein content, protect surface and groundwater, and decrease production costs. Field Sampling and Mapping Strategies for Balancing Nitrogen to Variable Soil Water Across Landscapes. by Michael C. Roberts A THESIS submitted to Oregon State University in partial fulfillment of the requirements for the degree of Doctor of Philosophy Completed July 16, 1991 Commencement June 1992 Approved: Redacted for Privacy Professor Emeritus of Crop Science in charge of major Redacted for Privacy / f C/ Head of Department of Crop and Soil Science Redacted for Privacy Dean of Gra School Date thesis is presented July 16, 1991 Typed by Michael C. Roberts Copyright by Michael C. Roberts March 16, 1991 All Rights Reserved Acknowledgements I would like to express gratitude to my wife, Darlene, and my children, Bryce, Brooke, Breahna, Brandon and Brett for their patience and faith with me. I would like to thank my family, friends, and fellow graduate students for their support and encouragement with my thesis. I would like to thank Dr. Floyd E. Bolton for support and assistance with this thesis. ii TABLE OF CONTENTS INTRODUCTION 1 LITERATURE REVIEW 4 Identifying Yield Variability 4 Soil Water and Fertility Remote-sensing 4 4 Assessing Spatial Variability 12 Classical Statistics Geostatistics 12 13 Variable Field Mapping 20 Geographic Information Systems Digital Terrain Models Basic Cartography 20 22 23 CHAPTER I. FERTILIZING WHEAT WITH NITROGEN BALANCED TO SOIL WATER VARIABILITY ACROSS LANDSCAPES 24 . . . ABSTRACT 24 INTRODUCTION 25 MATERIAL AND METHODS 28 The Pendleton study area The Wasco study area Sampling methods Analyses 29 31 32 34 RESULTS AND DISCUSSION 36 REFERENCES 43 iii TABLE OF CONTENTS CONT. CHAPTER II. FIELD SAMPLING STRATEGIES BALANCING COST AND PRECISION 45 ABSTRACT 45 INTRODUCTION 46 MATERIALS AND METHODS 47 RESULTS AND DISCUSSION 48 REFERENCES 54 CHAPTER III. MAPPING STRATEGIES TO REPRESENT AND MANAGE YIELD VARIATIONS ACROSS LANDSCAPES . . 55 ABSTRACT 55 INTRODUCTION 56 MATERIALS AND METHODS 57 Aerial Photography Digital Elevation Models Soil Management Map Units RESULTS AND DISCUSSION Digital Elevation Model Accuracy Reconnaissance Sampling of Landscape Positions Creation of Soil Management Maps REFERENCES 57 58 59 59 59 60 61 67 CONCLUSIONS 69 BIBLIOGRAPHY 71 APPENDIX 77 iv LIST OF FIGURES Figure I.1. Schematic representation of landscape positions (A), and grain yield (B) along transect 2, Wasco study area. 40 Figure 1.2. Wheat grain yield correlated to crop available water (A),ula.total nitrate (B) at Wasco study area. indicate significance at the 0.05, 0.01, and 0.001 probability levels, respectively. 41 Figure 1.3. Crop available water correlated to soil profile depth (A); wheat grain yield correlated to soil profile depth (B), and organic matter (C), Wasco study area, transect 2. 42 Figure II.1. Location of grain yield sampling sites along transects, Pendleton study area 51 Figure 11.2. Grain yield semivariogram, four transects, Pendleton study area 52 Figure 11.3. Grain yield semivariograms indicating anisotrophy; across slope (A), and down slope (B) 53 Figure III.1 Digital elevation model for the Pendleton study area derived from the Barnhart USGS 7.5 Minute Quadrangle. View direction is 310°, 30° above horizon, vertical exaggeration 1:1 (no vertical exaggeration). 63 Figure 111.2 Digital elevation model for the Wasco study area derived from the Wasco USGS 7.5 Minute Quadrangle. View direction is 110°, 30° above horizon, vertical exaggeration 0.7:1. 64 Figure 111.3 Soil management map units for the Pendleton study area. Legend, table III.1. 65 ' Figure 111.4 Soil management map units for the Wasco study area. Legend, table III.1. . 66 V LIST OF TABLES Precipitation for fallow-crop cycle near Table I.1. Wasco and Pendleton study areas. 39 Table III.1 Soil management map unit legend with family or higher taxonomic class. 62 vi LIST OF APPENDIX FIGURES Figure A.1. Wasco (Moore) study area transects with sample locations plotted. 78 Figure A.2. Pendleton (Timmerman) study area transects with sample locations plotted. 79 Figure A.3. Wasco (Moore) study area transect 2, soil profile characteristics. 80 Figure A.4. Wasco (Moore) study area transect 2, core 1, soil water (A), and nitrate (B), profile data. 81 Figure A.S. Wasco (Moore) study area transect 2, core 2, soil water (A), and nitrate (B), profile data. 82 Figure A.6. Wasco (Moore) study area transect 2, core 3, soil water (A), and nitrate (B), profile data. 83 Figure A.7. Wasco (Moore) study area transect 2, core 4.5, soil water (A), and nitrate (B), profile data. 84 Figure A.8. Wasco (Moore) study area transect 2, core 6, soil water (A), and nitrate (B), profile data. 85 Figure A.9. Wasco (Moore) study area transect 2, core 7, soil water (A), and nitrate (B), profile data. 86 Figure A.10. Wasco (Moore) study area transect 2, core 8, soil water (A), and nitrate (B), profile data. 87 Figure A.11. Wasco (Moore) study area transect 2, core 9, soil water (A), and nitrate (B), profile data. 88 Figure A.12. Pendleton study area (Timmerman) transect 1, soil profile characteristics. 89 Figure A.13. Pendleton study area (Timmerman) transect 1, core 1, soil water (A), and nitrate (B), profile data. 90 vii LIST OF APPENDIX FIGURES CONT. Pendleton study area (Timmerman) Figure A.14. transect 1, core 2, soil water (A), and nitrate (B), profile data. 91 Pendleton study area (Timmerman) Figure A.15. transect 1, core 3, soil water (A), and nitrate (B), profile data. 92 Pendleton study area (Timmerman) Figure A.16. transect 1, core 5, soil water (A), and nitrate (B), profile data. 93 Pendleton study area (Timmerman) Figure A.17. transect 1, core 6, soil water (A), and nitrate (B), profile data. 94 Pendleton study area (Timmerman) Figure A.18. transect 1, core 7, soil water (A), and nitrate (B), profile data. 95 Pendleton study area (Timmerman) Figure A.19. transect 1, core 8, soil water (A), and nitrate (B), profile data. 96 Pendleton study area (Timmerman) Figure A.20. transect 1, core 9, soil water (A), and nitrate (B), profile data. 97 Pendleton study area (Timmerman) Figure A.21. transect 1, core 10, soil water (A), and nitrate (B), profile data. 98 Pendleton study area (Timmerman) Figure A.22. transect 2, soil profile characteristics. 99 Figure A.23. Pendleton study area (Timmerman) transect 2, core 1, soil water (A), and nitrate (B), profile data. 100 Pendleton study area (Timmerman) Figure A.24. transect 2, core 2, soil water (A), and nitrate (B), profile data. 101 Figure A.25. Pendleton study area (Timmerman) transect 2, core 3, soil water (A), and nitrate (B), profile data. 102 viii LIST OF APPENDIX FIGURES CONT. Figure A.26. Pendleton study area (Timmerman) transect 2, core 4, soil water (A), and nitrate (B), profile data. 103 Pendleton study area (Timmerman) Figure A.27. transect 2, core 5, soil water (A), and nitrate (B), profile data. 104 Figure A.28. Pendleton study area (Timmerman) transect 2, core 6, soil water (A), and nitrate (B), profile data. 105 FIELD SAMPLING AND MAPPING STRATEGIES FOR BALANCING NITROGEN TO VARIABLE SOIL WATER ACROSS LANDSCAPES INTRODUCTION Some areas of the world are currently faced with the challenges of food overproduction, while others face critical shortages. While political and distribution problems may be beyond the control of researchers, a strong awareness of the need to increase food production exists. Increased food production will require optimal management of the arable land resources of the world. The amount of land projected to be brought under cultivation will probably only balance the land removed from agricultural production because of urbanization, increases in salinization of irrigated land, desertification, and the depletion of groundwater aquifers by irrigated agriculture. The production of food and fiber will be increasingly relegated to marginal lands. These marginal -- as well as some prime -- lands, are sometimes quite variable in production potential within a field or farm. As increases in efficiency of subsistence agriculture involves optimal management of limited resources, plantation agriculture as well could optimally manage soil resources on a soil basis within a field as opposed to treating fields as a uniform unit. This approach to within-field management would conserve 2 resources, reduce the potential of non-point source pollution of streams by agricultural chemicals, and increase production efficiency. Within a few years all satellites comprising the global positioning system (GPS) will be in orbit. Ground vehicles using real-time, differential GPS systems can now be guided to within 1.5 meters of the intended target or travel line. Algorithms soon to be released will increase the three-dimensional accuracy to 0.1 meter. The Federal Aviation Administration (FAA) is nearing approval of GPS for commercial aircraft navigation and glide path navigation for rural airports without instrument landing systems. Recent aircraft accidents on runways are also hastening FAA approval of GPS. Conflicts with the Department of Defense concerning selectiveavailability (purposely induced signal degradation for security reasons) is delaying FAA response. Differential- GPS accuracy is not affected by selective availability. Without selective availability, only one receiver is required for one meter accuracy. This accuracy would economically allow FAA to manage aircraft in flight and on the ground at any airport (Montgomery, 1991). The benefit of the FAA approval to agriculture is reduced cost. As more receivers are sold, the development cost is recaptured, and the receiver cost is reduced. The GPS navigation system would allow precise location 3 tracking and guidance of farm machinery within the fields. Integrating the yield production potential maps and the precise tracking of the farm machinery with computer controlled equipment to distribute chemicals, fertilizers, or water would allow optimum crop production on variable soils. This document is divided into three chapters or papers. The first paper correlates yield variations to stored soil water across landscape positions. The yield- water correlation will help farmers balance nitrogen fertilizer with stored soil water so that overfertilization, groundwater and grain protein problems will be alleviated. The second paper focuses on classical and geostatistical soil sampling strategies acceptable to farmers. Farmers apply fertilizer based on estimates of future rainfall or available water -- estimates that are correct less than half the time. The farmer must know where to sample so that a few samples will provide adequate, precise information upon which to base fertilizer management decisions. The third paper discusses using degital terrain models (DTM), geographic information systems (GIS), and aerial photographs to map landscape units to generate soil management maps that correlate yield with variable soil water storage across landscape positions. 4 LITERATURE REVIEW Identifying Yield Variability Soil Moisture and Fertility Farmers and researchers first observed yield variations associated with landscape position. et. al. Miller, (1988) observed wheat yield differences on landscape positions in California. Roberts (1987) reported yield variation associated with changes in landscape position and loess depth. Differences in soil fertility across landscape positions have been reported by Mulla and Hammond (1988), and Veseth (1989). Veseth (1989) and Wilson, et. al. (1991) described soil water redistribution on landscape positions in dryland wheat production areas. More soil moisture is available in the foot and toe slope positions. The amount of water available to the crop influences yield. Aerial photographs are valuable for identifying low yields associated with shallow soil (Roberts 1987). Remote-sensing By definition, remote-sensing involves detection, identification, and analysis of objects or features using sensors at remote positions from the objects studied. distance may range from a few meters to many kilometers (Avery and Berlin, 1985). Remote-sensing applications to agriculture involve The 5 measuring the electromagnetic spectrum (EMS) as it interacts with crop canopies and soil. The most frequently sensed portion of the spectrum is visible light (0.4 to 0.7 um), near or reflected- infrared (0.7 to 1.1 um) and thermal or emitted-infrared (8 to 14 um). Sometimes active or passive microwave (0.1 to 1.0 meter wavelength) radiation is used for soil moisture studies (Myers, 1983). Some sensors are photographic, producing a picture (the visible light and reflected infrared), others are analog devices which produce an image of the sensed object or features. Advanced sensors record the electromagnetic spectrum in numerical or digital format, for later conversion to image form. Many factors influence the reflectance of light or other portions of the spectrum from leaves. Some of the more important factors are chlorophyll density, water content, and internal leaf structure. One of the leading factors contributing to reflectance is the air-cell wall interface. As energy (EMS) impinges on the leaf, it is reflected, absorbed, or transmitted by the epidermis. The transmitted energy is then scattered primarily by the aircell wall interfaces of the palisade parenchyma of the mesophyll. At this point much of the visible light is absorbed by chlorophyll and other plant pigments in the chloroplasts (Knipling, 1970). The plant leaf has low reflection in the visible portion of the spectrum. In the 6 near-infrared portion of the spectrum plants exhibit high reflectance and transmission, with little absorption because of internal leaf scattering (Knipling, 1970). Opposed to reflected-infrared or near-infrared energy, plants emit energy in the thermal-infrared portion of the spectrum as heat. The heat emitted is a function of plant temperature and is regulated by convection, reflection of incident energy, and transpiration (Myers, 1983). Remote-sensors primarily record reflected energy from crop canopies instead of individual leaves, although, individual leaves make up the canopy. The canopy variables that influence the reflected energy are leaf angle and geometry, leaf area index (LAI), total biomass, percent ground cover, and background reflection from the soil (Knipling, 1970; Gates, et al., 1965; Myers and Allen, 1968; Myers, 1983; Daughtry, et al., 1980). Most important are leaf area index (LAI), total above-ground biomass, and soil background reflection-percent soil cover. Many environmental interactions influence crop spectral reflectance, but recording data in conditions as similar as possible helps keep the interactions constant. Leaf area index (LAI), total biomass, and percent ground cover influence reflectance primarily because more leaves reflect more energy. This is particularly the case in the reflected- or near-infrared portion of the 7 spectrum. As near-infrared energy enters the canopy approximately half of the energy is reflected by the upper leaves and the remaining near-infrared energy is transmitted to lower leaves (Knipling, 1970). The lower leaves, likewise, reflect about half of the near-infrared energy toward the upper leaves. Depending on leaf area index (LAI), biomass, and percent ground cover, the lower leaves transmit half the near-infrared energy to still lower leaves or to the ground. As the infrared energy reaches the ground, soil conditions determine what percentage is reflected up through the canopy. As the infrared energy is reflected up through the canopy, internal leaf scattering again reflects some energy back to the lower leaves and ground, and transmits some energy back up through the canopy to the sensors. Thus, the higher the total biomass, leaf area index (LAI), and percent soil cover the greater the sum of all possible increments of infrared reflection from a crop canopy. This is known as the infrared-enhancement effect (Knipling, 1970). Relatively more near-infrared energy than visible light is reflected by canopies with high leaf area indices (LAI's); and relatively less near-infrared energy than visible light is reflected by canopies with low leaf area indices. Near-infrared reflectance, then, results in more contrast between canopies of high and low biomass or leaf 8 area index (LAI) than visible light reflectance. (Knipling, 1970) Soil properties that influence reflection of the electromagnetic spectrum (EMS) are primarily related to soil color. The visible and reflected portions of the spectrum are most useful for distinguishing differences in color and, therefore, soil properties. Remote-sensing is useful for distinguishing soil properties that influence the color of the soil surface; and the soil surface color influences the background, or soil, reflection component of crop canopies (Myers, 1983; Myers and Allen, 1968). The Munsell color system is used in soil classification and soil surveys as a descriptive color system of hue, value, and chroma. Hue is the dominant spectral color (ie. green, red, or yellow,) and corresponds to the wavelength of the spectrum measured by sensors. Value is the relative brightness and refers to the total amount of light reflected. Chroma is the relative purity of the color, ie., the relative shades of a color. (Myers, 1983). Soil factors that influence color and reflection include water content, organic matter content, color of parent material, mineralogical composition, particle size, and texture (Myers, 1983). Past research estimating leaf area index (LAI) (Ahlrichs and Bauer, 1983; Wiegand, Richardson, and 9 Kanemasu, 1979), above-ground phytomass (biomass) (Asrar, et al., 1985), and dry matter production (Aase and Siddoway, 1981) of wheat by remote-sensing vegetation index methods (digital ratios of reflected red to nearinfrared increments of the electromagnetic spectrum) report accuracies (R2 values) of 0.70 to 0.95. These reports show almost a 1:1 relation for estimated to measured biomass, although some methods (vegetation indices) had higher standard errors than others. Daughtry, et al. (1980) report that crop cultural practices influencing biomass, such as planting date, moisture stress, crop development, and, in some instances, fertility rate, can be identified by remote-sensing methods. The agronomic factors influencing canopy biomass the most, therefore, causing a greater spectral response, are crop development stage and moisture status. High grain yield was also correlated with high canopy biomass. Asrar, Kanemasu, and Yoshida (1985) estimated leaf area index (LAI) from spectral reflectance measurements of red and near-infrared reflectance for wheat at three geographical locations. The actual measured LAI was not significantly different (p = 0.01) from an indirect LAI estimate or LAI estimated by a regression method (both estimates for LAI were made from digital vegetation index formulas: the indirect from an absorbed photosynthetically active radiation index, and the 10 regression from a red to infrared ratio). Spectral reflectance measurements were recorded for different combinations of planting dates and irrigation frequencies. Standard errors ranged from 0.2 to 0.78 (units of LAI). The above results were from Phoenix, Az. and Manhattan, Ks. At Obregon, Mexico excessive standard error estimates attributed to experimental variance and high within field variation caused estimated and measured LAI to be significantly different. Measurements were corrected for background reflection in some cases. Dozier and Strahler (1983) reviewed ground investigations supporting remote-sensing data collection. The purpose of the ground investigation, sometimes referred to as ground-verification or ground-truthing, was to correlate the remotely-sensed data with actual conditions found on the ground. Ground investigations usually fall into four categories: calibration/correction, interpretation of properties, training, and verification. Calibration/correction is essential to ensure that correct data are obtained and account for variations caused by uncontrollable interactions. For correct interpretation of properties, ground investigations ensure that the relation between the reflected data and the property to be estimated are correct. For digital analysis of remotely-sensed data, a training area is selected where the ground features are 11 correlated with the digital data. Any interpretation procedure from remotely-sensed data must be verified by actual ground observations because many anomalous interactions may lead to erroneous interpretations. The ground-verification should include examples of the entire range of the interpretation. Aerial infrared photography was selected to quantify canopy biomass variation in this study. Daughtry, et al. (1980) reported that high grain yield was also correlated with high canopy biomass. Subjective interpretation of infrared photography used in this study may not be as accurate as the digital vegetation index used by Daughtry, et al. (1980). Color infrared photography measures both the amount of infrared energy (0.7 - 1.1 um) and the green and red visible light (0.5 0.7) reflected by a crop canopy (for photographic infrared review, see Lillisand and Kiefer, 1979). In general, areas of a crop canopy reflecting higher amounts of infrared energy have larger leaf area indices (LAI's), the canopy architecture is more erect, and the canopy has greater biomass. Of course, many interactions may contribute to anomalies. Reviews of canopy infrared reflectance are provided by Knipling (1970); Gates, et al., (1965); Myers and Allen (1968); and Myers (1983). Infrared sensitive color photographic film records 12 reflected infrared energy at the red sensitive dye layer. In final prints (there are many uncontrollable processing and printing variables) bright red areas are associated with large biomass areas in the field. Light or white areas on the print are associated with very low biomass, Interpretation of color or bare spots in the field. infrared photographs is subjective; the photo interpreter uses knowledge of all areas of crop culture, growing conditions, and soil characteristics to identify crop variations on the basis of biomass. Ground verification of photo interpretation is essential to the interpretation. Assessing Spatial Variability Classical Statistics For some time research has addressed soil variability influences on crop yields. Difficulty has been encountered in assessing the actual magnitude and field area influenced by soil heterogeneity. Field plot design attempts to remove soil effects from treatment effects by blocking (Petersen, 1985). Sampling plans are designed to increase precision by stratifying homogeneous areas within heterogeneous fields to ensure adequate sampling coverage of all portions of the field (Petersen and Calvin, 1965). Soil survey personnel involved in mapping soil variability over regions must resolve the objectives of the survey 13 (land use) with the accuracy and cost of the soil map. Soil map accuracy depends on how well the description of a sampled site (pit or borehole) estimates the soil characteristics of the site neighborhood (Beckett and Webster, 1971). Classical statistical methods assume that the population from which the samples are drawn is normally distributed and that each sample is independent of all other samples. Because soil characteristics are often related to landscape position, close samples as well as distant samples may be correlated to each other depending on the characteristic. The population from which each sample is drawn is not normally distributed and samples are spatially dependent. Geostatistics A statistical analytical method, one that takes into account the interrelation of soil samples over landscape position, has been adapted to soil variability studies from geology. geostatistics. This statistical approach is known as Geostatistics was developed by Matheron in a theory called the "Theory of Regionalized Variables" (Vieira, et al., 1982). The first step in geostatistical analysis of spatial variation involves assessing the variability over the region or area to be studied. Samples are obtained at a 14 defined, regular interval along the length of a transect. Both very closely spaced samples (approximately one meter), and widely spaced samples are obtained. If geomorphic landscape relations to soil genesis are known, such as fluvial, loessal, or erosional features, transects are established and sampled both along and across the geomorphic pattern. Multiple sample transects provide an estimate of the variability of the soil characteristic as a function of distance and direction. Isotropic variation (variation independent of the direction of measurement) and anisotropic variation (variation dependent on the direction of measurement) can be assessed by sampling transects established in different directions across the landscape. The first step, or reconnaissance stage, of a geostatistical survey allows for the graphical analysis of soil variability derived from the sampled transects. This graph is known as a semivariogram (Burgess and Webster, 1980 a) and is a measure of variance among sample sites (dependent variable, or y-axis) as a function of distance (independent variable, or x-axis). Semivariograms are constructed for each transect direction. The variance between very close samples is called the "nugget" variance, it is the point on the graph where the relation crosses the y-axis (Burgess and Webster, 1980 b). The variance increases with distance (sometimes linearly or 15 spherically) eventually reaching a maximum variance known as the "sill". Nugget and sill are terms adapted from geologic mining referring to mining related variance (ie. nugget variance is the ore variance associated with the size of a gold nugget within a sample, and sill variance is the ore variance associated with a block or area of the mine). The "range" is distance (actual measured distance on the transect) relating the distance between samples necessary to obtain an estimated variance, or spatial dependence of sampled points. The range is the distance along the transect, from the nugget variance to the sill variance. Once the spatial dependence, or the variance increase with distance, has been accurately developed from intensive sampling to produce the semi-variogram, a sampling strategy is developed. Depending on the objectives of the study and the desired level of precision required, the density of sampling for the survey can be estimated from the semi-variogram by selecting the distance between samples needed to obtain the desired precision or variance estimate (Burgess, Webster, and McBratney, 1981). The most efficient sampling design will be one with the fewest samples, or maximum distance between samples, necessary to obtain the desired precision. Usually a square grid at a defined density is easiest to operate in the field. 16 After the survey is conducted, an interpolation procedure known as kriging is used to connect inferred similar values together with isarithms, similar to topographic contour lines connecting points of equal elevation (Burgess and Webster, 1980a). The kriging procedure is named for D.G. Krige who did much to develop the method for South African gold fields. Kriging is a form of weighted local averaging (Burgess and Webster, Kriging uses the spatial dependence from the semi- 1980). variogram to weight values at unrecorded or unsampled locations as a function of distance from recorded locations. Kriging provides an interpolated estimate at an unrecorded point that is unbiased, with known variance, and minimum variance, (Burgess and Webster, 1980a) therefore, it is a statistically sound and optimal method. If soil properties have large nugget variances (in other words, if the soil is very heterogeneous, if samples near each other have large variance) the punctual kriging isarithms (the contour-like lines on the map) are very erratic. In this case block kriging is described by Burgess and Webster (1980b) as a means of grouping areas together and averaging properties to produce a smoother map. The precision may necessarily be lower; however, the map may reveal a regional pattern. Sometimes a soil property is very difficult to measure or may require large outlays of resources (time, 17 labor, or money) to sample at an intensity required for the desired precision for interpolation methods. Another variable may be found to be highly correlated to the property of interest. If both variables are found to be spatially interdependent and exhibit the same anisotrophy or isotrophy, McBratney and Webster (1983) provide a method of cross semivariograms and co-kriging to interpolate one variable from measuring another. Actually, both variables are measured, the easiest to measure variable at a higher density or frequency than the more difficult to measure variable. Semivariograms are made from intensive reconnaissance surveys for both variables, compared for anisotrophy or isotrophy, and spatial dependence. From the semi-variograms, survey method and sample density are determined for both variables. Spatially correlated and co-regionalized variables, one sampled intensively (the easiest and most economical to measure), the other sampled much less intensively, can produce a map of the spatial variability of the variable of interest much more economically than intensive sampling of the difficult to measure variable. Remotely-sensed subsidiary variables are in many instances more economically measured than the primary variable of interest. The spatial variability of the remotely-sensed variable is often highly co-regionalized and has potential for co-kriging (McBratney and Webster, 18 1983). Examples provided by and Vieira, et al., McBratney and Webster (1983) (1982) include, temperature measured with remote-sensing techniques correlated to soil water, or organic matter content. One assumption of the geostatistical interpolation method is that the boundaries of the soil parameter to be measured cannot be recognizable by distinct geomorphic breaks and must have very gradual gradients across the isarithm. In other words the interpolation techniques do not estimate soil properties across a cliff (Burgess and Webster, 1980a). Bouma (1985) points out the value of using traditional soil survey methods combined with reconnaissance geostatistical methods. First, the aerial photography and geomorphological associations used to identify preliminary soil boundaries could be used to identify subpopulations or homogeneous areas based on vegetation, landform position, or topography. Next, the subpopulations could be used for initial high intensity reconnaissance sampling transects to assess the variability within the subpopulation. A group of similar subpopulations could be used to assess the variability within the region using the semi-variograms of the transects. Then, the estimate of the variance from the semi-variograms could provide the sampling density and method needed for both local and regional surveys at a 19 desired precision. Combining the ideas of Bouma (1985), Burgess and Webster (1980), McBratney and Webster (1983), and Vieira, et al. (1982), subpopulations, locally and regionally, could be identified based on remotly-sensed subsidiary variables associated with grain yield. The variance of the subpopulation (estimated from the previously obtained reconnaissance semi-variogram for the region) could provide the grower with a sampling plan and density for assessing the variability of grain yield within his farm. GIS systems could overlay several remotely-sensed subsidiary variables accelerating the identification of regional subpopulations and providing maps of the spatial variability of grain yield. It should be pointed out that with the regionalized estimate of the variance of the soil property to be measured, geostatistical methods may increase precision over classical sampling designs, at a given cost, by reducing the total number of samples, even though the reconnaissance survey is sampled intensively. By integrating the classical soil survey, remotesensing, and geostatistical methods, an economical approach may be formulated for assessing the regional and local variation in grain yield for the Columbia Plateau in eastern Oregon. Warrick and Gardner (1983) and Russo (1984) provide papers dealing with assessing the spatial variability of 20 yield (a parameter with many interactions, therefore, hard to measure) and improved crop management by spatially varying irrigation. Bresler, et al. (1981) correlated crop yield of peas (total dry matter, pod, and hay) with soil water content. Excellent reviews of geostatistical procedures of soils are provided in a five part series of papers by Burgess and Webster (1980, 1980b), Webster and Burgess (1980) Burgess, et al. (1981) and McBratney and Webster (1983), including examples of analyses. Vieira, et al. (1982) include a review of geostatistical procedures, with examples and computer code. Neilsen and Bouma (1985) edited a workshop proceedings on various aspects of soil spatial variability. Variable Field Mapping Geographic Information Systems Remote-sensing systems characteristically collect volumes of data relating to large geographical areas. Extracting the relatively small amount of relevant data from the total data set presents a formidable task. The development of computer systems to handle large volumes of geographic or spatial (referring to geographic arrangement in space) data is essential for optimum use of remotesensing systems (Marble and Peuquet, 1983). Geographic information computer analysis systems 21 (GIS) are designed to handle large volumes of spatial data, obtained from remote-sensing systems as well as ground derived data such as soil surveys, streams, or geological surveys, in either map or attribute (ie. stream flow volume) data format. The GIS provides an efficient means of storage, retrieval, manipulation, analysis, and presentation of spatial data, usually in map form (Marble and Peuquet, 1983). Unlike other data analysis systems, GIS systems handle and analyze spatial data as well as non-spatial attribute data associated with the spatial data. To preserve the data's spatial dependence, raster (a gridlike network of x-y coordinates) or vector (a series of points, line segments, line nodes, and polygons) formats (or a combination of both, raster and vector) are used for data management and analysis (Marble and Peuquet, 1983; Eastman, 1990). Future combinations of remote-sensing and GIS systems may provide efficient means where ground investigation derived GIS data can be imported into remote-sensing classification systems, thereby improving the accuracy of interpretation and classification. Presently, the remote- sensing information is transferred to the GIS, where, in combination with ground data, it is analyzed. The GIS analysis is similar to overlaying one map onto another map, finding where areas of interest from 22 both maps intersect, or where areas are spatially dependent, then producing a map of the areas of intersection, or the spatial dependence. In a sense, GIS analysis is a form of visual association of spatially dependent geographic areas, that allows for the analysis of non-spatial attribute data associated with the areal (referring to geographic area) data. For example, when combining a map of soil associations in which the soil types differ in sewage disposal attributes with a map of utility services, the combination map may identify future sites for rural housing subdivisions. Marble, et al. (1984), Marble and Peuquet (1983), and Starr and Estes (1990) present information, examples, and technical details of geographic information systems. Digital Terrain Models Contour surface models of fields can aid farmers in identifying spatial patterns in fields. Slope and aspect are correlated to yield potential. Digital elevation data can be obtained from aerial photographs by photogrammetric processes, from USGS 7.5 minute quadrangles, or by "field digitizing" with global positioning system receivers. The photogrammetricly produced digital terrain model (DTM) from 70 mm, photography though useful, does not provide the detail of the quadrangle data. 23 Basic Cartography Aerial photos can be updated in the field to indicate low yield areas. Aerial photos are not maps, the delineated areas on the photos can be transferred to an accurate base map in the office with a zoom transfer scope. (Pain, 1981) The USGS quadrangles provide an economical and accurate base map (scale 1:24 000). function of scale. Map accuracy is a Variable fertilizer application with vehicular guidance systems may require increased base map accuracy or precision, requiring a scale of 1:5 000 or 1:10 000. Field boundaries and map units can be "field digitized" with GPS receivers. Marking the map units with flags may be necessary to enable allowing the farmer to drive along the map unit and field boundaries to field digitize the field base map at the required scale for precision. 24 CHAPTER I. FERTILIZING WHEAT WITH NITROGEN BALANCED TO SOIL WATER VARIABILITY ACROSS LANDSCAPES ABSTRACT Potential grain yield varies within fields -- even within delineations of soil map units. The variation is often associated with landscape position. This study was conducted to evaluate yield variations across landscape positions correlated with changes in soil water. Rainfall is often redistributed, with more stored soil water on concave landscapes, and less on steeper, convex landscapes. Two fields were selected to represent the regional wheat yield variability on the Columbia Plateau. One field was situated in a high (30-50 cm) precipitation area and no noticeable shallow soils. The second was located in a low (25-30 cm) precipitation area and areas with shallow (<50 cm) soils. Soil moisture, nitrogen, and grain yield were sampled along reconnaissance transects traversing landscapes. yield was correlated (r = 0.96 ) Grain to crop available water (CAW) and to total soil nitrate nitrogen (r2 = 0.67*). Using these estimates farmers can establish a transect crossing landscape positions, measure crop available water 25 and nitrate nitrogen, then, balance nitrate fertilizers to crop available water. INTRODUCTION Farmers and researchers know that yield differences exist within fields. Dryland winter wheat farmers in the Columbia Plateau of Eastern Oregon apply nitrogen fertilizer in the spring of the summer fallow year or just prior to planting in mid-September. Hoping for the highest yield possible -- and more rain than usual during the coming crop year -- farmers often over-fertilize with nitrogen. Farmers know that the soft white wheat produced in the region is higher in protein than the consumer prefers. They also understand the potential for excessive nitrogen to foul surface and groundwater resources. Some growers soil profile depth, then apply more nitrogen on deep soils and less on shallow soils. Others revert shallow soil areas to rangeland. Researchers across the United States have identified agronomic causes of yield variation. In glacial till soils of the northern midwest, drainage differences within fields have been documented with aerial color infrared photography (Fairchild and Knutson, 1989). Photographic maps of the fields enable farmers to test and vary 26 fertilizer according to drainage class. In the rolling hills of the central plains, surface nitrate differences are detected with real-time sensors attached to fertilizer spreaders (Colburn, 1986). The rate of nitrogen fertilizer is varied according to soil nitrate levels. In the high organic matter soils of Texas, real-time photocells detect changes in soil chroma correlated to organic matter content (Van Scoyoc, et al., 1989). Organic matter content is correlated to the amount of mineralized nitrogen. Both agricultural chemical and fertilizer rates are adjusted with soil organic matter. Differences in subsurface fertility were identified using soil sampling and geostatistical techniques (Mulla and Hammond, 1988; Hammond, Mulla, and Fairchild, 1988) in pivot-irrigated fields in eastern Washington state. The source of yield variation determines the field methodology for identification, measurement, statistical analysis, and application of chemicals and fertilizers. In the past, field plot type statistical analysis of variance methods were used to evaluate yield variations. On landscapes, yield often changes gradually as sample position progresses across the landscape. Two questions could be asked about yield variation across landscapes. First, how many locations should be individually sampled for yield? Second, are landscape positions considered as treatments or varieties, in the analysis of variance? 27 Presuming landscape positions are considered treatments, paired observations are collected on many landscape positions. Careful use of mean separation procedures (Hatfield, 1990) suggest comparisons among landscape positions be made in progressive order (as are treatments) across the landscape. This procedure may indicate statistically nonsignificant differences among landscape positions. Conversely, if landscape positions were considered to be varieties, random means could be compared. Comparing random means, a south shoulder position could be significantly different from a north back slope, yet not different from the summit. Both interpretations could be correct for identical data collected from one landscape transect. The assignment of landscape position in the analysis of variance depends on the objectives of the study. With the use of geostatistical procedures and transect experimental designs (Nelson and Buol, 1990) yield variation studies across landscapes are easier to interpret. When using geostatistical procedures, the researcher must use care not to violate geostatistical hypotheses. When collecting data along transects that cross landscapes, a landscape may not have characteristics known to the researcher (a priori knowledge) to be radically different from other landscapes. (Burgess and Webster, 28 1980a) Nelson and Buol (1990) point out that the transect designs are inherently low in precision because replication is constrained by inability to randomize individual landscape positions. This study was conducted to evaluate yield variations associated with landscape position correlated to available soil water. Low precision reconnaissance transect sampling and experimental design methods were used. The preliminary yield-soil water correlations can be reevaluated later with additional high precision experimental designs. MATERIAL AND METHODS In the summer of 1988, two fields were selected to represent the variability within the Columbia Plateau winter wheat producing region. The Pendleton study area is in a higher (40-50 cm) rainfall and potential (4-6 Mg ha 1) yield part of the region. The Wasco study area is in a lower (25-30 cm) rainfall area and has shallow (<50 cm) soil areas. Transects were established across landscapes for sampling and plot experiments. by use of landmarks, The transects were sited, so sample and plot sites could be repeatedly visited without unnecessary physical markers in the fields. 29 The Pendleton study area The field in the Pendleton study area is about 4 km west by northwest of the Pendleton airport. The approximately 18 ha field is located in the SW1/4 of the NW1/4 of section 15, Township 3 North, Range 31 East Willamette Principle Meridian. The field is bordered on the south by the county road, west by a field road, east by a driveway, and north by a gulch. Transect 1, (Figure A.2) starts at the southwest corner of the field and angles northeast (40) to a utility pole on the horizon, intersecting the driveway about 100 meters south of the house. Transect 1 traverses a gradual (3-7%) north slope, crosses an erosion control terrace, traverses a steeper (7-12%) north slope, intersects a second erosion control terrace and the drainage way, proceeds up a steep (12-20%) southwest slope to a level-to-north slope ridge to the driveway. Transect 1 crosses two soil mapping units, 114B and 114C (Johnson and Makinson, 1988). Transect 2, (Figure A.2) starts approximately 40 meters north of the second (lower, or north-most) erosion control terrace. Transect 2 parallels the west boundary field road, traversing northward, offset 15 meters east of, the west boundary field road, on the west border of the field. Transect two is in mapping unit 114C (Johnson and Makinson, 1988). 30 Topography can be obtained from the Barnhart, Oregon USGS 7.5 minute quadrangle (USGS staff, 1979). Mapping unit descriptions, soil survey area Mapping unit 114B (Johnson and Makinson, 1988), is Walla Walla silt loam, 1 to 7 percent slopes, classified as coarse-silty, mixed, mesic Typic Haploxerolls. These soils are deep well drained soils on broad summits. Included in this unit are small areas of Anderly and Hermiston soils, Walla Walla soils that have a hardpan at a depth of 100 to 150 cm or are eroded, and Vitrandepts (now classified Vitraxerands). Also included are small areas of Walla Walla soils that have slopes of 7 to 12 percent. Included areas make up about 15 percent of the total area. (Johnson and Makinson, 1988). Map unit 114C, is Walla Walla silt loam, 7 to 12 percent slopes, again classified as coarse-silty, mixed, mesic Typic Haploxerolls. Included in this unit are small areas of Anderly and Hermiston soils, Walla Walla soils that have a hardpan at a depth of 100 to 150 cm or are eroded, and Vitrandepts (Vitraxerands). Also included are small areas of Walla Walla soils with slopes of 1 to 7 percent, and 12 to 20 percent. Included soils make up about 25 percent of the total area. Anderly soils are coarse-silty, mixed, mesic Typic Haploxerolls. Hermiston soils are coarse-silty, mixed, mesic Cumulic Haploxerolls. 31 (Johnson and Makinson, 1988). The Wasco study area The study area, approximately 40 ha, near Wasco, Sherman County, Oregon, is located about 8 km southeast of Wasco, Oregon, and about 8-10 km northwest of Moro, Oregon. The field location is the SE1/4 section, of section 24, Township 1 North, Range 17 East Willamette Principle Meridian. The triangular shaped field is bordered on the north by a county road, then an unimproved road. The east boundary is by the same unimproved road bending south east, then south to highway 206. The highway (206) is the south west boundary. Transect 2 starts 390 meters east, and 30 meters south, of the north west corner, of the SE1/4, of section 24. Transect 2 proceeds directly south approximately 300 meters to within 50 meters of highway 206. traverses soil map units: Transect 2 WaBN, WaA, and WmBS. (Mayers, 1959). Topography of the field can be obtained from the Wasco, Oregon USGS 7.5 Minute Quadrangle (USGS staff, 1973) . The first mapping unit (Mayers, 1959) traversed by transect 2 is WaBN, Walla Walla silt loam, very deep, to 20 percent north slopes. 7 The Walla Walla soils are coarse-silty, mixed, mesic, Typic Haploxerolls. About 10 32 percent of this map unit is other Walla Walla soils. After proceeding south up the north slope, transect 2 traverses a delineation of the WaA mapping unit. The WaA mapping unit is Walla Walla silt loam, very deep, 3 to 7 percent slopes. About 10 percent of this mapping unit (Mayers, 1959) consists of the more shallow or more sloping Walla Walla soils. These soils occur on broad flat ridge tops. After crossing the ridge top, transect 2 proceeds across a delineation of the WmBS mapping unit. mapping unit is a The WmBS Walla Walla silt loam, low rainfall, moderately deep, 7 to 20 percent south slopes. This soil occurs on outer edges of broad ridge crests, ranges from 20 to 90 cm in depth to basalt or caliche. This mapping unit contains rock outcrops and field drainage-ways. About 10 percent of this map unit consists of Walla Walla soils that have low rainfall and Starbuck soils. Starbuck soils are shallow, less than 20 cm deep, stony to very stony (basalt fragments) loamy, mixed, mesic, Lithic Xerollic Camborthids. In the study, Transect 2 traversed all the soils described above by Mayers, 1959). Sampling methods In the fall of 1988, soil profile cores were obtained with a Giddings soil probe mounted on a four-wheel drive truck. Intact cores (or as nearly as possible with a 33 Giddings probe) were sampled along the transects at intervals of about 1.5 meter of vertical relief. Cores were sampled to lithic contact or to a depth of 4 meters. Some indurated layers were disturbed by augering. The cores were described with soil survey field methods (Soil Survey Staff, 1975). After cores were obtained, the farmers planted winter wheat with normal cultural and fertility practices. In early April, soil cores were obtained within one meter of the original core position. content was measured. Gravimetric water Nitrate nitrogen (30 cm depth intervals to 1.8 meter) and percentage of organic matter (surface 30 cm) were measured at the OSU Soil Test Lab. Soil fertility soil-test-calibration plots were established at original core sites, normal to the transect. These plots were 2.5 meter2 in size. At the Pendleton study area, the farmer applied in the fall about 100 kg ha-1 nitrogen, therefore, three treatments were applied, a control - no nitrogen, 25 kg ha-1, and 50 kg ha of solution 32 in water. At the Wasco study area, the farmer applied about 60 kg ha-1 of nitrogen on north slope positions and 45 kg ha-1 of nitrogen on the remainder of the field. An additional 3.5 kg ha-1 of nitrogen was applied by air in the spring along with a herbicide at the Wasco study area. At the Wasco study area three treatments were applied, a control (no nitrogen), 100 kg 34 hat, and 150 kg ha of nitrogen as solution 32 in water; however, the last two core sample sites on the transect with noticeably shallow soils received treatments of control (no nitrogen), 25 kg ha-1, and 50 kg ha of nitrogen as solution 32 in water. At harvest one m 2 samples were hand harvested from the center of the fertility plots. Samples were cleaned and yield, test weight, and 1000 kernel weight were obtained. Percentage of grain protein was measured with a Technicon 400 Infra-Analyzer. The Technicon was calibrated for percentage of grain protein by the microKjeldahl procedure described by Nelson and Summers, 1973. The simple correlation coefficient of the percentage of grain protein measured by the Technicon as a function of micro-Kjeldahl measured percentage of grain protein was 0.97, and standard error of the estimate was 0.45 percentage point of grain protein. After harvest soil core samples were obtained within 1-2 meters of core sites, and gravimetric water content (30 cm intervals) was measured to 1.8 meters. Volume bulk density measurements were made along the transect at foot slope, back slope, and summit positions on transects, at approximately 20 and 60 cm depths. Analyses Change in soil water content from April to August was 35 measured and expressed as cm depth of water in 30 cm increments of soil profile depth (equivalent depth; Hillel, 1982) and summed to 1.5 and 1.8 meters of soil depth. Crop available water (CAW) was calculated from equivalent depth water content (April measurement), minus the published wilting point (-1.5 Mpa) water content of Walla Walla soil (Huddleston, 1982). Necessary conversions used measured volume bulk density and equations of Hillel (1982). Rainfall received after April in most years does not wet the soil to a depth sufficient for root uptake. Yield measurements from the three nitrogen level plots were plotted and regressed on total nitrate content summed to 1.8 meters. fitted. A linear or quadratic function was In the linear model the highest yield was identified as the maximum fertilized yield. The first derivative of the quadratic model was interpreted to be the maximum fertilized yield. Maximum fertilized yield was regressed on total nitrate nitrogen, and compared to similar linear models obtained from the fertilizer guide (Gardner and Goetze, 1980). Soil residual nitrogen and change in soil water content were plotted against soil profile depth and studied for anomalous conditions. Grain yield as a function of soil profile depth, crop available water, total nitrogen, soil profile depth, and 36 organic matter were analyzed with simple (linear) regression procedures. Crop available water (CAW) as a function of soil profile depth, was similarly analyzed. RESULTS AND DISCUSSION Grain yield correlation to crop available water (CAW) was highly significant (Figure 1.2) at the Wasco study area and non significant at the Pendleton study area. Pendleton received considerably less precipitation than Wasco in December and June of the summer fallow year (Table I.1). Dry soil conditions at the Pendleton study area during planting resulted in delayed crop emergence and development, and reduced yield. Grain yield correlation to total nitrogen was very similar (Figure 1.3) to the correlations obtained from the fertilizer guide (Gardner and Goetze, 1984). The objective of the nitrogen fertility calibration trial on landscape positions was to answer questions concerning extrapolating fertility recommendations obtained on uniform, representative portions of fields to heterogeneous, previously unsampled areas. This research supports the fertilizer guide research conducted across regional potential yield and environment variability. The nitrogen required for maximum fertilized yield may result in excessive grain protein content. The three control treatments with highest CAW had protein contents 37 between 9.2 and 9.8 percent. Cerrato and Blackmer (1990) indicate that quadratic models may overstate nitrate levels for optimum corn yield. Additional research investigating the interaction of CAW and nitrate on wheat yield and protein is needed. Crop available water and grain yield (Figure 1.3) are not correlated well with soil profile depth probably because of moisture redistribution (run-off on up-slope and run-on at lower slope positions, Figure 1.1A). Water redistribution on landscapes is another area of additional research, both to predict yield and protect groundwater. Grain yield was not correlated well to surface organic matter content (Figure 1.3). The lowest organic matter points on the plot (Figure 1.3) correspond to samples taken in the shallow lithic xerollic camborthid soils. The nitrate nitrogen concentration of the surface 30 cm was uniform along the transect. Real-time organic matter or nitrate sensors would not be feasible for variable fertilizer application in this region. The soil core descriptions obtained on the transects contained no discrepancies of map unit integrity concerning descriptions of included areas. Also, particle size classifications of described cores indicated no departure from soil water data provided by Huddleston (1982) . Farmers can establish transects crossing landscape 38 positions, measure crop available water and nitrate nitrogen, then, balance nitrate fertilizers to crop available water using Figure 1.2. Applying a starter fertilizer rate at planting, then, applying the balanced nitrogen rate early in the spring when soil moisture flow has slowed may prevent contaminating surface and groundwater with nitrates. Balancing nitrate to water may lower percentage of grain protein. It is important to note the standard error of the estimate associated with the grain yield prediction models in Figure 1.2. The prediction of grain yield from CAW is plus or minus about 0.7 Mg ha-1; and grain yield from total nitrogen, about 1.7 Mg ha-1. In this region a map based system of spatiallyvariable application seems appropriate given the causes of grain yield variation. The Wasco study area could conceivably have five soil management map units: toeslope, footslope, backslope, shoulder-summit-south shoulder-south backslope, and south backslope (Figure I.1) . As noted previously, these data and results are reconnaissance in nature and are of low precision. correlations may be re-evaluated with more precise experimental designs. The Table 1.1. Precipitation for fallow-crop cycle near Wasco and Pendleton study areas. Fallow Crop 1987 mo. Wasco 1988 Pend. mo. Wasco 1988 Pend. 1989 Wasco Pend. 0 14.22 0.5 N D 63.75 5.58 7.87 2.54 54.86 9.39 mo. Wasco Pend. F 33.78 19.55 M 48.51 47.24 34.54 43.68 39.87 37.33 14.47 2.28 mo. MM A S 0 1.77 0 0.25 N D 16.76 82.04 1.27 J 0.76 0 19.3 31.24 F M A M J J A 40.64 5.33 31.75 56.13 13.97 25.9 1.01 0 47.27 3.04 24.13 62.73 39.62 7.87 0.25 0 S J A 21.33 M J J 23.11 0 0 Figure 1.1. Schematic representation of landscape positions (A), and grain yield (B) along transect 2, Wasco study area. A 0 0 DO F O 0 0 B T F B North 00 So Su oo - So T toeslope footslope backslope shoulder summit Su cr 0.00 0 0 Su So 0 O B -00 F B 0 _ _ lithic (basalt) contact o So O soil surface = D o 4 So cc3 Q) 0 O 100.00 200.00 "" Transect Distance (m) O CD 300.00 0 0 coo 100.00 200.00 Transect Distance (m) 300.00 Figure 1.2. Wheat grain yield correlated to crop available water (A) and total nitrate (B) at Wasco study area. *'**,*** indicate significance at the 0.05, 0.01, and 0.001 probability levels, respectively. A Y, = 1.43 + 0.206 X OCCx:X) Spring of Cropping Year Fertilizer Guide Data r` = 0.96 ." Y = 1.007 + 0.03 x Stnd. Error of Est. = 0.351 o 01._-= csi 0 0 - 0 Y, = 1.81 + 0.02 x r2 0.67 (Prob. Level = 0.012 Stand. Error of Est. = 1.12 0- 0- 0 0 cioo takkeirte April of Cropping Year 5 00 10.00 15.00 20.00 25.00 Crop Available Water (cm) 30.00 0.00 1 50.00 1 100.00 Total NO3 (kg ha I 1150.00 ) 200.00 Figure 1.3. Crop available water correlated to soil profile depth (A); wheat grain yield correlated to soil profile depth (B), and organic matter (C), Wasco study area, transect 2. A E8 owl Y 2.65 + 0.106X (prob. level - 0.044) .e 0.51 6.46 Stnd. Error of Eat. 11E8 O 3 N 8 o.).; 0 73- 8 >6 aoo 06 8 50.00 100.00 150.00 Soil Profile Depth (cm 200.00 C B 3.2 + 5.71X - 0.45 (prob. level 0.065) Sind. Error of Eel. 1.35 Y. = 1.09 + 0.016X 0.40 (prob. level Stand. Error of Eat. 40.00 80.00 120.00 0.08) 1.08 160.00 Soil Profile Depth (cm) 200.00 0. 0 1.00 1.20 Organic Matter (se) 1.40 43 REFERENCES Burgess, T.M., and R. Webster. 1980. Optimal interpolation and isarithmic mapping of soil properties: I. The semi-variogram and punctual kriging. J. of Soil Sci. 31:315-331. Cerrato, M.E. and A.M. Blackmer. 1990. Comparison of models for describing corn yield response to nitrogen fertilizer. Agron. J. 82:138-143. Colburn, J. 1986. (Aguila Corp.) R&D on a fertilizer sensor and control system. U.S. Department of Energy, Washington, D.C. DOE/ID/12581-1 (DE87014929). Fairchild, D.S., and R.D. Knutson. 1989. Innovative equipment for application of agricultural inputs by soil variations. p.315 in Agronomy abstracts. ASA, Madison, WI. Gardner, H. and N.R. Goetze. 1980. O.S.U. Cooperative Extension Service. Fertilizer guide for winter wheat (non-irrigated -- Columbia Plateau) FG 54. Oregon State University, Corvallis, Oregon. Hammond, M.W., D.J. Mulla, and D.S. Fairchild. 1988. Development of management maps for spatially variable soil fertility. Proc. 39th Annual Far West Regional Fertilizer Conference, Bozeman, Montana. July 11-13, 1988. Hatfield, J.L. (ed.) 1990. Instructions to Authors (statistical methods). Agron. J. 82:171-172. Hillel, D. 1982. Introduction to soil physics. Press, Orlando, Fl. Academic Huddleston, J.H. 1982. Soils of Oregon: summaries of physical and chemical data. OSU Extension Service, special report 662. Oregon State University, Corvallis, OR. Johnson, D.R., and A.J. Makinson. 1988. Soil Survey of Umatilla County Area, Oregon. USDA-SCS. U.S. Govt. Print. Office, Washington DC. Mayers, L.R. 1959. Soil Survey of Sherman County, Oregon. USDA-SCS. U.S. Govt. Print. Office, Washington DC. 44 Mulla, D.J., and M.W. Hammond. 1988. Mapping of soil test results from large irrigation circles. Proc. 39th Annual Far West Regional Fertilizer Conference, Bozeman, Montana. July 11-13, 1988. Nelson, L.A., and S.W. Buol. 1990. Experimental designs to evaluate crop response on adjacent soil mapping units. Soil Sci. Soc. Am. J. 54:841-849. Nelson, D.W., and L.E. Sommers. 1973. Determinations of total nitrogen in plant materials. Agron. J. 65:109112. Soil Survey Staff. 1975. Soil Taxonomy. U.S. Dept. Agr. Handbook No. 436. USDA-SCS. U.S. Government Printing Office, Washington D.C. 754 p. U.S. Geological Survey. 1973. Wasco, Oregon Quadrangle - Sherman Co. 7.5 Minute Series. (Topographic). N4530--W12037.5/7.5 U.S. Govt. Print. Office, Washington DC. U.S. Geological Survey. 1973. Barnhart Oregon Quadrangle - - Umatilla Co. 7.5 Minute Series. (Topographic) N4537.5--W11852.5/7.7 U.S. Govt. Print. Office, Washington DC. Van Scoyoc, G.E., D.G. Schulze, L.D. Gaultney, J.L. Nagel, and J. Schonk. 1989. Variable rate application of chemicals using a soil organic matter sensor. p. 319 in Agronomy abstracts. ASA, Madison, WI. 45 CHAPTER II. FIELD SAMPLING STRATEGIES FOR BALANCING COST AND PRECISION ABSTRACT Farmers are aware that wheat yields on the Columbia Plateau loess region vary across the landscape. Some have noticed changes associated with loess mantle depth. Previous research has indicated that yield increases with increasing stored soil water, and optimal yields are obtained by balancing nitrogen fertilizer to stored soil water. The farmer, faced with the expense of field sampling, asks, "how many sites must be sampled?" Four reconnaissance yield transects were sampled in a 25 ha field. Grain yield was evaluated by geostatistical semivariance techniques. Results from one crop year indicate that the field be sampled with a 158 meter square grid. Anisotropic analysis indicate a rectangular grid, 68 by 43 meter would increase precision. For field orientation the grid should consist of transects traversing slopes (spaced 68 meters apart), sampled every 43 meters. Many farmers may consider this sample number to be excessive and are willing to sacrifice precision for cost. If sample number must be reduced, precision-sample number may be optimized by eliminating alternate transects while sampling each landscape position. 46 INTRODUCTION Farmers are aware that wheat yields vary across the landscapes. Most farmers are aware of the interaction effect of stored soil water and fertility on grain yield. Some farmers have noticed more moisture storage with increasing loess mantle depth. The loess mantle depth changes across landscape positions. Previous research has indicated that increased yields are associated with increased soil water storage on foot and toe slope position (Figure I.1). Because of the high cost of field sampling, the farmer asks, "where and how often must I sample"? Some farmers have chosen representative landscape positions, sampled depth of loess mantle, and managed fields by shallow, moderate, and deep soil depths. The traditional approach recommended for soil test sampling was to obtain a number of subsamples for each strata to be subsequently bulked (Peterson and Calvin, 1965). as one unit. The field was managed With increased knowledge of the fields, stratified sampling methods have been proposed (Roberts, 1987) . Independence of samples is considered with all statistical procedures. Recently, the theory of Regionalized variables (Matheron, 1971) has been introduced to soil science. The basic difference of 47 Regionalized variable theory and classical statistics is that samples obtained h distance apart, if the distance is small, samples may not be independent. A graphical analysis known as the semivariance is performed. The semivariogram indicates the range in distance h, that the samples are dependent and at what distance h, the samples are independent. (Nash and Daughtery, 1990). Once a semivariogram has been analyzed for a reconnaissance transect, a sampling scheme can be devised with known precision given sampling spacing (Burgess, Webster and McBratney, 1981). The objectives of this study were to (i) evaluate geostatistical procedures to estimate the number of grain yield samples required for various precision estimates, and (ii) to compare the variance estimate with that of an intensively sampled uniform strata. MATERIALS AND METHODS This study site was located about 5 km northwest of the Pendleton, Oregon airport. The grain yield data were collected by hand sampling one meter2 plots within the control plots of fertilizer calibration trials on transects one and two. Grain yield data were obtained with a commercial field combine between the plots along the transect. Approximately 50 meter2 samples were 48 harvested with the combine. Transects three and four were harvested with the combine. Harvested sites were measured along the transects and plotted on an aerial photo of the field. digitized and plotted (Figure I.1). The sites were The x and y coordinate of each plot with the associated yield were analyzed using GEOPACK (Yates, 1989). Yield data were plotted and found to be normally distributed. semivariogram was determined. The The automatic model fit was used and modified with the manual model fit function to obtain a model representing the semivariogram. Resulting semivariograms were analyzed by geostatistical procedures for anisotropy. The yield data in transect 3 was analyzed by the descriptive statistics package of GEOPACK for the mean and sample variance and compared to the semivariogram of all four transects. RESULTS AND DISCUSSION The wheat stand was thought to be delayed because of dry planting conditions which also resulted in poor yields. This year the stored soil water was poorly correlated to yield. This field is fairly uniform with low slopes and deep soils. The degree of anisotrophy is lower than in many fields but should still be recognized. 49 The semivariogram for all data points with a width angle of 90° is shown in figure II. 2. Semivariograms with a width angle of 45° and a direction angle of 90°, across slope, and 0°, down slope are indicated in figure II. 3. The sampling plan using the non-isotropic semivariogram would be to sample on a 158 meter grid. This would be acceptable for this field. There would be two across slope positions sampled, and two or three down slope sample positions. The frequency of slope changes is about that distance (158 meter), see figure III.1. Care should be taken to assess the anisotropy of all fields, particularly those with complex slopes. There is usually not as much variability within a given landscape or landscape position (the across slope direction) as among landscape positions (down slope direction). As figure 11.3 indicates, the sampling plan for anisotrophy would be a rectangular grid 68 by 43 meters. For field orientation the grid should consist of transects spaced 68 meters apart normal to the landscape, sampled every 43 meters. This arrangement would ensure adequate sampling of most landscape positions. The precision for these sampling plans is at least one half a megagram per hectare. This is far more precise than many of the decisions farmers make such as decisions based on future rainfall. The farmer may conclude that 50 the sample number can be decreased, saving money, but optimizing management of fields. If the decision is made to decrease sample number, the anisotropic semivariograms indicate that alternate transects could be eliminated and all landscape positions on remaining transects be retained. This may equate with stratifying all landscape positions in a stratified random sampling plan. The standard error of transect 3 is 0.27 Mg ha-1, and variance is 0.07 (n = 22). This error data can be used to compute sample number - costs estimates for stratified sampling schemes (Cochran, 1977). The semivariance estimates for the across slope for the entire field is 0.54 Mg ha-1; this is greater than the standard error (0.27 Mg ha-1) of stratifying the landscape of transect 3, but precise for most management decisions. With geostatistics 80% of the total sampling costs occur in the reconnaissance phase (sampling transects to construct the initial semivariogram). The farmer could conceivably select some well sited reconnaissance transects, then extrapolate the sampling density of the initial semivariograms to the entire farm or group of fields. Figure 11.1. Location of grain yield sampling sites along transects, Pendleton study area. N 5 N 5066814 E E 350081 + + ++ + + ++ + ++ ++ + + + "C 34972T 4- + + ,++ Transect 3 + +' ++ ++ Transect 1 TAT + Transect 4 1 Transect 2 N 5067207 E 50 350013 meters 100 Figure 11.2. Grain yield semivariogram, four transects, Pendleton study area. 0 0 0 Eo C0 O cn l 11111111111111111111111111111111111111111) Oo N0 0 O O Spherical Model L_ 0 o E0 Range = 158.8 m Sill = 0.48; Sill Nugget = 0.18 Nugget = 0.30 Width of Angle Class: 90° Angle for Direction: 0 (f) 0 0 C.00 100.00 200.00 Lag Distance 300.00 Meters 400.00 Figure 11.3. Grain yield semivariograms indicating anisotrophy; across slope (A), and down slope (B). A B 0 oo 0 co d Spherical Model 0 Ed %%% 111111111111 119 11111111111111111111 II a CJ) 0 o 11) EcO 0) Co C.= Range = 68.3 Sill = 0.54; Sill Nugget = 0.32 Nugget = 0.22 Width of Angle Class: 45° Angle for Direction: 90° 100.00 200.00 Lag Distance 0 0 0 0 ci.00 Width of Angle Class: 45° Angle for Direction: 0° 0)0 Gaussian Model .20 Nugget = 0.11 CD 0 D0 0 C0 0 0 Range = 43 m Sill = 0.26; Sill Nugget = 0.15 300.00 Meters 400.00 o 0 llllllllllllllllllllllllllllllllllllllllllllllllllllll 1111111111111 9 :! E6 0 0 Cf) 0 0 doo 50.00 100.00 150.00 Lag Distance 200.00 Meters 250.00 54 REFERENCES Burgess, T.M., R. Webster, and A.B. McBratney. 1981. Optimal interpolation and isarithmic mapping of soil properties: IV. Sampling strategy. J. of Soil Sci. 32:643-659. Cochran, W.G. 1977. Sampling Techniques. John Wiley & Sons, New York. 428 p. Matheron, G. 1971. The theory of regionalized variables and its applications. Ecoles des Mines, Fontainebleau, France. 211 p. Nash, M.H. and L.A. Daughety. 1990. Statistical comparison of soil map-unit boundaries. Soil Sci. Soc. Am. J. 54:1677-1681. Petersen, R.G., and L.D. Calvin. 1965. Sampling. pp. 5472. in C.A. Black (ed.) Methods of Soil Analysis, Part 1. Agronomy Monograph No. 9, American Society of Agronomy, Madison, Wi. Roberts, M,C. 1987. Remote-sensing and geographic information system techniques to map spatial variation of wheat grain yield. M.S. Thesis, Oregon State University, Corvallis, Or. Yates, S.R. and M.V. Yates. 1989. A Users Manual for the GEOPACK Geostatistical Software System. USDA/ARS U.S. Salinity Laboratory, Riverside, CA. 55 CHAPTER III. MAPPING STRATEGIES TO REPRESENT AND MANAGE YIELD VARIATIONS ACROSS LANDSCAPES ABSTRACT Varying fertilizer rates according to yield potential variation within fields can be a formidable task for farmers. Real-time surface organic matter or nitrate sensors are not useful in controlling variable fertilizer application in arid regions because in dryland cropping it is the water stored in the soil profile that determines yield. Map based systems must be used for soil management in this region. In the present study, digitized 7.5 minute quadrangle data were found to provide more accurate digital terrain models (DTM) than medium format (70 mm) aerial photography data. However, landscape position units delineated from aerial photographs (70 mm) were used with a geographic information systems (GIS) to generate soil management maps that correlate yield with soil water storage and landscape position. The GIS can be used to provide an initial DTM (three dimensional surface model of field topography) map for soil sampling each slope, updating the field map during data collection, and producing the final soil management unit map. 56 INTRODUCTION Farmers are aware of yield potential differences within fields. Management strategies for variable fertilizer application fall into two categories: time or map based. real- Real-time systems employ a soil surface organic matter or a soil nitrate sensor attached to the fertilizing machinery (Schueller, 1991; 1988; Gaultney et al., 1988; Grifis, 1985). Gaultney, Changes in soil nitrate or organic matter are relayed to a microprocessor that adjusts fertilizer application rate to a preset total soil concentration (Schueller, 1991). If there is no correlation between crop yield and soil surface characteristics such as surface nitrate or organic matter, then map based systems are required. In dryland areas crop yield is often correlated with characteristics of the entire soil profile, not just the surface characteristic. Water redistribution on the landscape and subsequent storage in the soil profile was correlated to winter wheat yield (Roberts, 1991; Veseth, 1986; and Wilson et al., 1991). Map based systems require digital maps and a vehicle guidance system in addition to the microprocessor variable control unit. Methods for creating digital field maps are common and relatively inexpensive. Accurate guidance systems using the global satellite positioning system are 57 becoming available. Attempts to apply technology already developed in cartographic, geographic, and remote sensing fields to large (1:10 000) scale soil management mapping are limited to a few examples. This technology was developed to study large geographic areas (small map scale, 1:250 000), such as the Great Basin of the western United States. However, Roberts (1987) and Borgelt (1988) have used GIS to analyze within field variability of yield potential and soil acidity. The objectives of the present study were: (i) to compare relative accuracies of digital terrain models (DTM) created with low level medium format (70 mm) photography data and with digitized 7.5 minute quadrangle data, (ii) to use a geographic information system (GIS) in the analysis of DTM's and landscape units derived from aerial photographs for the purpose of optimizing the sampling of landscape positions and developing variable soil management maps. MATERIALS AND METHODS Aerial Photography Aerial color and color infrared photography (scale 1:5 000 and 1:10 000) with 35 and 70 mm formats were obtained before seeding and during late vegetative growth for the Wasco and Pendleton study areas. 58 Areas with low vegetation and light colored soils were classed as low potential yield, delineated on transparent overlays using photo interpretation methods (Pain, 1981) and entered into a GIS computer system in digital format (digitized). Landscape positions were delineated on other transparent overlays and digitized. Elevation data were obtained from the photography on a twenty meter grid using the above mentioned methods (Avery and Berlin, 1985; Pain, 1981). The northing (y) and easting (x) control coordinates (state plane coordinate system - meters, USGS, 1973, 1979) were calculated by triangulation (Avery and Berlin, 1985; Pain, 1981) and digitized. The northing, easting, and elevation data were used to create digital elevation models (DEM's) using the IDRISI (Eastman, 1990) GIS system. Digital Elevation Models Field digital elevation models were obtained by digitizing elevation contours from USGS 7.5 minute quadrangles (USGS, 1973, 1979) and by the air photo method mentioned above. The northing and easting state plane (meter) coordinates were digitized. The IDRISI (Eastman, 1990) system was used to create a DEM. used to create slope and aspect maps. The data were Actual commands and procedures are similar but will vary with each field mapped. The IDRISI tutorials contain examples and 59 procedures for analysis of DEM's (Eastman, 1990). Soil Management Map Units Soil management map units were defined at a larger scale (1:5 000 for Pendleton and 1:10 000 for Wasco) using soil water (Roberts, 1991), the landscape position overlay, and the existing soil survey. The soil management map units reflect detailed phases of slope and aspect, landscape positions, and small soils areas not similar in yield potential to the soil mapped, known as inclusions, (table III.1), at a larger scale than the 1:20 000 scale of the soil survey. RESULTS AND DISCUSSION Digital Terrain Model Accuracy The DTM derived from low altitude (1000 meters) medium format (70 mm) photography was not accurate due to turbulence encountered during aerial photography. Turbulence causes aircraft and camera tilting that result in photographs not accurately representing the ground surface. Accurate large scale DTM's require special camera and photo analysis equipment capable of correcting errors caused by tilting of aircraft. The DTM obtained by digitizing elevation contours from the USGS 7.5 minute quadrangles provide an accurate maps (Figure III.1 and 111.2). The 1:24 000 scale 60 quadrangle is essentially enlarged to 1:5 000 or 1:10 000 -- with the inherent accuracy limitations of a 1:24 000 map. The DTM produced from 70 mm aerial photography was not acceptable because of elevation data errors near the photo edge. The DTM derived from quadrangle data also has some edge irregularities. Figure II1.1 indicates obvious regions of southwest aspect, > 12% slopes (white) along the western edge that do not exist when field verified. All maps may have some edge irregularities. These areas can be marked in the field, and the map updated. Questions of map resolution, scale, and accuracy must be addressed. Large scale (1:10 000) maps are required to represent detailed (resolution) soil units occuring in fields -- and therefore, require equally detailed (same scale) topographic and soil characteristic data to create the map. In the future, DTM's could be produced by tracking northing, easting, and elevation with the satellite navigation methods. Reconnaissance Sampling of Landscape Positions To save time, soil and yield sampling transects can be delineated across hills on the surface plot (DTM) of the field (Figure III.1, and 111.2), before visiting the field. This will ensure that all necessary landscape 61 positions representing the variability in the field can be sampled. Once in the field, the location of the transects and sampling plan can be verified quickly, even in large - 300-400 ha -- fields with many hills. Creation of Soil Management Maps Soil management maps were created from landscape positions delineated on aerial stereo-pair (Pain, 1981) photographs and digitized with the IDRISI (Eastman,1990) GIS (Figure 111.3 and 111.4). Lower than normal rainfall depressed yield at the Pendleton study area (Figure 111.3), therefore, soil management units represent phases of slope and aspect where yield differences likely will occur in normal rainfall years. At the Wasco study area, soil management map units (Figure 111.4 and Table III.1) indicate north and east foot (WaEF, WbN F) and north back slopes (WdNB) requiring additional fertilizer, and shallow soil (CaS) areas requiring no fertilizer. be fertilized with current management rate. Other areas can (Roberts, 1991) The value of this system is that once a map is made, soil water can be measured on every landscape position or soil management unit each year and fertilizer can be adjusted to the soil water (Roberts, 1991). Every year a soil management map can be updated with current soil data economically using the GIS. 62 Table II1.1 Soil management map unit legend with family or higher taxonomic class. Pendleton Study Area Symbol Name Coarse-silty, mixed, mesic Typic Haploxerolls WaA WaBN WaCN WaDN WaFTN WaCW WaDW WaDRN Walla Walla silt loam, 0 to 3 percent slopes Walla Walla silt loam, 3 to 7 percent north slopes Walla Walla silt loam, 7 to 12 percent north slopes Walla Walla sift loam, 12 to 20 percent north slopes Walla Walla silt loam, 0 to 6 percent foot and toe north slopes Walla Walla silt loam, 7 to 12 percent west slopes Walla Walla silt loam, 12 to 20 percent west slopes Walla Walla silt loam, 0 to 10 percent drainage way north slopes Wasco Study Area Symbol Name Coarse-silty, mixed, mesic Typic Haploxerolls WaA WaEF WaNT WbE WbEB WbN F WcS WcNSO WdNB Walla Walla silt loam, 0 to 3 percent slopes Walla Walla silt loam, 0 to 3 percent east foot slopes Walla Walla silt loam, 0 to 3 percent north toe slopes Walla Walla silt loam, 3 to 9 percent east slopes Walla Walla silt loam, 3 to 9 percent east back slopes Walla Walla silt loam, 3 to 7 percent north foot slopes Walla Walla sift loam, 7 to 12 percent south slopes Walla Walla silt loam, 7 to 12 percent north shoulders Walla Walla silt loam, 12 to 20 percent north back slopes Loamy, mixed, mesic Lithollic Xerollic Camborthids CaS Starbuck silt loam, 3 to 12 percent south slopes 63 a) 0 O N 0 0 r-1 0 , C1 W > W $.4 .Q ,1 0> W r0 0 fd 0 gj ° W 0 (-1 g$ 0 >4o ,--1 Tr 'SI 4..) to k-1 cc 00 Oq 0 im .E_n4 . .4.) ,-I , :5 w t Lr) CI, -Pro (1) ..1 .4,a) o4;cir4 0 4 4..) Zr4 cr,t31 ,3) ,-1 al a) te Pl.> r-itt 0 Oty,r/t 0 4-4 ,I a)' 0,-I 7:5 Oz 2 0 _ Z c rocti -,...1 cd P ..1 °I la) N . 4 -, W v.-. 4 (U -F.4 0 A Z Figure 111.2 Digital terrain model for the Wasco study area derived from the Wasco USGS View direction is 110°, 30° above horizon, vertical exaggeration 7.5 Minute Quadrangle. 0.7:1. 0 100 meters N 5047142 E 684633 N 504715U N 50J E 94 68528 5-3.01 Figure 111.3 Soil management map units for the Pendleton study area. 111.1. N 5066814 E 350081 N 5066833 E 349729 IV N 5067207 E 350013 0 50 meters 100 WaDN WaFTN Legend, table Figure 111.4 Soil management map units for the Wasco study area. N 5047142 E 604633 Legend, table 111.1. N 5047159 Reference Power Pole E WaNT 605301 WdNB We S Hwy 206 -) CaS N WcS 0 50 100 meters CaS _---- WhEB-C-1--WaNT N 5046394 \ E 685289 67 REFERENCES Avery, T.E., and G.L. Berlin. 1985. Interpretation of Aerial Photographs, Fourth ed. Burgess Publishing Co. Minneapolis, Mn. 554 p. Borgelt, S.C. 1988. Geographic information systems for on-farm use. ASAE paper 88-1606 Colburn, J. 1986. R&D on a fertilizer sensor and control system. US Dept. of Energy DOE/ID/12518-1. Eastman, J.R. 1990. IDRISI a grid based geographic information system. Graduate School of Geography, Clark Univ., Worcester, MA. Gaultney, L.D. 1988. Soil physical properties sensing. Research Activities. Purdue Univ. Dept. Agric. Eng. 45. J.K. Schueller, J.L. Shonk, and Z. YU. Automatic soil organic matter mapping. ASAE paper 88-1607. , 1988. Grifis, C.L. 1985. Electronic sensing of soil organic mater. Trans. ASAE 28(3):703-705. Pain, D.P. 1981. Aerial photography and image interpretation for resource management. John Wiley & Sons, New York. 571 p. Roberts, M.C. 1991. Fertilizing wheat with nitrogen balanced to soil water variability across landscapes. in press. Ph.D. Thesis, Oregon State University, Corvallis, Oregon. 1987. Remote-sensing and Geographic Information System Techniques to Map Spatial Variation of Wheat Grain Yield. M.S. Thesis, Oregon State University, Corvallis, Or. Schueller, J.K. 1991. A review and integrating analysis of spatially-variable control of crop production. Fertilizer Review, in press. U.S. Geological Survey. 1973. Wasco, Oregon Quadrangle - Sherman Co. 7.5 Minute Series. (Topographic). N4530--W12037.5/7.5 U.S. Govt. Print. Office, Washington DC. 68 U.S. Geological Survey. 1973. Barnhart Oregon Quadrangle -- Umatilla Co. 7.5 Minute Series. (Topographic) N4537.5--W11852.5/7.7 U.S. Govt. Print. Office, Washington DC. Veseth, R. 1989. Variable fertilizer application improves profits and conservation. in Veseth, R., and D. Wysocki (eds) Steep Extension Conservation Farming Update. Steep extension program, Univ. of Idaho, College of Agriculture, Moscow, ID. Wilson, J.P., S.P. Sandor, and G.A. Nielsen. 1991. Productivity index model modified to estimate variability of Montana small grain yields. Soil Sci. Soc. Am. J. 5:228-234. 69 CONCLUSIONS Realizing that variable stored soil water across landscape positions is the first constraint to potential yield, farmers can sample soil water at each landscape position along a few well placed transects. Fertilizer nitrogen can be adjusted to stored soil water. Residual nitrogen, soil water, and landscape position data can be managed with a geographic information system, and soil management maps produced. With some simple maps farmers may be able to fertilize each landscape position with nitrogen adjusted to stored soil water. Economical vehicle guidance systems may soon be available to automate variable fertilization. Spatially-variable management may be applied to other crop production practices. Plowing gently sloping convex landscape positions may suppress downy brome infestations. No-till planting on steep back slopes and shoulders may help control erosion. Planting rate may be increased in high yielding areas of fields and decreased in low yielding areas. Varieties may be chosen for specific yield potential areas. Barley could be planted on shallow soil south slopes. Maturing earlier than wheat, barley may use limited soil moisture more efficiently. Weed populations, such as downy 70 brome, often are more severe on high potential yield areas. Applying herbicides only to portions of fields with threshold weed populations may reduce pesticide use. Farmers may be able to store low protein (higher quality) soft white wheat harvested on high yielding north slopes separately. After a few years of variable fertilizer management, excessive residual soil nitrogen may be lowered to acceptable levels and grain protein will be acceptable for wheat grown on all landscape positions within the field. Farmers using this information can save money on fertilizer, chemicals, and fuel, while protecting surface and groundwater resources. 71 BIBLIOGRAPHY Aase, J.K., and F.H. Siddoway. 1981. Assessing winter wheat dry matter production via spectral reflectance measurements. Remote Sensing of Environment 11:267277. Ahlrichs, J.S., and M.E. Bauer. 1983. Relation of agronomic and multispectral reflectance characteristics of spring wheat canopies. Agron. J. 75:987-993. Asrar, G., E.T. Kanemasu, R.D. Jackson, and P.J. Pinter, jr. 1985. Estimation of total above-ground phytomass production using remotely sensed data. Remote Sensing of Environment 17:211-220. and M. Yoshida. 1985. Estimates of leaf area index from spectral reflectance of wheat under different cultural practices and solar angle. Remote Sensing of Environment 17:1-11. , Avery, T.E., and G.L. Berlin. 1985. Interpretation of Aerial Photographs, Fourth ed. Burgess Publishing Co. Minneapolis, Mn. 554 p. Beckett, P.H.T., and R. Webster. 1971. Soil variability: a review. Soils and Fertilizers 34(1):1-15. Borgelt, S.C. 1988. Geographic information systems for on-farm use. ASAE paper 88-1606 Bouma, J. 1985. Soil variability and soil survey. pp 130-149 in D.R. Neilsen, and J. Bouma (Eds). 1985. Soil Spatial Variability. Proceedings of a workshop of the ISSS and the SSSA, Las Vegas, Nevada. Bregt, A.K., J. Bouma, and M. Jellinek. 1987. Comparison of thematic maps derived from a soil map and from kriging of point data. Geoderma, 39:281-291. Bresler, E., S. Dasberg, D. Russo, and G. Dagan. 1981. Spatial variability of crop yield as a stochastic soil process. Soil Sci. Soc. Am. J. 45:600-605. Burgess, T.M., and R. Webster. 1980. Optimal interpolation and isarithmic mapping of soil properties: I. The semi-variogram and punctual kriging. J. of Soil Sci. 31:315-331. 72 1980. Optimal interpolation and isarithmic mapping of soil properties: II. Block kriging. J. of Soil Sci. 31:333-341. , and A.B. McBratney. 1981. Optimal interpolation and isarithmic mapping of soil properties: IV. Sampling strategy. J. of Soil Sci. 32:643-659. , Cerrato, M.E. and A.M. Blackmer. 1990. Comparison of models for describing corn yield response to nitrogen fertilizer. Agron. J. 82:138-143. Cochran, W.G. 1977. Sampling Techniques. John Wiley and Sons, New York. 428 p. Colburn, J. 1986. (Aguila Corp.) R&D on a fertilizer sensor and control system. U.S. Department of Energy, Washington, D.C. DOE/ID/12581-1 (DE87014929). Daughtry, C.S.T., M.E. Bauer, D.W. Crecelius, and M.M. Hixon. 1980. Effects of management practices on reflectance of spring wheat canopies. pp. 1-20. AgRISTARS no. SR -PO- 00458, NAS9-15466; Technical Report May 1980, Laboratory for Applications of Remote Sensing, Purdue University, West Lafayette, Indiana. Dozier, J., and A.H. Strahler. 1983. Ground investigations in support of remote sensing. pp. 959-986. in R.N. Colwell (ed.), and D.S. Simonett, F.T. Ulaby. (Vol. 1, (eds.) Manual of Remote Sensing. V (1). Amer. Soc. of Photogrammetry. Falls Church, Va. Eastman, J.R. 1988. IDRISI a grid based geographic information system. Graduate school of geography, Clark University. Worcester, Ma. (loose leaf) Fairchild, D.S., and R.D. Knutson. 1989. Innovative equipment for application of agricultural inputs by soil variations. p.315 in Agronomy abstracts. ASA, Madison, WI. Gardner, H., and N.R. Goetze. 1980. O.S.U. Cooperative Extension Service. Fertilizer guide for winter wheat (non-irrigated-- Columbia Plateau) FG 54. Oregon State University. Gates, D.M., H.J. Keegan, J.C. Schleter, and V.C. Weidner. 1965. Spectral properties of plants. Applied Optics 4(1):11-20. 73 Gaultney, L.D. 1988. Soil physical properties sensing. Research Activities. Purdue Univ. Dept. Agric. Eng. 45. J.K. Schueller, J.L. Shonk, and Z. YU. 1988. Automatic soil organic matter mapping. ASAE paper 88, 1607. Grifis, C.L. 1985. Electronic sensing of soil organic mater. Trans. ASAE 28(3):703-705. Hammond, M.W., D.J. Mulla, and D.S. Fairchild. 1988. Development of management maps for spatially variable soil fertility. Proc. 39th Annual Far West Regional Fertilizer Conference, Bozeman, Montana. July 11-13, 1988. Hatfield, J.L. (ed.) 1990. Instructions to Authors (statistical methods). Agron. J. 82:171-172. Hillel, D. 1982. Introduction to soil physics. Press, Orlando, Fl. Academic Huddleston, J.H. 1982. Soils of Oregon: summaries of physical and chemical data. OSU Extension Service, special report 662. Oregon State University, Corvallis, OR. Johnson, D.R., and A.J. Makinson. 1988. Soil Survey of Umatilla County Area, Oregon. USDA-SCS. U.S. Govt. Print. Office, Washington DC. Knipling, E.B. 1970. Physical and physiological basis for the reflectance of visible and near-infrared radiation from vegetation. Remote Sensing of Environment. 1:155159. Lillisand, T.M., and R.W. Kiefer. 1979. Remote sensing and image interpretation. John Wiley & Sons, New York. 612 p. Marble, D.F., H.W. Calkins, and D.J Peuquet. 1984. Basic readings in geographic information systems. Spad Sys. LTD. Williamsville, N.Y. 1 V. (loose leaf). and D.J. Peuquet. 1983. Geographic information systems and remote sensing. pp. 923-958. in R.N. Colwell (ed.), and D.S. Simonett, F.T. Ulaby. (Vol. 1, (eds.) Manual of Remote Sensing. V (1). Amer. Soc. of Photogrammetry. Falls Church, Va. , 74 Mayers, L.R. 1959. Soil survey of Sherman County, Oregon. USDA-SCS. Government Printing Office, Washington D.C. pp. 1-104, sheet no. 14-15. McBratney, A.B., and R. Webster. 1983. Optimal interpolation and isarithmic mapping of soil properties: V. Coregionalization and multiple sampling strategy. J. of Soil Sci. 34:137-162. Miller, M.P., M.J. Singer, and D.R. Nielsen. 1988. Spatial variability of wheat yield and soil properties on complex hills. Soil Sci. Soc. Am. J. 52:1133-1141. Montgomery, H. 1991. GPS takes off. GPS World. Aster Publishing Co. P.O. Box 10460, Eugene, OR. February, 1991 pp 18-19. Morkoc, F., J.W. Biggar, D.R. Nielsen, and D.E. Meyers. 1987. Kriging with generalized covariances. Soil Sci. Soc. Am. J. 51:1126-1131. Mulla, D.J., and M.W. Hammond. 1988. Mapping of soil test results from large irrigation circles. Proc. 39th Annual Far West Regional Fertilizer Conference, Bozeman, Montana. July 11-13, 1988. Myers, V.I. 1983. Remote sensing applications in agriculture. pp. 2111-2228. in R.N. Colwell (ed.), and J.E. Estes, G.A. Thorley (Vol. 2, eds.) Manual of Remote Sensing. V (2). Amer. Soc. of Photogrammetry. Falls Church, Va. and W.A. Allen. 1968. Electrooptical remote sensing methods as nondestructive testing and measuring techniques in agriculture. Applied Optics 7(9):1819, 1838. Nash, M.H. and L.A. Daughety. 1990. Statistical comparison of soil map-unit boundaries. Soil Sci. Soc. Am. J. 54:1677-1681. Neilsen, D.R., and J. Bouma (Eds). 1985. Soil Spatial Variability. Proceedings of a workshop of the ISSS and the SSSA, Las Vegas, Nv. 243 p. Nelson, L.A., and S.W. Buol. 1990. Experimental designs to evaluate crop response on adjacent soil mapping units. Soil Sci. Soc. Am. J. 54:841-849. 75 Nelson, D.W., and L.E. Sommers. 1973. Determinations of total nitrogen in plant material. Agron. J. 65:109112. Pain, D.P. 1981. Aerial photography and image interpretation for resource management. John Wiley & Sons, New York. 571 p. Petersen, R.G. 1985. Design and analysis of experiments. Marcel Dekker, Inc. New York. 429 p. and L.D. Calvin. 1965. Sampling. pp. 54-72. C.A. Black (ed.) Methods of Soil Analysis, Part 1. Agronomy Monograph No. 9, American Society of Agronomy, Madison, Wi. in Roberts, M.C. 1991. Fertilizing wheat with nitrogen balanced to soil moisture variability across hilislopes. in press. Ph.D. Thesis, Oregon State University, Corvallis, Oregon. 1987. Remote-sensing and geographic information system techniques to map spatial variation of wheat grain yield. M.S. Thesis, Oregon State University, Corvallis, Or. . Russo, D. 1984. A geostatistical approach to the trickle irrigation design in a heterogeneous soil: 2. A field test. Water Resour. Res. 20(5):543-552. Star, J., and J. Estes. 1990. Geographic information systems an introduction. Prentice Hall, Englewood Cliffs, New Jersey. 303 p. Schueller, J.K. 1991. A review and integrating analysis of spatially-variable control of crop production. Fertilizer Review, in press. 1988. Machinery and systems for spatiallyvariable crop production. Paper no. 88-1608. ASAE, St. Joseph, MI. . Soil Survey Staff. 1975. Soil Taxonomy. U.S. Dept. Agr. Handbook No. 436. USDA-SCS. U.S. Government Printing Office, Washington D.C. 754 p. U.S. Geological Survey. 1973. Wasco, Oregon Quadrangle - Sherman Co. 7.5 Minute Series. (Topographic). N4530--W12037.5/7.5 U.S. Govt. Print. Office, Washington DC. 76 U.S. Geological Survey. 1979. Barnhart Oregon Quadrangle -- Umatilla Co. 7.5 Minute Series. (Topographic) N4537.5--W11852.5/7.7 U.S. Govt. Print. Office, Washington DC. Van Scoyoc, G.E., D.G. Schulze, L.D. Gaultney, J.L. Nagel, and J. Schonk. 1989. Variable rate application of chemicals using a soil organic matter sensor. p. 319 in Agronomy abstracts. ASA, Madison, WI. Veseth, R. 1989. Variable fertilizer application improves profits and conservation. in Veseth, R., and D. Wysocki (eds) Steep Extension Conservation Farming Update. Steep extension program, Univ. of Idaho, College of Agriculture, Moscow, ID. Vieria, S.R., J.L. Hatfield, D.R. Nielsen, and J.W. Biggar. 1982. Geostatistical theory and application of variability of some agronomical properties. Hilgardia 51(3):1-75. Warric, A.W., and W.R. Gardner. 1983. Crop yield as affected by spatial variations of soil and irrigation. Water Resour. Res. 19(1):181-186. Webster, R., and T.M. Burgess. 1980. Optimal interpolation and isarithmic mapping of soil properties: III. Changing drift and universal kriging. J. of Soil Sci. 31:505-524. ., 1985. Quantitative spatial analysis of soil in the field. in ed., Advances in Soil Science vol. 3. Springer Verlag. New York. Wiegand, C.L., A.J. Richardson, and E.T. Kanemasu. 1979. Leaf area index estimates for wheat from LANDSAT and their implications for evapotranspiration and crop modeling. Agron. J. 71:336-342. Wilson, J.P., S.P. Sandor, and G.A. Nielsen. 1991. Productivity index model modified to estimate variability of Montana small grain yields. Soil Sci. Soc. Am. J. 5:228-234. Yates, S.R. and M.V. Yates. 1989. A Users Manual for the GEOPACK Geostatistical Software System. USDA/ARS U.S. Salinity Laboratory, Riverside, CA. APPENDIX 77 Location of Study Areas The Pendleton study area The field in the Pendleton study area is about 4 km west by northwest of the Pendleton airport. The approximately 18 ha field is located in the SW1/4 of the NW1/4 of section 15, Township 3 North, Range 31 East Willamette Principle Meridian. The field is bordered on the south by the county road, west by a field road, east by a driveway, and north by a gulch. The Wasco study area The study area, approximately 40 ha, near Wasco, Sherman County, Oregon, is located about 8 km southeast of Wasco, Oregon, and about 8-10 km northwest of Moro, Oregon. The field location is the SE1/4 section, of section 24, Township 1 North, Range 17 East Willamette Principle Meridian. The triangular shaped field is bordered on the north by a county road, then an unimproved road. The east boundary is by the same unimproved road bending south east, then south to highway 206. The highway (206) is the south west boundary. Wasco (Moore) study area transects with sample locations plotted. Figure A.1. N 5047142 E 684633 N 5047159 Reference Power Pole 685381 E ransect 2 1 + Traect. 1 4 ++ +++ + 4-i++++4 iNT Hwy 206 - 0 50 100 Meters N E 504639N4 605209 .1N Figure A.2. Pendleton (Timmerman) study area transects with sample locations plotted. N5 N 5066814 E 349729- E 350081 ++ r ++ +++ ++ +++ Transect 3 +++ +++ + + Transect 1 ++ + ++ TAT 41- ++ + + Transect 4 I + Transect 2 N 5067207 E 50 350013 meters 100 + + + + + Figure A.3. Wasco (Moore) study area transects 2, soil profile characteristics. Core Number 50.00 8.00 6.00 4.00 2.00 0.00 0.00 10.00 O 0 0 12.00 00 16.00 14.00 0 A 0 0 A 100.00 O 0 na) A 150.00 0 L_ IL A 200.00 0 A A A O 0 0 C) °COCO Lithic Contact 6.AA6v1 Effervescence Indurated MM. U) 250.00 2nd Indurated Figure A.4. Wasco (Moore) study area transects 2, core 1, soil water (A), and nitrate (B), profile data. A Equivalent Depth H2O 0.00 0.00 1.00 3.00 2.00 4.00 5.00 5.00 3.00 4.00 1.00 2.00 0.00 0.00 ..1111,111,11iii..ili,.111111111,111 6.00 0. Do_41es! E Residual NO3 (kg ha-1) (cm) E 50.00 - 50.00 6 cI. _c 100.00 100.00 CL a ia) CD 150.00 - 150.00 :7= 4- 0 L_ O 0- 200.00 0 CL 200.00 -1 ()COCO Initial H2O MOOD Final H2O H2O 0 Available Water (17 444446-1.5 Moo U) 00000 Crop 250.00 trtdotra Residual NO3 250.00 co Figure A.5. Wasco (Moore) study area transects 2, core 2, soil water (A), and nitrate (B), profile data. B A Equivalent Depth H2O 0.00 0.00 E 2.00 4.00 6.00 8.00 Residual NO3 (kg ha-1) (cm) 10.00 0.00 0.00 12.00 E 50.00 50.00 _c 100.00 _C 100.00 CL CL 0) 0) U 12.00 8.00 I 11t 1 it111111111iim11ii ii iii16.00 4.00 150.00 150.00 0) 0 0 200.00 a_ 200.00 0Q0Q0 Initial H2O 000:11:/ Final H2O 0 44Ls44 1.5 Mpa (/) 00000 Crop 250.00 H2O Available Water 0 *frtotrtr Residual NO3 to 250.00 Figure A.6. Wasco (Moore) study area transects 2, core 3, soil water (A), and nitrate (B), profile data. A Equivalent Depth H2O 0.00 0.00 E U c 3.00 4.00 2.00 1.00 1111111111111111111111111111111 5.00 Residual NO3 (kg ha-1) (cm) 6.00 0.00 0.00 7.00 lllll 111111 lllll 11111111111111 50.00 - E 10.00 6.00 8.00 4.00 11,111111111ii1111111111iimitu,1111111 2.00 50.00 c 100.00 100.00 4-, - CIL a) a) 0 150.00 150.00 a) a) O O L_ CL 200.00 200.00 -_ OCOCX) Initial H2O CIDCLI Final H2O 0 44444 -1.5 Mpa U) 00000 Crop 250.00 H2O 0 Itntraitati Residual NO3 Available Water 250.00 - Figure A.7. Wasco (Moore) study area transects 2, core 4.5, soil water (A), and nitrate (B), profile data. A Equivalent Depth H2O 0.00 0.00 E 0 0) 1.00 4.00 3.00 2.00 Residual NO3 (kg ha-1) (cm) 0.50 0.00 5.00 E 50.00 1.00 1.50 2.00 2.50 3.00 50.00 : 0 150.00 150.00 0 /11 0 0 200.00 -.= CI- 200.00 000C10 Initial H2O 7..= 001:11:113 Final H2O 0 4444? -1.5 Mpa V) 06000 Crop 250.00 H2O Available Water 0 Ltot*tdr Residual NO3 V) 250.00 03 Figure A.8. Wasco (Moore) study area transects 2, core 6, soil water (A), and nitrate (B), profile data. A Equivalent Depth H2O 0.00 0.00 E 1.00 2.00 3.00 4.00 Residual NO3 (kg ha-1) (cm) 5.00 0.50 0.00 6.00 E 50.00 1.00 1.50 2.00 2.50 3.00 50.00 0 0 _c 100.00 100.00 b a a) 150.00 150.00 a) O U- 200.00 200.00 00000 Initial H2O EICICI0 Final H2O 0 44Atit, -1.5 Mpa (f) 00000 Crop 250.00 triatrtitr Residual NO2 H2O Available Water 250.00 Figure A.9. Wasco (Moore) study area transects 2, core 7, soil water (A), and nitrate (B), profile data. A Equivalent Depth H2O 0.00 0.00 E 1.00 2.00 3.00 B Residual NO3 (kg ha-1) (cm) 4.00 0.00 0.00 5.00 E 50.00 2.00 4.00 6.00 8.00 50.00 0 0 100.00 0_ a) 150.00 150.00 a) 4- 0 0 200.00 Q_ OCOQD Initial H2O = OO Final H2O :L17 0 44LA,4, 1.5 MDa (J) 00000 Crop 250.00 200.00 H20 Available Water 0 tatitcAntr Residual U.) 250.00 1%,10a Figure A.10. Wasco (Moore) study area transects 2, core 8, soil water (A), and nitrate (B), profile data. A Equivalent Depth H2O 0.00 1.00 E iiiiii 2.00 0.00 3.00 Residual NO3 (kg ha-1) (cm) 4.00 0.00 5.00 6 A 50.00 -- 0 2.00 4.00 6.00 8.00 10.00 0.00 E 50.00 0 _c 100.00 100.00 CL CL CD (1) 150.00 (i) a) 150.00 4= 0 0 200.00 7 - 0 _ _ (f) - 250.00 0_ 200.00 0:1000 initial H2O 130000 Final H2O 44444 -1.5 Mpa H2O 00000 Crop Available Water tatotatntr Residual NO3 In 250.00 Figure A.11. Wasco (Moore) study area transects 2, core 9, soil water (A), and nitrate (B), profile data. A Equivalent Depth H2O 0.00 0.00 1.00 2.00 3.00 --ID Residual NO3 (kg ha-1) (cm) 0.00 0.00 5.00 4.00 1.00 1.50 2.00 2.50 3.00 -0 11 E 0.50 50.00 E 100.00 _c - 50.00 0 100.00 = I. CL 0) (1) 0 (i) 150.00 150.00 Z4-7- 0 L. O 200.00 CI- ooh Final H2O 0 kzerere. -1.5 Mpa H2O Crop Available Water (/) WOO 250.00 200.00 = 00030 Initial H2O 0 *Jr*** Residual NO3 CT) 250.00 :- Figure A.12. Pendleton study area (Timmerman) transects 1, soil profile characteristics. Core Number E:- 0 100.00 6.00 4.00 2.00 0.00 0.00 A A A A A A A 8.00 10.00 A A A 0 0 12.00 0 I=1 0 o o 0 O 0 O Cl) 0 00000 Lithic Contact ,6A6A6, Effervescence LI I Indurated I 400.00 0 I _I 00000 2nd Indurated 03 V) Pendleton study area (Timmerman) transects 1, core 1, soil water (A), and Figure A.13. nitrate (B), profile data. A Equivalent Depth H2O 0.00 0.00_ iiiii E 1.00 2.00 3.00 Residual NO3 (kg ha-1) (cm) 5.00 4.00 0.00 0.00 6.00 Il111111111 E 50.00 20.00 40.00 60.00 80.00 100.00 120.00 50.00 C) r 100.00 _c 100.00 CL a) 0 150.00 150.00 (1) 0 0 4- 0 L O L_ 200.00 OCOOD Initial H2O 00000 Final H2O 0 44A1^. 1.5 Mpa (I) 00000 Crop 250.00 (Eq. Depth) (Eq. Depth) H2O viEq. Depth) Available ater (Eq. Depth) CI- 200.00 O tritrtctitr Residual NO3 250.00 Figure A.14. Pendleton study area (Timmerman) transects 1, core 2, soil water (A), and nitrate (B), profile data. A Equivalent Depth H2O Residual NO3 (kg ha-1) (cm) 0.00 0.00 6.00 5.00 4.00 2.00 3.00 1.00 0.00 0.00 ttttuutluttuutltutttut luutttttlutuuuluuutttl E E 50.00 - 20.00 40.00 60.00 80.00 100.00 120.00 50.00 0 0 4) c 1 o o .00 8 150.00 150.00 a) 0 a) 4= 14.= U- 200.00 (1- 200.00 0 O L._ 0 021000 Initial H2O 00OLIO Final H2O 444,40, -1.5 Mpa H2O 00000 Crop Available Water (!) 250.00 Ittto&rtr Residual NO3 O cn 250.00 Pendleton study area (Timmerman) transects 1, core 3, soil water (A), and Figure A.15. nitrate (B), profile data. A 0.00 0.00 E 1.00 2.00 3.00 q 0.00 0.00 6.00 E 20.00 40.00 60.00 80.00 50.00 0 0 s 5.00 4.00 50.00 - Residual NO3 (kg ha-1) (cm) Equivalent Depth H2O _C 100.00 100.00 -F, CL -4- 0 150.00 N 150.00 4_ O 0 a- 200.00 = 0 200.00 0000D Initial H2O 00000 Final H2O 4eiLIA4, -1.5 Mpa H2O Crop Available Water 0 ***tett Residual NO2 000 (f) 250.00 250.00 Figure A.16. Pendleton study area (Timmerman) transects 1, core 5, soil water (A), and nitrate (B), profile data. A 0.00 0.00 E Residual NO3 (kg ha-1) (cm) Equivalent Depth H2O 4.00 5.00 0.00 0.00 6.00 2.00 3.00 1.00 111111111111111 lllll i111111111111111t11111111J11111 b 50.00 U 10.00 20.00 30.00 lllll 40.00 50.00 60.00 50.00, 6 _c 100.00 100.00 4, 0 CL a) (1) 9 150.00 150.00 El 0 (1) 0 200.00 = Q_ CXX)C0 Initial H2O 01:1000 Final H2O 4444,0, 1.5 Mpa 00000 Crop 250.00 200.00 = = H2O Available Water 0 totdatio:r Residual NO3 to 250.00 to L.) Pendleton study area (Timmerman) transects 1, core 6, soil water (A), and Figure A.17. nitrate (B), profile data. B A Equivalent Depth H2O 0.00 0.00 E 4.00 2.00 Residual NO3 (kg ha-1) (cm) 0.00 0.00 6.00 E 50.00 10.00 20.00 30.00 40.00 iiiii50.00 50.00 0 0 4A 150.00 150.00 0) 0) A 0 L_ 0 L._ IZ 200.00 O 0_ 200.00 OCXXX) Initial H2O 011:100 Final H2O 444.44, Cf) 1.5 Mpa 00000 Crop 250.00 H2O Available Water 0 Litot7Idt Residual NO3 (r) 250.00 :1 Pendleton study area (Timmerman) transects 1, core 7, soil water (A), and Figure A.18. nitrate (B), profile data. A Equivalent Depth H2O 0.00 0.00 1.00 2.00 3.00 4.00 Residual NO3 (kg ha-1) (cm) 5.00 0.00 0.00 6.00 E 50.00 20.00 40.00 60.00 80.00 50.00 - U I.Q .c 4, 100.00 0 0 150.00 150.00 a) 0 0 L O 0 100.00 200.00 CI- 200.00 00000 Initial H2O COCCID Final H2O 44444 -1.5 Mpa O 00000 Crop 250.00 - 1120 Available Water 0 **tat& Residual NO3 (f) 250.00 Pendleton study area (Timmerman) transects 1, core 8, soil water (A), and Figure A.19. nitrate (B), profile data. A '"'"'"I N ..E 0 A 50.00 20.00 40.00 60.00 80.00 -1) 100.00 120.00 0. EC- E 0.00 0.00 8.00 6.00 4.00 2.00 0.00 0.00 Residual NO3 (kg ha (cm) Equivalent Depth H2O 50.00 0 \ A _C 0 /2 150.00 0 \ \ A 100.00 N 0 a) \ 150.00 \ 0 O O a_ 200.00 0 200.00 00000 Initial H2O 00000 Final H2O 44/10.Li, -1.5 Mpa H2O 004500 Crop Available Water 250.00 0 1VdcArtr Residual NO3 01 250.00 Pendleton study area (Timmerman) transects 1, core 9, soil water (A), and Figure A.20. nitrate (B), profile data. B A 1.00 E 50.00 o .. 5.00 \ d % A ,/ L. - - di E I ....- --- I -.. 100.00 -4---, o a 50.00 i1111111111 t / . A / .0 4--- 50.00 - CL b 150.00 40.00 _c 100.00 '0 N CL a) 30.00 0.00 10.00 20.00 111111111111111111111111111111 0.00 6.00 9 A . .,---", 4.00 3.00 2.00 Residual NO3 (kg ha-1) (cm) Equivalent Depth H2O 0.00 0.00 0 o a) 150.00 a) O O eL 200.00 0 200.00 00000 Initial H2O 00000 Final H2O 44644 1.5 Mpa U9 00000 Crop 250.00 1-120 Available Water 0 litc*Cr Residual NO3 00 250.00 Pendleton study area (Timmerman) transects 1, core 10, soil water (A), and Figure A.21. nitrate (B), profile data. A omo 1[11 ttttt 5.00 4.00 3.00 2.00 Residual NO3 (kg ha-1) (cm) Equivalent Depth H2O 0.00 tttttt 40.00 30.00 10.00 20.00 0.00 1111111111111111111111111111111111111111 0.00 50.00 / E A 50.00 50.00 = Q. 0 0 I c ..... 100.00 A 4 CL _c ...._ --. a) 0 U) 13, Q d a) 0 150.00 I b 100.00 - CL . ci \ / 150.00 *,;:= d) 7...= 0 0 0 +, .... .... 200.00 0 0 000Q0 Initial H2O 00000 Final H2O 40444 -1.5 Mpa 060() Crop 250.00 200.00 = totteokft Residual NO3 1120 Available Water 250.00 Figure A.22. Pendleton study area (Timmerman) transects 2, soil profile characteristics. Core Number 0.00 0.00 1.00 2.00 F 0 100.00 A 3.00 A 4.00 5.00 A 0 0 O 300.00 O 0 7.00 A A a) 6.00 O 0 O 0_ alto Lithic Contact O (r) O 400.00 O AAAAA, Effervescence I Indurated I I I I I 00000 2nd Indurated to Figure A.23. Pendleton study area (Timmerman) transects 2, core 1, soil water (A), and nitrate (B), profile data. A 2.00 1.00 0.00 0.00 11111 liallitiltiattlittillitilmitlittl R E g i ` 1 \ \ R II 50.00 E \ 10.00 20.00 30.00 40.00 50.00 60.00 70.00 lit1uuuluuuutt uuuutluu111111uu11i 50.00 0 ........., \ _c 0.00 ,------, Q U 0.00 6.00 5.00 Ittili11111 4.00 3.00 Residual NO3 (kg ha-1) (cm) Equivalent Depth H2O 100.00 \ _c 0 1 CL a) a 150.00 \ d 4= o / 0 100.00 - CIL Q) ? -.7-._ 150.00 -(1.) ..... 0 t_ 0 Cl- 200.00 CI- 0:XXXX) Initial H2O 00000 Final H2O 0 444.44 -1.5 Mpa Ui 0000 250.00 Crop 200.00 ;7= H2O Available Water 0 tuktritintr Residual NO3 (1) 250.00 Pendleton study area (Timmerman) transects 2, core 2, soil water (A), and Figure A.24. nitrate (B), profile data. A Residual NO3 (kg ha-1) (cm) Equivalent Depth H2O 5.00 4.00 3.00 0.00...I.1.00 iiiii 2.00 II iiiii 1.1111.111.1.1 0.00 0.00 50.00 0 1 b A / i b b.\ / b 0 a) 150.00 0 I I I V ,f? 60.00 50.00 0 ..,, _c 100.00 4-, CL I a) . --0 a (!? \ 0 0 150.00 14= 0k 0 21- 50.00 / / I CL E ... P14 _c 100.00 40.00 9 4 . E 30.00 20.00 10.00 iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii 1111111111111111111111 0.00 200.00 0 00000 Initial H2O 00000 Final H2O 40.NO4 1.5 Mpa H2O 00000 Crop Available Water 250.00 a- 200.00 :7= 0 Residual NO3 (/) 250.00 Figure A.25. Pendleton study area (Timmerman) transects 2, core 3, nitrate (B), profile data. soil water (A), and A Equivalent Depth H2O 4.00 2.00 Residual NO3 (kg ha-1) (cm) 6.00 0.00 8.00 0.00 aaaaa 10.00 20.00 30.00 40.00 50.00 111111111111111111111111111(111ft 0.00 E 50.00 E 50.00 I. a 100.00 c 100.00 (i) CL a) 150.00 150.00 -o 4- O 200.00 O 200.00 000C0 Initial H2O 0C10013 Final H2O 44444 -1.5 Mpa 00000 Crop 250.00 -7 ItA*Antr Residual NO3 1-120 Available Water 250.00 -7 Pendleton study area (Timmerman) transects 2, core 4, soil water (A), and Figure A.26. nitrate (B), profile data. A Residual NO3 (kg ha-1) (cm) Equivalent Depth H2O 40.00 20.00 0.00 0.00 IIIIII11111111111111 6.00 5.00 4.00 3.00 2.00 1.00 .fls.iss.lui.misi.I.11.11,11.1 0.00 0.00 E 50.00 iiiii60.00 IIIIIIII iiiii80.00 100.00 50.00 0 0 100.00 b a. a) 150.00 150.00 a) El 0 a) 4-E o *4-= 0 s_ CI- 200.00 200.00 °woo Initial H2O DOME/ Final H2O O 444.40, -1.5 Mpa H2O 0000 Crop Available Water (f) 250.00 ittstrtatra Residual NO3 O (J) 250.00 Pendleton study area (Timmerman) transects 2, core 5, soil water (A), and Figure A.27. nitrate (B), profile data. A 0.00 0.00 E 1.00 2.00 3.00 Residual NO3 (kg ha-1) (cm) Equivalent Depth H2O 4.00 5.00 50.00 40.00 30.00 20.00 10.00 0.00 0.00 1111111(11IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII 6.00 E 50.00 50.00 0 0 iP _c 100.00 _c 100.00 CL 0 150.00 150.00 A 4- 0 200.00 200.00 0Q000 Initial H2O 00000 Final H2O 44444 -1.5 Mpa O (I) 60000 Crop 250.00 H2O Available Water 0 to:dotria Residual NO3 Cf) 250.00 Pendleton study area (Timmerman) transects 2, core 6, soil water (A), and Figure A.28. nitrate (B), profile data. A Equivalent Depth H2O 0.00 E 4.00 2.00 Residual NO3 (kg ha-1) (cm) 6.00 0.00 0.00 8.00 E 50.00 50.00 100.00 150.00 200.00 50.00 0 s _c 1 00 .00 - 100.00 Cl a) 150.00 150.00 a) 4= O 0 CI- 200.00 Cl- 200.00 030Q0 Initial 00000 Final H2O H2O 440,4wok 1.5 Mpa 1120 40000 Crop Available Water O C/) 250.00 Itctrtak Residual NO3 O (../) 250.00