6.11   Spatial Applications for Agriculture: Educational Case Studies

Principal Investigators

Scott A. Shearer, Associate Professor, Biosystems and Agricultural Engineering
Thomas G. Mueller, Assistant Professor, Agronomy |
Carl R. Dillon, Associate Professor, Agricultural Economics
Samuel G. McNeill, Assistant Extension Professor, Biosystems and Agricultural Engineering

Introduction

            Favorable economic conditions during the mid to late 90s and the lower cost DGPS equipment allowed many producers to initiate site-specific management of crop production.  Perhaps most significant was the evolution of the yield-monitor, and the attendant yield maps that illustrated the variability that exists across the soil landscape.  In response many producers began soil-sampling programs with the hope that variable-rate fertilizer would bolster productivity.  Today, low crop prices coupled with drought conditions in 1999 are causing many producers to question the profitability of many precision agriculture practices.  To this end we, a team of multidisciplinary educators and researchers, propose to develop a series of educational modules oriented towards answering many of the question posed by producers, and to illustrate new opportunities for adopting and utilizing appropriate site-specific management practices in production agriculture.  Many of the proposed modules are focused on quantifying the economics of precision agriculture.

Objectives

1.                   To develop 15 educational modules for use in teaching precision agriculture concepts to undergraduate students at UK, and for use in training Kentucky agricultural extension agents, service providers, and producers.

2.                   To deliver these instructional modules to interested parties via classes at UK, short courses, and the Internet.

Background

            College courses that focus on helping the undergraduate student to develop an understanding and appreciation for site-specific crop management are just now beginning to emerge.  Shearer and Barnhisel (1997) offered one of the first college courses in precision agriculture.  Similar efforts were underway at other four-year institutions in the U.S. (Ess and Morgan, 1997) and at community colleges (Brase, 1997).  These courses, as well as contemporary primers (Morgan and Ess, 1997), focused on explaining the technology to novice and would-be users.  Data to support the profitability of this technology was just beginning to emerge.  Today, we know more about the potential returns to these management practices, and most practitioners now recognize these returns depend heavily on the nature of the variability that exists across the soil landscape, as well as the approach they choose to manage this variability.  Further, that the profitability of these practices vary from state to state, farm to farm, and even field to field.

            Holt and Sonka (1995) provide a framework for transferring agricultural technology in the 21^st century.  They identified a four-tier approach to research and development (R&D).  The four-tiers are basic research, developmental research, adaptive research, and technology transfer.  With this approach they advocate an integrated approach to R&D of which technology transfer is a key focus.  They contend that full integration of technology transfer within the R&D structure saves time and helps R&D personnel recognize potential problems sooner.  Tim (1999) provides a model for integrated technology in precision agriculture via the Internet.  The model environment is composed of instructional materials, study guides, activities, virtual field trips, chat pages, and links to other sources of information.

Procedures

            In Kentucky we propose a series of educational modules with supporting background and hands-on activities to be developed to foster the understanding and implementation of precision agriculture practices for the spatial management of crop production.  This will be a three-year effort with a minimum of five educational modules developed and evaluated each year.  Each module will be packaged for delivery to each of the groups noted above, with provisions to utilize a variety of data sources.  At a minimum the following educational modules are proposed:

Year 1:

1. Coordinate Conversion and Datums - Coordinate conversion and specification of datums often create problems for practitioners of precision agriculture who utilize data from numerous sources.  The intent of this educational module is for practitioners to develop a background in coordinate conversion that enables them to manipulate GIS coverages, thereby allowing them to resolve registration problems when using multiple layers of data from numerous sources.

2. Boundary Mapping and Area Determination - Points, arcs, and polygons; the basic building blocks of GIS packages will be discussed.  The properties of each will be reviewed.  An exercise, based on Green's Theroem, will be developed to guide practitioners through the process of calculating areas from projected coordinates.  This is a basic and fundamental element of any GIS package, and this exercise should serve to foster a better understanding of the capabilities of these packages.

3. Yield Map Generation - Yield map generation, for the most part, has become a routine task for most producers.  Unfortunately, the yield map generation task is entrusted to agricultural software developers.  This educational module will be designed to instruct practitioners in data filtering techniques that enable them to eliminate spurious data, and then to generate yield maps using a standard convention that allows for visual interpretation. 

4. Soil Sampling - Soil sampling to describe variability is far from a science, as implemented today it is more of an art.  Routinely, an aligned grid is used for simplicity, with resolution driven by what the market will bear.  This module is intended as a review of possible alternatives to the more traditional approach.  Yield maps will be used in conjunction with other sources of data to compare and contrast a "smart" sampling approach with grid sampling.

5. Interpolation Techniques for Map Generation - Many precision agriculture practitioners routinely use interpolation techniques to generate maps for visual interpretation.  Unfortunately this is done with little knowledge of which method is most appropriate.  This exercise will detail the development of kriging, inverse distance, minimum-curvature, nearest neighbor, triangulation, and polynomial regression interpolation techniques.  The intent will be to illustrate variations in interpolated surfaces in relation to the methods used to generate these surfaces, and the accuracy of interpolation techniques for predicting parameter values at unknown locations.

Year 2:

6. Land-Use Determination form Historical Yield Data - Perhaps the most significant opportunity with respect to spatial management of crop production is the identification of regions that are not profitable for crop production, or those where revenue can be improved through participation in government program such as the Conservation Reserve Program (CRP).  Historical yield data will be analyzed to determine the best course of action for a producer to pursue given the current value of grain and production costs.  Yield stability from year to year and crop to crop will be used as a basis for making land-use decisions.

7. Profit Map Generation - Perhaps the most important coverage to be generated with regard to the spatial management of crop production is the profit map.  This unit will focus specification of fixed and variable productions costs, gross receipts of marketed grains, and the spatial distribution of profits.  Profit maps form multiple years will be generated so that students can identify the regions of the field that are consistently profitable, in an attempt to distinguish these from areas that are not.   

8. Variable-Rate Fertilizer Recommendations - Variable-rate fertilizer application is most frequently based on the spatial distribution of soil fertility levels and the anticipated benefits of fertilizer and soil amendment application.  Soil fertility data will be interpolated to produce a surface of the respective attributes.  AGR-1 soil fertility recommendations will then be utilized to generate application surfaces.  Surfaces will then be translated into management zones for generation of the application maps that are downloaded for application control.  Application file features will be discussed along with standards for their generation.

9. Variable-Rate Fertilizer Application - Many producers view variable-rate fertilizer application as the epitome of precision agriculture practices.  Potential errors relating to variable-rate application may nullify many of the potential profit enhancing characteristics of this activity.  This case study will be designed such that practitioners appreciate the potential for errors associated with variable-rate application.  Further, application errors will be modeled for spinner and air-boom applicators, and in turn used to predict the actual application surface.  Predicted application surfaces will be compared with the desired application surfaces.  Model parameters will then be adjusted so that practitioners gain an appreciation for how "look-ahead" rate changes, application overlap, and control systems response affects application accuracy.

10. Economics of Variable-Rate Fertilizer Application - The economics of variable-rate application will be projected from crop response curves to estimate the potential returns to management.  Inherent in any economic analysis is the effect of errors on the marginal returns.  To this end soil parameter interpolation errors will be combined with application errors to determine the overall effect of sampling and management grid resolution on net returns.  Students will be guided through a strategy for optimizing marginal returns by adjusting the various management parameters.

Year 3:

11. Remote Sensing - Remote sensing offer potential benefits to Kentucky producers.  This data has proven useful for predicting biomass accumulation and yield.  Other possible uses include determination of soil moisture contents and organic matter contents.  This exercise will focus on importing satellite images, registering these with digital rectified orthoquads, field boundaries, and yield maps.  Various common vegetative indices will be compared to yield monitor data to assess the suitability of these parameters for predicting yield.  Existing Landsat 5 images will be used in this activity.

12. Variable-Rate Seeding - Evidence is mounting to support variable-rate seeding in accordance with topsoil depth, where higher rates are seeded in locations with thicker topsoil.  Response functions will be generated for yield increases as a function of topsoil thickness.  This relationship will be applied to topsoil depth maps and historical yields to determine if variable-rate seeding of corn is warranted.  Further, marginal returns maps will be generated to visualize the spatial variation in increased profits as compared with fixed-rate seeding. 

13. Conducting On-Farm Investigations - The technologies attendant to precision agriculture (GPS and yield monitoring) now afford producers the opportunity to conduct on-farm investigations.  This unique situation can be easily misused producing less than desirable results.  This module will concentrate on conceiving and carrying out field investigations that allow producers to draw meaningful conclusions.  

14. Variable-Rate Nitrogen Application - Variable-rate nitrogen application in accordance with historical yields may be profitable on well-drained karstic soils.  Data from actual field investigations will be presented to practitioners, and they will be guided in the analysis of this data using linear regression to establish a response function.  This function will be applied to historical data from similar investigation sites to predict crop response and the marginal returns associated variable-rate application.   

15. Soil Conductivity Correlation with Landscape Features - Soil conductivity measurements are relatively simple non-intrusive measurements that yield information about subsurface features.  These features may range from clay pan depths to salinity concentrations.  Conductivity data will be presented along with soil landscape features for correlation analysis.  Practitioners will be instructed in elementary data analysis techniques that will allow them to determine if conductivity, at varying depths, is a reasonable predictor of one or more of these features.

These educational modules will be developed and implemented using platforms that include Microsoft's Excel spreadsheet, ESRI's ArcView GIS, and Golden Software's SURFER for Windows.  The intent is to utilize analysis tools that are commonly available within the university community and through the Cooperative Extension Service.  Where appropriate Avenue scripts for ArcView will be developed to streamline data manipulation and analysis.  These scripts will be distributed as public domain software.

Expected Benefits 

            The anticipated benefit of this educational effort will be better-informed students and clientele groups.  The primary educational goal is to focus on practices and opportunities to increase producer profitability through the implementation of spatial crop production management.  Similarly, these educational efforts are intended to help producers and managers recognize in which settings the cost of implement precision agriculture crop production practices may outstrip potential returns.

Deliverables

            The primary deliverable from this investigation will be a series of 15 educational modules with supporting background materials to guide practitioners in the acquisition of spatial data management analysis skills with a focus on agricultural crop production in Kentucky.  These modules will be made available to all interested parties via the University of Kentucky's Precision Agriculture web page, and through regularly scheduled short courses and undergraduate course offerings.  Short courses will be scheduled on an as needed basis, with a minimum of one workshop to be held per year for county agriculture agents, producers and service providers.  Parallel extension publications will be generated for each of the educational modules.  These publications will be stand-alone documents that detail the procedures for replicating the same analyses using a generalized approach.