Using, mapping, and modeling detailed Farm Service Agency agriculture data:

Harold Robinson, Greg Easson, and Stephen Threlkeld

Abstract:
Data regarding agricultural practices, both crop type and agrichemical usage, is readily available from state and local government offices. The available data is usually aggregated as values for a given state, county, parish, or other political unit, with little or no spatial information. This gives the field researcher only a broad regional picture of the state of agriculture for the area of interest. A scientist seeking data on the effects of agriculture on the environment must conduct field research in areas that would be affected by agriculture and additionally in areas that would not be affected for experimental control. The data available at the state or county level is of such a coarse scale to be of limited use to selecting sites for experimental study or even general long term monitoring.

As part of an EPA funded Ecosystems Indicators Project designed to evaluate the effects of multiple stressors on agricultural wetlands, it was necessary to characterize in significant detail the agricultural practices for the study area. The types of information needed included wetlands, crops and acreage planted, location of planted crops, and agrichemical application rates and practices. The area of study is the Little Tallahatchie River drainage basin in northern Mississippi (figure 1). The basin encompasses an area of approximately 2000 square miles and is contained in parts of nine different counties.

Location Reference
FIGURE 1: Location Reference

The aggregated data could not be used to characterize the agricultural practices in the basin for several reasons. There are only nine data points, representing the nine counties, with which to characterize the basin. The data would represent a whole county, whereas the study basin only represented a fraction of the area of the county. Finally, agricultural practices vary significantly throughout the basin, dependent on topography and geology; in some cases the agriculture represented by the county numbers might not even be in the basin.

The scope of the project included an evaluation of the combined effects of the chemicals; characterization of the basin sought to identify sites in which the given chemicals might occur in specific combinations. The primary criterion for choosing a wetland site for sampling was the likelihood that the site might be affected by specific agricultural practices. This project is investigating potentially interacting stresses caused by four commonly used chemicals: Methyl Parathion, Clorpyrifos, MSMA, and Atrazine.

To more accurately characterize the basin, farm data was obtained from the local county Farm Services offices located in each county. The farm data contained the following key items:
  • Identification numbers - which could be used to cross reference data from previous years as well as provide a unique identifier (consisting of a farm and tract number
  • Type of crop planted and acreage planted
  • Public Land Survey System (PLSS) location for the farm

This data for the individual farms and tracts was the key to developing a more precise view of agricultural practices across the basin. Farmers participating in federal agriculture programs are required to report information concerning crop types and acreage planted. Data is collected by tract for each farm in a county. Farms may have more than one tract, which are reported separately, but a given tract identification will occur only once per county. The local county agencies then compile and report aggregated numbers to State agriculture offices.

The PLSS location reference contained on the data sheets for each farm tract provided the means to spatially locate the reported agriculture practices. The PLS System is a common land survey system used through most of the U.S. In consists of a series of north-south trending Meridians and east-west trending Base Lines. These Meridians and Baselines are the used to reference columns and rows called Townships and Ranges, each having a width of 6 miles. Each Township and Range cell or block is composed of 36 Sections each 1 square mile. The legal definition of land location using the PLSS system can further divide the Sections into quarters or other fractional unit. The PLSS location reference for each farm tract was to the nearest section, effectively giving researchers a spatial location accurate to within one square mile. Agricultural characterization of the basin was completed based on the PLSS grid (nominally one square mile) increasing the number of data points to over 2000.

The farm and tract data was collected from each of the nine county Farm Service offices for each PLSS section contained in the study basin. Data was then entered into a database with attributes linked to a unique identifier created by the farm tract number and a code for each county. Using the sum feature provided by ArcView (Esri), data was aggregated into total values for each of the PLSS sections in the basin. The resulting data set provided a more detailed description of the distribution of the agriculture crops planted across the basin. Figure 2 shows the distribution of Cotton acreage for the study basin.

Example Crop Distribution map
FIGURE 2: Example Crop Distribution map

Agrichemical usage data was then generated from accepted application rates for each crop type combined with acreage of the crops planted. Application rate data was obtained from local County Extension offices. An example would be – section XX had 200 acres of cotton planted, with an MSMA application rate of 2.53 kg/acre for cotton yields and expected chemical loading of 506 kg could be expected for that section. The total amount for each of the chemicals was then divided by the area of the section (acres). If section XX were of the nominal 1 square mile (640 acres) then the loading rate for the section would be 0.79 kg/acre. Figure 3 is a map of the usage of MSMA for the study basin. The final step served to normalize the chemical data by the area of each section. This was necessary because the actual PLSS grid has some inconsistencies resulting in sections that are not of the nominal size. The best example of this occurs in the western portion of the study basin, where the PLSS grid is reset along a northwest-southeast trending line - an old Indian treaty boundary (figure 4).

Example Chemical Distribution Map
FIGURE 3: Example Chemical Distribution Map

Example of PLSS Irregularity
FIGURE 4: Example of PLSS Irregularity

A ranking of NONE, LOW, or HIGH chemical loading was assigned to each section for each of the given chemicals, based on the maximum loading for a given chemical in any section in the basin. A value of NONE was assigned to sections with no expected usage of a chemical. A Value of LOW was assigned to sections with an expected usage of less than 25% of the maximum. HIGH was reserved for sections with expected usages exceeding 25% of the maximum. Using a simple technique of assigning a numerical value for each rank, a composite of the total chemical loading was created. For instance the NO, LOW, or HIGH ranking for MSMA was assigned to single digits (i.e. 1, 2, or 3), where as, Atrazine was assigned to double digits (i.e. 10, 20, or 30). When simply added, the resulting number represents the ranking of the four chemicals for the given sections. An example would be 2321: Methyl Parathion – low ranking (2000), Chlorpyrifos – high ranking (300), Atrazine – low ranking (20), and MSMA – none present (1).

Figure 5 illustrates the distribution of potential agrichemical combinations in the Tallahatchie River Basin by section. This compilation of FSA data allowed researchers to locate sections with specific chemical combinations to be able to study the effects of multiple stressors. From these data, sections were selected to locate wetlands for sampling. The GIS database contains the necessary information for biologist to be able to evaluate the results of biological testing of wetland ecosystems.

Completed Chemical Distribution Map
FIGURE 5: Completed Chemical Distribution Map

ACKNOWLEDGEMENTS:

University of Mississippi – Oxford, Mississippi
Project: EPA/NCERQA Ecosystems Indicators Program: EPA Award R826595-01-0
Project Coordinator: Dr. Stephen T. Threlkeld (Email: stt@olemiss.edu)
Project Web Page: http://www.olemiss.edu/projects/epa-eig

AUTHOR INFORMATION:

Harold D. Robinson Jr.
Graduate Student/GIS Technician
University of Mississippi
Department of Geology and Geological Engineering
118 Carrier Hall
University, MS 38677
Phone: (662) 915-7498 – Dept.
Email: HRobin1499@aol.com

Dr. Greg Easson, Associate Professor
Director, University of Mississippi Geoinformatics Center (UMGC)
University of Mississippi
Department of Geology and Geological Engineering
118 Carrier Hall
University, MS 38677
Phone: (662) 915-5995 – Office
Email: geasson@olemiss.edu

Dr. Stephen T. Threlkeld, Professor
University of Mississippi
Department of Biology
318 Shoemaker
University, MS 38677
Phone: (662) 915-5803
Email: stt@olemiss.edu