Mapping USDA-NASS Crop Data

Introduction

One third of the U.S. population lives in the Northeast, including the Mid-Atlantic, which creates an excellent market for agricultural products. However, there is not enough farmland or cheap labor in the region to provide all of the food people demand. The available agricultural land is farmed as intensively as possible, and in close proximity to dense human population.

Agriculture in the Mid-Atlantic region accounts for a significant portion of land use. Table 1 lists "land in farms" as a percentage of "total land area" for 1997. For most states, between 1/5 and 1/3 of the total land was considered "land in farms." According to the Census of Agriculture glossary, "The acreage designated as 'land in farms' consists primarily of agricultural land used for crops, pasture, or grazing. It also includes woodland and wasteland not actually under cultivation or used for pasture or grazing, provided it was part of the farm operator's total operation." (U. S. Department of Agriculture, 1997) The portion of land that is used for crops is determined on a yearly basis using surveys, and reported by the National Agriculture Statistics Service (NASS), which is a part of the U. S. Department of Agriculture (USDA).

The purpose of this study is to define agricultural regions so that areas of conflicting land use can be identified. In order to determine the spatial distribution of different types of agricultural activity, it is useful to look at NASS data at the county level. By focusing on a few major classes of crops, the balance of "land in farms" that is not used for pasture or grazing can be estimated. The crops selected for this discussion are economically important to the region, and are grown in ways that maximize production. The intensive cultivation has the potential to put water sources at risk through chemical, nutrient and sediment runoff. Groundwater, rivers and eventually open water such as the Chesapeake Bay are the ultimate destination of materials carried away from agricultural fields.

Methods

The NASS Published Estimates Database gives users access to statistical data at the county level for every year, as opposed to state level data collected during Ag Census years (those that end in "2" or "7"). This is important for producing crop area numbers for the smallest possible administrative unit over a number of years. For this study, the crop area data from the years 1990-2000 were averaged to produce an area number that was "typical" of a county for each crop.

The crops selected were small grains (barley and winter wheat), corn (for grain or silage), soybeans and hay (all types). Table 1 shows that these selected crops make up a large portion of "land in farms," and the bulk of the remaining acreage is probably pasture and grazing land that is not intensively cultivated. Agricultural data is readily available in tabular form through the NASS website, but putting it in a spatial context (such as a chloropleth map) requires some additional steps. The most critical piece of information linking spatial data with new attribute data is the five-digit "FIPS" code identifying a state and county.

An Excel tool was developed using Visual Basic for Applications (VBA) to create modified files using data available from the NASS website. The modified file contains:

• The standard five-digit "FIPS" code (state + county) for "joining" records of statistical data with spatial data in ArcView;
• Only the data types (data fields) selected by the user (area harvested, production, yield, etc.);
• New descriptive data field headers containing the data type, crop and year.

This tool facilitates the combination of tabular and spatial data. It is generalized to allow it to work with data downloaded from the NASS Published Estimates Database website for any crop from any combination of states and any combination of years. It can be run multiple times for multiple crops, as in the case with this study. The NASS data can be "joined" with any spatial file containing standard five-digit numerical FIPS codes, and several such spatial files are available from the Environmental Systems Research Institute (ESRI), makers of ArcView. If you are interested in learning more about this Excel tool, or obtaining a copy for your own use, please procede to the description and download page.

Once the data has been downloaded and reformatted, it can be manipulated to give general trends of agriculture in the Mid-Atlantic. Using Splus software, a principal components analysis was performed to illustrate where the different crop types are located together. The conceptual framework, for those familiar with principal components analysis of imagery, is that counties are "pixels" and crops are "bands." The Mid-Atlantic has 368 "pixels" and four "bands" in this case.

To determine the "loading" of each crop for each component, the percentage of area harvested for each crop was calculated for each county. This had the effect of "normalizing" harvested area based on the size of the county. Once the "loadings" for each crop were calculated (Table 3), a single value for each component was calculated for each county. For example, for Kent County, MD, component #1: 0.376 = ((0.114 * 0.607) + (0.325 * 0.525) + (0.228 * 0.596) + (0.028 * 0.000)). For some components, a particular crop had no impact on the county value (Hay = 0.000 in Component 1, for example). The values for each component were associated with the appropriate "FIPS" code for the state and county, and "joined" to the ESRI spatial dataset in ArcView.

Results

The following chloropleth maps of percentage of harvested area are the result of "joining" NASS data with ESRI spatial data using ArcView. The Susquehanna River flows from the north, and the Potomac River from the west.

The chloropleth maps below are the result of "joining" the results of the principal components analysis with ESRI spatial data using ArcView.

The principal components analysis also yielded an interesting picture of agriculture in the Mid-Atlantic. Table 2 shows how Components 1 and 2 explain 91.3% of the variation observed in the dataset. Table 3 shows the "loading" of each crop type for each component.

Table 3. Principal component loadings
	Comp. 1	Comp. 2	Comp. 3	Comp. 4
Small grains	0.607	0.161	-0.286	0.724
Corn	0.525	-0.372	0.763	0
Soybeans	0.596	0.215	-0.355	-0.687
Hay	0	-0.889	-0.457	0

Conclusion

Looking within the Chesapeake Bay watershed, there are a number of notable features displayed by the chloropleth maps of the different crop types. First, the intensity of small grain and soybean farming on the Coastal Plain is obvious. Also, corn acres stretch from the Coastal Plain up into the Shenandoah River Valley. Hay can be found in the Potomac and Shenandoah River Valleys, but it is absent on the Coastal Plain.

This same trend is evident from the chloropleth maps of the first two components of the principal components analysis. Examining Table 3 along with the map of Component 1 shows that the values of counties on the Coastal Plain, where small grains, corn and soybeans are present, are going to be relatively high and positive. The values of counties in the Shenandoah Valley are not as large as the Coastal Plain because they do not have much soybean acreage.

Examining Table 3 along with the map of Component 2 shows that the values of counties in the Shenandoah Valley where corn and hay are present are going to be driven toward the negative side. The values of the counties on the Coastal Plain are higher because they are buoyed by the presence of small grains and soybeans.

The maps produced by this study show the intensity of agriculture in the headwaters and adjacent to the Chesapeake Bay. This puts the Bay at risk of non-point source pollution from agricultural activity.

Discussion

Maryland has taken steps toward solving the agricultural runoff problem with its Nutrient Management Law. However, it is not currently possible to verify compliance due to the size and distribution of fields. If a current map of the locations and types of fields was available to nutrient management specialists, they could market their services to farmers with the most at-risk fields. The only way a current map can be produced is through remote sensing, as it has been done in Kent County, Maryland, and other states such as Arkansas.

Towards the goal of using remote sensing to do crop classifications, the category of "small grains" was chosen instead of "wheat" and "barley" because these crops have similar growing seasons and physiology, so they subsequently have a similar "spectral signature." They are both usually double-cropped with soybeans, although the barley's slightly shorter growing season makes it a better candidate. The area of small grains that are double cropped with soybeans is currently unknown. However, Ray Garibay, State Statistician for Maryland, estimated that 40% of soybeans are double-cropped with wheat in Delaware and Maryland. The use of remote sensing could answer this question more definitively on an annual basis.

The "hay" category encompasses all types of hay (legumes and grasses) because these species are often mixed in hay fields anyway. Hay represents a challenge to remote sensing because the "spectral signature" changes frequently during the growing season due to multiple cuttings. It may be possible to detect "hay" by a process of exclusion, although it will probably be difficult to detect it directly since its appearance changes unpredictably, depending on the weather.

There are several additional sources of data that could be used to clarify the picture of agriculture in the Mid-Atlantic. Elevation data would probably be a good indicator of where agriculture is present in a county. County animal and livestock populations (dairy cattle, beef cattle, broilers, turkeys, hogs, horses, etc.) combined with NASS crop data would provide a more detailed description of what type of agriculture is taking place. The current online NASS database is not fully populated, although access to this type of yearly livestock data may eventually be available.

The addition of livestock data would show that the Eastern Shore of Maryland is an excellent example of intensive agriculture taking place in a confined space. There are not enough acres to dispose of all the animal manure, because all of the fields (and then some) are being used to produce feed for the animals. Phosphorus runoff from manure has fouled the surface waters, and nitrogen used to fertilize the crops has leached into the ground and polluted the aquifer. A detailed description the agricultural regions of the Mid-Atlantic is an important first step in an effort to protect the waters of the Chesapeake Bay.

References

National Agricultural Statistics Service, various. Production Estimates Database. U. S. Department of Agriculture, 1997. Census of Agriculture. Weber, R., 2001. 50 States & Capitals. Weber Publications.