Classification and Mapping of Agricultural Land for National Water-Quality Assessment

By Robert J. Gilliom and Gail P. Thelin

U.S. Geological Survey Circular 1131

---

SUMMARY AND CONCLUSIONS

Agricultural land use is one of the most important influences on the water quality of the nation. Although there is great diversity in the character of agricultural land, variations follow regional patterns influenced by environmental setting and economics. These regional patterns can be characterized by the distribution of crops. A new approach to classifying and mapping agricultural land use for national water-quality assessment was developed by combining geographic information on general land- use distribution with information on crop patterns from county agricultural census data. Two separate classification systems were developed, one for row crops and another for orchards, vineyards, and nurseries. These two general types of agricultural land are distinguished from each other in the general land-use classification system of Anderson and others (1976), which was used for the U.S. Geological Survey national Land Use and Land Cover (LULC) database.

Classification of cropland was based on the areal extent of crops harvested, rather than income production or a specific management practice. The acreage of each crop in each county was divided by total row-crop area or total orchard, vineyard and nursery area, as appropriate, thus normalizing the crop data and making the classification independent of total cropland area. The classification system was developed using simple percentage criteria to define combinations of 1 to 3 crops that account for 50 percent or more or harvested acreage in a county.

The classification system consists of 21 level I categories and 46 level II subcategories for row crops, and 26 level I categories and 19 level II subcategories for orchards, vineyards, and nurseries. All counties in the United States with reported harvested acreage in the 1987 Census of Agriculture are classified in these categories. The distribution of agricultural land within each county, however, must be evaluated from general land-use data. This can be done at the national scale using "Major Land Uses of the United States" (U.S. Geological Survey, 1970); at the regional scale using more detailed data from the LULC database; or at smaller scales using locally available data.

The classification system has many of the attributes required for national and regional water-quality assessment, but also there are several remaining questions that need to be resolved as applications are tested. The system is discussed below in relation to the desired attributes listed at the beginning of the circular.

1. Relevance to Management Practices - The classification system is based on the assumption that management practices, such as irrigation, cultivation, chemical usage, and fertilization rates are directly related to the crops grown in an area. Thus, crop mixtures are used as a proxy for the distribution of many management practices, which can be specifically evaluated for a given class either nationally or for individual areas or regions, as needed. As the system is applied to water-quality studies, this assumption will be tested.

2. Appropriate Scale - The scale of the classification is determined by choice of percentage criteria and by choice of the minimum number of counties or area required to qualify as a class. Most counties in most classes occur in relatively contiguous areas. For national and regional-scale water-quality assessment, the scale and resolution of the county-based classification is acceptable in many parts of the nation. For analysis of relations between agriculture and water-quality conditions within individual NAWQA Study Units and other drainage basins, the classifications by county may not be detailed enough, but can be improved with local data using the same classification system.

3. Transferability Among Scales - The system developed can be applied in the same way across a wide range of scales. The county unit is arbitrary. If data are available on crop distributions at a higher resolution, such as by irrigation district, township, or drainage basin, then these areas can be classified as if they were counties. Caution must be exercised, however, if areas are so small that relatively few individual fields are included. The system is based on the assumption that an area classified is large enough to include representation, at the time data on crops were collected, of the proportional role of individual crops in typical rotation schemes.

4. Stability Over Time - Percentage criteria, based on proportional areas, yield a simple classification system that will not change with changing time periods unless data analysts choose to make changes. Multi-variate statistical methods, such as cluster analysis, yield different definitions of clusters, or classes, with any major change in data.

5. Practicality - The classification system developed can be readily applied using existing data. Suitable existing data include consistent, nationally available data (Census of Agriculture and GIRAS) and similar locally available data with higher resolution. Crop statistics are regularly updated at national, state, and county levels by various agencies, and the Census of Agriculture is conducted every five years. Land-use data, such as GIRAS, are not regularly updated at the present time. A major effort is now underway by several agencies, however, to update land-use/land-cover characterization for the nation at a resolution similar to GIRAS. The NAWQA program is participating in this effort and will be developing updated land use characterizations for all study units.

---

Next - References

---

Back to:
Table of Contents