Analyzing Site Sustainability For
Animal Waste Utilization Using GIS
Elizabeth A. Cook and Daniel P. Silberberg
USDA-Natural Resources Conservation Service
Missouri state regulations define areas suitable to receive animal wastes as soil nutrient supplements using characteristics of the soils, slope and land cover, and limit the spread of manure in areas not meeting the criteria. Mapping and quantifying areas where all spreading criteria coincide within a landscape is time-consuming when done on-site, and is a well-suited application of geographic information systems (GIS) analysis. The technique was piloted for Loose Creek watershed, Osage County, Missouri.
Loose Creek is a fifth-order stream with a watershed of 58,300 acres. The area has about 50 concentrated animal feeding operations
(CAFOs) with more than 30,000 head of swine and 810,000 poultry, producing an estimated 2.2 million cubic feet per year of manure. The landscape is mostly rolling and forested, limiting acres suitable for the use of wastes produced in the watershed. State soil and water conservation and 319 Clean Water Act funds were acquired through local initiatives to provide incentives for better waste management. The purpose of the GIS analysis was to quantify spreadable acres in relation to the locations and manure production of the
CAFOs, thus targeting areas of concern where incentive dollars could be used to best advantage.
Sites suitable for spreading animal waste are defined in state regulations as open land with less than 15 per cent slope, soil permeability not exceeding 2 inches per hour in any soil horizon, soil texture less than 50 per cent gravel and soil depth greater than 60 inches. Digital soil mapping units with related tabular data, topographic contour lines, land cover and CAFO locations were required to conduct the analysis. Ancillary data, such as roads, section lines and a watershed boundary, were also used for cartographic presentation of the analysis results.
The 1:24,000-scale county soil survey was digitized by scanning mylar separates of the data, with subsequent editing and attributing. Tabular data pertaining to soil properties and interpretations, already stored in a digital database, were linked to the spatial data using the GIS software. The 1;24,000-scale contour lines were also scan digitized. Point locations of CAFOs were visually located on digital raster graphics
(DRGs) of the 1:24,000 topographic maps using information provided by local project leaders. Land cover information was generated from digital image processing of Landsat Thematic Mapper satellite imagery covering the project area. The watershed boundary was delineated using visual interpretation of topographic contour lines on
DRGs. Other digital data used for map production were collected from various sources.
Each site suitability criterion was interpreted from these source data. Open land, identified as crop or grassland, was subset from the land cover data. A digital elevation model (DEM) with 10-meter grid cells was generated using a
hydrologically-correct interpolation algorithm, available in the GIS software, applied to the contour data. The DEM was then used to classify slope ranges, with areas less than 15 per cent considered suitable. Soil permeability was an attribute provided in the soils properties database for each layer of a soil profile. Soils with permeability not exceeding 2 inches per hour in any layer were considered suitable. Soil texture classes from the soils properties database were used to address gravel composition. Because a soil texture class with 'skeletal' in the name is only given to soils with gravel greater than 50 per cent by volume, soils with skeletal texture classes were excluded from suitability. Soil depth is also an element in the soils database, allowing for subsetting of areas with soils greater than 60 inches as suitable sites. These data were then unioned together to allow for selection and mapping of areas meeting all five criteria of site suitability.
Circular buffers were generated around each CAFO to estimate the amount of spreadable acres within an economically-feasible transport distance. Also, the buffers were intended to show areas of the watershed currently not within transport distance for any
CAFO, thus suggesting an opportunity for manure brokering to those land owners. Local project leaders suggested distances of 1 mile for hog operations and 3 miles for poultry operations based on their discussions with operators in the watershed.
The land cover classification from satellite data showed that 30,779 acres (53%) of the watershed is forested, 170 acres is surface water, 573 acres (1%) is urban and the remaining 26,779 acres (46%) is comprised of crop or grassland and thus available to receive animal wastes. A total of 34, 498 acres (60%) of the watershed has less than 15% slope. Soil permeability is suitable for 50,107 acres (86%), soil depth for 43,886 acres (76%) and soil gravel composition is least restrictive, with 55,794 acres (96%) being suitable. An overlay of these five criteria yields 15,662 acres (27%) of the watershed as suitable. The transport buffers showed the unexpected result that no area of the watershed is outside of an economically-feasible transport zone. Spreadable acres within transport buffers for each CAFO has not yet been quantified.
Discussion and Conclusions
The quality of this analysis was directly dependent on the quality of the input data. For land cover, the satellite data were considered marginally adequate. The dates of the imagery and crop rotation practices in the watershed made it difficult to separate crop and grassland. Because initial screening of spreadable acres considers both grass and cropland, aggregating them into an open land category was acceptable. Of course, for individual farm nutrient management planning, a more accurate land cover layer would be necessary. Heads-up digitizing of land cover from a digital
orthophotograph, for example, would yield better land cover data but would be time-consuming.
The slope data generated from the 10-meter DEM was, if anything, too detailed. Fields tended to get divided into many slope categories by the computer analysis, whereas on-site estimates would generalize slight variations. Some of the slope variations could have been 'smoothed' using filtering techniques. Digitizing contours is very time-consuming. The use of USGS DEMs with 30-meter grid cells would work in some types of terrain and where no other data are available.
Most problematic of the analysis was interpreting the intent of the state criteria for spreadable acres. For example, soil permeability not exceeding 2 inches per hour was vague as to whether it meant of the top soil layer only, all layers or some other interpretation. Also, soil gravel not exceeding 50% was unclear as to meaning by weight or by volume. This work pointed out some limitations in the wording of state regulations and the need to discuss this pilot project with experts in several fields for suggested refinements.
Acceptance of the technique and results of this analysis by local project leaders was good, at least for initial planning and report writing. Visual assessment of the CAFOs overlaid on the spreadable acres map does point to some having a shortage of acreage in the near vicinity. A serious limitation of the analysis was not having a digital plat (ownership) map available. The transport buffers are of limited use because they do not take road travel and field ownership into consideration.
The GIS assessment conducted for Loose Creek did help project leaders target the best incentives to offer for manure management. For example, comparing the spreadable acres to volume of manure produced showed that the watershed does have adequate acres for spreading under current regulation. Of course, managing the location and timing of the spreading is still an issue, but out-of-watershed transport need not be seriously pursued at this time.
Timing of the GIS analysis is also important. This project began after field staff had been doing on-site evaluations for some time. Therefore in many ways the analysis was too late. Also, Loose Creek watershed is small enough for the field staff to evaluate it reasonably well on-site. The GIS analysis would be of even more value in a larger watershed where a full field-level assessment is impractical and screening is of particular importance.