GIS: a tool for siting farm ponds.
Our objectives were to determine and acquire the necessary databases, determine the best sources for the data to be compiled into GIS format, and to develop a procedure for calculating runoff from a specified watershed using the created GIS databases. The methodology and data were applied to determine possible pond sites and estimate rainwater harvesting potential for a 172 ha (424 ac) farm in west central Pennsylvania.
Several governmental agencies are using GIS for runoff predictions (Rosenfeld 1992; Van Blargan 1989). River drainage area projects typically assign Natural Resources Conservation Service (NRCS) curve numbers (CN) for the river basins using remotely sensed land use and landcover (LULC) source material (Stuebe and Johnston 1990). Computer modeling for land management has been used for some time (Burrough 1987) but much of it is designed for large land areas and not for the private farm or ranch. The research and analysis covered in this project were performed using a family farm as the study area. The farm's main enterprises [TABULAR DATA FOR TABLE 1 OMITTED] are corn and alfalfa production with an associated beef cattle grazing operation.
Methods and procedures
GIS was used as a screening tool to identify optimal farm pond sites using farm-scale data. The criteria for pond site selection included soil suitability, slope suitability, current land use, and rainwater harvesting potential. For this project, the NRCS curve number method of determining runoff was used.
The first coverage (GIS map layer and associated attribute data) created was FARMGEO, the georeferencing coverage. The four points chosen as referencing tics had known State Plane Coordinates on both the United States Geological Survey (USGS) and NRCS source maps. All coverages contained these four tics. Table 1 lists the original digitized coverages, their source maps, and a brief description of the coverage. All digitized coverages were vector coverages with associated attribute files.
The process of creating FARMGEO, which gave an approximation of the area of the study, is typical of that used for all the coverages derived from source maps. First the coverage was traced onto mylar, then digitized into ARC/INFO. After the coverage was cleaned, built, and checked for digitizing errors, it was transformed from the map units of the digitized topology to real world coordinates. Once the individual digitized coverages were created, they were edited, buffered, and combined [TABULAR DATA FOR TABLE 2 OMITTED] [TABULAR DATA FOR TABLE 3 OMITTED] to create the final coverages used for site selection. In the case of ROADS, STREAMS, and UTILITIES coverages (Table 1), line data were buffered to create polygon coverages in order to give areal extent to physical features which were originally defined as lines on the source maps. Table 2 lists coverages developed from the original data layers.
Several attribute tables were developed. Three-digit numerical codes were generally used in the attribute tables for pond suitability and land use and landcover (LULC). Table 3 is an example of the three-digit codes developed for pond suitability. Pond suitability for each soil mapping unit is based on information from the soil survey and considers that soil's potential to contain water and the depth to the water table. Table 4 is a sample of the LULC attribute file, which contains a two-digit code for CN. The code was developed based on the NRCS criteria for CN, which is related to the hydrologic soil-cover complex (U.S. Soil Conservation Service 1984; U.S. Department of Agriculture 1972; U.S. Department of Agriculture 1985; U.S. Department of Agriculture 1986).
A triangular irregular network (TIN) was created using elevation data from the ELEVATION, HIGHPOINTS, and PONDS coverages. From this TIN, slope and aspect data were obtained. The slope coverage derived using the TIN was used to determine pond site suitability from a topographic perspective.
After the unusable areas (utility lines, roads, and streams) were identified, the remaining farm area was screened to locate potential pond sites. This screening was based on the following three criteria: pond suitability of 300 or greater, (Table 3) a LULC of 300 or greater, (Table 4) and slopes of 8% or less. The soil characteristics related to the pond suitability values were ranked so that the more desirable the soil was for pond construction the greater the number. Soils with suitability rankings less than 300 had some limitation that made that site unacceptable for a pond. LULC values were similarly ranked; the lower the value, the less acceptable the land use was for pond construction. LULC values from 100 to 199 were unacceptable because the land use was simply not suitable for ponds (because of farm roads, buildings, etc.). LULC values from 200 to 299 were reserved for cropped areas. Because the economic value of the land for cropping is greater than its value for ponds, [TABULAR DATA FOR TABLE 4 OMITTED] these areas were removed from consideration as pond sites. Potential sites were further screened to exclude those areas less than 0.2 ha (0.5 ac) in size. The suitable sites were then located on the elevation coverage.
For each potential pond site, the watershed draining to the site was hand delineated and then scanned into the GIS. Using INFO functions within ARC/INFO, the watershed area and weighted average curve number were calculated. The area and curve number values were used later with spreadsheets to calculate potential runoff, and each site was ranked according to its ability to accumulate runoff. The soil antecedent moisture condition for all runoff calculations was assumed to be average. This decision was made to simplify the initial screening process and to eliminate the need to recalculate the curve numbers for each soil on a daily basis.
Daily climate data were generated for a 10-year time span using CLIGEN, which is a component of WEPP, a public domain erosion prediction model (Flanagan et al. 1991). CLIGEN was used to give representative data for the sire and to provide continuous daily data for the simulations. Also, this procedure could be readily used for other locations within the United States. For this project, all precipitation was assumed to be rain and all runoff events assumed to occur on the date of the precipitation.
Runoff for each site was calculated using the following equations:
S = 25400/CN - 254
Q = [(P - [0.2.sup.*]S).sup.**]2/(P + [0.08.sup.*]S)
Q = direct surface runoff (mm depth over the watershed);
P = storm rainfall (mm depth over the watershed);
CN = curve number (dimensionless)
S = potential abstraction of water by the soil (mm depth over the watershed).
S is related to the soil and cover condition of the watershed through CN, where CN ranges from 0 to 100 (Stuebe and Johnston 1990). Volume of runoff was determined by multiplying runoff depth by the watershed area. A pond with a surface area of 0.4 ha (1 ac) was assumed in calculations of evaporation losses and pond rainfall interception. This standard pond size was used so as to allow easier comparison between sites. Evaporative losses were calculated using the Jensen-Haise equation (School and Gander 1983).
After calculating the water yield for 10 years of simulated climate data, each pond site was ranked according to its potential to collect runoff and its location relative to animal pastures. Because of the lack of detailed data and to simplify the analysis, groundwater contributions to inflow, pond seepage to groundwater, and emergency overflow were ignored in the water balance calculations. To simplify the process, the water collected was expressed as the number of beef cattle that could be watered if each drank 0.057 cubic meters (15 gallons) per day.
Hardware used for this project included IBM RISC 6000 Model 320, Calcomp Digitizing Tablet, Calcomp 1043 Plotter, Macintosh Computer, Hawtek Scanmaster scanner, Alphatronix Rewrite Optical reader and optical disk. Software used included [TABULAR DATA FOR TABLE 5 OMITTED] IBM AIX Version 3.1 for RISC 6000, ARC/INFO Version 5.0.1 (Environmental Systems Research Institute 1990), MacScan-It 1.41, and Lotus 123 Release 2.2.
Results and discussion
The GIS analysis identified 13 locations that appeared suitable for ponds. These numbered sites are shown in Figure 1. Of these 13 sites, four (sites 10, 11, 12, and 13) were separated from one of the other nine sites only by utility lines. Therefore, these were included in the drainage area of the nearest downslope site. This reduced the potential pond sites to nine.
The water yield of potential pond sites is summarized in Table 5 and the site characteristics are presented in Table 6. As seen in the last column of Table 5, if a pond was constructed at each site, water would be available for almost 1,500 cattle per year over the 10-year period simulated. This assumes that grazing management would permit cattle access to the ponds so that overflow from the ponds does not occur and that all water collected is consumed by the animals.
When the number of cattle that can be watered on a year to year basis (yearly data) is compared to the running average [TABULAR DATA FOR TABLE 6 OMITTED] (cumulative data), the desirability of providing water storage to moderate fluctuations in supply can readily be seen. For the years 1984 and 1986, calculations indicate that approximately 617 and 632 cattle could be watered if only the water yield for these years is considered. When the running average is considered, the number of animals that can be watered is 1571 and 1467 respectively. Because it is unlikely that 1500 head of beef cattle will be grazed on the farm, it would be impractical to develop all sites based solely on animal watering requirements. Therefore, the best approach is to rank them according to desirability and to select the most favorable sites for development.
The last column in Table 6 ranks the potential sites based on their location relative to current pastures, difficulty of constructing ponds, and water harvesting potential. For practical purposes, Sites 3, 4, and 7 would be unacceptable because of their locations. Sites 3 and 7 are located on the farms property boundaries and are likely to present accessibility problems. Site 4 is adjacent to a public, unpaved road, which already has difficulties with excess water, drainage, and erosion. Locating a pond at Site 4 would only aggravate the road drainage problem. Sites 1 and 2 would require some major revisions in current grazing practices and considerable effort in site clearing, pasture development, and fence construction. Site 9 would be acceptable; however, there is already a spring and a well which usually provide all the needed water for that small pasture. Site 8 would be acceptable but Site 6 would collect more runoff and Site 5 would be more accessible. Therefore, Sites 5 and 6 would be the two most desirable sites for locating ponds according to current grazing practices.
A possibility for increasing water yield, should water requirements not be met by sites identified in the initial screening, is to relax site selection criteria. Relaxing the slope restriction to 12% increases the possible pond sites, as shown in Figure 2. In this figure, Sites 4, 5, 6, 7, 8, 11, 12, and 13 are enlarged areas of the respectively numbered sites in Figure 1. Site 14 is in the same segment as Sites 5, 11, and 12, which can be considered as one large area divided by utility rights-of-way. Site 15 is located within an existing pasture which has a spring at that location. Site 18 is an extension of Site 4 and would be unacceptable for development for the same reasons as Site 4. Sites 16 and 17 are possible additional sites that need a full evaluation. A water yield analysis, providing results similar to Tables 5 and 6, would result in a larger volume of water available for harvesting and hence an increase in the number of cattle that could be watered.
Conclusions and recommendations
Several precautions and modifications are needed before this methodology can be widely applied. The first limitation is application in karst terrain and land areas with subsurface drainage patterns. In these cases, runoff can move rapidly into the underground water table and not appear as surface runoff. Therefore, a hydrogeologic data layer needs to be added. This would permit additional analysis of surface water penetration to the groundwater and alleviate some of the runoff pattern distortions that exist in karst terrain.
Areas where detailed soil mapping units at a farm scale are unavailable would require that elevation data be used to determine natural drainage paths. Once these paths are defined, an interception site which drains some predetermined basin acreage would be selected. The watershed of this site would be used to estimate runoff. When the quantity of runoff is great enough to meet user-defined consumption, then the site would be field inspected and an assessment made as to its pond suitability.
Several improvements are possible. We suggest that GIS be used in conjunction with a Global Positioning System (GPS) (Hurn 1989) and remote sensing in developing the initial databases. This is especially applicable for the detailed land use and landcover coverage because the USGS maps may not be current and line width on the maps results in distortions in the location and physical width of features. Using a GPS and field work should result in more precise representation of land use and landcover and allow the GIS to be a more practical and powerful analysis tool.
To improve the hydrologic assessment capabilities of GIS, we recommend using greater resolution elevation data such as that acquired from digital ortho photos at a scale of 1:12,000 with 1m resolution. This would result in a more accurate representation of the surface (with a 1m ground resolution) and allow surface runoff patterns to more closely approximate reality.
Ideally, the method of determining acceptable pond sites would combine hydrologic considerations as well as pasture management practices. A grazing management component needs to be added to facilitate pasture rotation based on grass and water availability. An additional recommendation would be to add a component that varies animal water requirements to match climatic conditions.
GIS is a promising tool for identifying suitable pond sites and predicting water harvesting potential at a farm scale. The major difficulty, and most of the effort, lies with creating databases at the detailed farm scale. The major benefit is that once a GIS database exists it can be enhanced into a complete farm management GIS package.
Using commonly available farm-scale data and a geographic information system (GIS), potential pond sites and the amount of water provided at each site were identified for an actual farm in Pennsylvania. The approach involved developing appropriate digitized data layers and attributes; locating sites based on topography, soils, and land use; estimating the amount of water that could be harvested/stored at each site; and ranking the desirability of potential sites relative to one another. Potential runoff was predicted using the Natural Resources Conservation Service Curve Number Method and water balance calculations were made using a spreadsheet. The identified potential sites were compared and ranked as to most desirable.
Burrough, P.A. 1987. Mapping and map analysis: new tools for land evaluation. Soil Use and Management 3(1):20-25.
Environmental System Research Institute, Inc. 1990. Understanding GIS, The ARC/INFO Method. Redlands, California.
Flanagan, D.C., J.E. Ferris, and L.J. Lane. 1991. WEPP Hillslope Profile Erosion Model Version 91.5 User Summary. NSERL Report No. 7, National Soil Erosion Research Laboratory, USDAARS, West Lafayette, Indiana.
Hurn, J. 1989. GPS, A Guide to the Next Utility. Trimble Navigation Ltd., Sunnyvale, California. 76 pp.
Rosenfeld, C.L. 1992. Watershed management, fighting the effects of drought in West Africa. Geo Info Systems 2(3):29-39.
School, R.R., and G.A. Gander. 1983. Computation of runoff reduction caused by farm ponds. Water Resources Bulletin 18(3):529-532.
Stuebe, M.M., and D.M. Johnston. 1990. Runoff volume estimation using GIS techniques. Water Resources Bulletin 26(4):611-620.
U.S. Soil Conservation Service. 1984. Engineering Field Manual for Conservation Practices, Washington D.C.
U.S. Department of Agriculture. 1985. Soil Survey of Cambria County, Pennsylvania.
U.S. Department of Agriculture. 1986. Urban Hydrology for Small Watersheds (210-VI-TR-55), Soil Conservation Service, USDA, Washington, D.C.
U.S. Department of Agriculture. 1972. SCS National Engineering Handbook. Section 4. Soil Conservation Service, USDA, Washington, D.C.
Van Blargan, E. J. 1989. A Geographic Information System for Watershed Growing and Hydrograph Simulation, Ph.D. Dissertation, University of Maryland, College Park, Maryland.
Cheryl F. Vorhauer is a Ph.D. student in the Department of Agricultural Engineering, Texas A & M University, College Station (formerly a graduate student in the Department of Agricultural and Biological Engineering, The Pennsylvania State University). James M. Hamlett is an associate professor of Agricultural Engineering, The Pennsylvania State University, University Park, PA 16802.
Specific product names are included for the benefit of the reader and do not imply specific endorsement by the authors of The Pennsylvania State University.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Special Issue: Global Change & Terrestrial Ecosystems; geographic information systems|
|Author:||Vorhauter, Cheryl F.; Hamlett, James M.|
|Publication:||Journal of Soil and Water Conservation|
|Date:||Sep 1, 1996|
|Previous Article:||A GIS procedure for automatically calculating the USLE LS factor on topographically complex landscape units.|
|Next Article:||Conserving water and increasing alfalfa production using a tall wheatgrass windbreak system.|