ASAE Conference Proceeding
This is not a peer-reviewed article.
Modification of Curve Number Adjustment Technique for Prediction of Runoff
Samar J. Bhuyan, James K. Koelliker, Philip L. Barnes
Pp. 287-290 in Soil Erosion Research for the 21st Century, Proc. Int. Conf. (3-5 January 2001, Honolulu, Hawaii, USA), eds. J. C. Ascough II and D. C. Flanagan. St. Joseph, Michigan: ASAE. ,Pub. Date 3 January 2001 . ASAE Pub #701P0007
Adjustment of curve numbers were done on the basis of estimated antecedent moisture content (AMC) ratios from measured runoff, precipitation, and average curve number (CN) of watershed to predict surface runoff quantity data. Sub-watersheds of Cheney Reservoir watershed, Kansas was considered for this study. The measured stream flow data were obtained from different USGS gauging stations, located in these sub-watersheds and base flow for these data were separated using the USGS developed HYSEP2 model in order to obtain surface runoff amount that could be compared directly with Agricultural Nonpoint Source Pollution (AGNPS) model output during different storms events. The different model input parameters were extracted from land cover and other GIS layers using AGNPS-ARC/INFO modeling environment. The surface runoff depths during different storm events were predicted satisfactorily.Keywords. Model prediction, Landsat TM image, AGNPS-ARC/INFO.
The SCS curve number method is one of the most widely used methods to estimate the surface runoff coming out of a watershed. It uses the concept of dimensionless runoff curve number (CN) in this method. The CN is a function of land use, hydrologic characteristics of the watershed before a storm, and soil characteristics. The hydrologic characteristics of a watershed before a storm are difficult to define. SCS method adjusts these curve numbers on the basis of antecedent moisture condition, whether the watershed is dry, moderate, or wet. So the determination of the antecedent moisture condition (AMC) plays an important part in this process.
It is the purpose of this paper to come up with a technique to estimate AMC ratio for adjustment of runoff curve numbers of a watershed during different storm events. This technique was applied for adjusting the CN in the input file of the AGNPS (Young et al., 1987) model to predict surface runoff. The interactive AGNPS-ARC/INFO interface, developed by Liao and Tim (1997), was used to prepare the different input files for AGNPS (Liao, 1997).
Methods and Materials
The Cheney Reservoir watershed is located in south central Kansas covering area of 2,404 sq km. Over 99% of the watershed is agricultural and the average rainfall in the watershed is 696 mm/year. The USGS has divided the watershed into 5 (five) smaller sub-watersheds. They are: West Ninnescah (1,223 sq km), Silver Creek (497 sq km), Goose Creek (130 sq km), Red Rock Creek (135 sq km), and East Ninnescah (419 sq km) (Fig 1).
Different GIS layers required for the AGNPS-ARC/INFO interface were obtained from the Cheney Reservoir Water Quality Project (CRWQP) Office, South Hutchinson, Kansas. The coverages were: a composite boundary for the different sub-watersheds, soils, stream, location of feedlots and waste water treatment plants in the Cheney Reservoir watershed, impoundments within the watershed, the locations of conservation measures being taken, and 7.5-minute digital elevation model (DEM) files in grid format for different 7.5-minute quadrangles of the watershed.
The land cover coverages for the year 1997 and 1998 were prepared from Landsat TM images using unsupervised/supervised classification (Marzen et al., 2000). The land cover included high, medium, and low cover classes for wheat stubble and rangeland, other cropland (including corn and sorghum), woodland, water, and residential. For each land cover class, a value was assigned for C-factor, surface condition constant, overland Manning's coefficient, chemical oxygen demand, and fertilizer level as AGNPS input parameters. A cell size of 65 ha was selected to create the input files for AGNPS.
Figure 1. Location of Cheney Reservoir watershed with different sub-watersheds.
The daily rainfall data from eleven weather stations within and near the Cheney Reservoir watershed were obtained from the Weather Data Library at Kansas State University. Average rainfall over a sub-watershed was computed using the Thiessen Polygon method (Thiessen, 1911). Different storm events for running the AGNPS model for each sub-watershed were identified on the basis of the runoff hydrographs during which significant amount of runoff occurred. The number of storms selected was enough to total up to about 70 to 90% of the annual runoff amount on each sub-watershed. The East Ninnescah sub-watershed was not considered for this study, because stream flow data that represented this area were not available for this sub-watershed.
Base flow values were separated from the total stream flow to get best estimates of surface runoff amount for the daily stream flow record using the USGS Hydrograph Separation Program (HYSEP 2.2) model (Sloto and Crouse, 1996). For each storm, the surface runoff portion from the date of storm until the surface runoff was less than 0.025 mm/day was summed to get the total surface runoff for that event. The daily surface runoff depth (mm) was calculated from the daily surface flow divided by the drainage area.
Antecedent Moisture Condition (AMC) and Runoff Curve Number (CN) Adjustment
A CN polygon coverage for watershed consisted of combining the land cover and soil coverages and CN for AMC II (CN II ) were assigned based on the basis of land use and hydrologic soil group (Soil Conservation Service, 1968). These CNs were adjusted for individual storms depending on the hydrologic condition or AMC before each storm event. AMC ratios were developed from measured runoff and the average CN (CN av ) of the watershed as described below. After assigning the CN II to each polygon in the runoff CN coverage, the CN av of each sub-watershed was also computed by weighing with the area of each polygon.
Surface runoff depth is calculated on the basis of relationship between rainfall excess and total rainfall (on a 24-hour basis) as follows (SCS, 1968):
where Q is surface runoff depth, mm; P is precipitation, mm; S is the storage parameter, mm; and CN is the runoff curve number of the watershed. The actual CN (CN actual ) of the sub-watershed to produce the measured amount of runoff depth was determined by solving. This CN actual may be between AMC I and II or between AMC II and III. Thereafter, the AMC ratio was estimated (AMCratio), whichever is applicable, on the basis of CN actual , CN av , and CN for AMC I (CN I ) or III (CN III ). Note that the change in CN is not equal between AMC I and II range and AMC II and III range. Therefore, for, CN actual < can and for CN actual > CN av
This AMCratio was then used to adjust the CNs of each land use in the watershed for each event by linear interpolation between CN II and the appropriate CN I or CN III .
For AMCratio <2:
For AMCratio >2:
The AGNPS input files for different storms were made for each sub-watershed using the AGNPS-ARC/INFO interface. The interface computes the average CN ( CN II ) for each 65 ha (160 acre) cell in the AGNPS input file. These CN II values were changed with adjusted CN during different storms events.
Results and Discussion
The total measured and total AGNPS-predicted runoff data during different storm events of 1997 and 1998 are presented in Table 1. The coefficient of determination (R 2 ) based on a linear regression (zero intercept and 1:1 slope) of measured versus predicted runoff depths are also presented in the table. R 2 values of 0.988, 0.960, 0.789, and 0.781 were obtained for Red Rock, Goose Creek, Silver Creek, and West Ninnescah sub-watersheds, respectively, during 1997. In terms of total prediction difference (the difference between measured and predicted over the measured values), these values were only 1, 8, 37, and 15 % for the sub-watersheds, respectively.
Similarly during the 1998, R 2 values of 0.979, 0.992, 0.963, and 0.545 were obtained between the measured and AGNPS-predicted runoff data on these sub-watersheds. The prediction difference values were estimated as 2, 2, 7, and 20 % for Red Rock Creek, Goose Creek, Silver Creek, and West Ninnescah sub-watersheds, respectively. The predictions on Silver Creek and West Ninnescah sub-watersheds were not as good as on the other two smaller sub-watersheds. This may be due to the variation of rainfall of over the sub-watersheds.
Table 1. Total measured and predicted runoff on different sub-watersheds
Total runoff, mm
Total runoff, mm
Red Rock Creek
AGNPS assumes uniform rainfall over the watershed, which is certainly not true in case of the larger watersheds. The annual rainfall in the eastern portion is greater than the western portion of the watershed. This created a distortion in the estimation of average rainfall over the watershed using Thiessen polygon method. The USGS reports that about 186 sq km of the West Ninnescah sub-watershed is noncontributing (Christensen and Pope, 1997), which may be another reason which affected the prediction.
Christensen, V . G. and L. M. Pope. 1997. Occurrence of Dissolved Solids, Nutrients, Atrazine, and Fecal Coliform Bacteria During Low Flow in the Cheney Reservoir Watershed, South Central Kansas 1996. USGS Water-Resources Investigation Report 97-4153. 13 pp.
Liao, Hsiu-Hua and U. S. Tim. 1997. An interactive modeling environment for non-point source pollution control. Journal of the American Water Resources Association 33(3): 1-13.
Marzen, L. J., J. A. Harrington, Jr., S. J. Bhuyan, and J. K. Koelliker. 2000. Use of satellite imagery to determine land cover input variables in an AGNPS water quality model. ASAE Paper No. 00-2200. Milwaukee, WI: ASAE.
Sloto, R.A. and M. Y. Crouse. 1996. HYSEP: A Computer Program for Streamflow Hydrograph Separation and Analysis. U.S. Geological Survey Water-Resources Investigations Report 96-4040. 46 pp.
Soil Conservation Service. 1968. Hydrology, Supplement A to Sec. 4, Engineering Handbook. USDS-SCS, Washington, DC.
Thiessen, A. H. 1911. Precipitation for large areas. Monthly Weather Review 39:1082-1084.
Young, R. A., C. A. Onstad, D. D. Bosch, and W. P. Anderson. 1987. AGNPS, Agricultural Nonpoint Source Pollution Model: A Large Watershed Analysis Tool . Cons. Research Report. 35. Agricultural Research Service, USDA, Washington, D.C. 77 pp.