SIMULATION OF QUANTITY AND QUALITY OF STORM RUNOFF FOR URBAN CATCHMENTS IN FRESNO, CALIFORNIA

Rainfall-runoff models were developed for a multiple-dwelling residential catchment (two applications), a single-dwelling residential catchment, and a commercial catchment in Fresno, California, using the U.S. Geological Survey Distributed Routing Rainfall-Runoff model (DR3M-II). A runoff-quality model also was developed at the commercial catchment using the Survey's Multiple-Event Urban Runoff Quality model (DR3M-qual) . Data from the U.S. Environmental Protection Agency's Fresno National Urban Runoff Program for the rain seasons 1981-82 and 1982-83 were used to calibrate and verify the two models. The purpose of this study was (1) to demonstrate the capabilities of the two models for use in designing storm drains, estimating the frequency of storm runoff loads, and evaluating the effectiveness of streetsweeping on an urban drainage catchment; and (2) to determine the simulation accuracies of these models. Pipe capacities were exceeded at all three catchments when the 2-year design rainfall was input to DR3M-II. Two-year peak discharges also were estimated using the rational formula method for comparison with DR3M-II. The results of the two methods were similar only because the rainfall-runoff model cannot simulate flows greater than the calculated capacity of the pipes, and the runoff coefficients for the rational formula were too low. The runoff from the 2-year event probably would cause little or no pressurized flow or ponding at the multiple(1) catchment, but pressurized flows and ponding probably would occur at the single and commercial catchments. As a result, the 2-year peak discharge at the multiple(1) catchment probably is a reasonable estimate, but the estimated peaks at the single and commercial catchment probably are low. Simulation errors of the two models were summarized as the median absolute deviation, in percent, between measured and simulated values. Calibration and verification mad errors for runoff volumes and peak discharges ranged from 14 to 20 percent. A long-term time series of annual storm-runoff loads at the commercial catchment was produced by entering historical rainfall data to DR3M-II and entering the simulated discharges to DR3M-qual. Return periods for annual stormrunoff loads were determined by fitting a logPearson type III distribution to the time series of annual storm-runoff loads. The estimated annual storm-runoff loads from effective impervious areas that could occur once every 100 years at the commercial catchment were determined as 95 pounds per acre for dissolved solids, 1.6 pounds per acre for dissolved nitrite plus nitrate, 0.31 pound per acre for total recoverable lead, and 120 pounds per acre for suspended sediment. Calibration and verification errors for the above constituents ranged from 11 to 54 percent. The effectiveness of streetsweeping at the commercial catchment was evaluated by entering the long-term time series of discharges into DR3M-qual with various estimates for the model streetsweeping criteria. The analysis showed that a daily cleaning of all parking areas of the catchment by a 50-percent efficient sweeper could decrease the annual storm-runoff loads for dissolved solids and total recoverable lead by 27 percent.