A distributed non-point source pollution model:calibration and validation in the Yellow River Basin
A distributed non-point source pollution model: calibration and validation in the Yellow River Basin

The applicability of a non-point source pollution model--SWAT(soil and water assessment tools) in a large river basin with high sediment runoff modulus(770 t/km2 in the Yellow River) was examined. The basic database, which includes DEM, soil and landuse map, weather data, and land management data, was established for the study area using GIS. A two-stage "Brute Force" optimization method was used to calibrate the parameters with the observed monthly flow and sediment data from 1992 to 1997. In the process of calibration automated digital filter technique was used to separate direct runoff and base flow. The direct runoff was firstly calibrated, and the base flow, then the total runoff was matched. The sediment yield was calibrated to match well. Keeping input parameters set during the calibration process unchanged, the model was validated with 1998--1999's observed monthly flow and sediment. The evaluation coefficients for simulated and observed flow and sediment showed that SWAT was successfully applied in the study area: relative error was within 20%, coefficient of determination and Nash-Suttcliffe simulation efficiency were all equal to or above 0.70 during calibration and validation period.