A Semi-distributed Urban Rainfall-runoff And Non-point Source Pollution Simulation System

Urbanization process has led to significant increase of impervioussurface, so that the rainfall-runoff increases, runoff peak comes sooner withbigger flow, and the probability of flood increases. Washing the surfacesediments, rainfall-runoff creates more urban non-point source pollution; thequality of drinking water is threatened. Therefore, there is urgent need forresearching mechanism of urbanized area rainfall runoff and pollutants inorder to take effective measures to reduce flood risk and control of the urbannon-point source pollution.Currently distributed model is widely used in this area in foreigncountries, which has relatively high requirements for monitoring data.However, there is a lack of monitoring data of urban runoff and non-pointsource pollutants in China, thus, parameter determination often rely onoptimal calibration methods. Optimal calibration methods need complexinformation about the model parameters, the amount of calculation is great,and often causes unexpected problems. Compared with the distributed model,semi-distributed model has advantages because of its low requirements formonitoring data.Based on “A semi-distributed urban rainfall-runoff and pollutionsimulation system”, we introduce the concept of Clustering. Through themaster-slave parallel genetic algorithm, we achieve “A semi-distributed urbanrainfall runoff and pollution simulation system”. Firstly, TOPAZ model isused to deal with DEM data, then through clustering algorithm, the sub-basinis clustering into similar clusters, each cluster has respective definition of theerosion and other corresponding parameters; Finally, the master-slave parallelgenetic algorithms is used to speed up the process of parameter calibration.As a case of Beijing Wenyu River basin, the measured data were calibrated and validated. The result shows that the urban runoff and runoffpollution can be simulated well; the master-slave parallel genetic algorithm,which is implanted into the model, greatly improves the computationalefficiency level; the increase of impervious surface under urbanization cansignificantly increase the runoff at the same rainfall. Meanwhile, runoffpollution load is significantly higher in rainy season than that in the dryseason. To improve urban runoff quality, early storm water must be block totreatment.