Watershed Non-point Source Pollution Models, And Uncertainties

Miyun Reservoir is the source for drinking water in Beijing. As one of the catchment basins of Miyun Reservoir, Chaohe River plays a great role on water quality and supply of downstream reservoir. Related researches indicate that Non-point source pollution has become the main source of water pollution in Miyun Reservoir. With the increasing demand of drinking water, the research and control of non-point source pollution has become the urgent need. There is great significance on the protection of water quality of Miyun Reservoir and the safety of drinking water for Beijing.As Model simulation is an important method of non-point source pollution research, this paper mainly focused on the watershed non-point source pollution models, carried out comparative evaluation of the applicability of NPS models and uncertainty analysis of model parameters in Chaohe River basin. Then using the calibrated and verified model, the non-point source pollution loads and the characteristics of spatial and temporal variation of non-point source load in the Chaohe River basin were evaluated and summarized. The main conclusions are as follows:(1) Three kinds of non-point source models were compared by discussing the performances of applications from the literature. The results show that the performances of HSPF model for runoff and sediment simulation are usually better than the that of SWAT and AnnAGNPS in the same watershed conditions. As to the effect of nutrients Simulation, SWAT is slightly better than the other two models. In the view of model applicability scope, SWAT is relatively bigger than others. Three models generally have more reliable simulated results for long time series. But with the time step shortening, the precision of simulation decreased.(2) HSPF model had been calibrated and verified by using measured hydrologic and water quality data in Chaohe River basin. The simulation results matched the measured very well. A relative error and the Nash-Suttcliffe efficiency coefficient were calculated to evaluate the applicability of HSPF model. The Nash-Suttcliffe efficiency coefficient for the simulation of stream flow, sediment concentration and nutrients concentration were 0.88,0.74,0.76,0.63 accordingly. The analysis results indicate that the HSPF model is applicable for the simulation and prediction of non-point source pollution on Chao River basin. Then the first order error method was used to analyze the uncertainty of model parameters. The key uncertainty parameter for runoff, sediment and nutrient was determined. The CV values were calculated to evaluate uncertainty of each output variable. The degree of dispersion increases with an increase in the value of CV. Among the fore variables, sediment had the highest CV value(229.6%) and runoff had the lowest (12.99%). This suggests that the uncertainty associated with sediment was the greatest and the uncertainty associated with runoff was the least.(3) Based on the calibrated and validated model, the non-point source pollution load was quantitative evaluated, and the discipline of spatial and temporal variation of non-point source load was analyzed. The results showed that Changes of pollution load is relative with the rainfall during the year. Pollution load in the flood season accounts for a large proportion of the year. from the entire watershed, the spatial distribution of pollution load is greatly influenced by rainfall and runoff. The proportion for sediment loads is from 78% to 90%. The proportion for nutrients loads is from 30% to 50%. The distribution of pollution loads and rainfall are significantly correlated, especially in the high flow year. The role of rainfall in low flow years is not obvious. Pollution loads distribution is influenced by combined effects of the rainfall, land use, soil type and other factors.