Distribute SWAT Model And Multi-Forecast Study Of Chaohe Watershed In Miyun Reservoir

Miyun reservoir is one of the most important water resource for Beijing citizens, however, its run-off went lower in recent20years had a tendency of decay. In2012the run-off is less than0.4billion m3, decreasing by80%than60-70in last century, as we take it as research target for water forecasting to help researchers find reasons and make policy..Researches showed some reasons such as huge-scale planting, climate changing may clarify the facts, but there are various of combining effects among different explaining factors.This Paper uses SWAT model to make sub-basin research, on the other side, one way run-off forecast is a type of basic method to confirm the future. In text three different models: Multi-regression, BP-ANN, and ARIMA are introduced. Multi-Stepwise-Regression can do pre-analysis about independents:Collinear-Test gives that6independents are in some degree correlated, and it showed3independents are highly linear-correlated. The result of regression result is accepted. ARIMA’s model result takes the seasons, tendency and other random factors into considerations. ANN is better than the above2methods, because ANN is non-linear model and using BP structure.To analysis other factors than climate, the paper takes SWAT, a type of distribute water run-off forecast model which combined GIS information and considered details of all weather stations. SWAT model has a strength point of consisting of various sub-models and parameters, and do analysis of local HRUs, whereas, it makes model more complicated. At last, SWAT’s forecasting result is more accurate in recent forecast years, and the model can be used in other similar cases.Above all, previous forecast models can be taken as pre-analysis and focused on climate factors. SWAT model do analysis in the whole region, include climate factors, soil factors, land-use factors and so on. So research compared two different points and come into conclusions.