Study On Stage Forecasting Model And The Case Study Of Poyang Lake After The Project Of Re-conversion Of Reclaimed Lands To Wetlands (PRRLW)

Poyang Lake connects five rivers (Gan River, Fu River, Xiu River, Rao River,Xin River) in Jiangxi province, and is also maintained in a high water level byYangtze River and local inflow. In 1998, Poyang Lake undergone superflood, havebeen actualized PRRLW in Poyang Lake District from then. The boundary conditionof Poyang Lake and water level feature of typical gaging station changed because ofthis project. Study on water and soil comprehensive utilization of flood-strorageregion and flood prevention and drought resistance, we must know the change ofwater level. So stage forecasting model of Poyang Lake after PRRLW is established,water level feature of typical gaging station are analysised, the study can provides gistfor flood prevention and drought resistance of Lake region and comprehensiveutilization of lake region, at the same time, it can provide case for study of hydrologicof lake, affluent in theory of hydrologic of lake.Conclusions are the following:(1) Using stepwise regession, stage forecasting models of several typical gagingstations are established. These stations include HuKou station, XingZi station,DuChang station, TangYin Station, and KangShan station. WaiZhou station of GanRiver, LiJiaDu station of Fu River, MeiGan station of Xin River, DuFengKeng andHuShan station of Fu River, WanJiaBu and QiuJin station of Xiu River are shortenedfrom Fiver Rivers and Seven stations. Their mean daily discharge and men daily stageof HuKou station and local inflow are deemed to influence factors. Then water levelsof several typical stations are predicted according to different flood frequency. Butlocal inflow should be substituted for average daily precipitation. that Poyang Lake'surface and several typical rainfall stations around Poyang Lake region.(2) Using artificial neural network (ANN), water level prediction model isestablished. It be deemed to input variable that daily streamflow of five rivers andseven stations. It be deemed to output variable that daily water level of several typicalwater level stations. Using BP arithmetic of ANN, samples are trained; parameters ofmodel are achieved. According to these parameters, feature water levels of several typical stations are predicted through assumeing a set of input variable. The result iscompared with result of stepwise regression model. The precision and dependabilityof ANN model are analysed.(3) Optimization scheme that how operate flood protection dyke are framed. Theflood prevention measure of flood protection dam and region of flood storage isinvestigated.. Optimization scheme is established, so as to can diverse flood andreduce economic loss according to relational flood control of planning.