Study On Medium-long Term Hydrological Forecast Of The Inflow In The Dry Season At Yangtze Three Gorges

The Yangtze three gorges project is the largest water control project engineeringwith comprehensive benefit of flood control, power generation, navigation, irrigationand water supply etc. In order to meet the requirements such as the flood control, powergeneration and navigation, reservoir operation must implement optimized scheduling.Cuntan station and Wulong station are the inflow station of the three gorges reservoir,also are the main stations of the upper reaches of Yangtse River. Cuntan stands thecatchment area of866559km2, and controls the basic water regime of minjiang river,tuojiang river, jialing river and chishui river into the Yangtze river. Wulong stands thecatchment area of83035km2, is of wujiang river system. So Cuntan and WuLongstation can reflect the characteristics of the upper Yangtze river runoff.Runoff is a kind of natural phenomenon with a large number of uncertainty andcomplexity, it plays a leading role in the evolution of the hydrologic system, at the sametime affected by the weather, natural geography, and basin characteristics etc. Such asthe study of the dry season runoff at three gorges has many uncertain influence factor.At present it is hard to use mathematical and physical method to accurately describe thecomplete evolution and depicts the process, therefore this paper chooses time sequencemid-to-long-term hydrological forecasting method, in the analysis of time series basedon the variable, the use of certain mathematical method to establish forecasting model,and make the time trend to extension, to get the trend of the development of thesequence, determine the predicted variables.This paper studys the Three Gorges dry season runoff using three models in threedifferent prediction scale, gets the model with good practicability, high precision, andprovide reference for the Three Gorges dry season water resources reasonable allocation.This paper firstly analyzes the characteristics of dry season runoff of the upper reachesof the Yangtze river. From the analysis of the results, Cuntan dry interannual variabilityof runoff is not big, but is decreasing trend. Wulong dry interannual variability of runoffis big, and is increasing. Then building the prediction model at the control stations for10days daily runoff, Ten-day runoff and Month runoff hydrological forecast, choosingNearest Neighbor Bootstrapping Regressive Model, Multivariate threshold regressionmodel, and Mean generation function model. Nearest Neighbor BootstrappingRegressive Model don’t need to make some assumptions for dependencies form and theform of probability distribution of the research object, is a kind of data driven model. From the simulation results, the model can well simulate the different forecasting scaleof Wulong and Cuntan, runoff variation. Multivariate threshold regression model is topartition linear autoregressive model to describe the research object in the entire intervalnon linear characteristics. From the simulation results, the model in10senior highschool entrance examination into the runoff forecast average flow before and after thedependence, good characterization of flow in the timing of random variation, and itsforecast effect is better. Mean generation function model assumes that things past trendsinto the future, and as human nature change ability, the change of environment itself onhydrology rule created, its own rules may be changed, so the mean generating functionapplied to the runoff change small watershed. From the simulation results, the result ofCuntan station runoff forecast is good, on the Wulong station runoff forecast effect ispoor.