Research on the Monthly Precipitation and Runoff Forecastingfor the Heihe River Basin

摘要

One step ahead monthly forecasts of precipitation and streamflow in the Heihe River Basin in North — west China have been greatly improved by considering meteorological teleconnections. Enhanced values of explained variance for stream flows were obtained by incorporating antecedent values of the sea surface temperatures along the Pacific coast of China and the phase of the Southern Oscillation in addition to antecedent monthly precipitation air temperature and runoff into the forecast procedure. Although correlation analysis was employed to make the initial choice of independent variables , the forecast model was developed in the form of an Artificial Neural Network. The forecast model obtained was found to be capable of reproducing the behavior of low and medium sized flows to a high degree of accuracy. Nevertheless, the levels of explained variance achieved in this study by considering teleconnections were found to be high. This enhanced performance can be partly attributed to the regularity of the flow regime in the Heihe River Basin, for which the snowmelt in spring and the frequent interchanges between surface water and groundwater are particular features.