The impact of climate and catchment changes on streamflow and flow forecasting in the Qiantang River Basin, China

The purpose of this thesis is to investigate of the impact of climate and catchment changes on streamflow and flow forecasting in the Qiantang River Basin, China. Precipitation and discharge slightly increase during past decades at most gauging stations, and this trend becomes stronger from 1979--2000. Evaporation trends are not consistent and cultivated land area generally decreases over time. Precipitation is the dominant control of streamflow. Periodicities of 2--7 years in the time series may be related to the El Nino/Southern Oscillation. Reservoirs weaken the rainfall-discharge relationship and cause a muted seasonal flow pattern.; Monthly flows for three gauging stations were used to develop and test autoregressive integrated moving average (ARIMA) models. Stepwise models based on updates from observed data produced acceptable results but tend to slightly underestimate actual flows. Stepwise models based on updates from predicted values and non-stepwise models simply reproduce seasonal patterns and are not useful for forecasting. Long-term memory in the time series may affect the performance of the ARIMA models.; Three-layer feed forward backpropogation artificial neural network (ANN) and linear regression (LR) models were used for monthly flow forecasting based on precipitation inputs. The ANN provided acceptable estimates of discharge but extreme flows were not well estimated. The simpler LR models provided better estimates of discharge than the ANN. A global circulation model was used to provide predictions of monthly precipitation data that were used as input for forecasting flows for the period 2001--2050 using ANN and LR models. The LR model is recommended for long-term forecasting in the Qiantang basin but an ANN model would likely be a better choice in more non-linear watersheds or when there is likely to be a change in future precipitation.