Simulation of Climate Change Impacts on Streamflow in the Bosten Lake Basin Using an Artificial Neural Network Model

Abstract

Impacts of climate change on water resource in the Bosten Lake basin in the south slope of the Tianshan Mountains in Xinjiang, China, were evaluated using an artificial neural network model. The model was trained using the error backpropagation algorithm and validated for a major catchment that covers 82% of the Bosten Lake basin and has the only available weather and streamflow data. After validating the model it was used to examine the surface hydrology responses to changes of regional temperature and precipitation. Major results showed that because of an additional effect on glacier melt in the upper reach of the basin temperature increase can cause large increases of streamflow. Model results also showed that if the current climate trend continues, the annual streamflow would increase by 38% of its current volume, and the summer and winter streamflow would increase by 71.8 and 11.4% of their respective current volume in the next