干旱区生态环境知识资源中心

In recent years, the natural hydrology behaviors were greatly influenced by climate change. The relation between runoff and climate change are always the core of scientific hydrological study in arid region. This paper presents a multi-variate time series controlled auto-regressive (CAR) model based on hydrological and climatic data of typical tributaries Jinghe River in Ebinur Lake Basin of Xinjiang covering the period from 1957 to 2012. The aim is to study the climate change and its effects on runoff of the Jinghe River, Northwest China. The results showed the following: the runoff of the Jinghe River was unevenly distributed and has obvious seasonal changes throughout the year. It was concentrated in summer and has along dry season with less runoff. The monthly maximum river runoff was from June to September and accounted for 74% of annual runoff. The river runoff increased since the 1980s till the 1990s; in the 21st century there was a trend of decreasing. The oscillatory period of annual runoff series in the Jinghe River Basin was 21a and 13a, and these periods were more obvious, followed by 32a and 9a. The oscillation with a time scale of 21a and 13a was a full timed domain. The MRE is 6.54%, the MAE is 0.84 x 10(8) m(3), and the RMSE is 0.039. The CAR model passed the F-test and residual test, and the change trend of calculated and measured values of annual runoff is consistence, which means that the model was reasonable.