The effects of pre-season high flows, climate, and the Three Gorges Dam on low flow at the Three Gorges Region, China

The efficient operation of a multipurpose reservoir requires information on high and low flows. However, analyses of inflows for high flows and for low flows are typically done independently. In this paper, we considered the joint dependence of the low flow on the preceding high flow volume and duration for the wet season in the Three Gorges region of the Yangtze River Basin in China. High flow volume and duration were found to have a strong association with the annual minimum 7-day flow in Cuntan, Wanxian, and Yichang stations. Furthermore, we identified the Arctic Oscillation, Pacific Decadal Oscillation, and snow cover in the Tibetan Plateau to have statistically significant teleconnections with the annual minimum 7-day flow. Bayesian models that consider a different level of pooling of the site by site regressions were then developed for the annual minimum 7-day flow conditional on the climate indices and high flow volume (or duration). The full pooling model performed best, suggesting that a homogeneous regional response is best identified given the global climate predictors. Statistics such as the deviance information criterion and reduction of error, coefficient of efficiency, and coverage rate under cross validation indicate the good performance of the model. Snow cover in the western Tibetan Plateau and high flow volume were identified as the most influential factors of the annual minimum 7-day flow through their impact on water storage in the basin. Recent simulations since June 2003, when the Three Gorges Dam operation started, were used to analyse the effect of dam operation on the annual minimum 7-day flow. A comparison of observations and predictions during the post-dam period demonstrated that the dam operation effectively modifies the annual minimum 7-day flow period to have higher flows.