Abstract: As salt tide invasion in the Pearl River becomes increasingly severe in dry season, the risk of inhibiting the invasion is aggravated by uncertainty in the runoff, gravely lowering the safety of water supply across the watershed. This paper describes the uncertainty in the river runoff based on probability density distribution, and constructs a basic framework of single-source risk operation. In a case study of five reservoirs in the Xijiang watershed, we formulate a quantitative relationship of the runoff forecast error versus the systematic risk of inhibiting salt tide invasion, and reveal how the risk is transmitted in space and time. The forecast error is divided into three grades, and the threshold of each risk leapfrogging across the grades is determined. The results show that as the forecast error increases, the risk occurs earlier, the number of salt-inhibiting days and the risk value are increased. In time, the risk transfers from time to time in a cumulative way; in space, the risk transfers from upstream to downstream and from tributaries to the main stream. The risk leapfrogs from light to medium grade when the forecast error exceeds 16%, and from medium to heavy grade when it exceeds 21%. The results have significant application to risk regulation and control for the Xijiang watershed and to water supply security assessment for the Guangdong-Hong Kong-Macao Greater Bay Area.

Key words: inhibiting salt tide invasion, risk operation, uncertainty of runoff, forecast relative error, risk rank, critical threshold