Climate Change and Multi-Reservoirs Impacts on Extreme Flood: Nonstationary Regional Frequency Analysis and Uncertainty Estimation

The nonstationary impacts of climate change and multi-reservoirs on extreme floods cannot be ignored due to the potential risk to flood control and river security. Nonstationary regional flood frequency analysis (NS-RFFA) provides an effective way to consider these influences. But there is currently no standard framework of NS-RFFA and insufficient consideration of uncertainty in quantile estimates. In the study, we proposed an improved framework of NS-RFFA considering the influence of multi-reservoirs as well as climate change and involving uncertainty estimation. The framework was applied to analyze the frequency of extreme floods in the Huai River Basin (HRB) where multi-reservoirs have been constructed and to reveal the extreme flood variation under the impact of climate change and reservoir group. Results show that precipitation during flood season and reservoir index played a dominant role in the variations of annual maximum streamflow in the HRB. The impact of reservoirs was more significant in the upstream tributaries than in the mainstream. The nonstationary models with location parameters varying with time and/or main influencing factors perform better on fitting extreme streamflow in the HRB compared with stationary and at-site analysis. The quantile estimates and their uncertainty based on NS-RFFA are dynamic, which gives an expression of the changing environment effects on annual maximum streamflow. Thus, our work contributes an improved framework of NS-RFFA considering the influence of climate change and reservoirs, providing more accurate and reliable estimates for river flood risk management.