Abstract

Drought limit water level has been presented to guide reservoir drought resistance for years. However, current static drought limit water levels neglected the water transfer relationship among different periods. In this study, the Yuqiao Reservoir located in Tianjin, China, was taken as the study area. To avoid the excessive subjectivity brought by the conventional typical drought-year selection method, six indicators were used to analyse inflow intra-drought-season distribution characteristics, and a hierarchical agglomeration cluster method was used to propose a classification of the dry season inflow “shape.” To associate with the reservoir inter-month water transfer relationship, a number pair “early warning period/drought limit water level” format was proposed to represent the limit water level. The early warning period turned out to be related to initial water storage, inflow process in dry season and water supply rules. To overcome the uncertainty in inflow process during the dry season, a rolling computation mechanism was applied in determining the drought limit water level, thus dynamic regulation was realized. Finally, the dynamic reservoir drought limit water level was compared with the conventional method. The results suggest that drought limit water level has broad development potential in improving reservoir drought resistance guidance practices.