Understanding the variation of water level extremes with their potential drivers can provide insights for flood risk management. In this study, temporal variability of water level extremes is investigated across the plain river network region of the Taihu Basin. The driving force analysis on water level extremes is mainly conducted for atmospheric (rainfall, climatic index, and tide) and anthropogenic forcing. The quantile perturbation method is employed to examine variability of extreme values and the Spearman correlation analysis to identify potential drivers of extreme water level variability. Considering water level extremes in all seasons, the 1990s have statistically significant positive anomalies, while the late 1960s to the 1970s and the 2000s have significant negative anomalies. The oscillation pattern of anomaly in summer has a higher variability than that in the other three seasons. Significant correlations are detected between the anomalies of water level extremes and rainfall (tide level) during summer and winter. Water level extremes in summer and winter have a strong connection to the Pacific Decadal Oscillation and North Atlantic Oscillation/Arctic Oscillation, respectively. Conversely, no consistent significant correlations between water level extremes and climatic indices are found in spring and autumn, which is mainly related to hydraulic structure construction and operation.
K E Y W O R D Sdriving forces, quantile perturbation method, the Taihu Basin, variability, water level extremes