Can precipitation extremes explain variability in runoff and sediment yield across heterogeneous karst watersheds?

Abstract

Precipitation extremes are recognized as one of the main factors that influence annual runoff and sediment yield in specific watersheds. However, it is not clear whether precipitation extremes account accurately for the variability in runoff and sediment yield for different karst watersheds, especially in climate-sensitive areas, such as the karst critical zone of southwest China. The objective of this study was to quantify the relative importance of the 11 precipitation extremes variables that explain the total variance in annual runoff and sediment yield with a boosted regression tree (BRT) model for 40 typical karst watersheds in southwest China. Results indicated that the most important precipitation extremes for annual runoff and sediment yield were heavy precipitation amount, consecutive dry days, heavy precipitation days, consecutive wet days, and rainstorm amount. The BRT model accounted for 76% of the variation in runoff, which demonstrated that precipitation extremes were crucial factors for generating runoff. Although precipitation extremes had significant effects on sediment yield, the BRT model only explained 32% of the variability in sediment yield. Hydrogeological conditions and dam construction in these karst watersheds may be responsible for the difference in total variance explained by the BRT model between annual runoff and sediment yield. Because of the inherent nonlinearity present in many hydrological systems, the BRT model provided a robust capability to generate data-driven, objective, insight into complex relationships. We recommend this model for the application of non-stationary and nonlinear hydrological processes.