Enhancing hourly streamflow simulation in karst basins: Development of the modified DK-XAJ-EW hydrological model integrating epikarst water storage function dynamics

Abstract

The water storage function of the epikarst plays a critical role in influencing the precipitation-streamflow response in karst basins; however, this dynamic is often inadequately represented by a single saturation excess runoff generation mechanism. To effectively capture the spatial heterogeneity of the epikarst’s water storage capacity and its temporal dynamics in redistributing runoff, we developed a reservoir model module. This module performs storage calculations for runoff derived from the saturation excess mechanism, employing an exponential equation to represent the spatial variability of epikarst water storage. By integrating this module with the runoff generation and separation modules of the DK-XAJ model, we quantitatively describe both the vertical and lateral runoff processes within the epikarst, accounting for their temporal evolution driven by spatial heterogeneity, resulting in the modified DK-XAJ-EW model. RSA analysis shows that epikarst reservoir parameters enhance hydrological simulations across various objective functions. The DK-XAJ-EW model outperforms the DK-XAJ model in simulating hourly streamflow and flood events, with improvements in NSE, KGE, R², and RRE values of 0.12, 0.12, 0.12, and 1.05%, respectively. It more accurately simulates surface runoff, interflow, rapid-conduit runoff, and slow-matrix runoff, improving flood event simulations with rapid and slow rise-recession patterns. The average absolute RPE decreased from 13.3% to 6.4% for rapid events and from 19.2% to 6.7% for slow events, while the average absolute PTE decreased from 2 hours to 0.5 hours for rapid events and from 7.5 hours to 3 hours for slow events, highlighting the importance of epikarst water storage in modelling precipitation-streamflow dynamics. The results indicate that the DK-XAJ-EW model provides important insights for refining hydrological predictions in karst basins and has considerable potential for enhancing the accuracy of flood forecasting in these complex environments.