Hydrological modelling of large-scale karst-dominated basin using a grid-based distributed karst hydrological model

Abstract

Hydrological modelling is important to flood hazard prevention in the flood-prone karst area of Southwest China. Due to the special hydrogeological features with strong spatial heterogeneities and whose information is rarely obtained in large-scale karst-dominated basins; thus, it’s still a challenge to conduct accurate sub-daily distributed precipitation-streamflow simulation in these regions. To reveal the characteristics of hydrological response and the inertia of karst systems that are influenced by the spatial heterogeneity of the degree of karstification, a new Grid-based Distributed Karst Hydrological Model (GDKHM) is developed by this study to simulate streamflow in Li River Basin (LRB), which is a typical large-scale karst-dominated basin of Southwest China. The GDKHM considers the duality of hydrological behavior and the spatial heterogeneities of land uses types and karstification of each grid and river channels routing characteristics between different sub-basins. The results show the GDKHM using calibrated parameters could reflect the spatial distribution characteristics of hydrological response in the LRB. The mean NSE, KGE, R2 and RRE of the streamflow of all the sub-basin outlets during validation period reached 0.82, 0.84, 0.83 and 4.64%, respectively. The hydrograph simulation result of nested interior grid without recalibration was also satisfied. Furthermore, the GDKHM was able to reproduce the Auto-correlation function (ACF) shapes of simulated lateral inflow time series in sub-basins with different percentage of karst terrain areas (PKTAs) during validation period, and the ACF characteristics information including memory effect, slopes of rapid-conduit and slow-matrix flow phases could be captured, which indicates the model can characterize the spatial variability of karst system inertia in large karst basin. Due to the better performance in hydrograph and ACF of streamflow, GDKHM has the potential to successfully support flood forecasting in large-scale karst-dominated basins.