Modelling hydrologic processes of a karst region in Southern China using coupled hydrological models

In the karst area, the epikarst increases the infiltration of rainfall and the depressions and conduits make water exchange frequently between surface rivers and karst aquifers. To improve simulation accuracy, we proposed the Soil and Water Assessment Tool-Genetic Algorithm-Back Propagation (SWAT-GA-BP) model. Also, the SWAT-tank model was created to elaborate the runoff processes in different karst systems. Furthermore, we designed three scenarios to study the influence of land use changes on runoff. The Zuo River Basin was selected as a case study. Compared to SWAT, the determination coefficient of the SWAT-GA-BP increased by 4.8% and the Nash-Sutcliffe efficiency increased by 4.9%; the rapid flow accounted for more than 50% of the total runoff, and the slow flow for less than 10%; the contribution of the land use changes was 10.80% from 2008 to 2012 and 5.57% from 2013 to 2017. Therefore, the effect of land use change on runoff should not be ignored.