Elaborate simulations of floods in a karst trough valley basin with the simplified Karst-Liuxihe model

•A physically-based distributed Karst-Liuxihe model (KL) was improved and simplified effectively.•The rainfall-runoff generation and confluence algorithms were improved.•Better flood simulation results showed that the simplified and improved model is successful.•The simplified KL model requires less hydrogeological data when modeling in karst basins. The karst trough valley basin in Zhongliang of Chongqing is one of the most developed karst areas in Southwest China, with unique topographic features such as karst troughs and valleys that are prone to flash floods. Accurate simulation of flood processes in these areas using hydrological models can provide references for predicting future local runoff evolution trends. The main challenge for hydrological modelling in karst areas lies in building a model with limited hydrogeological data. In this study, the Karst-Liuxihe (KL) model was simplified and improved to simulate karst flood processes in the Zhongliang karst trough valley. The KL model consists of multiple, complex structures; a large amount of data is needed to model karst areas. To overcome the modelling data requirements, we modified the model structure and its parameters to propose a simplified KL (SKL) model. Additionally, we improved the runoff generation and confluence algorithms. Compared with the simulation results for 22 flood processes obtained by using the KL model, the results for this SKL model were better with less modelling data. For example, the mean value of the Akaike information criterion (AIC) index decreasing rate was 35%, and the process relative error and the peak flow relative error decreased by 10%, indicating that the technique of simplifying and improving the KL model is effective and that the simplified SKL model can quantitatively depict karst floods in karst trough valley areas.