Hydrological series data are non-stationary and nonlinear. However, certain data-driven forecasting methods assume that streamflow series are stable, which contradicts reality and causes the simulated value to deviate from the observed one. Ensemble empirical mode decomposition (EEMD) was employed in this study to decompose runoff series into several stationary components and a trend. The long short-term memory (LSTM) model was used to build the prediction model for each sub-series. The model input set contained the historical flow series of the simulation station, its upstream hydrological station, and the historical meteorological element series. The final input of the LSTM model was selected by the MI method. To verify the effect of EEMD, this study used the Radial Basis Function (RBF) model to predict the sub-series, which was decomposed by EEMD. In addition, to study the simulation characteristics of the EEMD-LSTM model for different months of runoff, the GM(group by month)-EEMD-LSTM was set up for comparison. The key difference between the GM-EEMD-LSTM model and the EEMD-LSTM model is that the GM model must divide the runoff sequence on a monthly basis, followed by decomposition with EEMD and prediction with the LSTM model. The prediction results of the sub-series obtained by the LSTM and RBF exhibited better statistical performance than those of the original series, especially for the EEMD-LSTM. The overall GM-EEMD-LSTM model performance in low-water months was superior to that of the EEMD-LSTM model, but the simulation effect in the flood season was slightly lower than that of the EEMD-LSTM model. The simulation results of both models are significantly improved compared to those of the LSTM model.

Global climate change and rapid urbanization increase the risk of urban flooding, especially in China. Climate change and the ‘heat island effect’ have increased the frequency of extreme precipitation. Affected by the backwardness of drainage facilities and the lack of drainage capacity, many cities have experienced large-scale waterlogging in low-lying areas, and ocean-like phenomena appear in cities. The public infrastructure was damaged and caused a lot of economic losses. Therefore, it is important to investigate the adaptability of drainage systems to the future in a changing environment. The Sixth International Coupled Model Intercomparison Project (CMIP6) and Storm Water Management Model (SWMM) were used to quantify the impact of climate change on Beijing's waterlogging under different rainstorm scenarios for the future 40 years. The quantile delta mapping method of daily precipitation based on frequency (DFQDM) is proposed to correct the daily precipitation of the climate model and which is proved to be feasible. After the annual precipitation and extreme precipitation index are corrected, percent bias (PBIAS) is significantly reduced. The PBIAS of the extreme precipitation index of the corrected model is all controlled within 6%. The corrected accuracy of CanESM5 is the best. The total flood volume (TFV) of the node increases with the aggravation of climate change. The TFV of SSP5-8.5 and SSP2-4.5 increased by 45.43 and 20.8% in the 100-year return period, respectively, and more than 94% of the conduits reached the maximum drainage capacity in different return periods. After the low impact development (LID) was installed, the improvement effect on the outflow with a smaller return period was significant, decreasing by about 50%. The LID can effectively reduce the overflow of the drainage system. The results of this study can provide suggestions for the reconstruction of the drainage system and the management of flood risk for Beijing in the future.

A groundwater model is needed to describe the complex groundwater confluence process of the groundwater system in karst areas. This is because surface water flows through dolines, grikes, and by other means and is directly exchanged with the groundwater. In this study, using the Xin’anjiang model, the conversion of surface water into groundwater and the influence of multiple series-parallel underground reservoirs on groundwater confluence through the generalization of dolines in karst areas were simulated. The water cycle process in the Sancha River Basin was simulated with measured data using multiobjective particle swarm optimization. Then, model parameters were validated with measured runoff data and compared with simulation results obtained using the traditional Xin’anjiang model based on its optimal parameters. The results showed that the determination coefficients of all hydrological stations over the study period were >0.76, and the Nash efficiency coefficient was >0.76, which were better than those for the improved Xin’anjiang model. Next, the simulation accuracy of the flood period in the karst area was analyzed. The model achieved a high fitting rate for the main flood peaks in a year, and the passing rate for the worst hydrological stations was 53%. Finally, the influence of karst development on the runoff was examined. The results indicate that different karst development stages and the heterogeneity of the karst in the basin have different effects on runoff.