Flow forecasting is an essential topic for flood prevention and mitigation. This study utilizes a data-driven approach, the Long Short-Term Memory neural network (LSTM), to simulate rainfall–runoff relationships for catchments with different climate conditions. The LSTM method presented was tested in three catchments with distinct climate zones in China. The recurrent neural network (RNN) was adopted for comparison to verify the superiority of the LSTM model in terms of time series prediction problems. The results of LSTM were also compared with a widely used process-based model, the Xinanjiang model (XAJ), as a benchmark to test the applicability of this novel method. The results suggest that LSTM could provide comparable quality predictions as the XAJ model and can be considered an efficient hydrology modeling approach. A real-time forecasting approach coupled with the k-nearest neighbor (KNN) algorithm as an updating method was proposed in this study to generalize the plausibility of the LSTM method for flood forecasting in a decision support system. We compared the simulation results of the LSTM and the LSTM-KNN model, which demonstrated the effectiveness of the LSTM-KNN model in the study areas and underscored the potential of the proposed model for real-time flood forecasting.

No abstract

No abstract

Free water storage capacity, an important characteristic of land surface related to runoff process, has a significant influence on runoff generation and separation. It is thus necessary to derive reasonable spatial distribution of free water storage capacity for rainfall-runoff simulation, especially in distributed modeling. In this paper, a topographic index based approach is proposed for the derivation of free water storage capacity spatial distribution. The topographic index, which can be obtained from digital elevation model (DEM), are used to establish a functional relationship with free water storage capacity in the proposed approach. In this case, the spatial variability of free water storage capacity can be directly estimated from the characteristics of watershed topography. This approach was tested at two medium sized watersheds, including Changhua and Chenhe, with the drainage areas of 905 km2 and 1395 km2, respectively. The results show that locations with larger values of free water storage capacity generally correspond to locations with higher topographic index values, such as riparian region. The estimated spatial distribution of free water storage capacity is also used in a distributed, grid-based Xinanjiang model to simulate 10 flood events for Chenhe Watershed and 17 flood events for Changhua Watershed. Our analysis indicates that the proposed approach based on topographic index can produce reasonable spatial variability of free water storage capacity and is more suitable for flood simulation.