Abstract
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.
Highlights
Streamflow simulation and flood forecasting are the main tasks in hydrological sciences and constitute the primary nonstructural flood prevention measures to avoid flood damages
5) reveal that the discharge simulations from the modelmodel were close the observations; itthus, is capable of computing both theboth hydrograph and the wereto close to the observations; it is capable of computing the hydrograph peakthe flow
Our results show that the Long Short-Term Memory neural network (LSTM)-k-nearest neighbor (KNN) model yielded relatively better performance in all three catchments, which demonstrates that the KNN algorithm, acting as an error updating model, could select useful historical data points that effectively reduced the error accumulation
Summary
Streamflow simulation and flood forecasting are the main tasks in hydrological sciences and constitute the primary nonstructural flood prevention measures to avoid flood damages. The forecasting models can be categorized into two types, process-based and data-driven models [1]. Process-based models describe, in a detailed manner, the different components and processes of the hydrological cycle. These models attempt to derive physical parameters that are useful for simulation or prediction tasks [2]. Data-driven models acquire the relationship between input and. The inherent complexity in the hydrological process and the impact of human activities make it challenging for process-based or conceptual models to capture highly nonlinear relationships in the hydrological processes for these areas. Data-driven models such as artificial neural networks (ANNs) gain an advantage in mimicking highly non-linear and complex systems as well as building models without a priori information [5]. Efforts to incorporate these models into real applications in the field of hydrology have been made
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
