Recently, weather data have been applied to one of deep learning techniques known as “long short-term memory (LSTM)” to predict streamflow in rainfall-runoff relationships. However, this approach may not be suitable for regions with artificial wat...
Recently, weather data have been applied to one of deep learning techniques known as “long short-term memory (LSTM)” to predict streamflow in rainfall-runoff relationships. However, this approach may not be suitable for regions with artificial water management structures such as dams and weirs. Therefore, this study aims to evaluate the prediction accuracy of LSTM for streamflow depending on the availability of dam/weir operational data across South Korea. Four LSTM models were designed for 25 streamflow stations. LSTMs #1 and #2 used weather data, and weather data combined with dam/weir operational data, respectively, maintaining consistent LSTM model conditions across all stations. Meanwhile, LSTMs #3 and #4 applied weather data, and a combination of weather and dam/weir operational data, respectively, while implementing unique LSTM models for individual stations. The Nash-Sutcliffe efficiency (NSE) and the root mean squared error (RMSE) were adopted to assess the LSTM’s performance. The results indicated that the mean values of NSE and RMSE were 0.277 and 292.6 (LSTM #1), 0.482 and 214.3 (LSTM #2), 0.410 and 260.7 (LSTM #3), and 0.592 and 181.1 (LSTM #4), respectively. Overall, the model performance was improved by the addition of dam/weir operational data, with an increase in NSE values of 0.182 – 0.206 and a decrease in RMSE values of 78.2 – 79.6. Interestingly, the extent of performance improvement varied according to the operational characteristics of each dam/weir. Particularly, the performance tended to increase when dams/weirs with high frequency and substantial amounts of water discharge were considered. Our findings confirm that the accuracy of streamflow predictions using LSTM is markedly improved with the inclusion of dam/weir operational data. Therefore, a comprehensive understanding of the operational characteristics of dams/weirs is crucial when utilizing their data to make reliable streamflow predictions with LSTMs.