Water Temperature Prediction Using Improved Deep Learning Methods through Reptile Search Algorithm and Weighted Mean of Vectors Optimizer

Precise estimation of water temperature plays a key role in environmental impact assessment, aquatic ecosystems’ management and resources planning management. In the current study, convolutional neural networks (CNN) long short-term memory (LSTM) network-based deep learning models were examined to estimate daily temperatures Bailong River China. Two novel optimization algorithms, namely reptile search algorithm (RSA) weighted mean vectors optimizer (INFO), integrated with both enhance their prediction performance. To evaluate accuracy implemented models, four statistical indicators, i.e., root square errors (RMSE), absolute errors, determination coefficient Nash–Sutcliffe efficiency utilized on basis different input combinations involving air temperature, streamflow, precipitation, sediment flows day year (DOY) parameters. It was found that LSTM-INFO model DOY outperformed other competing by considerably reducing RMSE MAE predicting temperature.