EVALUATION OF APPLICABILITY OF DATA AUGMENTATION METHOD FOR DAM INFLOW PREDICTION USING DEEP LEARNING

Abstract

 We attempted to improve the accuracy of dam inflow forecasting using deep learning by augmenting the training data, and validated the applicability of the proposed method to multiple watersheds. The proposed data augmentation method which assumes a steady-state condition of constant rainfall, and uses a theoretical data set of virtual rainfall-runoff data as the augmentation data, such that the total rainfall and dam inflow into the watershed are equal (the runoff coefficient is 1.0). As a case study, the applicability of the data augmentation method to recent large-scale runoff was validated for four dam basins: Terauchi Dam, Miyagase Dam, Nomura Dam, and Kanayama Dam. The prediction accuracy of the data augmentation method was confirmed for each dam. However, when the test flood was much larger than the study flood in the past, there was a limit to the improvement in reproducibility.