Uncovering Flooding Mechanisms Across the Contiguous United States Through Interpretive Deep Learning on Representative Catchments

• Published in: Water resources research 2022-01, Vol.58 (1), p.n/a
• Main Authors: Jiang, Shijie, Zheng, Yi, Wang, Chao, Babovic, Vladan
• Format: Article
• Language: English
• Subjects: Additives; Artificial intelligence; Catchment area; Catchments; Climatic conditions; Deep learning; flood; Flood predictions; Flooding; Floods; hydrologic modeling; Hydrologic models; Hydrologic processes; Hydrology; interpretive deep learning; Learning; LSTM; machine learning; Networks; Rain; Rainfall; Rainfall forecasting; Runoff; Runoff models; Snowmelt; Spatial variability; Spatial variations
• ISSN: 0043-1397; 1944-7973
• DOI: 10.1029/2021WR030185

Long short‐term memory (LSTM) networks represent one of the most prevalent deep learning (DL) architectures in current hydrological modeling, but they remain black boxes from which process understanding can hardly be obtained. This study aims to demonstrate the potential of interpretive DL in gaining scientific insights using flood prediction across the contiguous United States (CONUS) as a case study. Two interpretation methods were adopted to decipher the machine‐captured patterns and inner workings of LSTM networks. The DL interpretation by the expected gradients method revealed three distinct input‐output relationships learned by LSTM‐based runoff models in 160 individual catchments. These relationships correspond to three flood‐inducing mechanisms—snowmelt, recent rainfall, and historical rainfall—that account for 10.1%, 60.9%, and 29.0% of the 20,908 flow peaks identified from the data set, respectively. Single flooding mechanisms dominate 70.7% of the investigated catchments (11.9% snowmelt‐dominated, 34.4% recent rainfall‐dominated, and 24.4% historical rainfall‐dominated mechanisms), and the remaining 29.3% have mixed mechanisms. The spatial variability in the dominant mechanisms reflects the catchments' geographic and climatic conditions. Moreover, the additive decomposition method unveils how the LSTM network behaves differently in retaining and discarding information when emulating different types of floods. Information from inputs within previous time steps can be partially stored in the memory of LSTM networks to predict snowmelt‐induced and historical rainfall‐induced floods, while for recent rainfall‐induced floods, only recent information is retained. Overall, this study provides a new perspective for understanding hydrological processes and extremes and demonstrates the prospect of artificial intelligence‐assisted scientific discovery in the future. Key Points The expected gradients method was used to retrieve flooding mechanisms learned by long short‐term memory networks in 160 catchments in the United States Snowmelt, recent rainfall, and historical rainfall induce 10.1%, 60.9%, and 29.0% of the 20,908 identified flow peaks, respectively The additive decomposition method unveiled models' distinct behaviors in retaining and discarding information for different flood types