Deep learning rainfall–runoff predictions of extreme events

Details

• Journal Title:
  Hydrology and Earth System Sciences
• Personal Author:
  Frame, Jonathan M. ; Kratzert, Frederik ; Klotz, Daniel ; Gauch, Martin ; Shalev, Guy ; Gilon, Oren ; Qualls, Logan M. ; Gupta, Hoshin V. ; Nearing, Grey S.
• NOAA Program & Office:
  NWS (National Weather Service) ; NWC (National Water Center)
• Description:
  The most accurate rainfall–runoff predictions are currently based on deep learning. There is a concern among hydrologists that the predictive accuracy of data-driven models based on deep learning may not be reliable in extrapolation or for predicting extreme events. This study tests that hypothesis using long short-term memory (LSTM) networks and an LSTM variant that is architecturally constrained to conserve mass. The LSTM network (and the mass-conserving LSTM variant) remained relatively accurate in predicting extreme (high-return-period) events compared with both a conceptual model (the Sacramento Model) and a process-based model (the US National Water Model), even when extreme events were not included in the training period. Adding mass balance constraints to the data-driven model (LSTM) reduced model skill during extreme events.
• Keywords:
  Artificial Intelligence Extreme Events Machine Learning Rainfall Runoff
• Source:
  Hydrol. Earth Syst. Sci., 26, 3377–3392
• DOI:
  https://doi.org/10.5194/hess-26-3377-2022
• Document Type:
  Journal Article
• Funding:
  Grant no. NA19NES4320002
• Rights Information:
  CC BY
• Compliance:
  Submitted
• Main Document Checksum:
  urn:sha256:650d4697ae8fad3320c3cec392f6aa799ab74a0504af22aff4c579bc02aaf7ab
• Download URL:
  https://repository.library.noaa.gov/view/noaa/46523/noaa_46523_DS1.pdf
• File Type:
  [PDF - 1.57 MB ]