Published June 4, 2026 | Version v2.1

Global Machine Learning Prior Files for SWOT Discharge Estimation

Authors/Creators

  • 1. ROR icon University of Massachusetts Amherst

Description

Prior values for SWOT discharge estimation were derived from the ensemble mean discharge predicted by two machine learning models, δHBV2 and GRADES-hydroDL (see References for model descriptions).

 

The following prior variables were computed and replaced:

  • max_q
  • min_q
  • mean_q
  • monthly_q (Jan, Feb, Mar, Apr, May, Jun, Jul, Aug, Sep, Oct, Nov, Dec)
  • two_year_return_q
  • flow_duration_q (0.01, 0.06, 0.11, 0.16, ..., 0.91, 0.96)
  • logQ_sd

 

Name ML Discharge Peiord SWORD
MLprior_v1.0_SWORDv16 03/31/2023 - 02/28/2025 v16
MLprior_v2.0_SWORDv17 03/31/2023 - 02/28/2025 v17
MLprior_v2.1_SWORDv17 03/31/2023 - 12/31/2025 v17

NOTE: "ML Discharge Period" indicates the period of machine learning predicted discharge used to compute the prior variables

 

Files

Files (520.8 MB)

Name Size
md5:562f03c64f0d317dccbc82a0b74411c7
520.8 MB Download

Additional details

Related works

References
Publication: 10.1038/s41467-025-64367-1 (DOI)
Publication: 10.1029/2024WR039764 (DOI)