Webb et al. (2026) Antecedent moisture enhances early warning of atmospheric river flood hazards

Identification

• Journal: Nature Communications
• Year: 2026
• Date: 2026-02-12
• Authors: M. J. Webb, Christine M. Albano, Deniz Bozkurt, Rene Garreaud, Anna M. Wilson, Guo Yu, Michael L. Anderson, F. Martin Ralph
• DOI: 10.1038/s41467-026-69286-3

Research Groups

• Division of Hydrologic Sciences, Desert Research Institute, Reno, Nevada, USA
• Department of Meteorology, University of Valparaíso, Valparaíso, Chile
• Center for Climate and Resilience Research, Santiago, Chile
• Center for Oceanographic Research COPAS COASTAL, University of Concepción, Concepción, Chile
• Department of Geophysics, University of Chile, Santiago, Chile
• Center for Western Weather and Water Extremes, Scripps Institution of Oceanography, University of California San Diego, La Jolla, California, USA
• California Department of Water Resources, Sacramento, California, USA

Short Summary

This study demonstrates that incorporating antecedent soil moisture conditions into the atmospheric river (AR) scale significantly improves its ability to predict flood hazards in California and central Chile, nearly doubling its correspondence with peak streamflow.

Objective

• To investigate how land-surface conditions, specifically antecedent moisture, influence atmospheric river (AR) flood response and to develop a modified AR scale that incorporates these conditions to improve early warning of flood hazards.

Study Configuration

• Spatial Scale: 142 catchments (97 in California, USA; 45 in central Chile)
• Temporal Scale: 70,585 atmospheric river landfalls from 1950 to 2023

Methodology and Data

• Models used:
  • Global AR Database (tARget version 4 AR detection algorithm)
  • U.S. West Coast AR scale (original and modified)
  • Generalized linear models (gamma distribution with log-link)
  • Variance decomposition (Lindeman, Merenda and Gold method)
  • Lyne-Hollick filter (for baseflow separation)
  • Standardized Precipitation Index (SPI), specifically Seasonal Standardized Precipitation Index (SSPI)
• Data sources:
  • Daily gauged streamflow observations (GAGES II database, CAMELS-CL, USGS National Water Information System, Chilean Dirección General de Aguas)
  • Daily gridded precipitation, snowfall, snow water equivalent (SWE), and soil moisture (0-289 cm) from ERA5-Land reanalysis
  • Gridded data characterizing AR conditions from Global AR Database (based on ERA5 reanalysis)
  • ERA5 hourly Integrated Vapor Transport (IVT) from Copernicus Climate Data Store
  • Soil parameters from FAO-UNESCO Digital Soil Map of the World (DSMW)

Main Results

• The original AR scale provides limited insight into peak streamflow response and flood hazard potential, with only 5% of landfalling ARs leading to floods.
• Median Spearman rank correlations between interpolated AR rank and peak streamflow are modest (ρ = 0.29 in both California and central Chile).
• Runoff efficiency, primarily governed by antecedent soil moisture (7-day mean total-column soil moisture), explains the majority of peak streamflow variability not captured by the original AR scale (median 41% in California, 37% in central Chile).
• A simple modification to the AR scale, incorporating a seasonally adjusted 90-day Standardized Precipitation Index (SSPI), significantly improves flood classification.
• The SSPI-modified scale increases the proportion of flood-generating ARs classified as hazardous (AR4-5) from 63% to 81% in California and from 47% to 64% in central Chile.
• The median separation of flood response by AR rank more than doubles, reaching 60% in California and 80% in central Chile.
• Median peak streamflow differences between adjacent AR ranks increase by 2.1-fold in California and 1.7-fold in central Chile under the modified scale.
• The SSPI-modified AR scale improves the correlation with peak streamflow to a median ρ = 0.52 in California and ρ = 0.51 in central Chile.
• The modified framework demonstrates transferability to other mid-latitude, AR-influenced hydroclimatic settings.

Contributions

• Demonstrates that pre-existing soil moisture conditions are the primary explanation for divergences between atmospheric river (AR) rank and flood response in mid-latitude regions.
• Develops a physically grounded and operationally compatible modification to the AR scale by incorporating catchment-scale antecedent moisture conditions (using SSPI).
• Significantly enhances the AR scale's ability to classify flood hazards, nearly doubling its correspondence with peak streamflow and increasing the classification of flood-generating ARs as hazardous by over 30%.
• Extends the evaluation of the AR scale and its modification to central Chile, demonstrating cross-hemispheric transferability.
• Provides a structured approach for integrating land-surface conditions into early-warning hazard classification tools beyond ARs and flooding.

Funding

• National Science Foundation (NSF) Graduate Research Fellowship Program (Grant No. 1937966)
• National Science Foundation (NSF) International Research Experience for Students (IRES) program (Grant No. 1954140)
• Desert Research Institute (DRI) Maki Student Award
• USACE Engineer Research and Development Center Urban Flood Demonstration Program (Agreement No. W912HZ1920011)

Citation

@article{Webb2026Antecedent,
  author = {Webb, M. J. and Albano, Christine M. and Bozkurt, Deniz and Garreaud, Rene and Wilson, Anna M. and Yu, Guo and Anderson, Michael L. and Ralph, F. Martin},
  title = {Antecedent moisture enhances early warning of atmospheric river flood hazards},
  journal = {Nature Communications},
  year = {2026},
  doi = {10.1038/s41467-026-69286-3},
  url = {https://doi.org/10.1038/s41467-026-69286-3}
}