Dechow et al. (2026) Merging Regional Climate Models and Remote Sensing Observations to Better Estimate Mountain Snow Water Equivalent

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Water Resources Research
• Year: 2026
• Date: 2026-11-01
• Authors: Jack Dechow, Michael Durand, Melissa L. Wrzesien, Jessica D. Lundquist, Karl E. Rittger, Jeff Dozier, Tamlin M. Pavelsky, S. A. Margulis
• DOI: 10.1029/2025wr042242

Research Groups

Not specified in the abstract.

Short Summary

This paper introduces Blender, a variational data assimilation method that enhances modeled snow water equivalent (SWE) and snowfall time series by integrating remotely sensed snow cover fraction. Tested at 500 m resolution in the Tuolumne watershed, Blender significantly improved SWE spatial patterns and snow water storage estimates when applied to Regional Climate Model (RCM) and ERA5-Land priors, approaching the performance of a high-quality snow reanalysis.

Objective

• To develop and evaluate a variational data assimilation method, Blender, for improving large-scale, high-resolution snow water equivalent (SWE) and snowfall time series in mountainous regions by modifying snowpack mass and energy balance constrained by remotely sensed snow cover fraction.

Study Configuration

• Spatial Scale: Tuolumne watershed, Sierra Nevada, USA, at 500 m resolution.
• Temporal Scale: Water years 2013, 2016, and 2017 (representing low, average, and high snow accumulation).

Methodology and Data

• Models used: Blender (variational data assimilation method). Prior inputs from a Regional Climate Model (RCM) and ERA5-Land (ERA5L).
• Data sources: Remotely sensed snow cover fraction (for assimilation), 18 Airborne Snow Observatory (ASO) flights (for evaluation), Western United States Snow Reanalysis (WUS-SR) (for comparison).

Main Results

• Blender improved spatial SWE patterns for the RCM prior, reducing RMSEP from 17.2% to 11.5% and increasing R from 0.39 to 0.72.
• For the ERA5L prior, Blender improved spatial SWE patterns, reducing RMSEP from 13.9% to 12.9% and increasing R from 0.003 to 0.59.
• Blender's performance approached that of WUS-SR (RMSEP = 9.5%, R = 0.82) in terms of spatial SWE patterns.
• Snow water storage (SWS) estimates were improved for the RCM prior.
• For the ERA5L prior, SWS estimates were not improved, but the posterior SWS remained reasonably accurate with a median absolute error of 27%.

Contributions

• Introduction of Blender, a novel, flexible, and efficient variational data assimilation method for high-resolution, large-scale SWE estimation.
• Demonstrates significant improvement in modeled SWE spatial patterns and snow water storage when applied as a post-processing step to existing gridded snow datasets (RCM, ERA5L).
• Provides a pathway toward more accurate and computationally feasible SWE estimation in challenging mountainous environments.

Funding

Not specified in the abstract.

Citation

@article{Dechow2026Merging,
  author = {Dechow, Jack and Durand, Michael and Wrzesien, Melissa L. and Lundquist, Jessica D. and Rittger, Karl E. and Dozier, Jeff and Pavelsky, Tamlin M. and Margulis, S. A.},
  title = {Merging Regional Climate Models and Remote Sensing Observations to Better Estimate Mountain Snow Water Equivalent},
  journal = {Water Resources Research},
  year = {2026},
  doi = {10.1029/2025wr042242},
  url = {https://doi.org/10.1029/2025wr042242}
}