California WaterBlog

By Kyle Greenspan

. . .

We are well into the 2026 water year in the western United States, and so far, it appears to be an unseasonably warm one. Media across California, the western states, and the nation are covering the impact of our warm winter on snowpack and water resources. This post attempts to put headlines and current events into a longer-term context, explaining long-term changes we expect in snowpack water storage.

In late January and early February, the LA Times reported that snowpack in the Sierra Nevada was about 59% of average for that time of year. The San Francisco Chronicle noted a particularly significant snow deficit in the Northern Sierra, and that the culprit was largely not lack of precipitation, but rather early onset of melt. And the New York Times commented that a snow drought has gripped not only California but also the rest of the West, likely foretelling diminished runoff later this year. More recently, record spring heat across California and the West has driven a “snow-eater heatwave”, melting out much of the already-meager snowpack (Rhoades et al. 2025).

While this post is geared towards researchers and other folks familiar with climate change and water issues, I also penned another post earlier this year that covers similar topics and is geared toward all audiences. Both posts draw from research that my colleagues and I published back in July in Geophysical Research Letters, titled Preparing for Uncertain Water Futures: An Analysis of Intrannual Snowpack Processes in the Southern Sierra Nevada. 

We offer the five most important takeaways from our analysis for researchers below. In the study, we applied established methods for assessing model skill across a range of intrannual metrics—the snow water equivalent (SWE) triangle (Rhoades et al. 2018)—to a relatively new downscaled climate model dataset—the Western United States Dynamically Downscaled Dataset (Rahimi et al. 2024). Our analysis informs users of these data in California’s Fifth Climate Change Assessment and other researchers about where further improvements are needed to improve usability of the data for natural resource managers. Here are a few key terms to navigate this post:

• Model skill—A measure of a model’s accuracy in capturing observed outcomes.
• SWE triangle—The snow water equivalent triangle depicts snowpack accumulation and melt over the course of a water year (see Figure 1 and 2).
• Downscaling—Downscaling is the process of scaling down climate models from global to regional scales.
• Hindcast—The process of generating data from climate models for historical years.
• WRF-GCMs—The individual general circulation models that were downscaled using the Weather Research and Forecasting (WRF) model, a regional climate model.
• WRF-ERA5—The ERA5 reanalysis dataset downscaled using the same WRF methods as the WRF-GCMs.
• Instrumental record—Snow water equivalent measurements taken at 34 California Cooperative Snow Survey monitoring sites in the southern Sierra Nevada over our historical period (1986-2005).

1. Despite advancements, melt processes remain difficult for downscaled climate models and coupled land surface models to capture.

To plan for and operate downstream water storage, managers need accurate projections of snowmelt timing and rate. We found that the WRF-GCM ensemble hindcasts a melt date—the day on which most snowpack has melted—that is 19 days later than our instrumental record actually shows. In contrast, the WRF-GCM ensemble hindcasts a snowpack peak date that is only two days later than the instrumental value.

WRF-GCMs likely overestimate melt date because of biases introduced by the land surface and regional climate models used in the downscaling process. Nearly all WRF-GCMs underestimate melt rate by a similar amount. This is likely driven in part by Noah-MP, the land surface model coupled to these WRF-GCMs that is used to model melt processes (Von Kaenel and Margulis 2024). WRF-ERA5 overestimates melt date by a similar amount as the WRF-GCM ensemble, which indicates that some of the melt date bias is also introduced by WRF, the regional climate model used for downscaling. See our paper for a more detailed explanation of biases.

2. WRF-GCMs overestimate the amount of water stored in snow at each year’s peak.

Snowpack may store less water in the coming decades than our projections indicate. We found that the WRF-GCM ensemble overestimates the amount of water stored in snow at each year’s peak by about 19% compared to the instrumental record. Outlier WRF-GCMs may drive some of this snowpack water storage bias. Over half of the WRF-GCMs included in our ensemble hindcast a peak snowpack water storage value that is close to the instrumental median. Meanwhile, a smaller subset of WRF-GCMs overestimate this metric.

Cold bias at high-elevations that persists despite correction applied for this bias and biases introduced by the WRF downscaling process likely also drive WRF-GCM overestimation of snowpack water storage. We determined that there are likely multiple sources of snowpack water storage bias in play because the WRF-ERA5 value falls between the WRF-GCM ensemble and instrumental values. See our paper for a more detailed explanation of biases.

3. California is entering a future with a smaller water supply and greater flood hazard.

We highlight three main factors that are driving reductions in water supply and increases in flood hazard. First, the amount of water stored in snow is decreasing overall. We found a snow water equivalent loss rate of about 88mm per degree C of regional warming in the southern Sierra Nevada. In some conditions, our reservoirs may be able to make up for this loss of storage. But in extreme wet years or other conditions when reservoirs are full, this loss in snowpack water storage may be particularly impactful. Second, we found that interannual variability in snowpack water storage is decreasing while interannual variability in precipitation is increasing. This indicates that snowpack will not buffer our increasing precipitation extremes. Third, our water management system—both public policy and physical infrastructure—is not yet well-equipped to handle these new hydroclimate extremes without the assistance it historically received from snowpack. 

4. Snowpack will store less water during the dry times of year in the coming decades. 

In addition to diminished snowpack water storage, we project earlier, shorter snowmelt seasons. This shift in seasonality is problematic because it contributes to dry fuel conditions and landscape vulnerability to disturbances such as severe wildfire, longer periods of low flow in rivers and streams (which harms aquatic ecosystems), and less water for human consumptive use in dry times of year when supplementation of reservoir storage is most critical. The snowpack peak date is projected to shift 11 days earlier, the melt date 14 days earlier, and melt season 7 days shorter—all by mid-century.

5. Further improvements to WRF-GCMs will help California adapt to these changes.

If you are a climate scientist, hydrologist, or researcher in another related field, we encourage you to consider the biases in melt processes and snowpack water storage that we discuss here in your use of these data. For the modelers out there, know that improving the skill of downscaled climate models and coupled land surface models in these areas continues to be an important contribution toward improving the utility of these data for water managers. By continuing to refine our projections of snowpack water storage, the research community can help California adapt to its uncertain water future.

About the Author

Kyle Greenspan is a research associate at the Public Policy Institute of California Water Policy Center, where he focuses on solutions for the state’s wildfire challenges but has also worked on topics including Sustainable Groundwater Management Act implementation and greenhouse gas balance modeling for alternative uses of lands that will likely be fallowed due to groundwater cutbacks. He was previously a post-baccalaureate researcher at Claremont McKenna College’s Kravis Department of Integrated Sciences, where he conducted the climate and snowpack research described in this post. He holds a bachelor’s degree in environmental science, economics, and politics from Pitzer College.  

Further Reading

Becker, Rachel. “Record Heat, Melting Snow: What Does It Mean for California’s Reservoirs?” CalMatters. March 20th, 2026.

Greenspan, Kyle, Branwen Williams, Heather Williams, Stefan Rahimi, Alex Hall, and Lei Huang. “Preparing for Uncertain Water Futures: An Analysis of Intrannual Snowpack Processes in the Southern Sierra Nevada Under Climate Change.” Geophysical Research Letters 52, no. 15 (2025): e2025GL115768. https://doi.org/10.1029/2025GL115768.

Greenspan, Kyle. “California’s Snowpack Is the State’s Biggest Reservoir–and It’s Declining.” Public Policy Institute of California. September 2nd, 2025.  

James, Ian. “Halfway through winter, heat is melting the California snowpack.” The Los Angeles Times. January 30, 2026.

Porter, Greg. “Why California’s snowpack is melting even after a wet start to winter.” The San Francisco Chronicle. February 3rd, 2026. 

Rahimi, Stefan, Lei Huang, Jesse Norris, et al. “An Overview of the Western United States Dynamically Downscaled Dataset (WUS-D3).” Geoscientific Model Development 17, no. 6 (2024): 2265–86. https://doi.org/10.5194/gmd-17-2265-2024.

Rhoades, Alan M., Andrew D. Jones, and Paul A. Ullrich. “Assessing Mountains as Natural Reservoirs With a Multimetric Framework.” Earth’s Future 6, no. 9 (2018): 1221–41. https://doi.org/10.1002/2017EF000789.

Rhoades, Alan M., Joshua North, William Rudisill, et al. “Snow-Eater Heatwaves of the Western United States.” Preprint, In Review, September 11, 2025. https://doi.org/10.21203/rs.3.rs-7576317/v1.

Robbins, Jim. “Snow Drought in the West Reaches Record Levels.” The New York Times. February 2nd, 2026. 

Rojanasakul, Mira, Dance, Scott, and Kitajima Mulkey, Sachi. “Across the West, Record Heat Is Colliding With a Snow Drought.” The New York Times. March 21st, 2026.

Von Kaenel, Manon, and Steven A. Margulis. “Evaluation of Noah-MP Snow Simulation across Site Conditions in the Western United States.” Journal of Hydrometeorology 25, no. 9 (2024): 1389–406. https://doi.org/10.1175/JHM-D-23-0211.1.