Ask a Western water manager what keeps them up at night and the answers are familiar: a thin snowpack, a dry winter, a reservoir that refuses to fill. Wildfire usually sits lower on the list, filed under water quality, with ash in the intake and sediment in the reservoir. Two studies published in 2026 suggest that filing is too narrow. Fire also changes how a watershed spends its precipitation. That alters both how much water reaches a river and how confidently anyone can predict it.

This article sets a January 2026 synthesis by U.S. Geological Survey scientists beside a July 2026 analysis by University of California researchers. It asks a pointed question: can a region face more severe wildfire and become harder to forecast even when its annual precipitation barely changes? The evidence points toward yes, with real limits on what has actually been demonstrated.

Two Studies, Two Vantage Points

The first is a synthesis published on January 15, 2026 in Environmental Research: Water by a USGS team led by Brian Ebel, with John Hammond, Michelle Walvoord, Trevor Partridge, David Rey and Sheila Murphy. It is a narrative review of how fire alters hydrologic processes and streamflow generation across the western United States. Its opening admission is blunt: “Scientific understanding is currently insufficient” to deliver regionally accurate estimates of what wildfire does to water quantity. The authors say the reason is that the direction and magnitude of post-fire shifts in streamflow generation vary from one region to another.

The second is a paper in Hydrology and Earth System Sciences, published July 15, 2026, by Ziying Han, Han Guo, Michael Goulden and Roger Bales of UC Berkeley, UC Merced and UC Irvine. It integrates four decades of satellite-derived evapotranspiration, PRISM gridded precipitation, full natural flow and fire-perimeter data for five northern California watersheds. Those are the Trinity, Upper Sacramento, McCloud, Pit and Feather, which drain into Shasta Lake and Oroville Reservoir. Full natural flow is the runoff a river would carry without upstream diversions or reservoir storage. The analysis covers 37 fires larger than 10,000 acres between 1985 and 2020, with flow records running through the 2023 water year.

Evapotranspiration: The Water Fire Stops Using, For a While

Evapotranspiration, the water that evaporates from soil and canopy or is released by plants, is one of the largest ways a mountain forest spends its precipitation. When a fire kills trees, that spending drops. In the UC analysis, high-severity fires suppressed evapotranspiration by 100 to 250 millimeters in the first post-fire year. Across all 37 fires, the annual reduction ranged from 40 to 440 millimeters. The 2007 Moonlight Fire in the Feather basin burned 263 square kilometers and cut first-year evapotranspiration by 325 millimeters, about 85 million cubic meters of water the forest would otherwise have used. An earlier study led by Qin Ma at UC Merced, published in the Journal of Hydrology in 2020, found a similar pattern across the Sierra Nevada and southern Cascades. Evapotranspiration fell by 265 millimeters, or 36 percent of pre-fire use, in the first year, and the average reduction over 15 years was still 23 percent.

The recovery clock is where the story gets interesting. Thirty-one of the 37 fires reached 75 percent of pre-fire evapotranspiration in an average of 3.9 years, but the range ran from one year to twelve. The drier Pit basin, hit by repeated high-severity fires, needed about ten years on average and had recovered only 69 percent of its pre-fire water use after five years. The Feather basin reached 78 percent in the same time. The 1992 Fountain Fire in the Pit accumulated roughly 960 million cubic meters of reduced water use before crossing the 75 percent threshold in year twelve.

Re-burning resets the clock. In the Feather basin, recovery was interrupted by large fires nine and 21 years after the first burn, each pushing evapotranspiration down by about 100 millimeters again. The authors used 75 percent recovery rather than full recovery as their benchmark because evapotranspiration rarely returns completely in semi-arid and repeatedly burned systems.

Why Lower Evapotranspiration Is Not a Windfall

Less water used by trees should mean more water left over, and the California data show that in principle. According to a UC Merced summary of the work, cumulative burned area reached about one-fifth of the total watershed by 2023. That raised potential runoff by roughly 740,000 acre-feet per year, or about 5.4 percent of the combined average full natural flow, with the effect growing most markedly since 2019. Lead author Ziying Han described the underlying arithmetic as “a balance between vegetation mortality and regrowth.” The surplus is not a permanent gain. It is a moving difference between how fast forests die and how fast they return.

Two cautions keep this from being a true windfall. First, potential runoff is precipitation minus evapotranspiration, not measured water delivered to a reservoir. Second, at the scale of a whole basin the signal is easily diluted. The paper found that annual precipitation remained the dominant control on outflow, and most fires burned too small a share of a basin to register at the stream gauge. The authors cite earlier studies indicating that clear runoff responses typically emerge only when roughly 20 to 30 percent of a watershed has burned.

Where fire was extensive, the effect could cushion drought. In one sub-basin, labeled ANT in the paper, a 2007 fire burned 47 percent of the area and cut evapotranspiration by 31 percent. Precipitation that year fell 26 percent, yet full natural flow dropped only 10 percent. The authors read this as a partial offset from suppressed evapotranspiration, but they note that confirming it would require subsurface storage data they did not have.

The Feather basin shows the opposite problem. In 2022, evapotranspiration fell 58 percent and 54 percent in areas overlapping the 2020 North Complex and 2021 Dixie fire footprints. A modest rise in precipitation that year complicated the interpretation of a rise in basin outflow. Separating fire from weather is exactly the difficulty the USGS review lists among its ongoing challenges. It sits alongside water withdrawals, post-fire land management, overlapping disturbances and a scarcity of pre-fire data.

There is a subtler implication here, and it is an inference rather than a finding of either study. Any drought cushion supplied by burned ground shrinks as vegetation recovers. The buffer is therefore largest right after a fire and weakest just as the forest returns to its full thirst, potentially in the next dry spell.

Infiltration and Timing: What Annual Totals Cannot See

The UC analysis works in annual water-year totals, which helps in closing a budget but hides when water arrives. The USGS review fills part of that gap. Its conceptual model lists ten factors that govern how fire changes streamflow generation. They include precipitation seasonality, the synchrony of precipitation and potential evapotranspiration, the overlap between rainfall rates and soil infiltration, fire extent and severity, whether the burn scar sits in the headwaters or near the outlet, and the connectivity between soil and groundwater.

The synchrony factor is the least intuitive and possibly the most consequential for the West. The review summarizes earlier work showing that when annual precipitation or snowmelt lines up closely with peak potential evapotranspiration, streamflow has tended to increase. Strong asynchrony has tended to decrease it. For groundwater and baseflow, the relationship runs the other way: asynchrony consistently produces post-fire rises in groundwater levels and baseflow, while synchrony gives mixed results. In a Mediterranean climate like northern California's, where precipitation falls mostly between November and March and summers are dry, the same fire could send very different signals depending on when storms and snowmelt arrive.

Soil is the other half of the timing story. The UC authors note that canopy loss reduces interception and transpiration, while soil combustion and water repellency can reduce infiltration and push water toward surface runoff. They consider that mechanism more relevant in coarse-textured or shallow volcanic soils such as those in the Pit basin. That is the same basin with groundwater-fed baseflow and the slowest evapotranspiration recovery.

They also list accelerated snowmelt among the likely mechanisms behind fire-year budget shifts, though an annual analysis cannot isolate it. Earlier work in Southern California, cited in the paper, reported increased dry-season water yield in burned watersheds. Changes in timing can matter as much as changes in totals.

Groundwater: The Connection Fire Rewires

Groundwater is where the evidence is least settled. A USGS-led review, Disparate Groundwater Responses to Wildfire (WIREs Water, 2025), by Michelle Walvoord, Brian Ebel and colleagues found that post-fire studies report outcomes ranging from substantial increases to notable decreases in recharge and baseflow. Some find negligible or short-lived effects. The intermediate window of two to ten years after a fire shows especially high variability, which the authors link to five key factors.

A related study by David Rey, Martin Briggs, Michelle Walvoord and Brian Ebel used paired air and stream temperature signals. It detected changes suggestive of more shallow groundwater input mainly in streams that lacked a substantial pre-fire connection to deep groundwater.

Notice the overlap with the UC findings. The two-to-ten-year period when groundwater responses are most variable roughly matches the period during which evapotranspiration recovery unfolds. The UC authors acknowledge that quantifying compensating effects requires subsurface storage information they lacked. In other words, the moment a burned basin is hardest to characterize below ground is also when it is changing fastest above ground.

When the Water Budget Stops Closing

Over a full year, precipitation minus evapotranspiration should roughly equal the water leaving as river flow, once storage changes are small. The UC team tested that identity by comparing 13 years in which more than 3 percent of a watershed burned against 54 years without substantial fire. High-fire years showed systematically more negative residuals, meaning observed flow exceeded what precipitation minus evapotranspiration predicted by a larger margin (Welch's t-test, p = 0.017).

The authors are careful to say this does not point to a single mechanism. The residual mixes real storage changes with measurement problems. In one basin, raising gridded precipitation by only 3 percent closed the gap, consistent with known underestimation of mountain precipitation. In others, stream-gauge bias or uncertainty in reconstructing unimpaired flow was implicated.

The paper concludes that wildfires act as “short-term hydrologic shocks” that alter how precipitation is partitioned and distort modeled water budgets in fire-prone headwaters. It argues that steady-state models cannot represent these conditions. It calls for flexible evapotranspiration recovery functions, representations of vegetation change and nonlinear thresholds linking burn extent to runoff.

Can a Region Get Riskier and Less Predictable Without a Rainfall Shift?

Put the two studies together and the answer is a qualified yes. Neither paper claims that fire overrides climate; the UC data show precipitation still dominates basin outflow. What fire adds is a hidden state variable: the burn-and-recovery status of each watershed, which steady-state water budgets ignore. Four findings suggest why that variable erodes predictability even at unchanged precipitation:

• The recovery clock is not constant. Time to reach 75 percent of pre-fire evapotranspiration ran from one to twelve years, depended on moisture and severity, and restarts when areas re-burn.
• The sign of the response varies. The USGS review stresses regional variability in the direction and magnitude of streamflow shifts, and groundwater responses range from increases to decreases.
• Effects are thresholded and diluted. Small burn fractions may leave no detectable signal at the gauge, while large ones, roughly 20 to 30 percent of a basin or more, can.
• Budgets stop closing when fire is present. Fire years widen the gap between expected and observed outflow, and disturbance is entangled with measurement error.

What has not been shown matters too. Neither study directly measures whether reservoir-inflow forecasts have become less skillful, and the USGS authors flag the lack of pre-fire data and overlapping disturbances as obstacles to attribution. The unpredictability described here is a well-supported hypothesis built on measured components, not yet a measured decline in forecast skill. It is also not uniformly negative, since a burned watershed may temporarily pass more water downstream in a dry year.

Planning for Watersheds That Are Still Recovering

The UC authors argue that fire history, burn severity and evapotranspiration recovery patterns should be built into reservoir operations and water-allocation planning. Doing so would avoid overestimating water availability or underestimating infrastructure vulnerability. That matters in a state where the Central Valley Project and State Water Project depend on consistent outflow from snow-fed mountain basins, many now affected by high-severity fire.

The USGS review points to tools that could sharpen those estimates. They include gridded and remotely sensed precipitation and fire-effects data, geophysical, isotopic tracer and geochemical signatures to diagnose hydrologic change, advances in physically based and data-driven models, and comparisons of streamflow-generation recovery across diverse watersheds.

There is also a management lever upstream. Han Guo, a co-author at UC Merced, noted that mechanical thinning and prescribed fire can deliver comparable water benefits while also improving forest health. Roger Bales emphasized that reducing projected wildfire severity through fuels treatment carries wide co-benefits, from public health to erosion control. Gregg Verutes of the nonprofit Blue Forest framed the findings as making fuels treatment a water security investment, not only a risk-reduction measure. Whether or not one accepts that framing, the practical point stands. Western reservoirs are filled not just by rain and snow but by the condition of the forests that receive them, and that condition is one current models are only beginning to capture.