California's biggest reservoir
isn't a lake. It's snow.

Each winter the Sierra Nevada banks the state's water as snowpack, then releases it slowly through spring and summer — exactly when California runs dry. Climate change is breaking that timer.

↓ scroll to explore
Scene 1 · The Cause

It’s not how much falls.
It’s what falls as snow.

The Sierra’s cold season sits close to freezing, so a few degrees of warming flips a large share of winter precipitation from snow to rain — and warming bites hardest at the lower, milder elevations near the snow line.

CMIP6 prsn/pr · 2 GFDL models (GFDL-CM4, GFDL-ESM4) · cold season Nov–Mar. Regional Sierra-box mean (includes lower, warmer terrain), so this reads the phase shift, not high-country totals.
What winter precipitation falls as
Sierra Nevada · cold season (Nov–Mar) · snow vs rain
Snow Rain
Scene 2 · Historical Record

It’s already happening:
the snowpack is shrinking

April 1st snowpack — the benchmark water managers watch for “peak storage” — has trended downward across the Sierra since the mid-20th century. In these climate-model simulations the decline is already underway through the historical period, not just a future risk.

CMIP6 multi-model mean · historical simulation 1950–2014 + moderate future (SSP2-4.5) 2015–2023. April 1 snowpack (snow water equivalent), shown as inches above or below the 1981–2010 average.
Sierra Nevada · April 1 Snowpack (SWE)
INCHES ABOVE OR BELOW THE 1981–2010 AVERAGE
1950 2023
Showing 1950 – 2023
Sierra Nevada snowmelt — when it arrives, and how much
Monthly snowmelt · Sierra Nevada · water year (Oct → Sep)
Historical snowmelt (1970–2000)
Projected snowmelt · SSP5-8.5 (2070–2100)
Water demand · CA agriculture (illustrative)
Hover for monthly values
Step 1 of 5 · The water tower
The West's frozen reservoir
The Sierra Nevada banks California's winter precipitation as snow, then releases it slowly as meltwater. In these CMIP6 simulations, that melt pulse is strongest in late winter — long before the state needs the water most.
~30% of California's water supply
begins as Sierra snowpack
Step 2 of 5 · The rhythm
Most of the melt arrives
in late winter
Historically (1970–2000), the ensemble-mean melt peaks in March, then tails off through spring and summer. On a water-year axis (October → September) that peak sits near the center of the plot — the natural shape we're about to watch change.
Step 3 of 5 · The shift
Warming pulls the peak earlier
Under a high-emissions future (SSP5-8.5, 2070–2100), the melt peak slides from March to February — about a month earlier. Both curves are scaled to their own peak here, so watch when the melt arrives, not how high it is. The real amounts come next.
Peak month: March → February
(~1 month earlier)
Step 4 of 5 · The volume
And there's far less of it
Timing isn't the whole story. Now switch to actual amounts and the warm-future melt pulse collapses: peak melt nearly halves, from about 1.19 to 0.61 mm/day. The melt also concentrates into the deepest winter months, leaving even less for spring.
Peak melt: ~1.19 → ~0.61 mm/day
Nov–Mar share: 82% → 91%
Apr–Jul share: 16% → 9%
Step 5 of 5 · The mismatch
Earlier and smaller —
still out of step with summer
California agriculture draws the most water in mid-to-late summer, months after the melt peak. A winter pulse that is now both earlier and smaller means even more of the year's water has to be captured and stored to bridge the gap to peak demand.
Melt peak: Feb–Mar
Demand peak: ~July
Stored water bridges the ~5-month gap
Zoom out · The Sierra water network

The mountains are the reservoir.

The Sierra Nevada stores winter precipitation as snow, then releases it downstream as temperatures rise. It is a frozen reservoir that feeds far more than one river.

Step 2 · One source, many branches

One snowpack, several river branches.

That single snowpack splits into three distinct downstream systems — west toward California's farms and coastal cities, east into the Nevada basins, and south down the Owens Valley.

Step 3 · Western slope

The western slope feeds farms and cities.

The Sacramento and San Joaquin rivers drain into the Central Valley's farms, then on toward the Bay Area and Southern California water systems.

Step 4 · Eastern slope

The eastern slope feeds western Nevada.

The Truckee, Carson, and Walker rivers supply Reno–Sparks and the Lahontan Valley, terminating in Pyramid Lake and Walker Lake.

Step 5 · Owens Valley

The Owens River connects the Sierra to Los Angeles.

The Owens River flows into the Los Angeles Aqueduct — an engineered connection carrying Sierra snowmelt hundreds of kilometers to Southern California.

Step 6 · Mid-century · SSP2-4.5

Future warming changes the melt signal.

Scaling the same network by the modeled Sierra snowmelt under a moderate future (SSP2-4.5), the upstream signal weakens — the rivers thin even though the downstream demand does not.

Step 7 · Mid-century · SSP5-8.5

Under higher emissions, the network remains but the signal weakens.

Under a high-emissions future (SSP5-8.5) the downstream dependency network is unchanged — the same farms, cities, and lakes — but the upstream snowmelt that feeds it is markedly thinner.

Zoom Out: The Sierra Water Network
River thickness shows how much snowmelt feeds each river ·
Scenario
Highlight
Month
Scene 5 · The Consequence

Less and less reaches summer — right when we need it most

By the end of the century, far less of the Sierra’s snowmelt reaches the summer, when California needs it — partly because it never falls as snow, and increasingly because it melts too early to use.

Apr–Jul snowmelt as a share of the 1970–2000 summer mean · CMIP6, 2 GFDL models (futures 2070–2100). SSP2-4.5 ≈ current-policies trajectory (~2.7 °C by 2100). “Mistimed” water still flows — it is recoverable with storage.
Where the summer water goes
Share of the historical summer supply · where it goes

We still have time
to change our future.

The Sierra's summer water isn't lost yet. Today's emissions path leaves about half of it by late century, and the worst case only a quarter — but with aggressive cuts we could still hold onto four-fifths. The difference is the emissions path we choose.

Scene 6 · The Choice

How much water
each future keeps

Each dot is a future for the Sierra's snowmelt, measured against the 1970–2000 normal. The worse the emissions path, the less survives — and the distance between them is the water still within our control.

Apr–Jul snowmelt as a share of the 1970–2000 summer mean · CMIP6 snm · futures 2070–2100. Current ≈ SSP2-4.5, high = SSP5-8.5 (2 GFDL models); low = SSP1-2.6 (GFDL-ESM4 only).
Snowmelt as a share of the 1970–2000 normal
Before you go · Make it yours

Where does your water come from?

Pick the place closest to you. The Sierra snowpack feeds all of these — and the shift you just watched reaches every one.

Same snow, wrong schedule

The snow still falls — but it now melts before the system can use or bank it. Reservoirs have to keep room for floods, so an earlier, faster pulse partly can't be stored; it runs downstream and is lost. The fix isn't more snow — it's managing a schedule that no longer matches when we need the water.

▶  Watch the video walkthrough
Data & methods

Every chart uses CMIP6 climate-model output from the Pangeo Google-Cloud archive, averaged over a Sierra Nevada box (lat 36–40°N, lon 121.5–118.5°W), area-weighted by latitude, and reduced to monthly climatologies.

  • Snow vs rain (Scene 1): snowfall (prsn) and total precipitation (pr), cold season Nov–Mar, 2 GFDL models.
  • Snowpack (Scene 2): April 1 snow water equivalent (snw), CMIP6 multi-model mean; historical simulation 1950–2014 + SSP2-4.5 2015–2023, in inches vs the 1981–2010 average. These are model simulations, not station observations.
  • Snowmelt (Scenes 3, 5 & 6): surface snowmelt (snm), 2 GFDL models, on a water year (Oct→Sep); historical 1970–2000 vs end-of-century 2070–2100. Scene 6 adds the low-emissions path (SSP1-2.6), available from a single model (GFDL-ESM4).
  • Network (Scene 4): a modeled snowmelt index (5 models, 2050–2075); river widths are approximate, not measured river discharge.

Plain terms: “snowmelt” = modeled meltwater (snm); “moderate future” = SSP2-4.5; “high-emissions future” = SSP5-8.5; “water year” = Oct 1–Sep 30, the standard hydrology calendar. Summer demand is illustrative, not from CMIP6. The “~30% of California's water” figure reflects commonly cited state estimates.