Colorado River winter outlook
What the coming winter looks like for the basin that feeds Lake Powell.
The official ENSO forecast
The U.S. National Oceanic and Atmospheric Administration's (NOAA) Climate Prediction Center (CPC) issued an El Niño Advisory on 8 October 2026. It reads: “El Niño continues to strengthen, with a strong-to-very strong El Niño likely through January-March 2027 (remaining greater than an 83% chance). El Niño strengthened further last month, with sea surface temperature anomalies exceeding +4.0°C in portions of the eastern equatorial Pacific [Fig. 1] . Most Niño indices increased in the past month, reaching +2.1°C in Niño-3.4, +3.0°C in” For winter 2026-27 specifically, the official CPC probabilities (issued October 2026):
Probabilities are CPC's, rounded to whole percents — a displayed 100% or 0% does not mean certainty or impossibility. Each three-month season gets its own probability; these are odds of ENSO phase, not of any particular weather outcome in the basin. CPC also publishes a strength distribution for the event; strength does not guarantee regional impacts. Source: ENSO Diagnostic Discussion and official probabilities.
HydroStatus spring runoff outlook (experimental)
First forecast Nov 1, 2026, then weekly through April 1. HydroStatus forecasts April–July 2027 inflow to Lake Powell as odds of below-, near- or above-normal runoff, updated on the 1st, 8th, 15th and 22nd of each month as snow builds in the basin. Every update says how well forecasts from that date have done in past winters: early-season forecasts carry little skill, and skill climbs through February and March.
How good is each week's forecast? (tested on 26 past winters)
| Issue date | Skill | Versus even odds | 80% range caught the actual flow |
|---|---|---|---|
| Nov 1 | Low skill | cuts the error by 6% | 20 of 26 winters |
| Nov 8 | Low skill | cuts the error by 6% | 19 of 26 winters |
| Nov 15 | Low skill | no better | 17 of 26 winters |
| Nov 22 | Low skill | no better | 15 of 26 winters |
| Dec 1 | Low skill | cuts the error by 4% | 19 of 26 winters |
| Dec 8 | Low skill | cuts the error by 8% | 17 of 26 winters |
| Dec 15 | Low skill | cuts the error by 11% | 17 of 26 winters |
| Dec 22 | Low skill | cuts the error by 16% | 18 of 26 winters |
| Jan 1 | Low skill | cuts the error by 14% | 18 of 26 winters |
| Jan 8 | Low skill | cuts the error by 17% | 20 of 26 winters |
| Jan 15 | Low skill | cuts the error by 26% | 18 of 26 winters |
| Jan 22 | Moderate skill | cuts the error by 33% | 19 of 26 winters |
| Feb 1 | Moderate skill | cuts the error by 37% | 20 of 26 winters |
| Feb 8 | Strong skill | cuts the error by 39% | 18 of 26 winters |
| Feb 15 | Moderate skill | cuts the error by 43% | 17 of 26 winters |
| Feb 22 | Moderate skill | cuts the error by 42% | 16 of 26 winters |
| Mar 1 | Strong skill | cuts the error by 43% | 17 of 26 winters |
| Mar 8 | Strong skill | cuts the error by 48% | 17 of 26 winters |
| Mar 15 | Moderate skill | cuts the error by 55% | 18 of 26 winters |
| Mar 22 | Strong skill | cuts the error by 54% | 19 of 26 winters |
| Apr 1 | Strong skill | cuts the error by 57% | 18 of 26 winters |
How we tested: for each past winter from 2001 on, the model was rebuilt using only earlier winters and asked to forecast that spring from the snowpack, precipitation and Lake Powell inflow on the same date. "Versus even odds" compares its ranked probability score with 33/33/33 odds. A date is rated strong if it passed every check we fixed before seeing any result: beating even odds, persistence and the drying trend, with bootstrap evidence that the edge is not luck, sound odds in each category, ranges that caught the flow often enough, and skill since 2000. Low means the edge could be luck. The model, tests and thresholds were frozen before the test was run (freeze hs-ll-freeze-2.0). An earlier version built on Pacific sea-surface temperatures failed the same test and was never published. This is not an official forecast: CBRFC's water-supply forecast, shown above once issued, is the agency product. Method and category thresholds: data & methodology.
CPC's DJF 2026-27 outlook, mapped
The official seasonal outlook for precipitation (left) and temperature (right), with the Upper Colorado basin boundary drawn on. Probabilities are the chance of landing in the wettest/warmest third of 1991–2020 winters — not "percent more precipitation." Areas with no shading are "equal chances": no tercile favored, probabilities near climatological odds.
- Above-normal favored (shading deepens with probability)
- Below-normal favored
- Equal chances — no tercile favored
Issued 10/15/2026 by CPC (October 2026); shown values are CPC's native favored-category probabilities. The seasonal temperature outlook is background odds — it does not forecast individual storms' snow levels or melt timing. Historical skill for these outlooks by season and lead: CPC's verification portal. HydroStatus has not computed basin-specific reliability. CPC's own reasoning, from the accompanying prognostic discussion, appears below.
How the outlook evolves by lead
| Valid season | Precipitation (basin region) | Temperature (basin region) |
|---|---|---|
| NDJ 2026-2027 | Above 33% · Above 40% · Above 50% · Below 33% · Below 40% · EC 33% | Above 33% · Above 40% · Above 50% · EC 33% · Near 33% |
| DJF 2026-2027 | Above 33% · Above 40% · Above 50% · Above 60% · Below 33% · Below 40% · EC 33% | Above 33% · Above 40% · Above 50% · EC 33% · Near 33% |
| JFM 2027 | Above 33% · Above 40% · Above 50% · Above 60% · Below 33% · Below 40% · EC 33% | Above 33% · Above 40% · Above 50% · EC 33% |
| FMA 2027 | Above 33% · Above 40% · Above 50% · Below 33% · Below 40% · EC 33% | Above 33% · Above 40% · Above 50% · Below 33% · Below 40% · EC 33% |
| MAM 2027 | Above 33% · Above 40% · Below 33% · Below 40% · EC 33% | Above 33% · Above 40% · Below 33% · Below 40% · EC 33% |
| AMJ 2027 | Above 33% · Above 40% · Below 33% · Below 40% · EC 33% | Above 33% · Above 40% · EC 33% |
Adjacent three-month seasons overlap by two months — these are not independent forecasts. Categories listed are those CPC draws anywhere in the wider basin region at that lead.
Snow and water
SNOTEL basin index: active SNTL stations with official huc starting '14' (frozen; spec §5.3); formula: 100 * sum(values) / sum(medians), stations reporting both; suppressed when sum(medians) < 1.0; roster of 136 stations frozen 2026-08-11. CBRFC's daily ESP model guidance (distinct from the official forecast) for April–July 2027 is not yet available. The two series are never merged here. CBRFC's own verification for this forecast point: official interface — HydroStatus has not independently assessed its calibration.
Drought context
| State | D2+ (severe) | D3+ (extreme) | D4 (exceptional) |
|---|---|---|---|
| UT | 51.1% | 13.5% | 0% |
| CO | 28.5% | 12.6% | 4.3% |
| WY | 52% | 18.6% | 0% |
| NM | 21.5% | 3.2% | 0% |
| AZ | 6% | 0% | 0% |
US Drought Monitor, map date 2026-10-27 (updates every Thursday), percent of state area. The Drought Monitor is a synthesized drought depiction for context — it is not a runoff-efficiency predictor and is separate from the modeled soil-moisture state CBRFC uses in its forecasts. Seasonal drought outlook: CPC.
Does El Niño actually predict basin winters? How winter ENSO relates to basin precipitation
Every fall, coverage of the basin leans on ENSO. The basin sits in what climatologists call the ENSO transition zone — in composites, the typical tendency is wetter winters to the south and drier to the north during El Niño, with the Upper Colorado in between (NOAA's ENSO composites; composites are tendencies, not guarantees, and measure no forecast skill). So we computed the relationship directly, under a pre-registered protocol: winter ENSO state (DJF RONI, the Relative Oceanic Niño Index and CPC's official ENSO index) against the same winter's October–March basin precipitation (NCEI's official Upper Colorado watershed average), 1952-2026.
The answer is two-sided. For precipitation itself, there is a real but modest tilt: r = 0.235 (95% CI 0.083 to 0.393) across 75 winters — higher-index (El Niño) winters have averaged somewhat wetter, with about 6% of year-to-year variance shared. A tilt in the odds is not a forecast: at this strength, dry El Niño winters and wet La Niña winters remain common, as the scatter shows. And this is a historical association, not forecast verification — the winter's own DJF index isn't knowable in the fall. Sensitivity to the index choice (the legacy Oceanic Niño Index instead of RONI): r = 0.217, CI 0.067 to 0.37 — the same picture.
By the time the water reaches Lake Powell, even that tilt disappears. Against April–July unregulated inflow (1964–2026, a target further shaped by spring weather, soil moisture, and warming), the association is r = 0.075 with a 95% interval of −0.120 to +0.254 — little evidence of a meaningful linear relationship. The famous years scatter accordingly: 1983's super El Niño preceded 98th-percentile inflow, 2016's near-median inflow, and the La Niña winter of 2023 delivered 84th-percentile inflow. El Niño nudges the basin's precipitation odds; it does not tell you what the river will do.
Methodology pinned in committed code (Pearson r; circular block bootstrap, 5-year blocks, 10,000 resamples, fixed seed; ±0.3 language threshold is an editorial convention equal to 9% shared variance). Inputs are hashed; the latest index values are preliminary and revisable. Full protocol and robustness notes: data & methodology.
Other signals, measured against the same yardstick
The Pacific–North American (PNA) pattern
In the literature, this pattern is the standout: in a July-initialized statistical framework it was the only climate index with consistent prediction skill for following-March western-US snowpack — with regional variation, and with limited seasonal persistence of the index itself (Kapnick et al., PNAS 2018; the same study found PDO had no prediction skill). We put it through the identical protocol as ENSO: winter (DJF-mean) PNA against the same Oct–Mar basin precipitation, 1952-2026. Result: inconclusive — r = -0.198, 95% CI -0.414 to 0.034 (n = 75). The expected sign is there (ridgier PNA-positive winters lean drier), but in this basin the measured relationship is weaker than ENSO's — Kapnick's "regional variation" caveat in action. Two constraints keep this off the front of the page:
- Winter PNA is not knowable in advance: unlike ENSO, the index has limited seasonal persistence, so its current value (-0.64 for 2026-09, CPC monthly index) carries no demonstrated signal for the coming winter. In winter it becomes a real-time storm-track indicator, worth watching via CPC's teleconnection page and its ~2-week outlooks.
- The July-initialized skill in the cited study belongs to that full statistical framework, not to reading the index — and it was demonstrated for snowpack across the western US, strongest outside this basin.
The rest of the field
- Snowpack itself: once winter starts, SWE vs median plus antecedent conditions dominate runoff outcomes — the basis of CBRFC's forecasts. That is why this page's lead panel switches to SNOTEL from December through May.
- Emerging research: a late-summer sea-surface index near New Zealand correlates with southwestern-US winter precipitation better and earlier than ENSO — for CA/NV/AZ/UT climate divisions, not the Colorado headwaters (Mamalakis et al. 2018).
Research findings and our own measured associations — not HydroStatus forecasts.
Why an average winter no longer means an average river
One model study estimated roughly 9.3% lower annual mean Colorado River discharge per 1 °C of warming, driven mainly by snow-albedo loss and increased evapotranspiration (Milly & Dunne, Science 2020). Whatever this winter brings, it lands on that baseline — see the Lake Powell page for USBR's two-year scenario projections.
CPC's own reasoning, and sources
“The ongoing El Nino event continues to strengthen, and an El Nino advisory remains in effect and is favored to continue through Spring 2027. The seasonal outlooks for the upcoming winter and spring are heavily influenced by this very strong event. The Temperature Outlook for Nov-Dec-Jan (NDJ) 2026-2027 favors above-normal seasonal mean temperatures for eastern Alaska, the northern tier and western third of the contiguous U.S. (CONUS), south Florida, and the Ohio Valley and mid-Atlantic regions.” — CPC, Prognostic Discussion for Long-Lead Seasonal Outlooks, 830 AM EDT Thu Oct 15 2026 (full discussion)
The discussion is updated with each monthly release. In winter, CPC's official extended-range products (6–10 day, 8–14 day, and Weeks 3–4 outlooks) cover the subseasonal gap; we exclude raw MJO/QBO index forecasting because its interior-West skill lacks consensus (Mundhenk et al. 2018).
Sources: CPC (ENSO products, seasonal outlooks, prognostic discussion — public domain); NCEI nClimGrid HUC2 watershed averages; NRCS AWDB (SNOTEL); CBRFC (water-supply forecasts); US Drought Monitor; USGS WBD (basin boundary). Per-product issue dates shown throughout; products retain their agencies' vintages and are never blended. NMME model forecasts are not shown (structured access exists but was not verified for this page). Computation rules: data & methodology.