THE OUTLOOK
A seasonal forecast is a shift in the odds, not a description of the winter. Here is how big the shift actually is.
What a seasonal forecast is
AND WHAT IT IS NOTA weather forecast predicts a sequence: this storm, on Thursday, with that much rain. A seasonal forecast cannot do this and does not attempt to. The atmosphere's memory of its own initial state is exhausted after about two weeks — this is a hard limit and not an engineering problem — so nothing about the individual storms of February is knowable in October.
What can sometimes be known is a shift in the distribution. Slow boundary conditions — above all the temperature of the tropical Pacific — bias the odds of the whole season toward wet or dry without saying anything about its parts. A seasonal outlook is a statement about a probability, and it is right or wrong only in aggregate, over many seasons. A single wet winter after a dry outlook does not falsify it, in the same way that a coin coming up heads does not falsify a claim that it was weighted toward tails.
Because the skill lives in the odds rather than in the outcome, the only rational way to use a seasonal forecast is to change a decision whose cost is asymmetric — how much water to carry over, when to schedule the fuel-reduction burn, how much to hedge. It is not a way to know what the winter will be like, and treating it as one is how seasonal forecasting acquired its reputation.
How wide is the shift, really
THE FULL SPREAD INSIDE EACH ENSO CATEGORYThe ENSO page gives the mean outcome in each phase. A mean is exactly the wrong summary for a decision, because it hides the thing you need — how badly the forecast can be wrong. This shows the whole range instead: every water year in each category, the full span, and where the middle sits.
—
Reading an official outlook
WHAT THE PERCENTAGES MEANThe Climate Prediction Center issues seasonal outlooks as probabilities across three equally likely categories — the driest third, middle third and wettest third of the 1991–2020 record. Three things about that framing are routinely misread.
"40% chance of below normal" is not a 40% chance of drought
It is a 40% chance of landing in the driest third — against the 33% a coin-flip climatology would give. The forecast has moved the odds by seven points. That is a real, useful signal and it is not a prediction of a dry winter.
"Equal chances" is not "we expect normal"
EC means the forecasters have no skilful basis for tilting the odds at all, so all three thirds stay at 33%. It is a statement about the absence of a signal, not a forecast of an unremarkable season — and a near-normal season is in fact one of the less common outcomes in California.
The categories are terciles, not departures
"Below normal" means the lowest third of the recent record, not below the mean. In a distribution as skewed as California's precipitation — where a few very wet years drag the mean above the median — a year can be below the average and still sit in the middle tercile.
Where to get an actual forecast
THIS SITE DOES NOT ISSUE ONEThis page explains the mechanism and quantifies the historical skill. It does not contain a forecast for the coming season and never will. Issuing one would require an operational model pipeline, a verification record and a duty of care this site does not have — and the organisation that does have all three publishes for free.
For the official outlooks: the CPC seasonal outlooks (monthly and seasonal, temperature and precipitation), the ENSO Diagnostic Discussion (issued the second Thursday of each month, and the single best plain-language read on the state of the Pacific), and the NIDIS drought outlooks. For California water supply specifically, CDEC and the California Nevada River Forecast Center.