SPOILER - Boring explanation of weather forecasting follows but I'm going to do it anyway.
>how it's reported/displayed has nothing to do with the Met Office source data.
rtj is correct.
The output from an ensemble forecast is a range of possible outcomes with the distribution representing the uncertainty in the forecast. If atmospheric conditions are chaotic then the range of outcomes and therefore the uncertainty can be quite wide. Most of the outcomes though will tend to group around a particular value which is statistically the most probable outcome.
Forecasts for public consumption invariably present the most probable outcome. They rarely show you the whole range of possible outcomes.
For example, the BBC forecast will often say something like "overcast with occasional showers", they do not say "cloud cover between 6 and 8 okta in legacyland, precip is most likely to be around 3mm/6hr but there's a 25% chance it could be nothing or 6mm, and a 10% chance of 9mm".
Would it be better if Joe Public was given the whole range? Do they even care beyond "Is it going to rain this weekend?" I don't know.
>but it assume the model is still right.
The Met Office, like most forecast centers, run multiple models and generally two (or more) versions of each model. One version is the official "production" model, the other versions are run in parallel and contain slight tweaks to see if accuracy is improved. Only after a long test period will a tweaked version showing an improvement become the production model and the cycle start again.
Last edited by: Kevin on Sun 2 Sep 12 at 21:07
|