Despite popular opinion (or jokes) to the contrary, weather forecasting does not consist of flipping a coin, rubbing a lucky rabbit’s foot, or visiting the psychic down the street. It’s actually a real science with a proven method for success!
A good place to start any weather forecast is with a little history. The meteorologist may check out what the atmosphere has been doing since the previous day. This helps give her a sense of continuity. For instance, if the low was 50 this morning and the air mass, winds and sky have not changed, it might be reasonable to expect a similar low tonight. This is a very simple example, of course.
An accurate picture of present conditions is a crucial next step in preparing a forecast. This includes as a bare minimum a surface map and several upper level charts that show moisture, temperatures, winds and pressure surfaces (high and lows) as one moves upward. The speed, intensity, direction, and tendency of weather systems will need to be known. If that low pressure over Colorado is strengthening, then perhaps the winds in Dallas will be stronger than yesterday. Did the hurricane east of Miami change course overnight? Questions like these must not be left unanswered.
A satellite movie, or “loop” as well as an inspection of the radar is very helpful as well. And don’t forget to look out the window.
Next, he’ll move to the model output (See the Helium article on “How models create forecasts” for specifics on this). The models provide a rudder with which to “steer” the forecast.
The forecaster will compare the various models and establish a trend as to which one is closest to the observed weather. If yesterday’s XYZ computer output did not predict today’s rain, then it may not be a good one to trust. If, on the other hand, the XYZ model performed well, and it is forecasting the same weather for Friday as it did 24 hours ago for Friday, there’s something likely worth using.
One cannot forget the science behind the maps and the computer forecast products. An experienced forecaster learns the strengths and weaknesses of each model over time, and can then adjust accordingly. A model output is never an acceptable substitute to an alert weather prognosticator, who may realize the maps and the models don’t mix. When this happens, choosing the maps (along with hand analysis, experience and intuition) usually salvages an otherwise hopeless mess of a forecast.
Finally, there’s the verification. If there is a “busted” forecast, a dedicated weather professional will want to know what went wrong. He will go back over the charts and the model data and try to pinpoint the mistake. This allows the forecaster to improve when a similar setup comes along again.
Unlike the machine, which can only be improved by humankind, a scientist can improve their work by continued study and correction. Naturally, if we speak of a television weather forecaster, he or she will no doubt have any number of people happy to tell them when, and why, they’re wrong.