A place where stuff falls out of my mind and onto the electronic writing space

Thought of the day: Weather Forecasts

Disclaimer: I forecasted the weather in the USAF for 16 years and this is how I learned the trade.

The weather forecast is not binary: there is no will rain or will not rain, nor is there sunny or cloudy. Snow forecasts are not 0” or 5” of snow. In fact, it’s not fair to say forecasting provides a range without clarifying what that means.

Forecasting the weather provides a probability of something happening. Cloudy at 2PM? There is a percentage of that occurring at your location. Nothing is 100% or 0% but some probability in between. Even when something seems certain to a forecaster, it’s never really 100% likely, but some measure of value close to 100%, like 98% for example.

We see this all the time when we see precipitation forecasts. A 70% chance of rain at your location for a certain time is just that – a probability that rain should be falling at that time. It can also be applied to the amounts of forecasts, as we often see with snow forecasts.

Most forecasts predict a range of precipitation. These predictions are more accurate than specifying a very specific amount because the odds of that very specific amount verifying with the forecast are actually low. If I predicted 4.5” of snow, then technically a 4.0” snowfall is wrong. In reality though, is that the forecasted amounts can be predicted on a bell curve where the most likely outcome is at the peak of the curve and other possibilities are on either side.

In my previous example, I believe we should get about 4.5” of snow. In the example bell curve here, imagine the green zone © is 4” to 5” of snow received, and the next zones (yellow, D and blue, B) encompass the range of 3” to 6” of snow. That forecast encompasses an 86” likelihood of being correct and giving a reasonable range that one can expect based upon the current data.

Weather forecasting will never be perfectly accurate. We do not sample the atmosphere densely or often enough to get a complete picture of our environment. Also, computing rarely catches the unique perturbations that can occur among those unseen elements.

So give your weather forecaster a break and realize it’s a game of probabilities. That 20% chance of rain sounds like a remote chance, but you might be the one place that manages hitting the 20% jackpot. ;–)