# Room for Improvement

I was looking at some forecast statistics the other day and the numbers just didn’t add up. But then again I might be looking at them wrong. Then again, the average person probably doesn’t sit down and ponder exactly what a 30% chance of rain means. They look at it as purely “not very likely”. That’s not the ideal interpretation but I seriously doubt many people — let alone many forecasters — work out the real math.

There have been many, many days since the probability of precipitation (PoP) was dreamed up when it was pouring rain and the PoP was some number other than 100%, even when it was pouring rain over a wide area which included a rain gauge.  Perhaps the forecaster is thinking that somehow it is magically not raining and it is all a figment of the imagination.

On the other hand, I knew a forecaster who viewed the chance of rain as black and white, 0% or 100% and no in between. There’s actually some merit to this especially if the forecaster’s goal is to be “right” not just “close”.

One thing we do (at OrrWeather.com/ChrisOrr.org)  is to include the forecaster’s confidence as part of the forecast. For example, we’ll forecast scattered showers producing 0.10″ to 0.25″ between 3pm and 8pm with a 90% confidence level. And, yes, it is possible for a forecaster’s confidence level to be down in the 20% range. At that point you’re basically on your own.

An alternative is to include which way the forecaster thinks an error would be – the skew. Thus a forecaster could say that he/she is 90% confident that there is a 30% chance of rain, but if he/she is wrong the odds are that it will be a lower probability of rain.

Personally, I would like to see meteorology get to the point where the forecast is always right. I’ll tell you why that may be a long way off in a future post.