Thursday, June 12, 2008

Probability of precipitation

I'm finally glad to have this explained to me by someone from the Capital Weather Gang. This was part of a longer post in response to a Freakonomics attack on weather forecasting. The actual definition of POP is on the Freaknomics blog entry comment.

We do not take blind stabs at temp and rain possibilities. We do not regurgitate forecasts that are made by other entities (at least most broadcast outlets create their own from scratch). We independently create our own forecasts, based on data and our experience, making the best judgement call possible given the situation. And whoever said “We have no idea what is going to happen beyond three days out,” is completely wrong. The purpose of a seven day forecast is not so much its dead-set accuracy as it is to give viewers the ability to see trends. Are days 6 and 7 going to feature warmer conditions? Does it look unsettled? Things such as that. So let’s say we are experiencing a week of 80s in spring, and I predict a day 7 cooldown. I forecast 58 on day seven. It verifies as 52 degrees. I am off by six degrees, but I still told you seven days ago that a significant cooldown was expected in a week. Am I really wrong?

Another thing you do is grade the consistency of the forecasts, and you comment on how it changes from day to day. My question is, does that really matter? If I forecast 46 for a high on day 7 and the next day, it looks maybe a bit warmer, should I keep that 46 or bump it up to 49? You want the best, most accurate forecast, so unless you have significant reservations, you’re going to increase the temperature forecast. Weather is inconsistent. That’s why it’s so fascinating. So in my personal opinion, while you don’t want your meteorologist making 10 degree temperature adjustments every day, you do want to present the best forecast you can, and if that means making changes, then so be it.

We do not forecasts POPs where I am for this very reason. You’re better off verbalizing to a viewer that “there will be a few hit or miss showers around.” Some meteorologists prefer POPs. Others hate them. It all depends on your point of view. That said, although I hate them, I do not entirely believe that your grading method was fair. If a meteorologist says there is a 40% chance of rain and it does not rain at a specific point as you chose (KC Intl), does that really mean they failed? KCI is located several miles NNW of the city. So what you are saying is that if the met. went with a 40% chance, and it did not rain at KCI, but downtown KC received a thunderstorm, their forecast was incorrect. That is not fair to the forecaster or to people who will look at this and devise notions of their television meteorologist.

No comments: