That’s the title of a recent book…

WeatherBrains is the weekly netcast that’s all about weather. Each week, the show features some of the best minds in weather. Recent guests have included legendary radar expert Les Lemon, BBC Weather Presenter Hannah Bayman, famed Oklahoma City television meteorologist Gary England, severe weather expert Chuck Doswell and the Long Ranger from Accu-Weather, the incomparable Joe Bastardi.

This Monday night, one of the top television meteorologists in the country, Dave Freeman from KSN in Wichita, will be the Guest WeatherBrain. Dave is a leader in leveraging social science to improve the impact of weather forecasting. He will talk about killing sacred cows in weather forecasting.

One of his sacred cows is probability of precipitation (PoP). Most people want to have them in their forecasts, but few people really understand them. Heck, even us weather people don’t really understand them. In fact, you can ask most weather people and they can’t give you the real definition.

We have been running a poll on the blog for the past few weeks about the meaning of PoPs. The question asked: “There is a 60% chance of rain tomorrow. Which option best describes what the forecast means?” The answers were: (A) It will rain tomorrow in 60% of the area; (B) It will rain tomorrow for 60% of the time; (C) It will rain on 60% of the days like tomorrow and (D) 60% of weather forecasters believe it will rain tomorrow. There were 1,849 votes through Friday night.

The most chosen answer was A (44%). Dave did an identical poll on his website and got the same result. Most people interpret a 60% PoP to say it will rain over 60% of the area. That is not the technically correct answer, although the official NWS definition does define PoPs as the areal coverage multiplied by the forecaster’s confidence. So, if the forecaster believes there is a 100% chance there will be measurable rain in the forecast area, but believes there is going to only be 20% coverage, it would result in a PoP of 20%. This is a subjective way for forecasters to arrive at a number. The PoP number in a forecast is indeed a subjective thing.

Answer B was chosen 15% of the time. That answer is incorrect. PoPs have nothing to do with duration.

Answer D was chosen 9% of the time. While that answer is not correct, there is one interpretation of probability in meteorology that does relate to this concept. Powerful computers allow us to execute multiple runs of numerical weather prediction models using slightly different initial conditions. This results in a range of output. We call it ensemble forecasts. The relative number of times that one result occurs can be used to approximate the probability that that result will happen.

Answer C is the correct answer. It is also the least correct sounding one, so it was chosen 24% of the time. That means that the incorrect answer was chosen 76% of the time. A 60% probability means there is a 60% chance that there will be measurable rain. The convoluted answer makes sense when you consider flipping a coin. There is a 50% probability it will come up heads, and a 50% probability that it will be tails. Flip it twice, and it could be 50/50 or 0/100. Flip it ten times and a myriad of results could occur. Flip it one million times and the results will trend to 50/50.

One other way of interpreting probability is that it is the result of a function as the number of samples gets sufficiently large. Of all the times that all of the atmospheric conditions were like tomorrow, it rained 60% of the time. If we had an unlimited number of samples, we would not need a computer program, we would know the answer. The problem is, we can’t accurately know the initial state of the atmosphere enough, nor is our understanding of the physics of the atmosphere good enough for us to accurately predict the weather 100%. All weather forecasts have an element of uncertainty in them.

How do you use PoPs? Share your thoughts on the subject…

Follow my daily weather history tweets on Twitter @wxhistorian.