You Won’t See a 50% Chance of Rain November 8, 2011

 

Eric Bickel is a high school friend and is now leads the Graduate Program in Operations Research at the University of Texas.  The joke is we were probably the most unlikely pair from high school to get into science-related jobs.  One of the projects he did was compare the National Weather Service, The Weather Channel and a private forecasting company forecasts to each other.  The findings are interesting and can be found here:

http://journals.ametsoc.org/doi/abs/10.1175/2011MWR3525.1

The key to a good forecast is having sharpness and resolution.  Resolution, for example, is if you say there is a 40% chance of rain it will rain 40% of the time, or four out of every ten days you put 40%.  Sharpness is if you say it’s going to rain it will rain, or if you say it won’t rain, it doesn’t.  Sharpness is giving some certainty to the forecast.  In a normal year, without a drought, if I put a 30% chance of rain every day, by the end of the year my forecast will have good resolution because it will have rained on average once every three days.  However, I’ll have poor sharpness because I’m never really saying if it is going to rain or not.

One of the things that stuck out to me from Eric’s work was the Weather Channel avoids putting a 20% and 50% chance of precipitation.  The 50% omission is intentional.  The Weather Channel believes that users will interpret a 50% chance of rain as a lack of knowledge (after all there are only two possible outcomes), when, in fact, a forecast of 50% is more than twice the climatological average and thus a strong statement regarding the chance of precipitation.  Bickel shows in his paper how this policy degrades the quality of their forecasts.

To view past weather quiz answers click here:

Past Weather Quiz Answers

Advertisement

What Most People Think a 20% of Rain is January 18, 2011

 

Two answers and they are both wrong. It’s fascinating for me to see that most people in the United States view a weather forecast as deterministic, or if there is a rain percentage on the map, it definitely will rain. In other words, people interpreted 20% as it will rain 20% of the time or that 20% of the viewing area will get wet. The people conducting this study speculated that the general human tendency is to avoid complications of incorporating uncertainty in the decision process by ignoring it or turning it into certainty. Another example showed that people living in floodplains appeal to perceived cyclical patterns to predict the size of the next flood, rather than regarding it as uncertain.

The correct answer is if there is a 20% chance of rain today it may rain. In my experience, I’ve found that people in Southeast Texas view the chance as so small that it simply won’t rain. Most of the e-mails I receive from people upset about the forecast are on days with a 20% chance of rain and it does rain. The quote is usually, “You said it wouldn’t rain.”

Social science studies like this are important because it shows the correct message isn’t getting to the viewer. We as forecasters have to do a better job communicating the forecast. But the biggest problem is the forecast itself. The reason we use probabilities is because the future is uncertain. An 80% chance of rain means I feel strongly there will at least be a minimum amount of rain. As Frank Billingsley likes to say, “You only know what you know.”

To view past weather quiz answers click here:

Past Weather Quiz Answers

What Does a 30% Chance of Rain Mean? January 18, 2011

Hello Mr. Yanez.

I am currently a AP Stats student at Cinco Ranch High School, and right now we’re studying probability. Given that statement and the subject of this e-mail you might be able to guess where this is headed.

Exactly what do rain percentages tell us? If there is a 40% chance of rain, does that mean that 40% of the viewing area will receive rain, or that there is a 40% chance that any given location will get rained on?

I asked my teacher if he had any issues with weather men and their probability of rain, and he said that he would like to call one up and ask them to clarify. I’m sort of taking the liberty for him.

Likewise, if there is a 20% chance of rain in the morning and a 20% chance of rain in the afternoon, does that mean the chance of rain the entire day (morning AND afternoon) would be 4%? (.2 x .2 = .04) Or am I over-complicating things?

Thanks for taking the time to read this, and keep up the good work!

~Scott

Hi Scott,

Honestly I hate using percentages in the forecast. If this were an exact science we wouldn’t need to, we would simply say it will rain here and be dry here. Of course we are still limited in what we know about the weather and forecasting so we use percentages to show how likely it is to rain where you live.

This is how I was taught to use probability of precipitation. If I give a 30% chance of rain for Houston I am saying there is a 3 in 10 chance you will get wet. The 30% does not express how much rain will fall or how strong the storms will be if you get them. I have to do that separately.

The impression sometimes is a 20% means it will be light rain or not rain at all and this is simply not true. A 20% chance means it much more likely to be dry but definitely not a guarantee. If there are higher rain chances north or south of Houston I will always make a separate map showing the differences. The National Weather Service in Georgia expressed the math like this:

“The “Probability of Precipitation” (PoP) describes the chance of precipitation occurring at any point you select in the area.

PoP = C x A where “C” = the confidence that precipitation will occur somewhere in the forecast area, and where “A” = the percent of the area that will receive measurable precipitation, if it occurs at all.

So… if the forecaster knows precipitation is sure to occur ( confidence is 100% ), he/she is expressing how much of the area will receive measurable rain. ( PoP = “C” x “A” or “1” times “.4” which equals .4 or 40%.)

But, most of the time, the forecaster is expressing a combination of degree of confidence and areal coverage. If the forecaster is only 50% sure that precipitation will occur, and expects that, if it does occur, it will produce measurable rain over about 80 percent of the area, the PoP (chance of rain) is 40%. ( PoP = .5 x .8 which equals .4 or 40%. )

In either event, the correct way to interpret the forecast is: there is a 40 percent chance that rain will occur at any given point in the area.”

The last line is key. There is a four in ten chance you’ll get rain where you live.

I have heard other stations say 30% of the area, or there will be 30% coverage, of rain today. I think expressing the forecast like this is wrong. That forecast is basically stating 70% will be dry. What parts of the area will get rain and who will be dry? What happens if all of us get wet, doesn’t it make the forecast wrong? On a day when there is a 10% or 20% chance of rain and it doesn’t rain at all, it again makes the forecast wrong. Philosophically this explanation doesn’t work for me.

To answer your other question a 20% chance of rain in the morning and a 20% chance of rain at noon does not equal a 40% that day. For example, I make an hour by hour forecast that shows:

9am 20%

Noon 20%

5pm 20%

The chance of rain for the day is 20%.

But if I do this:

9am 20%

Noon 20%

5pm 60%

The chance of rain is 60% for the day, what I am saying is the chances to see early rain is very small but the evening looks like the time it would rain.

By the way, Frank Billingsley doesn’t like using percentages either. In fact, a few years back he got rid of using percentages past day three on the forecast. After hundreds of e-mails from people protesting, he put the percentages back on. People love seeing them even though not everyone knows what they mean.