Topic: Probability of Precipitation

[kcidwx]

The most common question I get asked is how do forecasters determine the probability of precipitation in a forecast. Most people are quite surprised when I explain it to them.

There are two values used in calculating the probability of precipitation. The first is the forecasters confidence that any precipitation will occur at all. The second is how much of the area covered by the forecast will get precipitation if it does occur. 

So if I issue a forecast and I'm only 20% confident that precipitation will occur and if it does 50% of the forecast area will get precipitation, then my probability of precipitation looks like this.

.2 x .5 = .1

So I would use a 10% chance of precipitation in the forecast.

Now let's look at a couple of ways a forecaster can arrive at a 10% chance of precipitation.

Confidence		Coverage		POP
20%		50%		10%
50%		20%		10%
10%		100%		10%
100%		10%		10%

[Jáchym]

Interesting, I never knew that.

[CW2274]

The most common question I get asked is how do forecasters determine the probability of precipitation in a forecast. Most people are quite surprised when I explain it to them.

There are two values used in calculating the probability of precipitation. The first is the forecasters confidence that any precipitation will occur at all. The second is how much of the area covered by the forecast will get precipitation if it does occur. 

So if I issue a forecast and I'm only 20% confident that precipitation will occur and if it does 50% of the forecast area will get precipitation, then my probability of precipitation looks like this.

.2 x .5 = .1

So I would use a 10% chance of precipitation in the forecast.

Now let's look at a couple of ways a forecaster can arrive at a 10% chance of precipitation.

Confidence		Coverage		POP
20%		50%		10%
50%		20%		10%
10%		100%		10%
100%		10%		10%

I've never heard it explained that way. The way I read was that the forecasted conditions that are expected to occur at the official place of measurement, usually the airport, and that over the past 30 years measurable precip. occured 10 of those years, POPs was placed at 30%.

[Old Tele man]

Lately, the NWS-TUC has been also providing side-bar "Weather Story" matrices showing *how* they arrived at their weather severity/probability forecasts, using a matrix similar to this example:



The *key* here is exactly as kcidwx mentions above: final result is the product of two values, A = probability and C = confidence, severity, impact, etc..  [ NOTE: often called a Risk Matrix ]

Unfortunately, what they DON'T tell us is what the numerical values are for the A and C groupings.

[CW2274]

Lately, the NWS-TUC has been also providing "side-bar" matrices showing *how* they arrived at their weather severity/probability forecasts, using a matrix similar to this example:



The *key* here is exactly as kcidwx mentions above: final result is the product of two values, A = probability and B = severity, dispersion, etc..

I get the concept, but this chart is not for POPS, nor do I know that there is one. Not saying there isn't, just that I've never seen one.

[Old Tele man]

...agreed, that EXAMPLE has absolutely nothing to do with WX, it's sole purpose was illustration of the 'matrix' that NWS-TUC has displayed recently during forecasts of wind/dust, rain/hail, severity, etc. probabilities during TS- & TD-related weather events heading our (Tucson, AZ) way.

[CW2274]

I just read the NWS definition of POPS and it's the same as the OP's. The way I see it is the NWS official version is more of a broad stroke, whereas mine is site specific.

[Old Tele man]

...quoted from the UK MetOffice:

"A single probability forecast is never right or wrong. We can only measure how good our probability forecasts are by looking at a large number of them over time. We do this by grouping together, for example, all of the 10% probability forecasts and checking that those weather events actually took place on one-in-10 occasions as predicted." [ emphasis is mine, not UK MetOff ]

[CW2274]

...quoted from the UK MetOffice:

"A single probability forecast is never right or wrong. We can only measure how good our probability forecasts are by looking at a large number of them over time. We do this by grouping together, for example, all of the 10% probability forecasts and checking that those weather events actually took place on one-in-10 occasions as predicted." [ emphasis is mine, not UK MetOff ]

I think that's kinda what I'm saying.

[Old Tele man]

...which, overtly implies, that FORECASTS are actually 'more' based upon HISTORICAL data (what happened in the past under similar conditions) than on PREDICTIVE data (projection from CURRENT data).

[CW2274]

...which, overtly implies, that FORECASTS are actually 'more' based upon HISTORICAL data (what happened in the past under similar conditions) than on PREDICTIVE data (projection from CURRENT data).

Now, that I know is what I'm saying.

[Old Tele man]

GREAT minds think alike...thus, NWS is actually PAST-casting rather than FORE-casting... (ha,ha)

[CW2274]

GREAT minds think alike...thus, NWS is actually PAST-casting rather than FORE-casting... (ha,ha)

[Jáchym]

LOL...

Isnt that what all national weather services do?

In our here in Czech we either pastcast or when we need a forecast we always use this... highly recommended and probably highest accuracy 

[Old Tele man]

WHERE can I get a set of those?

[CW2274]

LOL...

Isnt that what all national weather services do?

In our here in Czech we either pastcast or when we need a forecast we always use this... highly recommended and probably highest accuracy 

Oh jeez, I love those!

[Jáchym]

http://www.ebay.com/itm/Weather-Dice-Pack-of-6-Dice-Blue-18mm-for-Games-Six-Sided-Educational-D111-/221896825267?hash=item33aa14d9b3

http://www.ebay.com/itm/Elfball-Weather-Tokens-Dice-Set-MINT-/361399794321?hash=item54251b3a91

[Jáchym]

OK... now I admit... when it gets really serious or difficult we also use this

[CW2274]

OK... now I admit... when it gets really serious or difficult we also use this

I'll stick with the dice...

[Jáchym]

Sure, that will do for the normal forecast, as I said, we only take out the crystal ball when issuing severe warnings

[kcidwx]

I have to get a set of those dice for work. Perfect!

I just go with turning darker this evening and becoming lighter in the morning. Deal with a lot less flack that way. 

[CW2274]

I have to get a set of those dice for work. Perfect!

I just go with turning darker this evening and becoming lighter in the morning. Deal with a lot less flack that way.

Beautiful! George Carlin, AM/FM!

[Old Tele man]

In our Lil'Abner world, that "Crystal Ball" is black, has 3 holes drilled in it, and is manufactured by AMF (ha,ha)!