Question for weather geeks

[mattlistener]

The forecast from 11:25am this morning says southern New England will be hit by a strong, fast-moving storm that will develop off the coast of New Jersey this evening. It could bring gusts to 40mph and an inch of rain in 4 hours, causing localized flooding.

Can anyone tell me what makes that prediction possible? I find it kind of amazing that the behavior of a storm can be predicted before it's even formed.


Statement as of 11:25 am EDT on October 15, 2004

... Heavy rain and strong winds expected this evening...

A quick hitting but rather intense storm will affect southern New
England with a period of heavy rain and gusty winds this evening.

As much as an inch and a half of rain may fall within four hours.
This may cause localized street flooding... especially in areas
where fallen leaves have clogged drains. Sharp rises in smaller
rivers and streams are likely... but all rivers and streams are
expected to remain within their banks.

Some of the heavier showers and possible thunderstorms may produce
wind gusts to 40 mph... especially across Rhode Island and eastern
Massachusetts. This could be strong enough to bring down tree limbs
in a few locations... and cause isolated power outages.

The heaviest rain and strongest winds are expected to occur between
10 PM and 2 am... with improving conditions thereafter. The culprit
for the stormy weather will be low pressure developing off the New
Jersey coast this evening. This low will intensify rapidly as it
tracks north through eastern Connecticut... central Massachusetts and
then into northern New England overnight.

Comments

[mabfan.livejournal.com]

I'm not really a weather expert, but a lot of predictions are based on precedent, believe it or not. If you know that when two fronts which meet in a certain way have created a storm of a certain type 75% of the time in the past, it's usually safe to predict that same sort of storm happening again.

Of course, weather does not necessarily follow from initial conditions -- hence Chaos theory.

Maybe someone else can come up with a better explanation.

[cruiser.livejournal.com]

I'm a weather geek (as a matter of fact, someone called me that today, because I happen to know the URL for the National Weather Service's radar image for DC [ok, so remembering the "DS.p19r0/si.klwx" part is pretty geeky - so shoot me]), so I'll explain. Mabfan's "precedent" comment isn't too far from the mark. But all the precedents are added together to get models, which take into account all the "when this type of front hits that type of front with these initial conditions, this is the typical result"-behavior. The fact that not all the initial conditions are accounted for in the models, and the fact that in many cases we don't have a large enough baseline to account for every possible weather condition are why the predictions aren't always right. But a good meteorologist can give amazingly accurate weather predictions for a small enough area & a small enough window of time, assuming you're not asking him to predict a week ahead of time. When I worked with the SEALs, we had a meteorologist on the Commodore's staff who had about a 95% accuracy rate for "what's the weather going to be for this beach for the next 12 hours"-type questions. The telltale in the prediction you quoted is in the last paragraph, where it says "the culprit for the stormy weather will be..." The meteorologist (or his models) have seen pressure systems like this develop before, and they usually result in storms like the one described. A note, regarding predictions - meteorologists generally prefer to doomsay, rather than leaning toward the rosier picture. They do this because if the model says there's a 45% chance of this storm happening, and they go with the 55% odds, and the storm happens, it will have a much more adverse effect than if they predict it and it doesn't happen - if you say there might be flash floods, people prepare for flash floods - if you don't predict them, nobody prepares.

[mattlistener.livejournal.com]

Great commentary, thanks!

So the picture I get is that all the current data is fed into The Model, which when run over the next 12 hours shows a storm coming into existence and doing such-and-such. Meteorologists use their knowledge of the science and trends for the area to interpret it and make their forecast, but the level of detail is largely thanks to the model itself. Sound right?

[cruiser.livejournal.com]

Pretty much. But if you spend time watching hurricane predictions, you'll see them talking about 8 or 9 different models, all of which have different sorts of utilities for predictions. You'll see them say "LBAR predicts this and CLIPER predicts that, while NOGAPS says something else." Then based on their experience and their knowledge of the advantages and limitations of a given model, they'll make their prediction. But a good meteorologist is a model unto himself. He can look at the various observations around the area he's forecasting for, apply his experience and come up with a prediction. If he's got models to use, he'll look at their output & decide whether or not they actually match what he thinks will actually happen. The level of detail in any given prediction really depends on 1) the quality and quantity of observations in the prediction area, 2) how much time the forecaster has, and 3) the size of the prediction area (in time and distance). The more and better the observations, the more time the forecaster has, and the smaller the prediction area, the better the level of detail (and level of accuracy, if metro-boy knows what he's doing).

[cruiser.livejournal.com]

Ooops. I forgot to mention, when I mentioned non-hurricane forecast models, is that there are several of those too.

[cruiser.livejournal.com]

Oh, and when I was talking about hurricane models, what I forgot to add was "LBAR predicts this and CLIPER predicts that, while NOGAPS says something else. Historically, in situations like this, the wind speed will increase the way NOGAPS says it will, while the general track of the hurricane will usually fall somewhere between that predicted by LBAR and BAM."