Forecasting 101 - The Super Ultra Magical Blog-length Version

Contrary to popular belief, we don't flip coins or toss darts at a board that has labels on it like "cloudy", "snow", "sunny", etc

There's actually some science and a pantload of math.

Meteorologists use many tools, including experience, to help them make a forecast.  There's a danger in relying in any one tool too closely - a mistake that most often results in sucking.  Witness this past weekend where local mets (that's what we're called for short...has nothing to do with NL baseball in New York) spoke of a dusting to 1 inch of snow from Friday night through Saturday.

What were they thinking?

By Thursday night, numerous computer models were suggesting that the storm - which had been forecast to slide just to our south, scraping our area with light amounts of snow - was going to move more to the north than expected.  By Friday afternoon, this trend was evident on radar.

But local mets said to themselves (I paraphrase): "The models are wrong and the radar is smoking crack.  Our experience tells us that with the current atmospheric pattern, the low pressure center will definitely move south.  So let's stick with the forecast."

By the time they released their "revised" forecast for 4-6 inches in the DC area and issued a Winter Storm Warning, it was 3 in the morning.  VDOT loves it when they do that.  And so do you: the condition of the roads on Saturday was entirely the fault of the National Weather Service.

So, in meteorology, these forecasting tools are best used in combination.  And the urge to hype something as STORM OF THE CENTURY must be contained.

If you've tuned out by now, I understand.

A few words about computer models...

A long long long time ago, some smart people realized that if the entire atmosphere was based on physics and thermodynamics, then it must be possible to model the whole system with mathematics.

A bit later (12 noon on March 5, 1950 if you must know), the first computerized forecast model started running.  My favorite piece of trivia about this is that the machine running it took 24 hours to make a 24 hour forecast.  Very helpful.

Computer models have become MUCH more advanced since then.  And there's a crapload of them now.  They run quickly, at regular intervals during the day, using data from 2 daily weather balloon launches that are made at every NWS reporting station in the country.  Different models are better at different things.  You could teach 10 classes on the different computer models, so we'll just take them as they come along in my posts.

I'll do my best to help you understand what I'm seeing if you care.  But I'll also try to get to the point of each post quickly, in case you don't.