Thursday, March 08, 2007

More Evidence of Global Warming

Cold weather puts chill on February US retail sales
Thu Mar 8, 2007
By Nicole Maestri

[February] marked the coldest February since 1979, according to weather tracking firm Planalytics...

'Course, the source is Reuters, so caveat scholāris.

And speaking of predictions of global warming, which depend on forecasts from climate models:

Can This Weatherman See Your Future?
By Andy Raskin
August 1, 2003
Business 2.0

[All emphasis added]
[Ants Leetmaa], director of NOAA's Geophysical Fluid Dynamics Laboratory at Princeton, N.J., says it's "impossible" to achieve real accuracy in weekly, long-range forecasts. Leetmaa is a bit of a maverick himself--he's famous for having predicted the El Nino of winter 1997-98 six months in advance. "I said central California and Florida might not be ideal holiday destinations," he recalls. "Boy, did [the chambers of commerce howl]. But it rained for, like, 90 days straight."

Leetmaa explains that forecasters generally rely on either science or history to predict the weather. For short-range predictions, they use physics to model the atmosphere and simulate changes over time. Run on supercomputers like the National Weather Service's new 7.3-teraflop IBM, short-range models have "skill"--meteorology jargon for accuracy greater than flipping a coin--out to seven days. But going even a week out pushes the limits of dynamic modeling. Indeed, in 1963, MIT meteorologist Edward Lorenz, the father of chaos theory, famously asserted that weather is so complex that skill a month out would never be attainable.

For longer-range forecasts, meteorologists depend on statistical models that are less about physics than about historical patterns. Leetmaa's El Nino forecast showed that skill can exist beyond the Lorenz limit, but he was making only broad estimates for average temperatures and precipitation over a full season--in contrast to Planalytics's highly localized weekly forecasts. Then again, it's possible he just got lucky. NOAA's Climate Prediction Center makes year-ahead seasonal predictions every month; on a scale of 0 (no skill) to 100 (right in every region), NOAA's average score is 20...

But we already knew that it wasn't about the science...


Blogger Harry Eagar said...

Hawaii Public Radio once had a young lady reporter, Tracy Tong, who always gave the same weather report: "It will be a nice, sunny day."

Even when it was pouring rain.

She was right 85% of the time.

Beat that, NOAA.

March 08, 2007 9:52 PM  

Post a Comment

Links to this post:

Create a Link

<< Home