Is the weather man getting more accurate?
After Sandy made its way through Long Island, we can begin to look back at the adequate amount of forecast details we had regarding the storm. We knew when it would make landfall, how high the winds would be, including gusts and how much precipitation we would get. We also had computer models as to what the track or path of the storm would be. How do they do that so well nowadays?
In an IEEE Spectrum article “Predicting Hurricane Sandy” it was stated:
In 2009, the U.S. National Weather Service set out to improve hurricane forecasting by 50 percent within the decade. Just two years later, the first fruits of that effort were seen in the run-up to Hurricane Irene, when the service's 48-hour forecast was just as accurate as its 24-hour forecasts had been a decade earlier. This year, the NWS's Global Forecast System was able to move a bunch of algorithms from the research lab into its operational models, just in time for Hurricane Sandy. The new algorithms took temperature, humidity, and wind data collected by aircraft, weather balloons, satellites, and ground stations and extrapolated them onto a much finer grid pattern than was previously possible. This one change improved the accuracy of hurricane prediction models by 20 percent.
John M. Goff, National Weather Service, Burlington, VT has a presentation to describe how good forecasting has become using probabilistic analysis methods.
A commonly used method of gauging Probabilistic Precipitation (PoP) Forecasting is through the Brier Score—Accuracy of a probability forecast (The following is courtesy of Eumetcal-- The European Virtual Organization for Meteorological Training)
The Brier Score is probably the most commonly used verification measure for assessing the accuracy of probability forecasts. The score is the mean squared error of the probability forecasts over the verification sample and is expressed as:
where N is the sample size. The observations oj are all binary, 1 if the event occurs and 0 if it doesn’t. The Brier score ranges from 0 for a perfect forecast to 1 for the worst possible forecast. Although the score can be computed on a single forecast, the result wouldn’t be very meaningful because the observation is binary and the forecast is a probability.As an engineer, I am not as skeptical as I was before I read this scientific method of forecasting.
Take a look at what we saw on Long Island NY this week (Credit: Stephen Rathjen) as our prayers go out to families all over the region who have experienced severe loss and damages. Kudos to the First Responders all over the region!
An Angry Sea