Weather forecasting has come a long way in the past century! Almanac meteorologist Michael Steinberg walks us through his own experiences with forecasting, and introduces the National Weather Service’s latest forecasting system.
An Early Start
On my 15th birthday, way back in June of 1967, my family moved. And when I started 10th grade in the new school year, I discovered that my new high school had an “advanced weather program” that was not available at any other high school in the world.
We had access to the latest technology: “high-speed” Teletypes that received and printed the latest weather data at 30 characters per second, along with a weather radar that the school had received as military surplus.
By the time I graduated from high school (to eventually go on to major in meteorology in college and then graduate school), I already had two years of experience in doing the weather forecasts for a local radio station and newspaper.
Everyone Blames the Weatherman
Weather forecasting back then was almost entirely based upon experience and intuition. Some people were able to visualize how weather patterns would evolve based upon the current weather maps, and they were the best forecasters. We also had access to some of the first computer forecast models of the atmosphere, but they were so simple and inaccurate that they were of minimal value.
The primary problem with those forecast models was the computers, which had trivial capabilities compared with what is available today. To give you an example, the computer that helps your microwave to work has more computer power than the computers that were used to guide the Moon landing in 1969.
Since that time, more than 50 years ago, computers have gained many orders of magnitude in power, speed, and storage capabilities, and the computer forecast models that we use today yield much more accurate forecasts than even the most skilled meteorologists could ever hope to create without them.
Although we still complain about the accuracy of weather forecasts, they are more accurate than many other kinds of forecasts made today. The average error in the forecast for the eighth day is less today than it was for the second day back when I started forecasting the weather. And, while back then people laughed at a forecast of a blizzard coming the next day, now stores sell out of milk as people prepare for a big snowstorm that is forecast to occur in a few days.
A New, More Accurate System
In 2018, the U.S. National Weather Service had a major upgrade to its computer facilities, making its combined weather and climate supercomputing system among the 30 fastest in the world, with the ability to process 8 quadrillion calculations per second. This enabled them to make changes to their primary weather forecast computer model, which they tested and improved further over the following year.
On June 12, 2019, they put the new model into operation, as an upgraded version of the country’s flagship weather forecast model, the Global Forecast System. This is the first major upgrade in almost 40 years to the model’s dynamical core, which is a key model component that computes wind and air pressure for successful numerical weather prediction. The new model also brought about many other improvements and enhancements to make weather forecasts more accurate. This model is used not only by the National Weather Service, but also by TV meteorologists and commercial weather companies like AccuWeather, which use this and other models to generate their forecasts.
So, you may notice an improvement in the accuracy of weather forecasts covering the period from tomorrow through the next week or two, and this new computer forecast model is the main reason why. Of course, this does not affect the long-range weather forecasts in The Old Farmer’s Almanac, which are based on solar cycles and calculated more than a year in advance—and will remain as the most accurate long-range forecasts from any source.
