Why are forecasts so good these days? The simple answer is technology.

Cray super computer.

Cray super computer assigned to the European Model Output.

Back in the 1980’s meteorologists were doing well to get a two day forecast correct! In the 90’s that was extended to 3 days, and then 5. The 2000’s hit and 7 days were all the rage. Today, 10 to 15 days are now advertised. Granted, the forecasts that far out suffer from lower accuracy, but they can typically provide an overall trend. So did meteorologists suddenly get better at predicting the weather? Not exactly.

3-Dimensional grid.

Atmosphere is sliced into a 3-Dimensional mesh grid. The closer the grid spacing, the higher the resolution.

Back in 1995 a very vocal Dr. Chuck Doswell stated that computer models will one day replace the need for meteorologists, creating a loss of jobs. Everyone scoffed at the time, but in a sense he was right. The models have improved immensely, and displaying that data can be found just about anywhere on the internet. You can even get your forecast from those that have never stepped foot in a meteorology class. As long as they can read the output of a computer model, they can pass along that information to you.

The last 30 years of Meteorology has advanced tremendously. Not with the mathematical equations that govern the atmosphere, those are mostly set, but in our ability to define and weigh many different variables that were largely missed due to poor resolution. Those once missed variables can be inserted as side equations due to the tighter grid spacing of the models. Factor in the speed of computer processing ( 450 billion calculations per second) and the numbers can get crunched in record time. What used to take 6 to 12 hours to complete, can now be accomplished in just minutes. In addition, these super computers can run different variations and iterations providing tweaked outcomes as certain variables are nudged in different directions. This is called an ensemble which leads to a mean value of the weather forecast and thus greater confidence.

Each grid point captures the weather variables for that region.

Each grid point captures the weather variables for that region.

Will the models still fail on occasion? You bet, they do plenty of times. They’re still governed by the data being ingested; garbage in equals garbage out. There are also weaknesses assigned to the different types of models in the set of equations selected (hydrostatic vs non hydrostatic). Some utilize better schemes than others. In the end, the forecast I and others give you, is vastly improved and quite accurate most of the time. We can thank the computer age for this, your tax dollars, and continued academic research.

6/24/16 Edit: I came across this excellent in-depth discussion about the weaknesses and strengths of the various models out there which some of you might find interesting. In particular, the poor performance of the GFS (American Model) compared to the Canadian and European models. You can read more here: Model Wars

Liked it? Take a second to support Aaron Tuttle on Patreon!