The Colorado Sun’s Olivia Prentzel reports from Colorado SunFest that artificial intelligence is changing how meteorologists forecast Colorado’s famously unhinged weather. The panel featured Joel Gratz of OpenSnow, retired Denver7 chief meteorologist Mike Nelson, and 9News senior meteorologist Chris Bianchi, all talking about how AI-driven models are getting faster, more useful, and possibly better at helping forecasters see what is coming before Colorado does its usual “sunshine to sideways hail” routine.
The main takeaway is that AI is not replacing meteorologists, at least not yet. It is letting them run far more models, faster, with better probability forecasts. So, yes, the future of weather may be pretty faces reading smarter computers. Which, frankly, could mean the forecaster is both pretty and right. Around here, that would be a programming breakthrough.
The Bullet Point Brief
- Joel Gratz said the big AI advantage is speed, allowing forecasters to run 50, 100, or even 1,000 simulations instead of just a handful. Translation: fewer darts at the weather board, more actual math.
- Chris Bianchi said he now looks at AI models over older numerical models, a shift he said is new within the last couple of months. That is a pretty big confession from a TV weather guy. Somewhere, an old radar screen just asked for a cigarette.
- Bianchi also predicted that within five years, AI could produce pretty darn good forecasts up to two weeks out. In Colorado, a two-week forecast that is even halfway trustworthy would be like finding a unicorn in a Carhartt jacket.
- Gratz said one forecaster on his team trained an AI model on 40 years of tornado reports, lightning, hail, and wind data to predict specific weather events. That is useful. Also slightly terrifying, because now the computer knows when the sky wants to throw furniture.
- Mike Nelson said AI is the latest leap in forecasting technology, but humans still need to interpret the data and explain it clearly to the public. Good. Because a model can say “probability envelope.” A human can say, “Move your truck before the hail turns it into a golf ball.”
My Bottom Line
Wait a minute. So what they are saying is, with AI, weather forecasters may finally only need to be pretty? And now they might be pretty and right? What a change of pace. Science really is amazing.
In all seriousness, Colorado weather is not for amateurs. Northern Colorado can get wind, hail, snow, heat, smoke, and a thunderstorm that looks like it has unresolved childhood issues, sometimes in the same week. If AI helps forecasters sort through more data faster and give people better warnings, good. Use the tool. That is exactly what technology is for.
But Nelson is right that the human still matters. Data is not the same thing as judgment. Models are not the same thing as communication. A forecast has to be understood by actual people who are deciding whether to plant, travel, irrigate, pour concrete, drive a school bus, move cattle, or cancel a ballgame. The computer can crunch numbers. Somebody still has to translate the sky.
For my part, I like Kody the Weather Guy. He may not be the prettiest face in weather, but he seems to nail it for NoCo. His sense of humor is an acquired taste, like gas station coffee or county fair chili, but he tends to be right more often than the polished crowd. AI or otherwise, that is what matters. Pretty is fine. Accuracy is better. Pretty and accurate? Well, now we are just showing off.
Source: The Colorado Sun
Share your thoughts...