WEATHER/TECHNOLOGY
Looking at My Phone, I Think That the Weather App Is Using AI
The prediction models have looked weird for a while.
Over the last few years, haven’t you noticed that the weather apps and models seem slightly off?
I was on my phone to check the weather for the next 10 days, and I noticed something weird. For a couple of days early in May, the weather was forecast to be in the 50s with overnight lows getting into the 40s. That wasn’t the weird part.
That seems pretty normal for early May in Western New York. It was what the actual precipitation event would be at that temperature. A high of 54 and a low of 46 with rain and snow. What?
It might’ve snowed last year around that time when the temperature dipped lower, but it would never snow at that high of a temperature, right? The freezing temperature of water is 32 degrees, and the air temperature in this instance will be nowhere close to that.
Am I not accounting for something? Such as upper-level air temperature? How do you predict that at least a week out? I don’t buy it. The better explanation is that an AI weather model was possibly used to forecast this range of days on a year-to-year basis and estimate the temperature within 10 degrees.
After all, we already know the average annual temperature and what it’s supposed to be on every given day. The only thing the AI model would have to consider beyond that is sunshine, cloud cover, storm tracks, and pressure systems. 46 certainly gets close to the 30s when considering a 10-degree range anyway.
Also, there have been days in the last few months when our outdoor temperature sensor has found a nearly 10-degree discrepancy between the reported temperature and the real air temperature. You can’t tell me that the temperature sensor is broken. It’s right most of the time, and the discrepancies aren’t that occasional, but often enough to notice.
I’m not going to outright accuse these weather companies and weather forecasters of doing this. Based on what I’m measuring here on the ground, and the amount of time I’ve spent studying weather patterns and just being a huge weather nerd, I’m convinced that something isn’t the same as before when it comes to weather forecasting.
I considered becoming a weather watcher for one of the local news companies to give them a better eye on what’s on the ground at this point. I do understand the complexities and how hard it is to get things right. But we can’t just go to a technology that is notorious for getting many things wrong consistently to be an aid in forecasting something as important as the weather.
Many people depend on reliable forecasts to help them in various aspects of their lives. Farmers need to know weather patterns for harvest cycles or growing cycles. I don’t need to go through many more examples to explain how important this is.
We need to go back to relying on the tried and true technology, and also getting the input of the people on the ground before we turn to a fully automated forecasting system. Again, just a hunch, and not an accusation.
