Nov 24, 2021–When “official” weather predictions are less accurate than some random rodent’s, why do we keep paying attention to them? It seems to me that the accuracy of weather forecasting has plummeted like a barometer in Portland over the past few years.
For those of us who are not farmers, the main reasons for watching the weather at all is helping us decide whether to throw a light jacket in the back seat in the morning or if we need to park the car under the bank’s carport when there is a chance of hail.
But lately it seems the forecasts have been flat wrong every time. We go to bed with warnings of flash flooding and wake up to dewy grass. Predicted wind gusts and golf ball-sized hail arrive as gentle breezes and mist.
It seems the only forecasts in South Texas that are usually correct include highs in the 90s. Even my deaf dog could predict that.
Of course I wondered whether my forecasting skepticism is anecdotal. I’m a data-driven guy, and I know how perception can mislead the average human. So I looked it up. How accurate are weather forecasts?
Short answer: about half are correct. In other words, flipping a coin will get you the same level of accuracy as subscribing to WeatherBug.
This is flabbergasting, especially when you consider the vast global network in place for monitoring weather. Geostationary satellites hover 24/7, collecting images and data. The views available are nothing short of stunning, unimaginable to forecasters of the 1950s. We can now download bird’s-eye views of massive hurricanes, near constant lightning strikes world-wide, and even track Saharan dust crossing the Atlantic to make Texans sneeze.
But all a satellite can document is “current” weather. Predicting where that hurricane heads next is like trying to outsmart the roulette wheel in Vegas. Just look at the various tracks presented even 24 hours in advance of the next Hurricane Horace. The options can be off by a factor of two states.
Closer to the earth’s surface, we now collect real-time data from around the world. Weather balloons, satellites, water buoys. National Oceanic and Atmospheric Administration (NOAA) analysts measure wind speed, moisture content, ocean temperatures, and factor in the effects of El Nino, La Nina, andforest fires in Mexico. I can’t even take a bike ride without passing a weather data collector a mile from my house.
The Weather Channel recently published charts that described their forecasts as “mostly accurate” up to 9 days out. But they only measured temperatures, which anyone living in South Texas more than a year could predict.
A study in Britain measured the accuracy of the BBC’s weather forecasts. The one focused on the community of Cambridge determined the one-day forecast was 53% accurate. A schoolboy could do as well with a coin flip.
When they looked ahead four days, the accuracy dropped into the 30-40% range, or two out of three chances.
To put it in perspective, let’s slide down the scale, all the way to a below-the-ground weather prognosticator: Punxsutawney Phil (no relation).
According to the legend, whether Punxsutawney Phil sees his shadow on Feb. 2 means there are six more weeks of winter, or not. I can never remember the extremely technical shadow/winter formula. But the mangy groundhog has been doing this since 1887. How often does he get it right? Exactly 39% of the time. Which would place him neck and neck (if groundhogs had necks) with the BBC and within the margin of error of The Weather Channel.
Whether this is an indictment of state-run science or an endorsement of rodent prescience is up to you to decide.
Regardless, my cursory research confirms my thesis. Weather forecasts are random. In fact, they are worse than random. You would come out ahead betting on the weather forecast to be wrong, with as high as a 2 in 3 chance of collecting.
But so what? As I obnoxiously point out, unless you are baling hay or launching a space shuttle, the only time weather affects most of us is during the 30 seconds we walk from our door to our car.
Next time you are deciding whether to take your umbrella, just flip a coin.