On August 24 at 8:20 pm, a 44-year-old moonlighting meteorologist named Eric Berger was nearly finished writing a post for his Houston-centric blog, Space City Weather, titled “Harvey Late Night: Some Final Thursday Thoughts.” He was in his home office. He had just poured himself a glass of cabernet.
He had been looking at the online forecasts from the National Hurricane Center and agreed with their essential conclusions: Harvey was a well-organized storm that would land with hurricane force on the Texas Gulf Coast somewhere between Corpus Christi and Port O’Connor. Berger also backed the center’s belief that the winds would be strong in Houston that weekend, perhaps more than 40 miles per hour. But he was far more worried about the rain. The unanswered question is what happens to Harvey once it reaches the coast, Berger wrote. Where will it go, and will it go fast enough? Houston’s rainfall totals over the next five days depend on this, and we just don’t know.
The considerable majority of modern weather forecasting is aided by computer algorithm. Most hurricane tracking relies on data crunched by various public and private computer models, and the models, which take different variables (temperature, moisture, mass) and consider them in different ways, are not always in agreement. The National Hurricane Center takes input from several models to make its predictions, averaging out their differences, in part because it faces the tallest order in hurricane forecasting: It must say that the hurricane will go here. So must television meteorologists, one of the center’s principal conduits to a concerned public. TV, too, demands a singular answer.
Berger doesn’t have to draw a line. He is a certified meteorologist, but the weather is just a particularly absorbing hobby of his; his primary paying gig is writing about aerospace for Ars Technica (a site that is owned by Conde Nast, which also owns WIRED)1, and he blogs about the weather in his free time. That gives him two luxuries that most front-line meteorologists don’t have: He can value certain models and their ensembles much more heavily than others, untangling as many as 50 different versions of each forecast, and he can also admit doubt. He can explore the subtlety of the weather, marveling at its mysteries, the way he has for his small but loyal community of readers for years, but especially since he established his site in October 2015.
Berger does not generate his forecasts from scratch, pointing his licked finger into the wind and taking readings from the Galileo thermometer on his windowsill. He really does have a Galileo thermometer on his windowsill, but he works out of a home office that he hasn’t otherwise bothered to decorate, with a basic PC and a single monitor on which he toggles between tabs, from forecast to conflicting forecast. In Houston that evening, the US government forecast called for about 15 inches of rain. By then Berger had already begun to wonder.
There is one model he has come to trust and rely on more than any other: the European Centre for Medium-Range Weather Forecasts’ Integrated Forecasting System, more popularly known in the US as the European model. Funded by 22 EU members and 12 cooperating states, the European model is sometimes shockingly accurate, in part because it’s so well financed and its computing power is stronger than most.
With Harvey, it suggested that the storm would stall over Houston, dumping 25 inches of rain or more before eventually moving on. That synced with Berger’s own analysis of the weather patterns in the atmosphere. He detected a troublesome absence of steering currents, the forces that push hurricanes to wherever they’re headed next, and without those currents, the European model’s forecast of a stall made a lot of sense. Given the sum of the evidence before him, Berger felt confident in one fearsome prediction, and he wrote as much: Big-time floods are coming to Texas.
- >Nathan Hill
The Overwatch Videogame League Aims to Become the New NFL