Growing up in the marshy plains of the Texas Gulf Coast, Ellen Buchanan had seen her share of floods. But in 2017, when Hurricane Harvey dumped 40 inches of rain on her home in Silsbee, a suburb of Beaumont, even she was caught off guard.

“Harvey was a whole different thing,” Buchanan, 70, said. “It flooded places that had never flooded before. All the creeks and bayous that flow to the Neches River turned each community into its own little island.”

The Neches River, in turn, carried all of that water 15 miles south, to an already inundated Beaumont. There, it swamped the city’s main and secondary pump stations, cutting off water to 110,000 residents for more than a week. Without access to potable water, storm shelters full of shell-shocked evacuees were forced to seek safety elsewhere. Outside, they were greeted with the dank smell of sulfur dioxide, the result of hurricane damage to one of the region’s many petrochemical refineries.

In the aftermath, local officials, emergency responders, and residents like Buchanan wondered how they would prepare for the next storm. Could street-level structures like bioretention ponds and stormwater tunnels soak up the rain next time? Or did they need more significant interventions: levees, flood gates or even relocation assistance? Data-driven answers were hard to come by.

While meteorologists have gotten quite adept at predicting the path of massive storm systems like hurricanes, anticipating the associated impacts of rain, flooding and storm surges at the street level is often more complicated. That’s because existing computer-generated models don’t account for the block-by-block, neighborhood-by-neighborhood factors that impact a city’s microclimates.

“Most models have pretty coarse resolution of about 25 to 50 kilometers,” said Gary Geernaert, director of the climate and environmental sciences division at the Department of Energy (DOE). When researchers run model simulations to learn how even stronger storms might affect southeast Texas, towns like Silsbee and Beaumont — and all the creeks, parks, parking lots, and neighborhoods that form their microclimates — are obscured. “The model treats them kind of like a black box,” Geernaert said.

Without a clear picture of what’s to come, Buchanan has watched with concern as neighbors in her corner of the metro, a community with large Black and Hispanic populations, resumed buying the cheap houses that continue to be built in the flood plains.

Read the full article about improving climate models by Ashley Stimpson at Grist.