Austin, Texas—A wastewater treatment plant isn’t top on everyone’s sightseeing itinerary. But on a cloudless, breezy day in April, Agmed Weber, operations manager for the Walnut Creek plant—which handles more than half of the sewage here in Texas’s booming capital city—is an enthusiastic tour guide. He marvels at the foresight of those who built the jumbo plant a half-century ago, when Austin was a third of its current size. “They were really thinking about the future,” he says. Weber feels a similar pressure—to prepare for not only continued population growth, but also a hotter future. Austin is “in a tight space where we need to think ahead,” he says.

The plant is dominated by vast outdoor pools in which microbes chew through waste until it is clean enough to return to rivers. But in one corner sit two large tanks and a series of purple pipes that route some of the treated wastewater back to the city to water lawns or flush toilets. Based on a climate study that predicts more severe droughts in Austin’s future, the city is planning, by 2050, to increase the amount of reused wastewater by more than 10 times, as part of a $1 billion expansion of the plant. “That program is going to grow exponentially,” Weber says.

All too often, local climate projections skip this step, says Douglas Maraun, a climate scientist at the University of Graz. “The key aspect is model evaluation.” Often researchers use the model they are most comfortable with, whatever it says—or they use a global suite of data that has already been downscaled. Such global data sets are particularly problematic, as rarely does one model handle everywhere in the world well, says John Nielsen-Gammon, the state’s climatologist at Texas A&M University. “The users of downscaled information typically do not know what the strengths or weaknesses are.”

Read the full article on science.org