New paper in PNAS, joint with the wonderful Sustainable Development PhD student Stephan Thies as well as the usual suspects on my weather forecast work (Derek, Laura, Manuel). Headline findings: More accurate weather forecasts save lives, especially during extreme heat. Continued improvements in short-range temperature forecasts could reduce heat-related mortality in the U.S. by ~20% by the end of the century. That would offset nearly all of the increase in deaths under moderate climate change. But this is also “running to stay in place”: without climate change, these same improvements would lead to substantial declines in heat-related mortality. Surveyed operational meteorologists expect forecasts to keep improving, and realizing those improvements will require sustained investment in weather data and model development (including AI models).