Evaluation of Derecho Wind Gusts in Convection-Permitting Climate Simulations using WSR-88D and HRRR wind data

High-wind events are the most frequent extreme weather occurring in the Midwest and the most impactful regarding damage to electrical infrastructure. Historical data indicates that about 67% of all Iowa natural disasters from 1980 to 2022, incurring costs in excess of a billion dollars, have been due to severe winds. For example, an August 10, 2020, derecho storm swept through Iowa and adjoining states with wind speeds up to 140 mph across an area 40-60 miles wide from north to south. The damage was widespread, interrupting electric service for an estimated 585,000 customers for hours, days, and, in some cases, even weeks. Despite the significant damage, there are very limited observational wind data available to assess the impacts. Regional Climate models at grid spacings of dozens of kilometers provide valuable insights into the changes in large scale weather patterns, however they struggle to capture the complex dynamics and intensity of small-scale scale phenomena such as derechos, preventing us from better understanding these events’ impacts on energy infrastructure.

This study has three objectives. First, we develop effective algorithms to retrieve Doppler wind velocities from the WSR-88D radars during derechos. The WSR-88D data are at higher spatial and temporal resolutions (hundreds of meters at minute interval) than the High-Resolution Rapid Refresh (HRRR) model, making it more capable of capturing detail structure of mesoscale phenomena such as gust fronts and downbursts. Second, we collect all the recorded derecho events in HRRR data and extracts wind speed, direction, and associated larger-scale environment, including sea level pressure, near-surface air temperature, wind shears, and moisture profile. Third, we evaluate our state-of-the-art convection-permitted (a grid spacing of 4km) climate simulations for derecho events against HRRR and radar data to assess whether the downscaled climate simulations can capture the high-wind areas and associated synoptic environments as shown by HRRR. This evaluation will provide confidence for our next steps: (1) future projections of frequency changes for such high-wind events; and (2) downscaling the climate model simulated winds from kilometer scale to utility scale (100s of meter) using a machine-learning-based super-resolution approach. The product will be provided to infrastructure studies for decision-making and maintenance.