Allen Research Group

Current Research Projects

Project: Quantifying the Risk and Impact of Wind and Hail Storms in a Warming Climate

Funding Agency: NIST

Funded Period: July 2022 - June 2025

Project Details

Damaging windstorms and hailstorms are destructive convective hazards that continues to expose the vulnerability of structures, agriculture and infrastructure with increasing frequency in the U.S. In this proposal we focus on extreme wind and hailstorms, which are among the leading causes of property damage in the U.S.. Total losses due to severe convective storms have exceeded $10 Billion dollars every year for the past decade. Despite these impacts, hail is not a regular consideration in material design, and convective winds have only recently been included as a separate factor in the ASCE standards. The threat of either wind or hail can also be further magnified when they occur in concert, where winds can increase the impact energy of hail. These events represent a hazard to both electricity distribution and production networks (i.e. wind turbines and photo-voltaic systems), which can in turn lead to localized cascading outages of water supply, automotive fuel, and heating/cooling systems. The goals of this project are to determine spatially and temporally explicit probabilistic scenarios of hail and wind extremes, and assess decadal, and long term climate change impacts on the probabilities of hail and wind extremes. This work is involves a collaboration with the Institute for Business and Home Safety.

Group Members Involved

Dr. John T. Allen, Principal Investigator

Dr. Subhadarsini Das, Postdoctoral Research Fellow

Kaleb Clover, Graduate Research Assistant

Project Deliverables & Outcomes

We have extended a previously established extreme value approach to the local scale to assess the appropriate distribution to use. In this work, fitting of a generalized extreme value distribution was pursued for cities around the country, where observational records are representative (Allen 2022). This blended all available hail datasets (MPING, CoCoRAHS, SPC Storm Data), and then constructing block maxima for fitting the distribution. This has revealed that for smaller regions with high quality observations, the prior approach of a Gumbel distribution poorly fits the actual occurrence of hail, and instead, a heavy-tailed Frechet distribution would be more appropriate. This analysis also shows the return intervals where observations are saturated even to 10 km scales, the distributions can be approximated by using averaging of the nearby region to fit the distribution. These finding are important as they illustrate that a fitting approach using prior distributions may be a useful to address regions with limited observations, and that a more representative observed record can be developed for determining accurate hazard maps together with quantifying uncertainty.