Do urban areas intensify hail?

Hail represents a significant meteorological hazard and a source of insured property losses in multiple countries. Hence, there is interest in understanding the nature of this atmospheric hazard and in addressing questions such as whether feedbacks from urban land cover enhance the probability or size of hail, particularly considering global urbanization trends. Research presented herein is based on 20 years of data from dual-polarization scanning Doppler radar and suggests that, on average, urban areas modify both the frequency of radar-derived hail signatures and the probability of damaging hail and thus amplify the hazard magnitude. A paired sample analyses of radar-derived hail metrics over 16 major metropolitan areas of the contiguous US suggests that, on average, urban areas exhibit a statistically significant ( p ⩽ 0.05) higher probability of exhibiting radar-derived hail signatures than equivalent rural areas. The increase is evident for three-quarters of the conurbations studied and is due to a combination of an increased frequency of finite-duration (5–30 min) periods with any hail signatures and the number of hail signatures during periods with one or more hail signatures. On average, urban areas also exhibit a statistically significant higher probability of damaging hail (i.e. maximum estimated size of hail (MESH) >0.75 inch, or 19 mm). Most of the urban areas studied also exhibit a higher frequency of hail at the ground (probability of hail, PoH ⩾ 80%), but bootstrap resampling shows that the average impact is not statistically significant. The magnitude of the enhancement of hail frequency in the urban samples versus the rural samples across the 16 locations studied exhibits positive rank correlation coefficients ( ρ ) with regional hail frequency ( ρ not significantly different to zero), urban areal extent ( ρ ≠ 0 at the 80% confidence level), differences in convective available potential energy and aerosol abundance over the urban and rural locations ( ρ not significantly different to zero and ρ ≠ 0 at the 95% confidence level, respectively). The ratio of urban to rural frequency of radar-derived hail signatures also exhibits the largest dependence on urban aerosol enhancement.