Hail Detection Using 1-Minute GOES-14 and GOES-16 Satellite Imagery and Ground-Based Dual Polarization Radar

Large hail can lead to significant agricultural and property loss. Due to historical limitations of reporting and other factors, these events have not been studied as extensively as other hazards such as tornadoes. In the contiguous US, real-time analysis and detection of severe storms is largely based on information gathered by the ground-based NEXRAD (Next-Generation Radar) network, which obtains volumes at about 5-min intervals. Since 2013, the NEXRAD network has provided radar observations at dual-polarization, which has provided better indications of hydrometeor types and of hail events. After the launch of GOES-16 in November of 2016, the spatial and temporal resolution of geostationary satellite imagery over the US is now greater than that of the NEXRAD network. Previous studies have found several satellite signatures associated with deep and severe convection. For example, overshooting tops (OTs) show a clear and distinct updraft in visible imagery with a corresponding local minimum in infrared brightness temperatures, indicative of high cloud tops. Mesoscale atmospheric motion vector (mAMV) products have recently been used with 1-min GOES imagery to diagnose unique cloud-top motions that often precede severe weather. These satellite products, from both GOES satellites, may show improvement in identifying severe storms. However, to date, thorough assessments of the available radar- and satellite-based products’ ability to identify hail occurrence and hail size have been uncommon or, for some variables, do not exist. This study undergoes a comprehensive comparative analysis of numerous satellite- and radar-based products to determine the best indicators for both hail occurrence and hail size. First, products are evaluated based on storm analysis filtered by U.S. population density and available hail reports to establish the best indication of hail occurrence. The same process is then repeated to evaluate the potential for hail size discrimination from each indicator. Using 5-min NEXRAD data from 30 recent severe weather events (2013-Present) and more than 10,000 storms, we show that derived methods using only single-polarization or dual-polarization observations offer similar skill when identifying hail occurrence. However, when used in conjunction, the combination of both methods provides improved skill. Little potential exists for identifying hail from quantitative satellite parameters. Maximum Estimated Hail Size (MEHS) and tropopause-relative echo-top heights provide the best positive linear correlation to hail size. Proposed improved to MEHS will be discussed.