Estimating the probability of hail in Canada using a lightning proxy

Crowd-sourced severe weather reports are known to have reporting bias towards population centers, yet they often represent the only direct observation of high-impact weather such as large hail. We present a method to combine hail reports with lightning observations to build a climatology of hail. The probability of hail is inferred from Bayes’ rule by combining the interpolated prior probability of hail estimated at point locations with the likelihood ratio of lightning given hail observations over the lightning probability before observing any hail. We apply this methodology in Canada using a database of hail reports from Environment and Climate Change Canada, hourly METAR observations of hail occurrence at airports and lightning observations from the North American Lightning Detection Network. The relative frequency of non-severe hail (2- cm), severe hail (2-5 cm) and significant severe hail (5+ cm) is inferred from the fitting the tail of an exponential distribution to reported hail diameters in the hail report database. The derived hail climatology is compared to severe weather reports to assess reporting gaps. Finally, we discuss how the hail climatology can be used in contingency table-based verification of severe weather reports.

Figure: Hail climatology in Canada estimated from the lightning proxy (in hail days per year)