We were encouraged to post some recent work done by a forecaster within our office as others might find it interesting and be able to use it operationally as well. Tony Petrolito derived local probability of severe hail equations using the logistic regression technique. This was based on original work completed by Mark DeLisi in the late 1990s. The results of Tony’s work yielded three equations that are used operationally, a “VIL of the Day” equation (used prior to convection developing) and two other equations relating the height of the 50dBZ core of a particular thunderstorm to the 0oC and -20oC isotherm levels to estimate the probability of severe hail (used in real-time during a convective event).
A dataset of over 200 hail cases was used which included an unbiased sample of null events (no hail), small hail (less than 1.00 inch) events, and large hail (1.00 inch or larger) events. The following image is a screenshot of part of the hail database that was used.
To use this tool in operations, a simple Graphical User Interface (GUI) was created which provided the forecasters with the ability to adjust slider bars with the different predictors which would then provide the forecaster with an output severe hail probability. The following images are the “VIL of the Day” GUI and a GUI of one of the real-time equations that relates the height of the 50dBZ core to the freezing level.
The parameters used for the slider bars in each of the GUIs were the highest correlated predictors. Below is the abstract that was submitted to Eastern Region for a Technical Attachment publication that should be published sometime this Fall.
“In the late 1990’s, a probability of severe hail equation (LPSH75) for diameter greater than or equal to 0.75 in (1.9 cm) was developed for the Columbia South Carolina County Warning Area (CAE CWA) using a logistic regression methodology. This equation provided a “VIL of the Day” which forecasters used operationally as an aid in the warning decision making process. In early 2010, National Weather Service (NWS) criterion for severe hail was changed to a diameter greater than or equal to 1.00 in (2.5 cm). The purpose of this study was to develop new objective methods to estimate the probability of severe hail based on the new NWS definition. In addition to a new probability of severe hail equation (LPSH100) for diameter greater than or equal to 1.00 in (2.5 cm), two equations (LPSH0 and LPSH-20) were developed that relate the height of the 50dBZ core of a particular thunderstorm to the 0oC and -20oC levels to estimate the probability of severe hail. LPSH0 and LPSH-20 can be useful during storm interrogation by providing a non-subjective method to estimate the probability of severe hail in near real time. Local applications were developed to display the severe hail probability output to the forecasters in the Advanced Weather Interactive Processing System (AWIPS).”
– Hunter Coleman
WFO Columbia, SC