Amanda Murphy - February 23

Weather and Climate Systems

Discussion and Analysis of the GridRad-Severe Dataset and Methodology

Amanda Murphy

Wednesday, February 23

03:00 PM

Join Google Meet!

https://meet.google.com/hwd-ruyp-grx

Many studies have aimed to identify novel characteristics of storms that are indicative of current or future severe weather potential. Such fine-scale analyses are often limited to a handful of case studies due to how time consuming analyzing individual storms for such signatures is. Without a large dataset created specifically for severe weather analysis, attempting to diagnose such potential severe weather characteristics in multiple storms quickly becomes too time intensive. Therefore, we aim to create a dataset that allows for fine-scale investigation of many storms at once, leveraging understanding gained from past studies (e.g., best techniques for storm mode classification, supercell classification) and applying them to a large database of severe weather events.

Herein, we use the GridRad-Severe (GR-S) dataset, which includes a multitude of data for ~1000 severe weather days spanning 2010-2019, inclusive. Days and relevant spatiotemporal boundaries of data collection are determined based on the number of reports on a given day as well as the spatiotemporal distribution of reports. Sourced from within these relevant domains are composite radar data and storm report data, which can be used for objective storm tracking, supercell identification, and storm mode classification, to name a few. By matching these various characteristics to each individual storm that is tracked, we are able to answer more layered questions (e.g., what percent of tornadoes were produced by MCSs? were the largest hailstones reported associated with supercells? what times of the year, relative to other months, were single cell storms contributing the most to wind reports?) using a population of storms typically large enough to produce robust results despite the many categories the data may be divided into.

This talk will begin with a short discussion on the GR-S methodology, followed by demonstrations of the overall characteristics of the dataset and a discussion of the representativeness of the GR-S storm matched reports compared to the national repository of SED reports. After demonstrating its utility and ability to represent the total population of reports well, we will close with a comparison of GR-S conclusions to conclusions drawn in various other studies.