Is There a Total Lightning Precursor Signal for Non-supercell Tornadoes?

By the time of the AMS Annual Meeting it will have been nearly 35 years since the first documented case of a non-supercell tornado by Doppler radar; the Erie, Colorado tornado of 26 July 1985. The study of that tornado and other subsequent studies established the importance of a strong convective updraft positioned over a low-level circulation as the formative mechanism for this type of tornado, considerably more straightforward than the more complex tornado associated with a supercell storm. Unfortunately, the lack of a mid-level cyclone prior to tornadogenesis meant that any Doppler radar precursor rotational signal would be found close to the ground and generally be of far smaller diameter than a mid-level mesocyclone, making it very hard to see a developing non-supercell tornado if it was farther than 60 km or so from the radar. So, despite their easier to understand formative mechanism, the non-supercell tornado has proven to be quite difficult to warn for with much, if any, lead time.

The strong relationship of the non-supercell tornado to updraft strength raised the possibility that if one could diagnose an increasing updraft in real-time this might be a potential aid to helping to warn for these tornadoes. Research over the years with experimental ground-based lightning networks that measured both cloud-to-ground and in-cloud lightning established a relationship between in-cloud lightning and updraft strength. Many studies found that a distinct increase in lightning, often referred to as a “lightning jump”, was often seen to precede an increase in convective storm strength and in many cases subsequent production of severe weather. However, in-cloud lightning data was restricted to a few, limited area ground-based Lightning Mapping Arrays (LMAs). With the launch of GOES-16 (and more recently GOES-17), real-time measurements of total lightning (both in-cloud and cloud-to-ground) have been made with an instrument known as the Global Lightning Mapper (GLM). Meanwhile, while some issues have been worked through to make the GLM data operationally available to National Weather Service (NWS) forecasters, the NWS has purchased real-time data from a now nation-wide ground-based network established by Earth Networks (EN). EN displays both total lightning as well as the separate components of in-cloud and cloud-to-ground lightning. These two networks mean that NWS forecasters now have (or will soon have for all Weather Forecast Offices (WFOs) in the case of the GLM) real-time total lightning data, providing a relative measurement of updraft strength in a convective storm.

We began investigating the possible relationship between total lightning and non-supercell tornadogenesis in 2013, after the Boulder and Cheyenne WFOs became test sites for displaying GLM protype products on AWIPS (the workstation used by the NWS) using data from the Colorado LMA. Shortly after the test period began, a fortuitously placed non-supercell tornadoes occurred over the Denver International Airport (DIA), in close proximity to both the Denver NWS and Denver Terminal Doppler radars. This led to a detailed study relating the trend of total lightning to the vertical reflectivity and velocity characteristics of both the cell that produced the DIA tornado and a nearby cell that failed to produce a tornado, both storms forming along a pre-existing low-level boundary. The results of that study, which will be reviewed at the conference, gave us some confidence to pursue this total lightning/non-supercell tornado relationship. A month later a storm near Cheyenne, Wyoming produced a non-supercell tornado, and this one had a sharp increase in total lightning prior to tornadogenesis. Over time we have sought to find other non-supercell tornado cases over different locations to further investigate the relationship to total lightning. We now have several cases over the last couple of years from a variety of locations, and some with both GLM and EN data. This presentation will summarize these cases, some of which have certainly proven to be somewhat more complex than others. We will use our findings to suggest ways to use the data on AWIPS as a possible aid in issuing tornado warnings for non-supercell tornadoes.