247 Diagnosing and Predicting Thunderstorm-Induced Power Outages

Monday, 7 January 2019
Hall 4 (Phoenix Convention Center - West and North Buildings)
Matthew D. Eastin, Univ. of North Carolina at Charlotte, Charlotte, NC; and R. Cucinotta

Strong winds and lightning induced by thunderstorm activity are well known to be major causes of warm-season electrical power outages across the Midwest and Southeast. Strong winds (including straight-line gusts, microburst outflows, and tornadoes) often down trees and tree limbs, which can affect power transmission and distribution lines. Lightning strikes can also damage distribution line infrastructure (e.g., transformers, power-line poles, etc.). Restoring electrical power in a timely manner involves not only requesting and pre-positioning sufficient repair crews at relevant locations within the distribution grid, but also accurately anticipating the location and severity of such significant weather events. Historically, the prediction of warm-season thunderstorm activity up to 48 hours in advance has been limited to broad multi-state regions via daily probabilities of severe and non-severe convection with minimal information about storm type and intensity (i.e., scattered non-severe thunderstorms, severe squall lines, bow echoes, and/or tornadic supercells). As such, these broad convective outlooks provide energy forecasters and outage repair crews with limited guidance at the more critical intra-state spatial scales that could lead to more cost-effective outage mitigation decisions. However, the combination of advanced Doppler radars, increased scientific knowledge of internal thunderstorm dynamics, and recent advances in high-resolution numerical weather prediction, provides an opportunity to develop task-specific tools that can provide meteorologists with valuable forecasts of warm-season thunderstorm activity at the needed spatiotemporal scales.

The specific goals of this project are to (1) document spatiotemporal relationships between radar-derived metrics of convective vigor and reported power outages; and (2) develop a multi-regressive prediction model of power outage density that can be incorporated into a GIS framework. Five years (2013-2017) of warm season (March-October) Doppler radar and lightning observations combined with reported power outages, population density, and land cover/use data for three distinct regions (Midwest, Carolinas, and Florida) are used. All observations have undergone extensive quality control to remove spurious data while ensuring that only thunderstorm-induced outages are include in the developmental database. Details concerning the data, optimal metrics, and predictive models will be presented at the conference.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner