GC23E-1182
Characterization of Nighttime Light Variability over the Southeastern United States

Tuesday, 15 December 2015
Poster Hall (Moscone South)
Tony Cole1, Andrew Molthan2 and Lori A. Schultz1, (1)University of Alabama in Huntsville, Huntsville, AL, United States, (2)NASA Marshall Space Flight Center, Huntsville, AL, United States
Abstract:
Severe meteorological events such as thunderstorms, tropical cyclones and winter ice storms often produce prolonged, widespread power outages affecting large populations and regions. The spatial impact of these events can extend from relatively rural, small towns (i.e. November 17, 2013 Washington, IL EF-4 tornado) to a series of adjoined states (i.e. April 27, 2011 severe weather outbreak) to entire regions (i.e. 2012 Hurricane Sandy) during their lifespans. As such, affected populations can vary greatly, depending on the event’s intensity, location and duration. Actions taken by disaster response agencies like FEMA, the American Red Cross and NOAA to provide support to communities during the recovery process need accurate and timely information on the extent and location(s) of power disruption. This information is often not readily available to these agencies given communication interruptions, independent storm damage reports and other response-inhibiting factors.

VIIRS DNB observations which provide daily, nighttime measurements of light sources can be used to detect and monitor power outages caused by these meteorological disaster events. To generate such an outage product, normal nighttime light variability must be analyzed and understood at varying spatial scales (i.e individual pixels, clustered land uses/covers, entire city extents). The southeastern portion of the United States serves as the study area in which the mean, median and standard deviation of nighttime lights are examined over numerous temporal periods (i.e. monthly, seasonally, annually, inter-annually). It is expected that isolated pixels with low population density (rural) will have tremendous variability in which an outage “signal” is difficult to detect. Small towns may have more consistent lighting (over a few pixels), making it easier to identify outages and reductions. Finally, large metropolitan areas may be the most “stable” light source, but the entire area may rarely experience a complete outage. The goal is to determine the smallest spatial scale in which an outage can be detected. Presented work will highlight nighttime light variability over the southeastern U.S. which will serve as a baseline for the production of a near real-time power outage product.