Eos, Vol. 94, No. 5, 29 January 2013
VOLUME 94
NUMBER 5
29 JANUARY 2013
EOS, TRANSACTIONS, AMERICAN GEOPHYSICAL UNION
PAGES 53–60
Satellite Observations Monitor
Outages From Superstorm Sandy
In support of disaster response, NASA’s
Short-term Prediction Research and
Transition (SPoRT) Center developed an
experimental composite of VIIRS day-night
band imagery to help in the immediate
detection of outages and to monitor recovery.
The technique produced a false color
composite image combining nearly cloudfree, pre-event data from the early morning
of 31 August and data from the days
following Superstorm Sandy. Specifically, the
pre-event lights are assigned to red and
PAGES 53 –54
In late October 2012, Hurricane Sandy
traveled across Jamaica, Cuba, and the
Bahamas, then progressed northward along
the eastern seaboard of the United States,
resulting in numerous tropical storm
warnings along the coasts of Florida and
North Carolina. As the storm approached the
Mid-Atlantic region, interaction with an
upper-level low drew the cyclone inland,
with the center passing just north of Atlantic
City, N. J. In what media reports dubbed a
“superstorm,” Sandy produced hurricaneforce winds, significant coastal storm surge,
torrential rain, inland flooding, and extensive
damage over a vast area. Further west of the
cyclone center, strong winds increased wave
activity throughout the Great Lakes, and
heavy snowfall occurred across portions of
Tennessee, Kentucky, and West Virginia. As
of early November, more than 100 fatalities
had been attributed to Sandy in the northeastern United States, with total economic
losses of up to $50 billion [New York Times,
2012, and Walsh and Schwartz, 2012].
The high winds, heavy rainfall, and coastal
flooding caused widespread power outages
for millions of customers in the Northeast.
Particularly hard hit were New York City,
Long Island, and eastern New Jersey.
Although many utilities now provide
GIS-based mapping of known outages in
response to customer complaints, satellite
imagery can also be used to identify
large-scale power outages that result from
disasters such as Superstorm Sandy
(Figure 1).
A year after the October 2011 launch of the
Suomi National Polar-orbiting Partnership
(Suomi-NPP), the Visible Infrared Imaging
Radiometer Suite (VIIRS) carried aboard the
satellite was able to provide before-and-after
imagery of city, suburban, and rural lights
that was useful in identifying regions without
power following the storm. These images
were acquired from the VIIRS low-light
channel, or “day-night band,” which is
capable of detecting nighttime lights
produced by cities, wildfires, and other
human activities, in addition to moonlight
that is reflected from clouds and surface
features. Within the United States, products
from VIIRS are provided in near real time
through the satellite’s direct broadcast
capability, resulting in rapid acquisition of
imagery useful to emergency responders.
Fig. 1. False color composite of pre-event (31 August 2012) and post-event (1 November 2012)
imagery from the Suomi National Polar-orbiting Partnership (Suomi-NPP) Visible Infrared Imaging
Radiometer Suite (VIIRS) day-night band used to identify possible power outage areas.White
shades indicate points where lights are present in both the pre- and post-event imagery, while
yellow areas suggest reduced light in the post-event image. Large areas of blue shading correspond to cloud cover that was present only in the post-event image. Analysis of light-reduced
areas is limited to cloud-free areas in this compositing technique.The image was created by the
NASA Short-term Prediction Research and Transition (SPoRT) Center using real-time and archival
data provided by the Cooperative Institute for Meteorological Satellite Studies.
This paper is not subject to U.S. copyright. Published in 2013 by the American Geophysical Union.
Eos, Vol. 94, No. 5, 29 January 2013
green color intensities, with blue intensities
used for the current day.
The composite quickly highlights outages
during cloud-free conditions through the
resulting color combinations. For example,
lights that remained lit throughout the event
appear in both the before and after imagery.
These lights are assigned nearly equal values
of red, green, and blue to each pixel,
resulting in a white color. Lights present in
the pre-storm imagery but missing or
dimmed in the post-event imagery appear in
various shades of yellow. Some isolated areas
of cloud cover on 31 August that were not
present again on 1 November also appear
yellow but are generally limited to areas of
open water or outside the region of interest.
Caution must be applied in areas with cloud
cover so that a reduction or change in city
lights is not overestimated, but outage areas
are readily apparent when skies are clear.
An example of this experimental composite is shown in Figure 1, where yellow pixels
in New Jersey, lower Manhattan, and
portions of Long Island correspond to known
power outages that were reported by local
utilities and various media outlets. Though
not shown here, subsequent mornings of
clear sky imagery helped to monitor recovery
as yellow pixels were restored to white
shades corresponding to light emissions
comparable to pre-storm conditions.
Though an experimental product, these
false color comparisons were staged in the
U.S. Geological Survey (USGS) Hazards Data
Distribution System and the Federal
Emergency Management Agency data portal
for use by several federal agencies in
recovery efforts. Through collaboration with
USGS and NASA’s Applied Sciences Disasters
Program, these images were also provided to
other federal agencies that used yellow pixel
classifications to identify possible outages in
order to effectively stage generators, supplies,
and other relief items as part of their support
to the affected areas.
Future efforts will focus on the development of a VIIRS clear sky composite of city
lights to ensure that clear sky pre-event data
are readily available for analysis following
other disasters that lead to widespread power
outages, such as earthquakes, severe
weather, and damaged infrastructure.
Through development of clear sky composites in varying conditions, more quantitative
techniques can be applied to assess the loss
of power and the effectiveness of recovery
efforts. These types of analyses may also be
of benefit should power outages occur as the
result of other natural hazards, such as solar
This paper is not subject to U.S. copyright. Published in 2013 by the American Geophysical Union.
activity, earthquakes, winter storms, strong
winds, and tornadoes.
Acknowledgments
Data from the Suomi-NPP VIIRS instrument
were provided in near real time and archival
mode by the Cooperative Institute for
Meteorological Satellite Studies at the
University of Wisconsin in Madison.
Post-processing of the data was performed
by the NASA SPoRT Center using additional
routines provided by the Cooperative
Institute for Research in the Atmosphere at
Colorado State University in Fort Collins.
References
New York Times (2012), Mapping Hurricane
Sandy’s Deadly Toll, 17 Nov. [Available at http://
www.nytimes.com/interactive/2012/11/17/
nyregion/hurricane-sandy-map.html].
Walsh, M. W., and N. D. Schwartz (2012), Estimate
of Economic Losses Now Up to $50 Billion, New
York Times, 1 Nov. [Available at http://www.
nytimes.com/2012/11/02/business/estimate-ofeconomic-losses-now-up-to-50-billion.html?_r=0.]
—ANDREW MOLTHAN and GARY JEDLOVEC, NASA
Marshall Space Flight Center, Huntsville, Ala.;
E-mail: [email protected]