on the ground our study confirms that cloud-free composited DMSP nighttime lights overestimated
the area of lighting on the ground. This overestimation was the result of a combination of factors,
including (1) the large OLS pixel size; (2) the OLS's capability to detect subpixel light sources;
(3) overlap in the IFOV footprints of adjacent pixels, resulting in multiple pixel detections from
subpixel-sized lights; and (4) geolocation errors. These effects, present in data from single obser-
vations, were accumulated during the time-series analysis.
Three other mechanisms that may have enlarged OLS lights beyond the extent of surface lighting
under certain conditions were not explicitly explored in the current study. One was the scattering
of light in the atmosphere as it was transmitted from the Earth's surface to space. The second was
the reflection of lights off surface waves in cases where bright city lights were adjacent to water
bodies. The third possible mechanism was the detection of terrain illuminated by downward-
scattered light arising from very bright urban centers or gas flares.
Imhoff et al.
(1997) developed thresholding techniques to accurately map urban areas. The
disadvantage of these techniques is that they eliminate lights from small towns owing to their low
frequency of detection. We believe that it would be possible to reduce the overestimation of the
area of lighting based on an empirical calibration to the extent of surface lighting or via the
modulation transfer function (MTF) of the OLS nighttime visible band imagery.
The Defense Meteorological Satellite Program (DMSP) Operational Linescan System (OLS)
has a unique low-light imaging capability developed for the detection of moonlit clouds. In addition
to moonlit clouds, the OLS also detects lights from human settlements, fires, gas flares, heavily lit
fishing boats, lightning, and the aurora. Because all these lights are detected in a single spectral
band, and to remove the effects of cloud cover, time-series compositing is used to make stable light
products that depict the location and area of persistent light sources. This compositing is done
using data collected on nights with low lunar illumination to avoid the detection of moonlit clouds
and the lower number of lights detected due to the OLS gain settings during periods of high lunar
illumination. A number of studies have found that these stable lights products overestimate the size
of light sources present on the Earth's surface. This overestimation is due to a combination of
factors: the large OLS pixel size, the OLS's capability to detect subpixel light sources, and
geolocation errors. These effects, present in data from single observations, are accumulated during
the compositing process.
The NOAA NESDIS, Ocean Remote Sensing Research Program, funded this project. Informa-
tion Integration and Imaging LLC, Fort Collins, Colorado provided the Landsat imagery used in
this study imagery.
Elvidge, C.D., K.E. Baugh, J.B. Dietz, T. Bland, P.C. Sutton, and H.W. Kroehl, Radiance calibration of DMSP-
OLS low-light imaging data of human settlements,
68, 77-88, 1999.
Elvidge, C.D., K.E. Baugh, E.A. Kihn, H.W. Kroehl, and E.R. Davis, Mapping of city lights using DMSP
Operational Linescan System data,
Remote Sens. Environ.,
Photogram. Eng. Remote Sens.
, 63, 727-734, 1997.