The DMSP group at NGDC released version one of a pair of DMSP-OLS "nighttime lights of the world" images processed
specifically for the detection of change, covering the years 1992-93 and 2000. The images are in ENVI and Geotiff formats.
Our lab began work on methods and algorithms for producing high quality nighttime lights for use in change detection in early 2001. The digital
OLS archive extends to mid-1992 but has numerous gaps in the first year. Therefore we combined the 1992 and 1993 data for the early
product. The OLS detects lights from human settlements, fires, gas flares, and heavily lit boats (primarily squid fishing boats). We have
separated the four types of lights as best we could based on location, brightness / persistence, and visual appearance. In addition to these four
global cloud free products we also provide key intermediate products: number of valid coverages, number of cloud free coverages, number of
cloud free light detections, and the average DN of the cloud free light detections.
There are still some problems with the data. One problem that cannot be address is the saturation of the visible band in the bright cores of urban
centers. This is a characteristics of the operational OLS data, which forms the bulk of the archive. It is difficult using OLS alone to adequately
separate gas flares from human settlements. We are investigating the use of persistent MODIS fire detections as a guide for locating gas flares in
the OLS data. Likewise, the separation of city lights and fires is not entirely clean in some parts of the world with extensive fires. The other major
problem we note in the current product set are the effects of snow on the extent and brightness of the lights, which can be found in the northern
hemisphere above ~40 degrees latitude. A good place to look for the effects of snow is in the Great Lakes region of the USA. We plan to
reprocess the products in the next several months using an improved cloud detection algorithm and screen for snow using DMSP SSMI data.
Preliminary cloud-free snow-free product sets indicate that snow effects will be reduced in the next release.
Our plan is to process the entire digital OLS time series into annual nighttime lights composites and to develop a cross calibration for the four
OLS' that have produced nighttime data for the archive to date (on F-10, F-12, F-14, and F-15) using the prelaunch sensor characterizations
and empirical results from nights where lights were observed by multiple (2-3) OLS'.
The products released contain a number of interesting phenomena resulting from the expansion (and contraction) or lighting. Below are
four color nighttime lights change images produced from the released data.
Nighttime lights change images:
(The year 2000 compared with 1992-93)
The above images can be interpreted using the following key:
Cyan | background - no lights and offshore (land / sea mask applied). |
Black | bright lights detected in both time periods (at or near saturation). |
Red | Lights much brighter in 2000. |
Yellow | New lights in 2000. |
Light gray | Dim lighting detected in both time periods - little change in brightness. |
Blue | Lights dimmer or missing in 2000 (relative to 1992-93). |
Download 1992-1993:
Warning: Disk space required for compressed data is
~800 MB and uncompressed data is
~12 GB.
Download 2000:
Warning: Disk space required for compressed data is
~800 MB and uncompressed data is
~12 GB.
Each tar ball contains the data file, an associated header file, four jpeg images and a readme file. The data files have been zipped with gzip.