Southwest U.S. Change Detection Images from the EROS Data Center
Reno and Lake Tahoe, Nevada
by Kristi Sayler, U.S. Geological Survey
A matched pair of satellite images from different years can be processed
into a new change-detection image, which shows areas of increased or
decreased brightness, greenness, or wetness. These images can be used to
monitor landscape changes. These images of the Reno-Lake Tahoe area of
Nevada, show some effects of natural processes on the landscape:
- Burn area
- Extensive changes resulting from forest fires can be observed in
satellite images. Changes in greenness, displayed here as a dark blue
color, show an area of vegetation which has become blackened from burning.
- Conifer mortality
- Scientists at Boston University have made ground measurements to
quantify the decrease in the forest canopy as a result of drought. The
sustained drought conditions weakened the trees and made them more
susceptible to insect attacks. Satellite images can be used to estimate
the geographic extent of the damaged area.
- Forest clear-cutting
- Droughts increase the geographic extent of exposed lakebeds, where new
vegetation can be seen using satellite images. Changes in land use or
timber harvesting activities can also be seen in this forested area.
- Dry lake
- These images were taken at the beginning and culmination of drought
conditions in the western U.S. A broad shallow lake that was full of
water in 1986 was empty, and the entire lake bed exposed, in 1992.
|Reno/Lake Tahoe 8-5-1986||Reno/Lake Tahoe 8-5-1992|
|Change from 1986 to 1992
How these images were produced
The change detection contained in the following pages is done using Landsat
multispectral scanner (MSS) and thematic mapper (TM) data to identify and
characterize landscape change. The change analysis procedure involves:
- co-registering multiple dates of MSS or TM data,
- transforming the imagery to scene-based measures of brightness, greenness,
and wetness (TM data only),
- pairwise differencing of the brightness, greenness, and wetness measures
to compute change vectors (magnitude and direction) for each image pixel,
- encoding the change vectors using hue, saturation, and value (HSV)
for visualization, and
- formulating a signal-to-noise model by which to isolate areas of significant
The color key, shown here, is the model used to encode the calculated
change vectors produced for each pair of images. It can be used to interpret
the change images contained within the following web pages. An increase in
greenness is shown as green and a decrease in greenness is shown as magenta.
A decrease in brightness is shown as various shades of blue and an increase
in brightness is shown as orange/red tones. Any combination of brightness and
greenness changes will be shown somewhere on the color key. Shades of grey
imply no change or change that was not considered significant.
For a more detailed description of the techniques or algorithmns used consult
the following conference proceedings article:
Dwyer, J., Sayler, K., and Zylstra, G. 1996. Landsat pathfinder data sets for
landscape change analysis. Proceedings of International Geoscience and Remote
Sensing Symposium. Lincoln, Nebraska, May 27-31, 1996.