Long-term Vegetation Assessment in Wyoming using Landsat Multi Spectral Scanner (MSS) Data

This dataset was used for the purpose of estimating the change in vegetation
amount over time in Wyoming.

Dataset: North American Landscape Characterization Landsat Multi Spectral
Scanner (MSS) Triplicates. Complete coverage for Wyoming, 15 scenes. MSS has 79
m by 79m instantaneous field of view, data were resampled to 60m by 60m, and
geometrically corrected to a UTM projection 1:100,000 map scale. The aerial
extent of each scene is 185 km swath by 170 km along-track; however, orbit
differences between Landsats 1, 2 and 3 and Landsats 4 and 5 reduce the size of
each scene. Also available as a dataset is change of NDVI (see below). Images
and change in NDVI are in georeferenced TIFF format. Changes corrected for
precipitation variability will be available in the future.

What the dataset was used for: to estimate the change in vegetation amount over
time in Wyoming. We used three Landsat Multi Spectral Scanner (MSS) images, one
from the early-to-mid 1970s, one from the mid 1980s, and one from the early
1990s. Vegetation amount is determined from the Normalized Difference
Vegetation Index, defined as (Band 4 - Band 2)/(Band 4 + Band 2), where Band 4
covers near-infrared (NIR) wavelengths from 0.8 to 1.1 Fm and Band 2 covers red
wavelengths from 0.6 to 0.7 Fm. NDVI saturates with one-sided leaf area indices
(LAI) of 4 and higher; however, this is not a problem for Wyoming grasslands
and shrublands have low LAI. NDVI varies from -1 to 1, and are converted into
digital numbers (DN) by: 100 (NDVI+1), so the DN varies from 0 to 200.

Change detection is estimated as the NDVI(1990) - 0.5[NDVI(1980) - NDVI(1970)]
for each image, except for areas covered with snow, clouds, and shadows caused
by clouds. If two dates were available for these areas, we subtracted the NDVI
of the earlier date from the NDVI of the later date. The dates of acquisition
were usually collected towards the end of the growing season, and usually
during years of drought (for example: 1992, 1985, 1972). We have developed an
algorithm using the Advanced Very High Resolution Radiometer (AVHRR) data from
the EROS Data Center to adjust NDVIs to average precipitation amounts, but
these corrections were not applied to this dataset.