Image Processing Reference
In-Depth Information
Fig. 6.16
Trend of V-I-S from pre-earthquake to post-earthquake conditions, Izmit, Turkey (Kaya
et al.
2004
)
V-I-S model. SPOT-X (20m) data from 25 July 1999, and 4 October 1999, were
geometrically corrected to 26 ground control points and transformed to UMT
coordinates. Using the ISODATA (Interactive Self Organizing Data) classifier, 25
spectral classes were combined to form four classes: vegetation, impervious I (urban),
impervious II (collapsed buildings), and soil (neither vegetation nor urban). The
concept adopted by the investigators is that the collapsed buildings no longer functioned
as urban, where the term, impervious, implies buildings and roads with urban func-
tions. Rather they are treated as rubble and grouped with the soil class.
From the SPOT-X classification there was a slight loss in vegetation (−1.6%)
and a greater loss in impervious surface (−8.1%), and a significant gain in soil with
the addition of collapsed rubble (+9.7%). Some 7.1% of the original impervious
surface was reduced to “non-urban” function and added to the soil class. Figure
6.16
displays the results plotted on the V-I-S ternary diagram.
6.6.2
MSS/TM/ETM+ Data: Urban Growth, Expert System
Analysis
Hung (
2003
) employed Landsat MSS, TM, and ETM+ data to monitor land cover
change in Salt Lake Valley, Utah, 1972-1999. He used an expert system developed
earlier (Hung and Ridd
2002
) to determine sub-pixel composition. Seven cover
