Image Processing Reference
In-Depth Information
types were targeted: green grass, trees and shrubs, three classes of impervious, soil/
dry vegetation, and water. Water was separately treated, first by identifying all
water features, then masking them out of the terrestrial classification.
The green grass component is usually found in golf courses, baseball fields,
cemeteries, parks, and some residences. Green pastures and alfalfa fields are also
included in this class, representing peripheral active irrigated farmlands. The trees/
shrubs vegetation class is found in mountain foothills, residential areas, and parks.
The bright impervious class includes materials of high reflectance such as rooftops
typically of commercial/industrial buildings. The medium impervious class
includes many rooftops as well as concrete and weathered asphalt common in
commercial/industrial areas as well as residential areas. Dark impervious surfaces
of low reflectance include recently placed asphalt on roads and parking lots, com-
mon in commercial and industrial areas. This class also includes shadowed areas
occurring in all areas but especially in high rise downtown commercial and apart-
ment districts. In the V-I-S concept it is important to separate soil from impervious
surfaces because of their very different contributions to environmental processes
(see Fig.
6.1
). In this case, dry vegetation is included in the soil class as this is the
common condition of soil in the study area, and does not alter the energy/moisture
regime appreciably.
The six fractions show significant change across Salt Lake Valley in the 27-year
time span. Farm fields are substantially reduced under urban sprawl. Trees and shrubs
increased significantly in fringe area subdivisions during the period. Impervious
surfaces show an expansion of the downtown district, as well as an extensive
expansion of large industrial tracts to the west, replacing arid land soil and dry
vegetation. Many shopping centers also emerged in the outlying suburban and farm
areas undergoing transition. Through regression analysis, accuracies are expressed
in correlation coefficients. Most of the six classes show significance at the 0.01
level, and the rest at 0.05.
Figure
6.17
shows change over time for the three principal V-I-S components
in a linear-time plot (1972-1999), with soil diminishing while impervious sur-
face increases from 40% to 65%. Vegetation also increases for a time, and then
weakens. These data represent the University of Utah campus, as the medical
center complex and research park extend up into barren benches and foothills.
Figure
6.18
shows three methods of displaying change over time for an alfalfa
farm field being converted to a residential subdivision. From 1972, the green field
was cleared in preparation for urban development. In the upper chart, vegetation
drops sharply from near 80% to 25% in the first 3-year interval and then to 16% by
1981. From that point on, the balance between soil and impervious surface is inter-
changed as a sub-division takes shape. The last stage (1996-1999) shows vegeta-
tion beginning to rise as landscaping commences. The other two plots show the
same trend over the 27-year period. The lower left displays the same transition in a
Red/NIR feature space plot as a field of green vegetation suddenly approaches the
non-vegetation line and then proceeds generally toward the bright point to the right
as rooftops and concrete increase. The ternary V-I-S plot illustrates the same land
conversion. In due time as landscaping matures, a transition toward the vegetation
corner will ensue.
