Image Processing Reference
In-Depth Information
(Carlson and Boland 1978; Roth et al.
1989
). Nichol (
1996
) recorded a strong
correlation between air measurements and satellite measurements in low wind
environments in Singapore.
The majority of satellite measurements are restricted to the late mornings to
optimize conditions for the visible and near infrared regions. One exception is
NOAA AVHRR which records temperature in the night, but has a coarse resolu-
tion of 1.1 km per pixel at nadir. During mornings, urban materials will slowly
absorb heat. Urban heat islands are best observed in the night when maximum
radiation from urban materials occurs. The timing for satellites is therefore not
optimized for detecting urban heat islands, and perhaps more prone to the observa-
tion of heat sinks because of the heat absorption behavior of urban surfaces
(Nichol
1996
). The fact that satellite measurements are recorded only at particular
times makes the data suitable for comparative studies over long periods. However,
the behavior of urban heat or any environmental parameter during other times of
the day, apart from mornings, cannot be obtained from satellite data. Studies have
shown that the urban heat effect is dynamic over a 24-h period (Roth et al.
1989
;
Jauregui
1997
).
14.5
Comparison of Environmental Conditions in New York
City and Kuwait City
In this section, we present a case study to demonstrate environmental analysis of
two cities using satellite imagery. We illustrate differences in surface properties and
their influence on energy flux by comparing Landsat 7 imagery acquired for New
York City and Kuwait City. Surface temperature distribution and vegetation frac-
tion, or the areal proportion of vegetation within a pixel, distribution can be mea-
sured accurately and provide a synoptic view of spatial variations in environmental
conditions related to albedo, emissivity and evapotranspiration. These spatial varia-
tions are a primary determinant of the environmental con-
ditions that influence human comfort levels. Composite
surface temperature measurements are influenced by the
heterogeneity of materials within the sensor IFOV just as
optical measurements are. While spectrally mixed pixels
can be “unmixed” if spectral endmembers are known, the
same procedure cannot generally be applied to thermal
data as most satellite sensors collect only a single thermal
band and there is no evidence that thermal spectra mix
linearly. Consequently, most analyses of thermal data
assume that the target within the IFOV is thermally homo-
geneous and has uniform emissivity, which is assumed to
be near 1. However, when thermal imagery is used in con-
junction with optical multispectral imagery it is possible to
interpret the distribution of surface temperatures in the
spectral mixture
analysis cannot
generally be
applied to
thermal data as
most satellite
sensors collect
only a single
thermal band and
there is no
evidence that
thermal spectra
mix linearly
