Geology Reference
In-Depth Information
(b)
(a)
A
B
C
E
53°
85 GHz
37 GHz
19 GHz
Figure 7.5
Configuration of (a) a cross‐track scanning radiometer and (b) a conical scanning radiometer.
problem can be rectified by moving the lenses an infini-
tesimally small distance along the satellite track line in
order to account for its forward orbital motion. This
compensation can work only in the case of optical sys-
tems such as multispectral scanners. It does not work in
the case of microwave radiometers because of the large
aperture of the antenna. As a result microwave antenna
of cross‐track scanners usually scan successive lines in
one direction only. Alternatively, a conical scanner can be
used. These scanners are commonly used in space‐borne
passive microwave as well as TIR remote sensing systems.
It was used, for example, in the TIR channels of the
Along Track Scanning Radiometer (ATSR) onboard the
European remote sensing (ERS) satellites. A conical
scanner views the surface while looking forward or back-
ward at a constant incidence angle. The parameters in
Figure 7.5b pertain to the special sensor microwave
imager (SSM/I), which had four radiometric channels
operating in frequencies around 19, 22, 37, and 85 GHz.
More details about this sensor are presented in the next
section. An observation represents radiation integrated
over a specific “footprint” of the sensor, which is also
called the integrated field of view (IFOV). Its dimensions
are determined by the angular resolution based on 3 dB
beam width of the antenna pattern projected on the
Earth's surface. Scanning the scene at a fixed incidence
angle eliminates the effect of changing incidence angle on
the measured radiation. However, conical scanning
causes the IFOV to overlap more toward the end of the
swath. This point will be explained later.
The IFOV determines the spatial resolution of the sen-
sor on the ground, but the pixel size in the imagery data
is determined by the sampling rate of the observed radia-
tion. If the sampling takes place at an interval that corre-
sponds to ground distance shorter than the characteristic
dimension of the IFOV, the “pixel spacing” becomes
smaller than the IFOV and successive samples from adja-
cent pixels overlap (oversampling). This is different from
the overlapping of the IFOV near the end of the swath
due to conical scanning as mentioned before. This situa-
tion also arises when the data are gridded with grid spac-
ing less than the IFOV dimensions. In this case information
of emitted radiation from each IFOV is distributed over
an area of a few grid cells. The most authentic radiometric
information is found in swath (nongridded) data or when
the sampling rate matches the footprint of the sensor.
Quantitative retrieval of information from remote
sensing data requires radiometric calibration of the
imagery data. This entails converting the pixel value in a
given set of imagery data to the actual radiometric value
measured by the sensor or to a number proportional to
that value. The first process is referred to as absolute cali-
bration and the second is called relative calibration.
Absolute calibration accounts for sensor parameters such
as gain and offset, illumination and viewing geometries,
angular pattern of the illuminated signal, power spread
of the illuminated signal (in case of the radar sensor),
and finally the absorption and scattering by the atmos-
phere. This requires mathematical modeling to simulate
the observations. Remote sensing imagery data provided
to users are usually not absolutely calibrated. However,
absolute calibration can be performed using software
available in remote sensing image processing packages.
Relative calibration, on the other hand, is a simpler task
that aims at normalizing data from multiple passes of the
same satellite or multiple sensors on different satellites. It
usually involves adjusting the observation to match the
radiation from a “calibration” ground target of known
reflectance or emission, which is considered to be con-
stant over time. The adjustment is achieved by regressing
