ABSTRACT
The cameras we discussed so far have spectral selection either in one broad spectral band, usually covering the visible-near infrared (NIR), which is known as the panchromatic band, or in a number of narrow spectral bands at specific thematic locations in the visible-infrared (IR) region, which is referred to as multispectral imaging. Thus, the IRS PAN camera takes imagery in 0.5-0.75 μm, whereas LISS-3 collects data in four spectral bands (0.52-0.59, 0.62-0.68, 0.77-0.86, and 1.55-1.70 μm) within the visible to shortwave infrared (SWIR) spectral region. In multispectral imaging, we are sampling the reflectance spectra of the scene at specific wavelengths. This is adequate for many applications such as identifying the surface features and for generating thematic information. However, when there are certain narrow spectral characteristics in the detected spectra the coarse spectral bandwidth (which averages the spectral response within the band) and the sampling at a few points do not bring out these characteristics. Figure 8.1 gives the reflectance spectra of the mineral halloysite (from the U.S. Geological Survey spectral library) in the 0.3 to approximately 1.8 μm spectral region. Two specific absorption bands can be seen in the spectra. The dotted line in Figure 8.1 shows how the reflectance spectra will look in the 0.52-1.7 μm region, if measured using the four bands of LISS-3. The two absorption peaks are not brought out in this case. That is, the multispectral sensors like LISS, TM, and so on undersample the spectral information content available in the reflectance spectra and, hence, the fine spectral details are missed.
