Testing the Referencing Method on different cover type

Figure 40: are paired DMSV frames for a test site near Alice Springs before (LEFT) and after (RIGHT) correction from which land cover types were extracted for statistical comparison with increasing filter sizes. The following graphs demonstrate the relative effect of the referencing method for each spectral band.

Figure 41: shows the effect in the overlapping areas of contiguous scenes of enlarging the filter size on eight different cover types for DMSV Band 1. Position 1 is the overlap difference with the raw uncorrected data. Positions 2-6 indicate the relative improvement (or other) by increasing the filter. Note the maximum and minimum is for the trees and shrubs.

Figure 42: shows the effect in the overlapping areas of contiguous scenes of increasing the filter size on eight different cover types for DMSV Band 2. Position 1 is the overlap difference with the raw uncorrected data. Positions 2-6 indicate the relative improvement (or other) by increasing the filter. The divergence of trees and shrubs (likely to have the largest variability in shadow) is worth noting.

Figure 43: shows the effect in the overlapping areas of contiguous scenes of increasing the filter size on eight different cover types for DMSV Band 3 Position 1 is the overlap difference with the raw uncorrected data. Positions 2-6 indicate the relative improvement (or other) by increasing the filter

Figure 44: shows the effect in the overlapping areas of contiguous scenes of increasing the filter size on eight different cover types for DMSV Band 4. Position 1 is the overlap difference with the raw uncorrected data. Positions 2-6 indicate the relative improvement (or other) by increasing the filter.