Vegetation corrected continuum depths at 2.20 µm: An approach for hyperspectral sensors

Remotely sensing instruments, both airborne and spaceborne, of sufficient spectral resolution, can be used to identify absorptions that occur at 2.20 µm, which are indicative of AlOH minerals. In a pixel containing green and/or dry vegetation the depth of the AlOH feature at 2.20 µm is decreased. Since the depth of the feature at 2.20 µm is an indicator of the AlOH content it is desirable to correct the depth of the 2.20 µm feature in such a manner as to remove, or negate, the obscuring effect of the vegetation. This is achieved by using a multiple linear regression model where the coefficients of the linear model are produced via forward modeling, and where the independent variables are continuum removed band depth (CRBD) that are used to detect the presence of green and dry vegetation and the uncorrected AlOH CRBD. The proposed vegetation corrected continuum depth (VCCD) method was tested with synthetic datasets as well as hyperspectral data (HyMap) collected at Mount Isa in Queensland, Australia. The results of using the VCCD method on the uncorrected HyMap data were validated with vegetation free samples collected from the Mount Isa region. Improvements in the R squared statistics of the corrected 2.20 µm CRBD to the vegetation free CRBD, after application of the VCCD, were found to be 2–4 times greater than the uncorrected 2.20 µm CRBD. Visual inspection of large survey areas demonstrated that the uncorrected CRBD in vegetated areas were lower, and did not match adjacent vegetation free areas, and also produced false positives of high AlOH content.