I'm attempting to calculate a number of vegetation indices using hyperspectral AVIRIS data in ENVI. Some of these indices are normalized, such as NDVI, whereas others are not, such as the Transformed Chlorophyll Absorption Ratio (TCAR).
For the normalized indices such as NDVI, not doing anything to the bands produces an output raster within the valid range. However, in trying to calculate something like TCAR, not doing anything to the bands produces values well outside the valid range.
I have not been able to find anything definitive on when to rescale reflectance data when calculating vegetation indices. Am I correct that all normalized indices can take the unaltered reflectance data, while all other indices must have the reflectance data rescaled by multiplying the individual bands by 0.0001? In particular, is it true that indices that have independent parameters, such as the soil-adjusted vegetation index (SAVI), must also have reflectance data rescaled by 0.0001?