If I am using two different sensors e.g. Landsat-TM and Landsat-OLI for Iteratively Reweighted - Multivariate Alteration Detection (IR-MAD) change detection, is radiometric normalization of two data necessary? The IR-MAD literature (Nielsen A. A. 2007) states the method is invariant to gain and offset setting difference of the sensors as well as different atmospheric corrections applied.