I'm using Justin Braaten's excellent "Landsat ETM+ to OLI Harmonization" tutorial (https://code.earthengine.google.com/798ae7a268f8e2c8022433b9562785c0) to combine imagery from different Landsat satellites. Currently, y-intercept values for the regression coefficients are multiplied by 10,000 to match the scaling of USGS Landsat collection 1 surface reflectance data at line 25 and between lines 220 to 230 . With collection 2, the scale factor has changed to a multiplier and offset (0.0000275 + -0.2). I can't seem to determine the unscaled surface reflectance by inverting this process.
1 Answer
I would use that gain and offset to convert the scaled refelctances to actual reflectances [0,1] then directly apply the equations from Roy et al.