MODSCAG@NSIDC

These images compare Painter's Karakoram scene for 2005154 to the same scene processed at NSIDC using the two component MOD09GA tiles. Although the differences look unbiased, I'm suspicious of the patterns, I wonder if it's possible we used different spectral libraries? Based on the value of the respective center 'SolarZenith_1' pixel of each tile, I used z30 for both tiles.

Note: both images are thresholded at 20%. I know that when I do the "real" processing, I'll need to drop this to 15%. I just left it at 20% for purposes of the comparison.

Tom: what do you think?

Painter's MODSCAG output, Karakoram, 2005154 (June 3)

NSIDC MODSCAG output, Karakoram, 2005154, from MOD09GA tiles

Difference: NSIDC - Painter (ignores differences <= 1%)

Histogram of difference image including small differences

Histogram of difference image ignoring differences <= 1%