Reflectance to Radiance Sentinel-2 L1C Conversion Issue

Hi All,

I believe that I found a problem with the Reflectance to Radiance conversion for Sentinel-2 L1C data. The sun zenith angles do not appear to be interpolated correctly across the output radiance image (referencing the MTD_TL.xml file sun zenith value table). The upper left corner of the output radiance image looks correct, but errors build from there across the output image and get larger towards the lower right.

I utilized the conversion equation that I found in the source code, rad = rho * cos(radians(sza)) * Es * sc / (pi * d2), to solve for the solar zenith angle (sza) and back calculate the angle for every pixel using a reflectance L1C image and associated radiance image output from the s2tbx Reflectance to Radiance conversion. The calculated angle image had a much smaller angle range than the associated xml table.

I’d appreciate help in determining if this issue is limited to the dataset that I was using (S2B_MSIL1C_20201009T172139_N0209_R012_T14TPQ_20201009T204621, T14TPQ_20201009T172139_B03.jp2), is a general issue with the Reflectance to Radiance conversion, or if there is a setting (or something else) that I am missing. Thanks for your assistance!