I’m working with S2 tiles processed with the latest Sen2Cor. Command line processing seems to work very well and I get no error messages. The tiles are taken at very different times of day and different latitudes. One tile was acquired in much lower light conditions. How good is Sen2Cor at correcting for this effect? If there any documentation that could help me assess the potential scale of any differences? Can we expect that, post-Sen2Cor, the same land feature type will have a corrected reflectance at the same intensity in 2 tiles acquired at different times of day?
taken from a syn-synchronous satellite, the local sensing times should in general not be too different unless you combine ascending and descending orbit (which are off by about 12hours).
one of the goals with deriving bottom-of-atmosphere reflectances is indeed to correct for such illumination effects. How good that works depends quite a bit on the local conditions, e.g. the reflection properties of the surface (sen2cor assumes lambertian; any stronger deviations from that model will cause illumination dependent deviations), slope gradient and slope aspect of the terrain (I think, to some extend this is taken into account by sen2cor when you apply a DEM in the processing), etc.
you could test that out yourself a little by, e.g., comparing Sen2Cor results from data of the same location but taken from different relative orbits (i.e. in regions where swaths from different orbits overlap), hence are exposed to different illumination conditions.
as I just stumbled upon this, for ATCOR (which was the starting point to develop Sen2Cor from as I understand) the following is stated:
" For near nadir view angles (off-nadir angle < 10 degree), a flat terrain, and avoiding the specular and backscattering regions, an accuracy of the retrieval of surface reflectance of +/-0.02 (reflectance < 0.10) and +/-0.04 (reflectance > 0.40) is possible."
I’d expect that Sen2Cor is as good as that.
The atmospheric correction codes rely on radiative transfer codes and part of their accuracy is related to the accuracy of these codes. These codes suppose that the atmosphere is made of parallel and plane layers. This assumption results in large errors above solar angles of 70 degrees. Moreover, with such angles, directional effects (the surface reflectances depend on solar and viewing directions) are also very large.
As a result I would not trust much the surface reflectance obtained under low light conditions.