Topographic corrections, and BREFCOR

[rese_support]

We have a few questions about the functionality related to the (i) topographic corrections, and (ii) BREFCOR:
1. Topographic correction - As the terrain for our study areas is rugged, we used the ATCOR 3 module from the ATCOR menu (pre-defined sensor, as Landsat-5 and Landsat-7 data were used). Our main question regarding the topographic correction is that it appears to only partially correct for topography.
The correction looked good, but when we zoomed into other areas, we noticed that much of the topographic shadows remained, leading to the observation that the correction only worked partially.

The topographic correction is very sensitive to the geometric accuracy of both, the imagery and the DEM data. Offsets of a few pixels may lead to visible mis-corrections. So, you'd first need to check the accuracy of your inputs. If overcorrections become visible, you'd need to use the incidence BRDF correction for the topographic effects, which are selectable in the processing panels. Default settings are normally fine. 

2. BREFCOR – We had some questions regarding BREFCOR. We note that there are 2 types of BRDF corrections available in BREFCOR, namely model-based and empirical. When we tried using empirical, we repeatedly got an error (unfortunately, I did not save the error message), so we had to use the model-based option. Below, however, we can see that the BREFCOR output looked identical to the original output from ATCOR 3, although we believed that the BREFCOR output would have also been corrected for nadir-related BRDF effects. However, when we examined the pixel values of the BREFCOR outputs, they were also the same as the ATCOR 3 output, indicating that no correction had been done.

BREFCOR for satellites corrects for BRDF-induced differences between varying observation angles. To make it work, you'd need anumber of scenes (at least about 8) with different observation angles. There is no correction done within image as for small FOV imagery the observation angle is almost constant (as you saw in your image). I'll need to check regarding the error you encountered, however.

3. Nadir normalization – Since the BREFCOR results were identical to the original ATCOR 3 outputs, we also tried adjusting for BRDF by doing nadir normalization (which we assumed was a part of BREFCOR). While we got the nadir normalization to work, the results look like the following, but with the reflectance not varying smoothly across track.

Here, the same logic applies: Nadir normalization makes only sense for imagery with FOV's larger than about 10 degrees. The landsat 8 data should not show across track variations due to its small FOV and thus, it can't be corrected.