I am currently studying the Sen2Cor propagation of the S2 L1C uncertainty.
I am currently studying a small 2x2km area centered at Barrax (Spain), (lat/lon,39.049949, -2.099900)
I use the product S2A_MSIL1C_20180802T105621_N0206_R094_T30SWJ_20180802T132133.SAFE and modify the JPEG2000 images at each iteration with a sample error out of the uncertainty distribution. I use Sen2Cor 2.8 and only modified the altitude to 0.7km in the L2A_GIPP.xml.
Results are as expected wit the exception of the AOT retrieval.
The retrieval is identical and centered at the expected AOT value. However, there is a constant outlier result at ~0.184. I have identified that this is the result of lower L1C reflectance values and that the default VIS=40km is not updated.The tile has dark and water features but at a small percentage and the log file indicates that the normal DDV algorithm fails.
Any ideas why this outlier occurs? why the VIS is not updated?
Thanks in advance