Are Sentinel 2 images (level 2A) statistical comparabel?

Hello there,

I have two images (29.09.2017 (image A)) and (16.10.2017(image B)).
My goal is two compare some forest areas. (before and after a storm).
After calculating the NDVI via SNAP (optical>thermatic landprocessing>vegetation radiometric indices>NDVI Processor).
I receive strange results it turns out the NDVI is higher after the storm then before.

I did one preprocessing step atmospheric corection with Sen2Cor (third party Plugin).
Is any other preprocessing step necessary to receive comparable data?

For image A i have a black corner in the lower right corner (see picture bellow). I saw that for many Sentinel 2 images on the open HUB. Does this affect the pixel values or any other parameter of the image data.

how did you compare the images and see that the NDVI is higher after the storm?
The black edges don’t influence the reflectance of the pixels but maybe your histogram is shifted because the image statistics include more zeroes.

I used the statistic function of SNAP. All values (max, average…) turned out to be higher afterwards.

I imported Polygons (esri shapefile) of the areas i want to compare.