Sentinel 2 natural colours mismatch

Optical Toolbox
1

Hello, I hope this is the right place to ask this question.

I have two sentinel 2 products from the same region. But In one I can’t get the RGB natural colours to look realistic.
The two products are:

S2A_OPER_MTD_SAFL1C_PDMC_20160908T063241_R031_V20160907T011652_20160907T011651
Which consists of multiple tiles, I can’t get the realistic colour working on this one (it’s the larger of the two)
And
S2A_MSIL1C_20170611T010651_N0205_R131_T57WWS_20170611T010649

Both are from near Chersky in north east Russia. Thus with the evergreen forests and tundra the region should be a lot greener and far less red/pink. The difference in the lakes can be attributed to the difference between frozen and thawed lakes. Both images show the 4,3,2 RGB bands in snaps default natural colours setup.
Both are shown in the image below (I projected the smaller on top of the larger):

mismatch.jpg1200×1636 499 KB

I tried adjusting the histograms but have not been able to really improve on it. I read that large bodies of water could “mess” with the histograms but I’m not sure how to correct for that. Am I doing something wrong, or is there a reason this larger product is not suitable for natural colour representation?

Thanks,

Martijn

2

Hi @mpallandt

I have looked at the QI data from both images on the SciHub. This confirms that there is no issue in data quality (no Reports generated and the QUANTIFICATION_VALUE of both products is the nominal 10000).

So I would suspect the rendition of the ‘larger’ image ( a multitile product of 17 Tiles and~7GB size) by SNAP is the issue. Of course, you’ve managed to identify a relatively cloud-free (Cloud coverage is 0.2%) image as a comparitor :slight_smile: The single Tile product (T57WWS) contains ice, and - thus the opposite behaviour that occurs with larger dark areas (such as open water) affecting the rendition of visible bands in RGB the presence of highly reflective surfaces such as ice, high cloud and snow will make the surrounding surfaces (in this case tundra) darker.

Anyway, I’ll download the multitile and take a look.

Cheers

Jan

S2 MPC/CC Operations Manager

3

Hello Jan,

Thanks for answering my question, now I know its just the representation and with your explanation of the effect of ice this is something I should be able to figure out eventually.

Thanks for your time and the great tool,

Best wishes,
Martijn