3

I am dealing with Sentinel-2 (Level 1C) images over sea ice. I have downloaded some and was wondering: When I look e.g. at Band 8A, there are some images where the reflectance over the water between the ice has realistic values of 0.4 (this is at the same time the lowest reflectance value in this image). But in other images, the whole water has a reflectance of almost zero, partially even exactly zero which is a rather unrealistic reflectance for water.

How can this be and is there a possibility to correct the low values?

Improve this question
6
  • rToa = (float)(DN_L1C_band / QUANTIFICATION_VALUE); which QUANTIFICATION_VALUE did you use?
    – pLumo
    Apr 26, 2017 at 13:39
  • I used 10000 everywhere
    – julius
    Apr 26, 2017 at 13:49
  • That might be the problem, the value is not fixed at 10.000 and could have changed. Try to read the value from the Metadata.
    – pLumo
    Apr 26, 2017 at 14:11
  • I check this already. For all the images the Metadata say the value is 10000
    – julius
    Apr 26, 2017 at 14:20
  • Okay sorry. If no one can help you here yoou might want to ask in the Step forum (forum.step.esa.int)
    – pLumo
    Apr 26, 2017 at 14:24

2 Answers 2

Reset to default
4

I quite disagree with you regarding the standard reflectance values of water in the near infrared (Band 8a). As water absorbs light in these wavelengths, these values should be very low (of course not zero) except when the water is turbid or contains ice or foam, or is observed in the sunglint direction..

As a result, the unusual value, for me, is 0.4, most probably for one of the causes listed above.

Improve this answer
2
  • Thanks, this sounds helpful. The image with the minimum 0.4 value has also some clouds, could this affect the value? And is there a way to correct it in order to have a uniform range of values over several pictures?
    – julius
    Apr 27, 2017 at 9:00
  • Of course it could be clouds. It is very difficult to correct for clouds, because they are not homogeneous, and it is hard to separate them from the surface. There is an exception for cirrus clouds as mentioned by Aaron, even if it does not always do miracles, and does not correct for the shadows beneath the cirrus.
    – O. Hagolle
    Apr 28, 2017 at 16:07
3

You are using the level 1C product (top of atmosphere reflectance), which will, in many cases, have atmospheric scattering present. Atmospheric scattering will skew the true surface reflectance value due to atmospheric aerosols in the image (Figure 1). Additionally, atmospheric scattering makes it difficult to compare different images because aerosol levels may be different across images.

Theoretically, electromagnetic radiation in the near-infrared spectrum should exhibit complete absorptance in water, where bottom of atmosphere reflectance values over clear, deep water bodies should be 0. Any deviation from 0 will be due to factors such as algae growth, subsurface water conditions, sensor calibration issues, or atmospheric scattering.

To rectify this situation, I would recommend calibrating your top of atmosphere level 1C products to bottom of atmosphere level 2A products. You can do this with the Sentinel Application Platform (SNAP) Sentinel-2 Toolbox and the sen2cor processor

enter image description here

Figure 1: TOA Level-1C image data (left) and associated BOA image data (right) generated by the Sentinel-2 Toolbox

Improve this answer

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.