I have calculated at sensor reflectance for a multispectral image and performed atmospheric correction. The reflectance values for the visible and NIR bands are between 0.2 to 0.01 for surfaces, includng some vegetated areas.

Are reflectance values typically this low for multispectral imagery?

Ideally I'm trying to differentiate between types of vegetation and my spectral reflectance curves are almost flat. When I consult remote sensing books/papers I can see spectral reflectance curves that show a distinct difference between for example dry grass and forest.

  • Can you give some more details about your atmospheric correction method? Is this happening on multiple images or just one?
    – Erica
    Sep 19, 2014 at 14:36
  • 1
    I preformed DOS,however the at sensor reflectance values where already quite low in the 0.2 - 0.001 range, I did this for multiple images
    – Barry
    Sep 19, 2014 at 14:43
  • What sort of surface veg? Can you calculate NDVI on that and report to us what those values are. Because NDVI is dimensionless, it should cancel out any peculiar units which have made their way into your calculations.
    – BillyWill
    Sep 20, 2014 at 20:52
  • Grasslands and crops, for my NDVI I get a range between 0.3 to 0.02
    – Barry
    Sep 22, 2014 at 19:17
  • I suspect you have missed a scaling factor in the original data that needs to be applied. We would need the source of one of your images and metadata. It could additionally be a incorrect function in the correction but can you point us to the image and the associated data. Dec 15, 2015 at 19:16

1 Answer 1


Those values are definitely too low: green vegetation should be around 0.5 in the NIR.

In your case, my first advice would be to work without atmospheric correction (not necessary if you have a single scene). So you do not risk to loose radiometric precision.

Also DOS might not work because your dark object is not dark enough (e.g. glint effect on a water body). Have a look at MACCS for SPOT and LEDAPS for Landsat.

UPDATE: As mentioned by O. Hagolle, there is a new version of MACCS with improved atmospheric corrections, which is available for non commercial use on RedHat and Cent OS versions 6 and 7 (at the time of writing)


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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