I'm learning how to perform atmospheric correction manually, hence not using the automated methods available in ENVI

metadata (band 1) for my Landsat 8 scene is as follows;

RADIANCE_MULT_BAND_1 = 1.2344E-02

RADIANCE_ADD_BAND_1 = -61.72016

REFLECTANCE_MULT_BAND_1 = 2.0000E-05

REFLECTANCE_ADD_BAND_1 = -0.100000

SUN_ELEVATION = 48.12237303

So when I use band math to calculate TOA Radiance I type in this:

(0.012344*B1)-61.72016

And it yields a spectral statistics that look kind-of right SpectralStats

When I use any of the following band math formulae;

TOA Reflectance (excluding sun elevation adjustment)

(B1*0.00002)-0.1

TOA Reflectance (including sun elevation adjustment)

((B1*0.00002)-0.1)/0.74457226676389733207607359928648

Or calculate straight from the raw DN to TOA Reflectance (including sun elevation adjustment)

((((0.012344*B1)-61.72016)*0.00002)-0.1)/0.74457226676389733207607359928648

I end up with something like this:

TOA Reflectance

which looks incorrect.

I've googled and either my search terms are rubbish or no one uses manual band math to do this.

[EDIT]

Mikkel's answer below with the revised band math calculation (omitting the intermediate steps) yielded the following statistics:

enter image description here

which look somewhat more realistic.

Wondering if i should add 13.4305 to each band, or ignore the negative minimum?

[END EDIT]

[EDIT #2]

Mikkel was spot on again, in suggesting that the -13.4305 values were isolated to the "no data" areas around the image;

enter image description here

as you can see, isolated only to the northern edge of the scene.

[END EDIT #2]

up vote 7 down vote accepted

One flaw in your approach. You don't need to go through DN to radiance.
You can go straight to the DN to reflectance.
Just stick to ((B1*0.00002)-0.1)/0.74457226676389733207607359928648.

  • this worked pretty well! i suppose either it was never explained that TOA radiance / TOA reflectance excl. sun elevation weren't necessary steps in the correction sequence. – kyuss Oct 12 '15 at 8:09
  • per the edited post above, is it necessary to add 13.4305 to each band to bring them up to min = 0, or is that unnecessary?? – kyuss Oct 12 '15 at 8:16
  • @kyuss , try checking where those -0.134305 are located. I suspect that it is the no-data areas around the edge of the image. – Mikkel Lydholm Rasmussen Oct 12 '15 at 8:19
  • brilliant, i didn't even check anything but the southern borders where there werent negative values. thanks for all of your help. – kyuss Oct 12 '15 at 8:24
  • @kyuss now the next step in your process is "simply" to do the atmospheric correction. – Mikkel Lydholm Rasmussen Oct 12 '15 at 8:34

Here's a good read on the physics of atmospheric correction, and might help with the issue... I'm a button pushing guy, I feel your pain, at least you can just click that beautiful QUAC or FLAASH button once you're outside of academia.

http://modis.gsfc.nasa.gov/data/atbd/atbd_mod08.pdf

We're looking for long answers that provide some explanation and context. Don't just give a one-line answer; explain why your answer is right, ideally with citations. Answers that don't include explanations may be removed.

  • thanks Steven. i have - as you've inferred - learned much during my study but that modis document may well help fill out my understanding gaps re: radiative transfer. – kyuss Oct 12 '15 at 8:18

Your Answer

 

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

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