Hi Matteo,

Two methods can be applied.

The first one being correction by dark target subtraction and the second one the application of an atmospheric Radiative transfer Model like SMAC, 6S, MODTRAN and other atmospheric RTF's.

In many image software packages these methods are implemented commonly for Landsat imagery since the last 20 years.

Pick a software package that has the atmospheric correction included and there you go. A small detail when applying an RTF, is that you will need atmospheric variables like aerosol load (OD550), water vapour and ozone mixing ratio's besides the model as well. That's in fact the most difficult data to get hold of, and when they are produced by wild guesses, you correction will be a wild guess as well.

These data are not needed when you perform a dark target subtraction correction. The only thing you need in that case is a dark target (a dense forest or a water surface for example). Without these dark targets in your image, the correction method will not be applicable neither.

CHeers

Frank

Hi Matteo,

Two methods can be applied.

The first one being correction by dark target subtraction and the second one the application of an atmospheric Radiative transfer Model like SMAC, 6S, MODTRAN and other atmospheric RTF's.

In many image software packages these methods are implemented commonly for Landsat imagery since the last 20 years.

Pick a software package that has the atmospheric correction included and there you go. A small detail when applying an RTF, is that you will need atmospheric variables like aerosol load (OD550), water vapour and ozone mixing ratio's besides the model as well. That's in fact the most difficult data to get hold of, and when they are produced by wild guesses, you correction will be a wild guess as well.

These data are not needed when you perform a dark target subtraction correction. The only thing you need in that case is a dark target (a dense forest or a water surface for example). Without these dark targets in your image, the correction method will not be applicable neither.

CHeers

Frank

Hi Matteo, as Frank mentioned, one option is to use a radiative transfer model for the correction of the atmospherical effects over satellite imagery. From the mentioned ones, MODTRAN is a paid product, SMAC (http://www.cesbio.ups-tlse.fr/us/serveurs4.htm) is free and relatively easy to use, but to my knowledge it doesn't have the required coefficients for Landsat 8 products yet. The 6S (second simulation of satellite signal in the solar spectrum) ( 6s.ltdri.org ) developed by Vermote et al. (1997) is free and also relatively simple to use, but as Frank mentioned you need extra variables to account for the atmospheric conditions (which ideally should be measured in the field), however, you can get this information from a set of MODIS products. Aerosol Optical Depth at 550 nm can be obtained from the MOD04_L2 product (Kaufman & Tanré, 1998), O3 concentration can be obtained from MOD07 product (Gao an Kaufman, 1998) and water vapor concentration can be obtained from MOD05 product (Seeman et al., 2002), you should consider that this products are of coarse resolution but are a better option that a wild guess. Aside from this products you will need the relative spectral response of the OLI sensor (on board Landsat 8), which you should be able to get from here: http://landsat.gsfc.nasa.gov/?p=5779 .

Best Regards,

Andrés

Hi Mateo. We've been working on LandCor atmospheric correction framework since 2008 It is a set of Matlab routines that extend the use of 6S onto multiple conditions found within the satellite scene. Within days we will release a standalone version. Please let me know if you'd like to try it out ([email protected]).