Symposium on Remote Sensing for Resources Development and Environmental Management / Enschede / August 1986 
A simple atmospheric correction algorithm for Landsat Thematic 
Mapper satellite images 
P.I.G.M.Vanouplines 
Royal Museum of Central Africa, Tervuren, Belgium 
ABSTRACT: An atmospheric correction algorithm for Landsat Thematic is developed on the basis of an algorithm 
proposed by Sturm. The algorithm needs only one meteorological variable, the horizontal visibility or meteoro 
logical range. The algorithm is tested with a sensitivity analysis, and shows to be well applicable in regions 
where the atmospheric conditions are stable, and well measured at observatories and airports. 
RESUME: Un algorithme pour la correction atmosphérique des données Landsat Thematic Mapper est développé, basé 
sur un algorithme proposé par Sturm. Il ne fait usage que d'une seule variable météorologique, la visibilité 
horizontale. L'algorithme est testé avec une analyse de sensibilité, ce qui montre son applicabilité dans des 
régions où les conditions atmosphériques sont stables et bien mesurées dans les observatoires ou les aéroports. 
INTRODUCTION 
For many applications in water quality research on 
oceans and estuaria it is necessary to apply an at 
mospheric correction on the radiation received by the 
satellites. Such atmospheric corrections were develo 
ped and applied for the Nimbus-7 Coastal Zone Color 
Scanner (CZCS) and for some other satellites. 
The disadvantage of these atmospheric corrections is 
that generally many meteorological variables should be 
known. If one obtains a satellite image, taken some 
months or years ago, it is often difficult to retrieve 
these variables. Therefore it should be interesting to 
have an atmospheric correction algorithm that needs 
only meteorological variables which are measured on a 
regular basis. In that case one has to retrieve the 
data from existing observation series in order to 
apply a more or less reliable atmospheric correction 
on a given satellite image. 
Sturm's paper (1981) describes such a "simple" 
atmospheric correction model for oceanographic appli 
cations. The algorithm is called simple,since it needs 
Only one meteorological variable : the meteorologi 
cal range. This variable may easily be retrieved from 
meteorological stations and airports. Sturm developed 
the correction for the Nimbus~7 CZCS which has a re 
solution of 825 meter. Adapting this correction for 
satellites with a higher resolution, such as Landsat 
Thematic Mapper (TM) or the SPOT satellite, with 
resolutions of respectively 30 and 20 meter in the 
multispectral bands, will provide for applications 
in the field of surface water research. These higher 
resolutions allow for water quality studies in lakes, 
estuaria, larger rivers and canals. 
In this paper the adaption of Sturm's atmospheric 
correction algorithm for TM will be developed. 
1 GENERAL FORMULATION OF THE CORRECTION ALGORITHM 
The signal L received from water surfaces by a remote 
sensor can be expressed as follows (see also figure 1) 
L = L + L + L + L (1) 
SG tlG w p 
where Lg^ is the sunglitter, i.e. radiance due to 
direct solar radiance from the water surface, L^q the 
skyglitter, i.e. radiance due to diffuse radiance 
reflected from the water surface, L^, the water lea 
ving radiance and L p the path radiance, i.e. radiance 
scattered from molecules and particles in the 
atmosphere. 
Figure 1. Processes in the atmosphere and at the 
water-atmosphere interface in remote sensing. 
Equation (1) is valid for a remote sensor at height 
z, observing at a wavelength X, with a view direction 
zenith angle y, and a view direction azimuth angle (j). 
A practical solution for NIMBUS-7 CZCS has been 
proposed by Gordon (1978). Following assumptions have 
to be made : 
1 the sunglitter term is absent; 
2 at a given reference wavelength the upwelling 
subsurface radiance Lp is zero; 
3 the path radiance term L can be separated into a 
term due to air molecule (or Rayleigh) scattering 
(L pr ) and a term due to aerosol (or Mie) scattering 
(L pa )• 
Equation (1) can then be written for a wavelength X 
and a reference wavelength Xq as : 
+ L 
+ l: 
PR 
+ L 
X 
PA 
L = L,, + L + L 
HG PR PA 
Equation (2) supposes furthermore that the aerosol 
scattering path radiance at two wavelengths is pro 
portional. 
787