DIRECT AND INDIRECT METHODS FOR 
CORRECTING THE AEROSOL EFFECT ON REMOTE SENSING 
Yoram J. Kaufman 
Laboratory for atmosphères NASA /Goddard SFC, Greenbelt, MD 20771 
and 
Didier Tanré 
Laboratoire d’Optique Atmosphérique, Université de Sciences et Techniques de Lille, 
Villeneuve d'Ascq, France 
ABSTRACT 
Aerosol scatters solar radiation before it reaches the surface and scatters and absorbs it 
again after it is reflected from the surface and before it reaches the satellite sensor. The 
effect is spectrally and spatially dependent. Therefore atmospheric aerosol (dust, smoke 
and air pollution particles) has a significant effect on remote sensing. Correction for the 
aerosol effect was never achieved on an operational basis though several case studies were 
demonstrated. Correction can be done in a direct way by deriving the aerosol loading from 
the image itself and correcting for it using the appropriate radiative transfer model or by 
an indirect way, by defining remote sensing functions that are less dependent on the 
aerosol loading. To some degree this was already achieved in global remote sensing of 
vegetation where a composite of several days of NDVI measurements, choosing the 
maximal value, was used instead of a single cloud screened value. The Atmospheric 
Resistant Vegetation Index (ARVI) introduced recently for EOS-MODIS is the most 
appropriate example of indirect correction, where the index is defined in such a way that 
atmospheric effect in a blue spectral channel cancels to a large degree the atmospheric 
effect in the red channel in computations of a vegetation index. Atmospheric corrections 
can also use aerosol climatology and ground-based instrumentation. These aspects of 
aerosol studies and remote sensing are reviewed in this paper. New advances in ground- 
based instrumentation and future satellite systems (including measurement of 
polarization) are discussed. 
KEY WORDS: Aerosol, Atmospheric corrections, radiative transfer, vegetation index 
1. INTRODUCTION 
The atmosphere affects remote sensing of the earth surface from space by scattering 
and absorption by its aerosol particles (e.g. micron size dust particles or submicron pollution 
particles suspended in the air) and molecules (nitrogen and oxygen in the atmosphere). The 
effect of aerosol and molecular scattering is stronger in the shorter solar wavelengths where 
the particle size is similar to the radiation wavelength. Absorbing gases (e.g. water vapor, 
ozone and oxygen) absorb in specific spectral bands (e.g. Tanre et al., 1992). Aerosol scattering 
increases the apparent surface reflectance over dark surfaces while aerosol absorption 
reduces the apparent brightness of bright surfaces. While the concentration and profile of 
atmospheric gases do not vary significantly, the concentration of aerosol and water vapor 
can vary by more than an order of magnitude. 
Correction for the effect of molecular scattering, ozone and oxygen absorption is 
important, since even though their concentration does not change substantially from 
climatologic values, their effect on the radiation detected by the satellite and on the 
apparent surface reflectance will vary as a function of the view and illumination direction 
(e.g. Tanre et al., 1992). 
7