Point of Beginning

ImageLinks Adds New Atmospheric Correction Algorithm to Image Processing Stream 11.11.2002

November 11, 2002
Technique sharpens imagery.

ImageLinks has implemented an advanced new atmospheric correction algorithm based on the 6S radiative transfer code. This correction will be adopted as one of the standard digital image processing techniques ImageLinks applies to aerial and satellite imagery. Atmospheric correction improves image clarity by compensating for distortions caused by gaseous absorption and molecular and aerosol scattering along the Sun-target-sensor path.

“This atmospheric correction technique estimates the true surface reflectance, thus enhancing the clarity of the digital imagery. This significantly improves the quality of ImageLinks products derived from imagery for flight simulation, land classification and change detection,” said Kathy Minear, ImageLinks’ Research Scientist. “We have implemented the atmospheric correction code for the Landsat sensor with positive results. Correction for various other commercial satellite sensors and digital airborne cameras are currently being added into our processing stream.”

Atmospheric correction is necessary during the processing of remotely sensed digital imagery because the recorded radiance measured by the sensor differs from the true surface reflectance for each area on the ground. This difference is due to interference by various gases and aerosols which are present in the atmosphere. The path of electromagnetic radiation (ER) from the Sun to the surface and from the surface to the sensor is attenuated and scattered by these gases and aerosols, which are mainly found in the troposphere and stratosphere.

Gases primarily responsible for absorption or attenuation of ER are water vapor and ozone. Absorption by water vapor is of particular concern in certain portions of the near-infrared range because this effect lowers radiance measured at the sensor. Pre-defined atmospheric models describing temperature, pressure and water vapor and ozone concentration as a function of location and season are built into the 6S code. These include tropical, mid-latitude summer, mid-latitude winter, subarctic summer and others. User-defined models are also allowed.

Rayleigh and aerosol scattering effects are more significant than absorption effects in the visible light spectral range causing an increase in the at-sensor measured radiance values. This scattering is wavelength-dependent with a greater effect at shorter wavelengths, such as the blue band. This results in a hazy-blue appearance in natural color imagery. Aerosol scattering is handled in the 6S code using pre-defined models for urban, continental, maritime and desert environments. User defined models also allow the user to specify percentages of dust-like, soot, oceanic, and water-soluble aerosol components.

“The atmospheric correction code offers three different options for determining the aerosol optical depth, an important variable used in the radiative transfer code,” said Minear. “Two iterative approaches, dark target detection and dark dense vegetation, have been implemented. These methods extract the information from the image. Also the option of attaining the aerosol optical depth from the AERONET online database is available.”

Since per-pixel calculations are too costly, the atmospheric correction algorithm is implemented using either autonomously derived or user-defined grid points. A bilinear interpolation on the output correction parameters is performed between the grid points.

The 6S radiative transfer code is an improved version of the 5S code originally created in the 1980s in the Laboratorire d'Optique Atmospherique in France. The term 6S stands for Second Simulation of the Satellite Signal in the Solar Spectrum.