X
50 Rate this article:
No rating

Internal: Some benefits of FLAASH over ATCOR for Imagine/PCI

Anonym

ENVI's FLAASH module, which provides a model-based atmospheric correction for spectral imagery, is often compared to the ATCOR atmospheric correction software. This help article discussed the main advantages that FLAASH has over the version of ATCOR that is sold as an add-on to ERDAS IMAGINE and PCI Geomatica software.

Please note that ATCOR-IDL is also available from ReSe Applications Schläpfer. This article does not address that version of ATCOR.
  1. FLAASH uses MODTRAN to model an atmosphere individually for each corrected image, while ATCOR interpolates its solution from a MODTRAN-derived database based on typical flight altitudes and solar geometries. Therefore, FLAASH's results are more closely adapted to each image's unique atmosphere. This can become especially important for difficult image collection conditions, such as off-nadir viewing geometry, very long baselines, extremely wet atmospheres, or heavy aerosol loads (for example, collection near a forest fire).

  2. ATCOR includes a much heralded topographic correction. There is no consensus in the scientific community about whether topographic correction is appropriate during atmospheric correction. At the very least you are introducing another source of error due to the estimation of the surface slope, which is unlikely to be at the same scale as the reflectors in the iFOV (instantaneous field of view). Further, many users want their corrected scene to actually look (on the landscape scale) like reality, and that would mean it has to include the effects of topography, not remove them. There are special cases where a topographic correction may be important, but these are exceptions.

  3. For hyperspectral data: Superior spectral polishing. Polishing is a linear renormalization method that reduces spectral artifacts in hyperspectral data using only the data itself. To remove such spectral artifacts, it is not adequate to simply smooth the spectra. You have to carefully minimize the noise without muting the real signals.