The applicability of radiative transfer models for atmospherically correcting airborne hyperspectral data in Antarctica

As part of a collaborative project between BAS, DRDC Suffield (Canada) and ITRES Research Ltd., (Canada) the first known airborne hyperspectral dataset was acquired over the Antarctic in February 2011. The simultaneous deployment of commercially available visible-near infrared and shortwave infrared spectrometers generated a dataset covering 0.35 to 2.5 µm spectral range at a spectral resolution of 9.6-14 nm. To enable quantitative analysis of surface properties using imaging spectrometry data the removal of atmospheric absorption and scattering effects is an essential pre-processing step. The implementation of a sufficiently accurate and robust atmospheric correction methodology is of critical importance in ensuring that the results from spectral and spatial analysis algorithms are as accurate as possible. However, whilst methodologies are well established for most environments, there is currently no published methodology for correcting airborne hyperspectral data in the Antarctic region.

This study presents initial results from an investigation into the applicability of the MODTRAN-5® radiative transfer model and the ATCOR-4 atmospheric correction package for producing atmospherically corrected hyperspectral data in the unique Antarctic environment; an environment that is cold, dry and has low levels of aerosols and atmospheric pollution. Initial results from radiative transfer modelling and atmospheric correction produce absolute reflectance spectra which are partially comparable to laboratory measured spectra. Improvements are seen with the hybrid approach of radiative transfer modelling and the empirical line method using in-scene ground targets. Residual noise remains present due to absorption by atmospheric gases and aerosols which are not appropriately modelled for this environment. Overall, this demonstrates that commercially available packages are not currently flexible enough to correct Antarctic hyperspectral data without the addition of in-scene ground calibration targets. The implementation of Antarctic aerosol and atmospheric profiles into the radiative transfer model would likely improve these corrections and remains an area of investigation for future hyperspectral campaigns in the region.