Atmospheric correction of HRSC-AX line scanner data using the Dark-Pixel-Profile (dpp) method: A summary after 2 years

Abstract

The HRSC-AX (High Resolution Stereo Camera – Airborne eXtended) systems are multiple line "pushbroom" instruments developed by the German Aerospace Center (DLR). The dpp-profile is a simple correction method to remove additive path radiance based on view-angle dependent histogram statistics; it neither requires an absolute calibration nor a co-registration of the image data. It was developed at the DLR in Berlin-Adlershof and has been applied to several thousand km2 of HRSC-AX data for more than 2 years. The results are very encouraging: the dpp-method was applied to more than 90% of the examined targets (cities and rural areas), in less than 10% of the targets the method was not used due to the ambiguity of the statistics and the risk of tampering the data. In all selected targets the blue and the green channel were corrected, whereas the red channel was only corrected in about 50% of the selected targets where the statistics agreed with the expected pattern. A degradation of the image quality after the correction was never observed. Depending on the amount of atmospheric effects the visual differences between corrected and uncorrected image data can be described from almost not visible, same quality to strongly improved. Also the statistics of the improved image mosaics are more conclusive and show less systematic fluctuations. The dark pixels, which are used in the statistics of the dpp-method, resulted to be in almost all examined targets dark shaded vegetation. In areas clear of vegetation other wide spread dark objects (like water bodies) have to be identified as suitable dark objects, which is not always possible. Further investigation concerning the influence of the dpp-correction on classification results and an extension of the dpp-method to derive atmospheric parameters like optical depth are currently in progress.