Benefits of robust Parameter Estimation Techniques in an Automated Geocoding Processing Chain

Accurate geocoding of Earth observing (EO) satellite imagery is one major prerequisite to monitor the Earth’s environment. The quality of the obtained geocoded images depends on the accuracy of the sensor model parameters and the accuracy of the used height information. As for monitoring applications automated image processing chains are desired this paper concentrates on the automatisation of the required measurement ground control points (GCPs). Moreover, several methods to make least squares adjustment a robust estimator are presented. To improve the performance of the robust estimation with respect to the reduction of computational effort and reduction of number of GCPs needed, the use of the median absolute deviation is introduced. The presented general geocoding workflow and the proposed enhancements are tested using an ENVISAT-ASAR scene of the highly active Neovolcanic Zone of Southern Iceland. This test site, prone to natural hazards like volcanic eruptions and massive glacial torrents is equipped with a network of artificial corner reflectors enabling a detailed verification of the proposed approach.