Automatical Geocoding of Sar Products

An automatical system for geocoding and calibration of Envisat ASAR products have been developed by NORUT IT in the Envisnow EC EESD FP 5 project. The project have used data from ESA AOE 785, Cat. 2458 and near-real time data from Kongsberg Satellite station. The software is adapted to the Envisnow snow covered area processing system, and works automatically for the Envisat ASAR modes IMP, IMS, WSM, APS and APP and the ERS modes SLC and PRI and Radarsat (Standard and Fine). The software is mainly implemented in the program languages Envi/IDL under a common framework. The output is a geocoded and calibrated SAR image and different kinds of value added products in a selectable map projection. The system is used as a module in the Envisnow production line to provide high quality geocoded SAR imagery for snow cover area classification. The accuracy of the geocoding is estimated to be on average less than 1 pixel (30 m for image modes and 100 m for Wide Swath). This is verified by cross correlation the final geocoded SAR with a simulated SAR image based on DEM and SAR viewing geometry. The results are confirmed by manually overlaying different vector layers and noticing any geometrical offsets in the SAR. Due to inaccurate orbit data and SAR processing from some PAF’s, a new correction algorithm has been implemented. The algorithm is based on using the first order geocoded product based solely on orbit parameters and DEM together with the simulated SAR image. The algorithm then picks one or several control points automatically by cross-correlating sub-images of the first order geocoded SAR and the accompanying simulated image. This is subsequently used to refine or update orbit/SAR parameters for the final geocoding of the SAR scene. The method will be demonstrated and the quality assessed.