The application of COSI-Corr to determine dune system dynamics in the southern Namib Desert using ASTER data

摘要

The Coregistration of Optically Sensed Images and Correlation (COSI-Corr) algorithm was used to estimate dune migration and sand flux rates from a series of remotely-sensed optical image pairs. Several areas of barchan and transverse dunes transport sand along definite pathways, feeding the southern part of the Namib Sand Sea from beaches and deflationary basins in the Sperrgebiet. We give a detailed description of the pre- and post-processing routines used for our COSI-Corr analysis, and we evaluate the sub-pixel dune migration results at the dune and dune-field scales. The best set of parameters for the application of the algorithm was systematically derived, resulting in methodological refinements of the cross-correlation windows. The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) sensor aboard the Terra satellite provided the remote sensing imagery. Few comparisons have been made between dune migration measurements using COSI-Corr and other types of change detection analysis, and few dune fields have served as test cases for COSI-Corr. We suggest that analysis and interpretation of COSI-Corr output from ASTER data is more valuable if combined with other image analysis techniques, such as manual digitization of dunes in a geographic information system (GIS), processing of write memory function insertion (WMFI) imagery and multispectral image (MSI) analysis of composition. The dune migrations estimated from COSI-Corr results were validated using GIS and WMFI, and later MSI analysis added an important regional, contextual framework of sand transport pathways. The synthesis of these works lends more confidence to understanding the Namib dune system's dynamics and provides a basis for future comparisons to other dune fields.