Satellite Data Fusion Techniques for Terrain and Surficial Geological Mapping

Increased satellite image data availability, with different spatial, spectral, and radiometric resolutions, present some challenges for image fusion techniques to be used in many application areas. In this study, we demonstrate an improved technique combining RADARSAT (SAR), Digital Elevation Model (DEM) and Landsat (ETM+) images with surficial geology data into two complementary composite image maps within Mackenzie Valley Pipeline Corridor, Canada. The method includes four processing steps. (1) The shaded relief derived from DEM was blended with SAR data to produce 3-D effect for terrain and geological structural interpretation. (2) Image fusion using a pan-sharpened IHS technique produced a higher resolution multi-spectral (MS) image. (3) The higher resolution MS image is blended with the SAR-DEM to produce an image terrain map. (4) Two complementary composites were then produced. The first image map contains of geomorphic features/surficial geology GIS layers, and 3D terrain information derived from the SAR-DEM image. The second image map is a composite of geomorphic features/surficial geology blended with the SAR-DEM-ETM+ image map. Integrated image maps generated in steps 1 and 3 provide standardized image base maps to display surficial geological and terrain information.

1. INTRODUCTION

Geological interpretation and mapping are based on direct and indirect interaction of information on types and distributions of material found at the surface, geomorphologic and structural characteristics, drainage systems and vegetation. Multispectral (MS) and synthetic aperture radar (SAR) remote sensed data provide capabilities to support geological mapping and mineral exploration [4], [8], [9], [10]. MS images contain information on spectral characteristics of land cover, while SAR backscatter detects morphology, structures and soil moisture. The interpretation of multisensor and multiresolution remotely sensed data are significantly improved by recent research and development using various image fusion techniques [2], [12], [13]. The potential of information extraction is increased by integrating remotely sensed data with ancillary data (DEM, geophysical, etc) [3], [7]. In this paper, we present a mapping technique to further enhance the interpretation potential of fused data for terrain/surficial geology. The study area is located in the Upper Mackenzie River Valley, North West Territories, Canada. The area incorporates uplands and low areas covered by thick morainal deposits, which consists mostly of till and related glacial features. Glacial lacustrine sediments are also widespread. The Mackenzie Mountains, located in the western part of the area, are characterized by presence of valley deposits of glacial outwash, lacustrine sediments, and till and post-glacial alluvial terraces, fans, and colluvium.