Northern Alberta Mosaic of RADARSAT-1 Principal Components Images Derived from S1/S7 Ascending/Descending Imagery

This investigation uses RADARSAT-1 standard beam imagery as a tool for mapping surficial geology and geological structures in northern Alberta. Between 1999 and 2002, the Alberta Geological Survey acquired RADARSAT-1 imagery for northern Alberta north of 55� latitude. They were then processed to obtain orthorectified, tiled and filtered images for its regional mapping program and for future applications in other areas of environmental and resource management. From September to December 1999, 280 scenes of RADARSAT-1 Standard Beam modes S1 and S7 were captured for ascending and descending passes, as shown in Figures 1 to 4. The autumn season was chosen to minimize the effect of vegetation and maximize the microwave reflectance from the ground surface.

Radar imagery has been commonly used for mapping geomorphology and geological structure (Lowman et al., 1987; Singhroy et al., 1993; Gupta, 1991; Singhroy and Saint-Jean, 1999; Smith et al., 1999; Paganelli et al., 2001; Paganelli et al., 2002; Grunsky, in press). The range of incidence angles, all weather atmospheric penetration and response to surface morphology give radar imagery significant advantages in measuring surface features relative to conventional, fixed-beam optical satellites. A review of RADARSAT-1 imagery in geoscience applications showed that most studies used individual radar images with a single incidence angle, polarization, frequency and resolution. However, radar imagery has also helped extract information from the integration of optical and geophysical imagery (Harris et al, 1994; Mustard, 1994). Masuoka et. al (1988) applied the technique of principal components analysis (PCA) on a radar image composite derived from Shuttle Imaging Radar (SIR-B) and Seasat. Although most studies use radar for delineating geological structure, it has also aided in mapping surficial geology (Graham and Grant, 1994).

Previous studies on the use of PCA with radar imagery (ERS-1 and Canada Centre for Remote Sensing (CCRS) C-SAR) found the technique useful for highlighting structural features in the Sudbury area of Ontario, Canada (Moon et al., 1994 and Harris et al., 1994). In the study by Moon et al. (1994), different incidence angles, look directions, frequencies and polarizations were able to highlight geological structure. This Geo-Note used two look directions and incidence angles; polarization and frequency were held constant.

The detection of geological structure is partially dependent on the look direction of the satellite (Harris, 1984; Lowman et. al, 1987). If a linear feature is parallel to the look direction of a radar image, then it may be nearly invisible. Studies have indicated (e.g., Harris, 1984) that the linear features show up as distinctive lines in radar imagery when the feature is within 20 degrees of the perpendicular to the look direction of the radar sensor. In the case of using multi-beam imagery for RADARSAT-1 data, the identification of linear features will be determined by a range of look directions that are different for each beam mode.

The use of radar imagery in geological applications and projected use of RADARSAT-1 has been discussed by Singhroy et al. (1993). More recently, Singhroy and Saint-Jean (1999) have shown that variation in RADARSAT-1 incidence angles highlights ground features based on relief and surface texture. In this Geo-Note, the strategy of adopting S1 and S7 imagery was to contrast the radar responses based on the incidence angle and look direction (ascending-east looking/descending-west looking). The difference in responses due to the incidence angle is a potential indicator of surface variation. Figure 5 shows the configuration of beam modes and look direction for the image integration used for this study. Influences on the response are backscatter, which can be attributed to both volume and surface conditions, and these are influenced by topography, vegetation and surface moisture (Raney, 1998). The nature of backscatter, due to the variation of incidence angle, is likely to have an effect on the response. Lowman et al. (1987) described the effect of incidence angle on the ability to detect EUB/AGS Geo-Note 2002-24 (November 2002).