7.2.1 Terrain Type

Stereo pair 9 - RADARSAT - Okanagan Valley, British Columbia, Canada
This image pair covers approximately 2/3 of a full Fine mode scene of the Okanagan Valley, B.C. area. This area was chosen to correspond to the area shown in the SPOT, Landsat and ERS-1 images of Stereo pairs 2, 7 and 8. Geometry inherent in a radar remote sensing system such as RADARSAT is illustrated very well in this example. Effects of foreshortening, layover and shadowing are clearly visible. Mt. Acland (just right of centre on both images) is compressed on the west face and appears to lean towards the sensor. Layover is visible along the ridge running parallel to Lake Okanagan. Due to a much smaller intersection angle vertical exaggeration in this stereo pair is much less pronounced than in the SPOT imagery (Stereo pair 2).

Much of the contrast on these images is also due to geometry. Sides of mountains facing the sensor appear bright while those facing away from the sensor are in shadow. There is a strong contrast between land and water. Cleared areas, such as the hydro line, roads and fields appear darker than surrounding forest coverage, which is brighter. Seasonal vegetation differences cannot be distinguished as the data was acquired 2.5 weeks apart. Speckle is this scale.

Stereo pair 10 - RADARSAT - Ottawa, Ontario, Canada

The area covered by the images over Ottawa is characterized by the Gatineau Hills to the north, the Ottawa River Valley and flat lands to the south. Change in elevation between the river valley and the Gatineau Hills is approximately 300 m. The relief of the Gatineaus Hills, is clearly visible as is the rise from the Ottawa River. Vertical exaggeration is pronounced. The Gatineau Hills appear to be almost mountainous. The two images do not contain a large amount of contrast, however there are variations in contrast between them. The S1 image shows a stronger contrast between land and water bodies than the S3 image. However, S3 shows stronger contrast between ground cover types than S1. These radiometric differences can be explained by the difference in acquisition dates - S1 on Dec. 12, 1995 and S3 on Feb. 13, 1996. Therefore land/water boundaries are less delineated on the S3 image due to ice. Fields are brighter on the S1 than the S3 image because the snow was very dry in Dec. 1995 and very wet in Feb. 1996. These differences in radiometry interfere slightly with stereo-viewing.

The S1 image shows more foreshortening and layover effects than the S3 image. This is due to the steeper S1 viewing angle. Speckle is minimal on both images.

Stereo pair 11 - RADARSAT - Tapagos Forest, Brazil

This area is characterized by a change in elevation of less than 50 m. Change in relief between the coastal area and the inland plateau can easily be seen. Differences in height between the top of the tree canopy and fields cleared for agricultural crops can also be seen.

The two images are very similar radiometrically. The images were collected at the same time of the year. The incidence angle has little impact on either image as the land surface is almost completely flat. There is a weak contrast on the images. Land and water bodies are sharply differentiated. Some contrast is due to shadow. The ridge of the plateau and small stream beds are delineated by shadow.

There is evidence of sensitivity to roughness in the difference between agricultural crops and forest. Forested areas are a little brighter than agricultural fields. The images contain some speckle, which interferes with sharpness on the images. Fields or clear cut areas would be more clearly delineated with less speckle. Both images were collected at the same time of the year therefore seasonal changes in vegetation cannot be identified.

Changing the order of this stereo pair should lead to a pseudoscopic stereomodel. However, with this stereo pair some people will still see a normal stereomodel. This is due to psychological cues (shadow on both images), which are stronger than physiological cues (binocular disparity) on images of flat terrain.