7.2.4 Beam Mode

The examples in this section include two data sets acquired over the Sherbrooke, Que. area. The first, used in Stereo pairs 12, 13, 12A and 13A was acquired in Fine mode. The second, used in Stereo pairs 14 and 15 is in Standard mode. Obviously, the Fine mode data shows us more detail than Standard mode. This is advantageous if users are required to identify fine details for their particular application. Fine mode data contain more speckle than Standard mode data because one-look rather than four-looks signal processing is applied.

Fine Mode Stereo Pair

This stereo pair (F5/F1) is generated from ascending path, fine mode images. It displays much more cartographic detail, due to SAR resolution cells of about 8m. For example, most of the roads are clearly visible and can be easily extracted with small omission errors (less than 10%) and with a 10m positioning error depending on the type and size of the roads (Toutin, 1998 b). Since the SGF RADARSAT format images were resampled with 6.25m pixel spacing, they are under-sampled when compared to the resolution cell. The SGX RADARSAT format images, which are oversampled (3.125m pixel spacing) should then be preferred for more precise cartographic applications. Fine mode images display a lot of speckle, due to the one-look processing. However this does not generate confusion in stereo viewing and extraction because the human depth perception acts as a "filtering process".

This stereo pair (S1/S4) is generated from descending path, standard mode images. It displays less cartographic detail due to a coarse resolution cell of about 26m in range by 27m in azimuth. For example, most of the roads are visible but with omission errors varying from 20% for main roads to 70% for "unclassified" city streets. However, the positioning accuracy is better than one resolution cell (Toutin, 1999). The four-look processing of the SAR data to generate the images has reduced the speckle but smoothed the linear features, which explains the larger omission errors. On the other hand, the oversampled image pixels, when compared to the SAR resolution cell explains the high positioning accuracy.