Accurate and consistent mapping is essential for the successful management of forests. Global forests provide invaluable benefits and resources, both of an ecological and economic nature, to the world's population. Not only to the forests provide a supply of fuel and building material, but they also retain soil, regulate run-off, minimize the siltation of water, and provide fruits, nuts, tree extracts and medicinal plants (Ahern 1996). As the large proportion of the Earth's surface is occupied by forests, which are continually changing, a management tool is required that is capable both of covering large areas of the Earth's surface, as well as revisiting those areas with some frequency.

The ability of spaceborne radar sensors to image the globe in a short period of time, unimpaired by such atmospheric effects as light rain and cloud, is well known. As with other applications, this ability is well-suited to forestry applications, as a large proportion of the world's forests are found in tropical areas where perennial cloud cover is a problem.

Within forested lands, radar data can provide valuable information for reconnaissance mapping as well as for commercial forestry. Activities within these applications include terrain analysis, forest cover type discrimination, the delineation of burned areas, and mapping of cleared areas.

The use of radar for topographic mapping for terrain analysis has been well documented (Simonett and Davis 1983; Williams 1983). The side-looking configuration of spaceborne SARs highlights relief. When relief information provided by radar is combined with optical data, an image is created that provides valuable terrain information. Optical data, such as that provided by Landsat Thematic Mapper or SPOT, provides vegetation mapping capabilities, while radar data such as that provided by RADARSAT provides topographic enhancement. Topographic enhancement is optimized when the incident angle is about 15 º larger than the greatest slope in the imaging area.

The ability of C-band radar to discriminate between some forest types has been shown in a range of forest environments. In areas of mixed deciduous and coniferous forests, separation of general forest type is accomplished through the use of multitemporal data; the leaf-on and leaf-off conditions of deciduous species results in a seasonal change in backscatter which contrasts with that of the coniferous species which do not experience leaf-off. It is important to note that in order to acquire useful information on general forest type, using radar satellites such as RADARSAT, stands will have to be large and of uniform composition (Ahern et al. 1993). Species differentiation is more difficult in general at microwave frequencies then with optical data because of overlap in signatures between species.

The use of radar imagery for the update of existing forest inventories has been identified by Ahern et al. (1993) as one of the most important roles RADARSAT could play for providing information to national and regional forest management authorities for commercial forestry applications. Information required for the update of forest inventories are provided through the delineation of burned areas and mapping of clear-cut areas.

Radar is particularly effective for the delineation of burned areas because of its sensitivity to differences in structure and moisture content. Several months after a forest fire, burnt forest areas dry out and leaves and small twigs drop from the trees which results in a contrast in structure and moisture content between burnt and unaffected areas. On radar imagery, this produces a contrast in backscatter, where the burnt areas appear darker than the unaffected forested areas.

C-band radar has shown to be an ideal tool for detecting and mapping recent forest clear-cuts because there is a high contrast in tone with surrounding forest areas on C-band imagery, as their lower and sparser vegetation appears smoother and returns less backscatter than typical forest canopies. The areas of bare ground surface prevalent in recent clear-cut areas also return minimal backscatter to the sensor, which is why logging roads as well as recent cleared areas appear darker than the surrounding forest on radar imagery. In temperate regions, clear-cut mapping may be most accurate when wet snow cover is present. Wet surface snow (which appears as dark tones because of little backscatter) in the cut areas results in a very low backscatter and still-greater contrast with the surrounding forest, which has a rougher surface and much higher level of backscatter. Cleared forest areas must be imaged within 1 to 2 years after cutting; after this time, the vegetation in the area begins to regenerate resulting in a decrease in backscatter contrast between forest and clear-cut, and the ability to detect the clear-cuts decreases.