SEASONAL-CHANGES IN RELATIVE C-BAND BACKSCATTER OF NORTHERN FOREST COVER TYPES

This publication reports the results of an experiment performed to investigate the changes of C-band microwave backscatter as a function of season in northern forests. The purpose was to determine whether seasonal changes can be used to increase the information content of single polarization C-band SAR data, which are now available from the ERS-1 satellite and will also be provided by ERS-2 and RADARSAT. The Russian Almaz satellite is also providing data at a slightly longer wavelength of 10 cm. Soon after it was commissioned, the CCRS C-band SAR acquired C-HH data in four consecutive seasons: August 10 and October 29, 1987, and February 25 and May 2, 1988. In each case the SAR was flown on the same east-west line with a pixel spacing of 3.9 m (azimuth) by 4.7 m (range) with incidence angles ranging from 45-degrees to 75-degrees. Although absolute radiometric calibration was not available, calibration was carried out within each scene, allowing seasonal changes in relative backscatter and absolute dynamic range to be studied. In addition, numerical tests of class separability were conducted to indicate the optimum season(s) to differentiate various classes. This investigation demonstrated that: 1) The entire dynamic range of mean C-HH backscatter values of forest stands was never more than about 6 dB. The range exhibited seasonal variations, ranging from only 3.5 dB in February, to 6.0 dB in May, 5.6 dB in August, and to 4.3 dB in October. The seasonal changes in dynamic range of the nondeciduous softwoods (spruce, jack pine, red and white pine) are hypothesized to be dominated by changes in the dielectric constant of the woody and foliar parts of the trees. 2) The separation among the three softwood types was greatest in August, decreasing significantly to May and October and then to February. The backscatter of the softwoods was ordered inversely to the needle length. Whether the needles are a causative factor in this effect is not known. 3) Deciduous trees (including tamarack) had higher backscatter relative to softwoods when they are without leaves than with leaves. Seasonal changes of deciduous backscatter relative to the softwoods allows multitemporal SAR data to be used to distinguish between hardwood and softwood species. 4) The separation between the nonforest areas and all of the forested test areas was greatest in February. The most plausible explanation is that the low and gradually varying dielectric constant of snow cover over frozen soil (which also has a low dielectric constant) reduces the backscatter of nonforest areas.