A multi-temporal classifier for SIR-C/X-SAR imagery

Biophysical systems in temperate regions undergo distinct changes in each annual cycle. For forested ecosystems this includes moisture and phenological changes. The dual-flight program (April and October) for the SIR-C/X-SAR instrument aboard the Shuttle Endeavor was designed expressly to acquire imagery at different stages of such cycles. This provides new opportunities in terrain classification where phenology and moisture changes may be used as additional information. Given multi-season SIR-C/X-SAR imagery, there are three possible approaches to classifier development: (1) ignore the multiseason availability and develop independent classifications for each scene using n features, (2) develop one classification for a set of x scenes using n features, with x times the number of samples per feature, and (3) develop a true multi-temporal classification where N of features equals n (number of features) times x (number of scenes). Each of these are applied and results for the first two of six selected scenes show the true multi-temporal and the October scene alone to have the highest accuracies (94% and 95% respectively) at a level II classification.