Abstract
Forest age class is essential to make forest policy and management decisions. Remote sensing is a viable alternative to time consuming field and aerial investigations to determine forest age. In this study, 76 stands of nine 20-year age classes of western red cedar from the interior of British Columbia, Canada were selected to test the ability of pan-sharpened SPOT-5 imagery to classify stand age. For each stand, a Structural Complexity Index (SCI) was calculated using principal components analysis of stand-level variables (diameter at breast height, crown diameter, basal area, and stem density). To further increase classification accuracy, three window sizes (5×5, 11×11 and 25×25) of first-order and second-order statistical measures of texture were calculated. Stepwise discriminant analysis and multivariate regression were used to determine the best explanatory model of forest age and SCI using the spectral and textural data. The most accurate model used a combination of these measures of texture and the SCI to predict age class.
Acknowledgements
This research was funded by the Natural Sciences and Engineering Research Council of Canada. We thank PCI Geomatics for the use of the PANSHARP module for this study. We would also like to take this opportunity to thank the British Columbia Ministry of Sustainable Resource Management for providing the forestry inventory and other ancillary GIS data.