| Summary: | Digital elevation models (DEMs), multitemporal Landsat TM images and forest inventory maps were use to study the spatial patterns of boreal mixedwoods in the Boreal White and Black Spruce zone of British Columbia. A Bayesian classification method was developed to produce a high resolution map of posterior probabilities of forest cover classes. The procedure employed the forest inventory maps as prior probabilities, while evidence was provided by the results of a maximum likelihood classifier applied to the satellite images. Topographic parameters, derived from the DEMs by fitting a conic surface equation, were used in a fuzzy-logic based classification to produce maps of terrain classes. A filtering technique, expressly developed to mitigate the systematic errors present, was applied to the DEMs before the classification. The resulting maps were employed to investigate the relations between terrain and vegetation using an appropriately modified version of Ivlev's electivity index. The analysis suggests that topography plays a role in the distribution of the different types of boreal mixedwood forest stands. As expected, black spruce (Picea mariana (Mill.) B.S.P.) and tamarack (Larix laricina (Du Roi) K. Koch) dominated stands were strongly associated with flat areas. Hardwood dominated stands were found significantly associated with convex slopes and ridges, white spruce (Picea glauca (Moench) Voss) dominated stands positively were associated with concave slopes and channels, and mixed stands were associated with neutral slopes. The forest cover maps were used to study the distribution of mixedwood and hardwood stands in relation to white spruce stands. The analysis employed a logistic regression model where topographic variables and stand age were used as covariates. The likelihood of a site being occupied by hardwoods instead of mixedwoods was found to be positively correlated with increasing distance from white spruce stands. This result suggests that the rates or the outcomes of succession in the mixedwood forest are influenced by the spatial arrangement of different stand types at the landscape level. The most likely explanation is the control of seed dispersal on white spruce seed availability in the landscape, although a direct causation cannot be inferred from this correlation analysis.
|
|---|
