Patterns of forest regrowth following clearcutting in western Oregon as determined from a Landsat time-series

The rate at which forest vegetation re-establishes dominance after clearcut harvesting can impact many ecological processes, such as erosion/sedimentation, nutrient and water cycling, carbon storage potential, wildlife habitat, and trophic interactions. Although knowing a forest stand's current state of succession is useful, a clearer understanding of the impact forest harvesting has on the aforementioned ecological processes can be achieved with a more dynamic characterization of the successional process. To more fully model the continuous nature of forest regrowth following clearcut harvesting we extrapolated percent tree cover data collected by the U.S. Forest Service Pacific Northwest Forest Inventory and Analysis program to a cross-normalized Landsat time-series using a date-invariant regression modeling approach. Using three periods of mapped clearcuts we extracted and classified the extrapolated percent tree cover data into four regrowth classes (little to no, slow, moderate and fast). These forest regrowth classes were used to develop frequency distributions describing the landscape patterns of postharvest forest recovery for two ecological provinces in western Oregon. The patterns of forest regrowth observed over the three clearcut periods indicated a much higher percentage of fast regrowth in the Coastal Range Province and a much higher percentage of little to no regrowth in the Western Cascade Province. For both ecological provinces we observed the propensity for faster regrowth on north facing aspects, shallow slopes and at low elevations. The forest regrowth classes and the frequency distributions indicated that a wide range of successional stages could be found in both ecological provinces 18 years after clearcutting. The extension of forest regrowth trajectories to the spectral space of Landsat provided an opportunity to use CART statistical analysis to more fully investigate the climatic and topographic drivers influencing the rate of postharvest forest regrowth. Based on the Kappa statistic, predictions from both CART models were in “fair” to “moderate” agreement with the test samples. Both classification trees yielded ecologically interpretable insights into the environmental attributes influencing forest regrowth rates after clearcutting. In both ecological provinces, elevation followed by potential relative radiation (PRR) explained the largest amount of variation in forest regrowth rates. To gauge the effectiveness of predicting more generalized postharvest forest regrowth rates we combined the four forest regrowth classes into two general “fast” and “slow” categories. Based on the Tau statistic, the CART models correctly classified 12% (CRP) and 26% (WCP) more combined test samples than classification of the four regrowth classes.