Can Ancient Stands of Cedar-Hemlock within Old-Growth Forests be Identified using LiDAR?

No overarching combination of metrics explained the variation between all categories. Additionally, variation within categories far exceeded that between categories, so a regression equation could not be established. A random forest classification of the data found that the distinction could only be accurately predicted between 25 and 60 percent of the time, and different iterations of the same model exhibited extreme variation.

The validity of the results was limited by reliance on estimated attributes from the Vegetation Resources Inventory. If this methodology were to be repeated with field verified measurements as input data, it may be able to mitigate these issues and provide the basis for a reliable predictive classification. This classification could provide an accurate and objective standard that would aid in the identification and conservation of ancient forests.