Predicting Forest Age up to 250 Years in Old-Growth Boreal Mixedwoods of Western Quebec, Canada, Using Airborne Laser Scanning Data and Modelled Tree Species Composition

Old-growth boreal forests play a key role in global conservation policies. Forest age maps, however, tend to lose accuracy as stands age, making it impossible to consider the full successional diversity of old-growth forests. Our aim was therefore to develop remote sensing models that can predict old-growth boreal forest ages up to 250 years. As ground-truth data, we used a network of 432 field plots located in the boreal mixedwood forest of eastern Quebec, Canada. Forest age (time since last fire) was known for all plots, ranging from 74 to 258 years. We extracted indices from airborne laser scanning (ALS) and tree species prediction data for these plots at 4 different resolutions (400, 1600, 4900 and 10,000 m2) to capture variations in structure and composition at different grain sizes. The models obtained showed high predictive accuracy, at all resolutions (R2: 0.574–0.759, RMSE: 28.686–40.026, Bias: −1.093 − 3.933). Predicted tree species composition data contributed the most to model accuracy. These models provided an unprecedented mapping of the successional diversity of old-growth forests, compared with available data. Our methodology, therefore, would be an essential resource for improving our forest management strategies, allowing better conservation of old-growth boreal forests.