Open Collections

Abstract

As boreal forests face increasingly frequent and severe natural disturbances, developing spatially explicit methods to quantify resilience and understand its underlying mechanisms is critical for forest management. This dissertation advances the operationalization of resilience concepts by leveraging multi-source remote sensing, including airborne laser scanning (ALS) and long-term Landsat time series, to characterize the forest characteristics shaping forest resilience response to disturbances. I demonstrated the capacity of bi-temporal ALS to detect fine-scale vertical structural changes, establishing that pre-disturbance forest structures modulate disturbance severity. Scaling to the landscape level, analyses of Landsat time series data revealed that the spatial arrangement of forest stands is a key determinant of disturbance patterns, specifically identifying host species configuration as a primary driver of disturbance likelihood. To bridge the gap between theoretical resilience frameworks and empirical monitoring, I investigated forest resilience dynamics over nearly four decades across three key satellite-derived vegetation indexes that capture distinct plant physiological processes. Moreover, I decoupled the role of forest structure, composition, configuration, and climate variables in shaping resilience responses to disturbances across these physiological processes using machine learning. Results revealed that, while forest composition and configuration are important, minimum summer and winter temperature, summer precipitation, disturbance history, and canopy cover ultimately govern long-term resilience outcomes. Notably, these findings were consistent across all three spectral indexes. Collectively, this dissertation demonstrates the power of remote sensing in providing a robust, multi-source framework for monitoring forest resilience under changing climate and disturbance pressures.