. Data integration from the tree to the satellite level to understand forest resilience across scales

Abstract
Forests of the Southern Rocky Mountains have recently experienced larger, more intense, and more frequent wildfires, increasingly severe and prolonged drought events, blowdown events, and widespread insect outbreaks, resulting in an increase in tree mortality and altered post-fire regeneration. Also, human land use alters forest dynamics and introduces more anthropogenic ignition sources. The interactions of fire with other disturbances and land management activities across large spatial scales is not well understood. A major challenge in exploring resilience following disturbance is the lack of widely available fine-scale forest data that cover a sufficiently large spatio-temporal extent. We present a framework to utilize Unmanned Aerial Systems (UAS) data to help address a fundamental question related to ecological resilience and stability: How do disturbance interactions alter forest recovery trajectories across the Southern Rockies? In this framework, species-specific spectral profiles from UAS are used to develop plant functional group (PFG) endmembers (i.e., coniferous forest, deciduous forest, shrub, herbaceous cover, and bare ground) to spectrally unmix Landsat data, resulting in annual maps of the percent of each Landsat pixel occupied by each PFG. Preliminary efforts using NEON AOP data demonstrate that random forest models allow cataloging the spectral signatures of individual species to ~80% accuracy. The unmixed Landsat data are paired with disturbance data under a Bayesian model to evaluate the change in the mixing proportion of each PFG for each pixel under different conditions, as well as project changes in vegetation states over time. Preliminary efforts using MODIS-derived forest state demonstrate, for example, that pine beetle infestation preceding fire slows forest recovery by ~7 years. This framework and the scalable sampling design leverages emerging technologies to help us meet the challenges associated with changing climate and disturbance regimes by identifying what factors most strongly influence land cover changes at large spatial scales.