The overarching objective of this research was to investigate changing vegetation dynamics across boreal North America, with a focus on Department of Defense (DoD) lands of interior Alaska. This project assessed, observed, and predicted changes in boreal tree composition through time and in a spatial context by linking field measurements with five decades of satellite observations and models. The analyses incorporated broad scale impacts of disturbance, particularly fire, and post-disturbance vegetation succession. The spatiotemporal analyses and modeling incorporate climate as a “press” disturbance as well as the “pulse” disturbance brought about by fire, each of which impact tree productivity and mortality. The project team then predicted change into the near future (20-50 years) using a combination of empirical and process-based models. The project team also interacted with DoD and other state and federal land managers to transfer the data products and analysis tools for their needs.

This project made extensive use of field measurements and multi-scale satellite observations, typically coupling the two in development of spatial models (maps) and change through time, while accounting for error and uncertainty. The field measurements incorporated species-specific responses to drivers and the remote sensing component addressed broader scale responses and changes in plant functional types including deciduous broadleaf trees, evergreen conifer trees, deciduous shrubs, and others. Two types of models were used to predict future change, informed with changes observed over recent decades. The first type included machine learning (statistical deep learning) models and the second type was a process-based model (University of Virginia Forest Model Enhanced) that simulates a wide range of vegetation composition, structure and function in response to a suite of environmental drivers and ecosystem processes (e.g. climate, competition, fire, permafrost thaw).

This project documented wide scale change across the boreal forest ecosystems of central Alaska, as well as a broader spatial domain into Canada, over the past several decades using a diverse and extensive range of field measurements combined with multi-source remote sensing data. These were augmented with spatial models predicting future change, all with the purpose of informing management decisions. A wide range of previously unavailable data sets were documented and made available via active data archive systems that are open access. In some cases, such as data available via Google Earth Engine, these are accompanied with workflows and source code allowing updates to the data sets. This project also developed an interface for launching scenarios of management actions and predicting the expected future outcomes of those actions. All of these outcomes were supported by a rich set of publications in peer-reviewed journals.

This project developed, in consultation and interactions with Alaskan DoD resource managers, multi-layered information that can be used in conceptually simplified management practices that incorporate directional changes in vegetation composition resulting from ongoing changes in climate combined with an increasingly severe fire disturbance. The map data products that have been developed and archived can inform applied management spatially across landscapes. The approaches, tools and outcomes will benefit management efforts for DoD lands within Alaska but also have utility for applications including other land management agencies and efforts.