Open Collections

Abstract

Vegetation productivity across the boreal ecosystem is undergoing rapid change in response to climate change, introducing uncertainty into the quantification of regional, national, and global carbon budgets. Conventional monitoring of these dynamics relies on the fusion of coarse remote sensing observations, sparsely distributed in-situ data, interpolated climate information, and various other geospatial datasets. The synergy of these datasets has conventionally limited the detail that can be achieved in modelled productivity estimates, making it difficult to accurately assess changes across the highly heterogeneous boreal landscape. Recent advances in satellite-based observations of key boreal productivity drivers offer a promising opportunity to address these challenges. Capitalizing on this opportunity, this dissertation investigates how estimates of boreal Gross Primary Productivity (GPP) can be enhanced through the use of multi-source remote sensing datasets and analysis approaches. To this end, a comprehensive review of current issues and opportunities in boreal productivity monitoring efforts was performed. Based on demonstrated limitations in a current remote sensing-based GPP modelling framework, I then proposed the new CANadian Temperature Greenness (CAN-TG) framework based on observations of surface temperature, vegetation greenness, soil moisture conditions, and soil freeze/thaw states. These variables represent key controls on boreal productivity, capturing information on vegetation physiology, hydrological constraints, and seasonal phenology at varying spatial scales. To illustrate how the particular use of freeze/thaw information may benefit this new framework, novel methods were then presented to use this data to characterize seasonal constraints on boreal GPP. Across a selection of boreal focus sites, I then assessed the ability of the new CAN-TG framework in constraining estimates of GPP, demonstrating improved r2 (+0.15) and reduced RMSE (-65%) compared to a benchmark model. Finally, CAN-TG was then applied across the northern Canadian treeline. Whilst maintaining high agreement with independent reference GPP (r2 = ~0.91), fine-scale estimates of GPP provided detailed insights into changing productivity trends at previously unavailable (30 m) spatial resolutions. Through this work, this dissertation demonstrates that GPP estimates in boreal environments can be meaningfully constrained and enhanced through the use of a robust, comprehensive, and scalable framework based exclusively on remote sensing-based datasets.