- Main
Terrestrial carbon fluxes and pools under a changing climate
- Nguyen, Bao Ngoc
- Advisor(s): Keenan, Trevor F
Abstract
Terrestrial ecosystems have captured 21% of human-emitted CO2 over the last decade, highlighting the importance of studying their carbon cycle dynamics under a changing climate. The terrestrial carbon cycle has two main parts: carbon fluxes and pools, which represent short-term carbon exchanges such as between biosphere-atmosphere, and long-term carbon storage such as in vegetation and soils. When climate becomes more variable with changing precipitation regimes, increasing atmospheric CO2 concentration, and warmer temperature, how terrestrial carbon flux and pools respond to these changes determines the ecosystems’ carbon storage potential.Throughout my Ph.D. dissertation, I utilize the global network of eddy-covariance CO2 flux tower measurements across diverse ecosystems (FLUXNET) together with tree-ring data, a geospatial foundation model, and Dynamic Global Vegetation Models (DGVMs) to answer important scientific questions about terrestrial carbon fluxes and pools under a changing climate. Specifically, my dissertation (1) models rain-induced CO2 pulses from global drylands, the processes that have been largely ignored in dryland carbon cycle modeling, through the FluxPulse framework, (2) spatializes CO2 fluxes at a 10-m resolution within tower measurement areas to support carbon monitoring of lands of interest, through the FluxLens framework, and (3) demonstrates that the recently observed short-term decoupling between photosynthesis and plant growth neither implies sink limitation nor constrains long-term growth.Firstly, climate change leads to increases in rainfall intensity and decreases in frequency, making water-limited ecosystems more vulnerable. In drylands, vegetation and soil microbes rely on intermittent rainfall to thrive and regulate ecosystem carbon dynamics, yet the carbon pulses resulting from sporadic rainfall events have been extremely understudied. In the first chapter, I utilize the FLUXNET global database of ecosystem carbon fluxes to identify a widespread underestimation of rain-induced carbon losses from global drylands. I build an unprecedented dataset of manually and machine-learning labeled rain-pulse events, including more than 1800 rain-induced carbon loss events across 323 site-years at 34 dryland eddy-covariance sites worldwide. I find that rain-induced carbon losses contribute significantly to annual net ecosystem productivity and ecosystem respiration, but existing eddy-covariance carbon flux models significantly underestimate the response of ecosystem respiration and vegetation productivity to rainfall events by up to 30%. Motivated by these results, I develop FluxPulse, a novel approach to resolve rain-induced biases in current carbon flux models.Secondly, in an era where precise quantification of land carbon uptake and release becomes more and more important for ecosystem carbon monitoring, I extend my focus from drylands to global terrestrial ecosystems and introduce the Flux observation Local ENvironment Scaling (FluxLens), a novel framework for spatializing carbon fluxes within flux tower areas at 10-meter resolution. FluxLens uses Google AlphaEarth Foundations (AEF) annual embedding datasets and climate datasets to predict monthly estimates of gross primary productivity (GPP) and ecosystem respiration (Reco) at 10-m spatial resolution within tower measurement areas, across 184 global eddy-covariance tower sites. Across sites, FluxLens captures well CO2 flux magnitude, seasonal dynamics, and visible spatial nonuniformity, with the mean predictions across all pixels consistent with flux-tower GPP and Reco estimates (Site-specific pixel-mean R2 GPP = 0.65-0.94 (25th-75th percentile); R2 Reco = 0.58-0.88).Finally, rising atmospheric CO2 has increased global photosynthesis by roughly 14% in the last 40 years, but whether this increase in carbon source activity translates into lasting woody biomass gains remains uncertain. Resolving this question is central to evaluating the potential of terrestrial ecosystems to continue slowing climate change in the coming decades. Recent studies have reported a short-term decoupling between photosynthesis and woody biomass growth and interpreted this as evidence of emerging or intensifying sink limitation. By combining tree-ring data, eddy-covariance observations, and an ensemble of global vegetation models simulating carbon fluxes and pools from annual to centennial scales, I show widespread decoupling between photosynthesis and growth. Crucially, however, this decoupling has no bearing on long-term growth outcomes under elevated CO2. My third chapter shows that short-term decoupling arises primarily from the dominant role of processes downstream of photosynthesis, and that its magnitude is unrelated to whether ecosystems are source- or sink-limited. These findings call into question previous reports of a strengthening sink limitation, and suggest that plants may not be as constrained under elevated CO2 variability as currently assumed.My dissertation answers important and relevant questions about the long-term relationship between carbon fluxes and pools under a changing climate, and develops two novel frameworks: FluxPulse for modeling rain-induced CO2 pulses from global drylands and FluxLens for spatializing tower-based carbon fluxes at 10-meter resolution. Chapter 1 provides the first global assessment of the impact and drivers of rain-induced carbon losses across global drylands since their first official documentation in the 1950s. Chapter 2 leverages an emerging geospatial foundation model to provide spatially explicit carbon flux estimates with promising applications in ecosystem monitoring, land management, high-resolution satellite product evaluation, and plot-to-tower comparisons. Chapter 3 challenges previous reports of a strengthening sink limitation, and suggests that plants may not be as constrained under elevated CO2 variability as currently assumed. Together, these chapters contribute to the understanding of terrestrial carbon fluxes and pools under a changing climate.