Constraining Northern Carbon Fluxes with Atmospheric Carbon Dioxide Observations

Abstract

Accurate estimates of land carbon fluxes at continental and regional spatial scales require increased understanding of site-to-site differences in seasonal carbon exchange and robust benchmarks against which to evaluate model performance. In this thesis, we use measurements of atmospheric CO2 to develop a new observationally-based metric related to the land carbon flux against which to evaluate model performance and we probe land-atmosphere carbon exchange across spatial and temporal scales to increase understanding of the spatial distribution of carbon sources and sinks. First, the northern extratropical growing season net flux (GSNF) is estimated using aircraft profiles of CO2 measured over the remote oceans. This GSNF is shown to be a robust model benchmark. The northern extratropical GSNF is estimated to be 5.7 ± 0.3 Pg C and coupled model intercomparison project phase 5 (CMIP5) and phase 6 (CMIP6) models are shown to underestimate the GSNF and overestimate the growing season length on average when compared to the observations. This result provides a new and robust observational target of large-scale land carbon flux for prognostic model evaluation. Second, an emergent constraint approach is applied to prognostic model GSNF to estimate northern extratropical annual fluxes for gross primary productivity (GPP), heterotrophic respiration (RH), and net primary productivity (NPP). Our GSNF-constrained value of 56 ± 15 Pg C for GPP is 8 Pg C larger than a commonly used estimates derived from upscaled flux towers. Our larger hemispheric GPP estimate in comparison to that from upscaled flux towers indicates that estimates for global GPP may be on the higher end of the current range. Third, we estimate the northern hemisphere seasonal cycle by upscaling the Total Carbon Column Observing Network (TCCON) to determine how representative the network is of the northern hemisphere. The results indicate that the TCCON is not yet representative of the northern hemisphere. This is likely due to sparse coverage in the high-latitudes. Increasing the coverage and of the network may improve the representativeness of the network. Lastly, to understand spatial heterogeneity in seasonal cycle amplitudes, we delved into diurnal cycle information from TCCON sites and spatially resolved variations in productivity from space-based data. We saw that saw that nearly all of the variability in the seasonal cycle can be explained by mean March potential temperature at 700 hPa, mean December diurnal cycle amplitude, and maximum annual solar induced chlorophyll fluorescence at a given site.