Constraining sources and sinks of carbon at the regional scale with aircraft observations.

摘要

Anthropogenic perturbations to the carbon cycle have led to increasing atmospheric concentrations of the radiatively active species CO₂. Quantitative knowledge of terrestrial carbon sources and sinks at the regional scale (102∼103 km)—the scale of political borders and ecosystems—remains highly uncertain. CO₂ measurements have generally been collected in the remote marine boundary-layer, where signals of terrestrial fluxes have already been attenuated. CO₂ concentrations over the continent contain strong terrestrial signals but exhibit large spatio-temporal variability that challenges current-generation analysis frameworks.; This dissertation seeks to address the current uncertainties in regional scale carbon sources and sinks. A quantitative characterization of the spatial variability of CO₂ was carried out based on an extensive dataset compiled from aircraft observations over the Pacific and North America. The variability was used to determine the deviation between point observations and gridcell-averaged value used by models (“representation error”). The higher variability of continental CO₂ led to significantly larger representation errors than over the ocean. The Stochastic Time-Inverted Lagrangian Transport (STILT) model was developed to resolve the variability in tracers like CO₂ and minimize the representation error. STILT simulates atmospheric transport backward in time using particle ensembles starting from the locations of point observations that link the observations to upstream sources/sinks. Empirical assessments of time-reversibility revealed that mass violation in the wind fields led to time-irreversibility; backward-time results were identical to those from forward-time simulations when mass conserving winds were used.; The constraint on regional sources/sinks provided by continental CO ₂ observations was explored in the CO₂ Budget and Rectification Airborne (COBRA) pilot study. A receptor-oriented analysis framework which merges the regional scale atmospheric constraint with detailed information from ground-based measurements is presented. The STILT model was further applied as a flight planning tool during COBRA to predict locations of upstream air parcels and implement Lagrangian experiments, which minimized uncertainties in upstream tracer values and imposed direct constraints on regional scale fluxes.; The remaining limitations to the presented work and the necessary steps forward are discussed within the context of a general information transfer problem.