Measurement of Atmospheric Concentration of CO2 in the Hudson Bay Lowlands: An Application of a Lagrangian Particle Dispersion Model (STILT)

Atmospheric CO2 concentrations are influenced by surface fluxes, as well as advection and vertical mixing on the way to the measurement tower. The capability of transport models to accurately represent air parcel trajectories and footprints is crucial in inverse analysis. This study employs the Stochastic Time-inverted Lagrangian Transport model (STILT), driven by meteorological inputs from the North American Regional Reanalysis (NARR), to simulate atmospheric CO2 in the Hudson Bay Lowlands. The primary objectives include: (1) Characterize daily, seasonal and interannual variations of atmospheric CO2 for a 5-year (2008-2012) period; (2) Evaluate the performance of the STILT model, and CarbonTracker flux estimates. STILT-modelled CO2 concentrations compare reasonably against observations. The mean model bias was -0.57 ppm at Churchill, and -2.44 ppm at Fraserdale. Smoothed seasonal curves fitted to the daily afternoon data revealed that model bias was highest during summertime, particularly over the Fraserdale region. This disparity between modelled and observed results are attributed to transport errors related to advection and PBL mixing.