Ecosystem CO2 exchange and atmosphere boundary layer (ABL) mixing are correlated diurnally and seasonally as they are both driven by solar insulation. Tracer transport models predict that these covariance signals produce a meridional gradient of annual mean CO2 concentration in the marine boundary layer that is half as strong as the signal produced by fossil fuel emissions. This rectifier effect is simulated by most global tracer transport models. However, observations to constrain the strength of these covariance signals in nature are lacking. We investigate the covariance between ecosystem carbon dioxide exchange and ABL dynamics by comparing one widely cited transport model with observations in the middle of the North American continent. We measured CO2 flux and mixing ratio using an eddy-covariance system from a 447-m tower in northern Wisconsin, mixed layer depths using a 915-MHz boundary layer profiling radar near the tower, and vertical CO2 profiles from aircraft in the vicinity of the tower. We find (1) that simulated and observed net daily CO2 fluxes are similar; (2) the simulated maximum ABL depths were too shallow throughout year; (3) the simulated seasonal variability of the CO2mixing ratio in the lowest layer of the free troposphere is 3 ppm smaller than that inferred from a mixed layer jump model and boundary layer observations; and (4) the simulated diurnal and seasonal covariance between CO2 flux and mixing ratio are weaker than the observed covariance. The comparison between model and observations is limited by the questionable representativeness of a single observing site and a bias towards fair weather observing conditions.