Three‐dimensional SF 6 data and tropospheric transport simulations: Signals, modeling accuracy, and implications for inverse modeling

Surface emissions of SF 6 are closely tied to human activity and thus fairly well known. They therefore can and have been used to evaluate tropospheric transport predicted by models. A range of new atmospheric SF 6 data permit us to expand on earlier studies. The purpose of this first of two papers is to characterize known and new transport constraints provided by the data and to use them to quantify predictive skill of the MOZART‐2 atmospheric chemistry and transport model. Main noteworthy observational constraints are (1) a well‐known steep N‐S gradient at the surface confined to an ≈40° wide latitude band in the tropics; (2) a fairly uniform N‐S gradient in the upper troposphere; (3) an increase in the temporal variation in upper troposphere Northern Hemisphere records with increasing latitude; (4) a negative SF 6 gradient in Northern Hemisphere vertical profiles from the surface to 8 km height, but a positive gradient in the Southern Hemisphere; and (5) a clear reflection in surface records of large‐scale seasonal atmosphere movements like the undulations of the Intertropical Convergence Zone (ITCZ). Comparison of observations with simulations reveal excellent modeling skills with regards to (1) large‐scale annual mean latitudinal gradients at remote surface sites (relative bias of N‐S hemisphere difference ≤ 5%) and aloft (≈10 km, relative bias ≤ 25%); (2) seasonality in signals at remote sites caused by large‐scale movements of the atmosphere; (3) time variation in upper troposphere records; (4) “faithfulness” of advective transport on timescales up to ≈1 week; and (5) the general shapes and seasonal variation of vertical profiles. The model (1) underestimates the variation in the vertical of profiles, particularly those from locations close to high emissions regions, and (2) overestimates the difference in SF 6 between the planetary boundary layer (PBL) and free troposphere over North America, and thus likely Eurasia, during winter by approximately a factor of 2 (STD ≈ 100%). The comparisons permit estimating lower bounds on representation errors which are large for sites close to continental outflow regions. Given the magnitude of the signals and signal variance, SF 6 provides a strong constraint on interhemispheric transport, PBL ventilation, dispersion pathways of northern midlatitude surface emissions through the upper troposphere, and large‐scale movements of the atmosphere.