Spatial and Temporal Representativeness of ATom Transects Using GEOS-5 and GMI–CTM Simulations

ATom provides a contiguous, global-scale dataset of the remote troposphere using transects through the middle of the Pacific and Atlantic Ocean basins. The GEOS-5 global model reproduces the observed background CO levels very well, including the timing and topography of CO enhancements during the ATom mission. In this study, we use these GEOS-5 CO simulations to investigate the spatial and temporal representativeness of the specific ATom flights during ATom-1 and -2. Our model analysis shows that probability distribution (PD) of simulated CO over regional Pacific latitudinal bands (e.g., 20˚N-40˚N, 20˚S-20˚N, 20˚S-80˚S) is similar to the ATom transects, demonstrating that ATom-sampled CO is likely representative of typical regional concentrations in August. Over the North Pacific (40˚N-80˚N), the CO distribution on the ATom-1 path peaks around 85 ppb, which is about 10 ppb less than that over the larger region. A second peak in the PDs low tail indicates a possible perturbation from stratospheric very-low CO air. Over the tropical and subtropical south Atlantic, CO is mainly driven by southern African biomass burning during ATom-1 period. CO along ATom transects over tropical Atlantic shows “skewed right" distribution with a higher peak compared to the broader region, indicating a stronger contribution from Africa to the transect than to the mean Atlantic basin. PDs for simulated CO along ATom flights are compared with the monthly-accumulated PDs to test the sensitivity of ATom results to episodic transport events. In addition to the GESO-5 CO statistics, we also take a quick look at the NASA Global Modeling Initiative – Chemistry transport model (GMI-CTM) results, which are driven by assimilated MERRA-2 meteorological fields for the ATom-1 and -2 flight periods. This full-chemistry simulation allows us to include similar statistical analyses of PDs for the ATom ozone, which is reproduced well by the GMI-CTM in magnitude and variability as suggested by the initial model-data comparison.