CO₂ baseline concept in 3-D atmospheric transport models

Abstract

Atmospheric transport models are now commonly used to estimate the distribution of sources and sinks of CO₂, given atmospheric measurements of this greenhouse gas. Yet, to properly assess large-scale flux patterns, one must employ only those measurements that are representative of large scale air masses. Typically the observed time series are filtered, according to different criteria, as a mean to select baseline conditions. In this paper we raise the problem of comparing selected data with the whole simulated time series. Aiming for better consistency with the observations, we propose four procedures to select the baseline periods in the transport models. Each is tested with a high resolution 3-D model (TM2_z), at 3 monitoring stations currently used by the modeling community (Cape Grim, Cape Meares, and Amsterdam Island). Results demonstrate that such selection is necessary, especially for coastal stations. By using information on the synoptic transport as estimated by either 5-days backtrajectories or radon-222 concentrations, the CO₂ synoptic scale variability (averaged over the year) is reduced by a factor of three. The seasonal cycle (peak-to-peak amplitude) is also reduced by about 30% for the two coastal stations. With these corrections, the model-data agreement improves dramatically. Moreover, annual averages can be affected by few tenths of ppmv. Finally, we also investigate synoptic events classified as non-baseline both in observations and simulations. The ability of the 3-D model to reproduce some characteristics of these data, points out the potential of such data to characterize the strength of regional fluxes.