Evaluating trace gas sampling strategies with assistance from a global 3D photochemical model: case studies for CEPEX and NARE O3 profiles

A technique is presented for evaluating sampling strategies based on in situ measurements and output from a 3D photochemical model. The technique quantifies the expected ability of various sampling strategies to give a representative picture of the actual mean trace gas concentration at the location being sampled. The representativeness is considered as a function of three parameters: (1) the number of samples, (2) the time between individual samples, and (3) the total time from the first to last sample. The technique is applied to analyzing the quality of various sampling strategies for ozone above five locations, ranging from the equatorial Pacific to the northern Atlantic. The analysis is based on data obtained by O₃ sondes during the CEPEX and NARE campaigns, along with the output from the global 3D photochemistry-transport model MATCH-MPIC. It is found that for all regions, the synoptic time scale of 3–5 days is a critical parameter for O₃. As long as the interval between soundings does not exceed this time scale, the number of soundings within a specific period does not strongly affect the representativeness of the sample mean profile; instead, the total period from the first to last sample is the most important parameter. For longer intervals between soundings, the behavior becomes more complex, with certain intervals resulting in better or worse representativeness than other intervals for the same total number of soundings. The model output is shown to generally be able to capture the synoptic time scale variability, and thus can be useful for establishing the relationship between synoptic meteorology and sampling strategies for various locations. However, the model underestimates the variability due to sub-gridscale processes (mainly convection), and thus must be used more cautiously in regions where this is critical, such as in the tropical upper troposphere. Future analyses such as this using previous observations and model output could assist field investigators in planning sampling strategies for campaigns or long-term monitoring.