Glass plate sampling efficiency for trace gases in the sea surface microlayer

Abstract

Many climate-active trace gases in the atmosphere are closely linked to production and consumption in the ocean, which are, in turn, influenced by the sea surface microlayer (SML). The SML is the upper most layer of the ocean with up to 1 mm thickness, often enriched in organics. Studies of trace gases in the SML aim to identify and quantify potential processes unique to the SML and to understand the SML's influence on the transfer between air and sea. Established sampling techniques of the SML (e.g., glass plate, mesh screen) are associated with high losses for the volatile trace gases. Despite the high losses, in this study we find that meaningful analysis of glass plate samples for the studied trace gases is possible. We experimentally determined the sampling efficiency for the short-lived trace gases dimethyl sulfide (DMS), isoprene, and carbon disulfide (CS2). Water temperature and trace gas concentration were the main drivers for sampling efficiency variability, while salinity and the number of dips of the glass plate were not significant. The (physicochemical) effect of surface-active substances (surfactants), modelled by the artificial surfactant Triton X-100, could not finally be untangled. Although our results are consistent within our experiments, we do not quantify a sampling efficiency to correct individual measurements, as our experiments did not encompass the full suite of environmental parameters normally encountered in the field, particularly limited to cases of oversaturation. Instead, we discuss using the identified sampling efficiencies of 0.13 ± 0.01 (±standard error) for DMS and isoprene, and 0.12 ± 0.01 for CS2 as thresholds to identify cases of enrichment in the SML. Future studies should extend to long-lived species (e.g., nitrous oxide, methane), encompass the full range of trace gas saturation, investigate a range of single surfactants as well as mixtures, include the effect of wind, and be repeated for the mesh screen. We hypothesize that a correction of individual measurements requires determining sampling efficiency as a function of environmental parameters, which were limited in the present study.