A top-down approach of sources and non-photosynthetic sinks of carbonyl sulfide from atmospheric measurements over multiple years in the Paris region (France)

International audienceCarbonyl sulfide (COS) has been proposed as a proxy for carbon dioxide (CO2) taken up by plants at the leaf and ecosystem scales. However, several additional production and removal processes have been identified which could complicate its use at larger scales, among which are soil uptake, dark uptake by plants, and soil and anthropogenic emissions. This study evaluates the significance of these processes at the regional scale through a top-down approach based on atmospheric COS measurements at Gif-sur-Yvette (GIF), a suburban site near Paris (France). Over a period of four and a half years, hourly measurements at 7 m above ground level were performed by gas chromatography and combined with 222Radon measurements to calculate nocturnal COS fluxes using the Radon-Tracer Method. In addition, the vertical distribution of COS was investigated at a second site, 2 km away from GIF, where a fast gas analyzer deployed on a 100 m tower for several months during winter 2015–2016 recorded mixing ratios at 3 heights (15, 60 and 100 m). COS appears to be homogeneously distributed both horizontally and vertically in the sampling area. The main finding is that the area is a persistent COS sink even during wintertime episodes of strong pollution. Nighttime net uptake rates ranged from -1.5 to -32.8 pmol m-2 s-1, with an average of -7.3 ± 4.5 pmol m-2 s-1 (n = 253). However, episodes of biogenic emissions happened each year in June-July (11.9 ± 6.2 pmol m-2 s-1, n = 24). Preliminary analyses of simulated footprints of source areas influencing the recorded COS data suggest that long-range transport of COS from anthropogenic sources located in Benelux, Eastern France and Germany occasionally impacts the Paris area during wintertime. These production and removal processes may limit the use of COS to assess regional-scale CO2 uptake in Europe by plants through inverse modeling