Inferring Gross Primary Production and Respiration From a Global Carbon Cycle Model Including Carbonyl Sulfide

Carbonly sulfide (COS), an analog of carbon dioxide is emerging as a useful atmospheric tracer of carbon cycle processes. Previous studies have shown that COS is taken up by leaves in reactions associated with photosynthesis and that the rate of its uptake is closely linked to the rate of gross primary production (GPP). However, unlike CO2, there is apparently no significant source of COS from terrestrial ecosystems. Therefore, changes in the concentration of COS in the atmosphere over these ecosystems reflects the rate of photosynthesis and is largely independent of the rate of respiration (RESP), while that of CO2 reflects the net sum, GPP + RESP = NEE. The potential significance of this can be seen by considering COS and CO2 exchange in a closed box containing either an ecosystem or a leaf from that ecosystem. For a leaf, the ratio of COS/CO2 uptake normalized by the ratio of COS/CO2 concentration (X) is observed to be about 2, while this ratio for an ecosystem (Y) can range from 3 -10. If we know X, and we can measure Y, we may calculate that, GPP = NEE*Y/X and, RESP=NEE*(Y/X-1). Thus, simultaneous measurements of COS and CO2 exchange could provide new information for carbon cycle studies. Toward this end, we have incorporated the biochemical and biophysical mechanisms controlling COS exchange into a land surface model (SIB) and we have used this model to simulate global COS and CO2 fluxes and transported these together with other known sources and sinks in a chemical transport model (PCTM). The model exhibits good skill in simulating observations of the seasonal cycle and vertical profiles of COS and CO2 concentration from NOAA and INTEX-NA over N. America. Though the major features of the observations are captured in the model calculation, discrepancies remain and illustrate some problems in the parameterization of COS fluxes. We use these "modeled data" to test the feasibility of COS-based GPP and RESP estimation.