Implementation of vegetation and soil carbonyl sulfide exchanges in the ORCHIDEE land surface model to better constrain ecosystem gross primary production

Better constraining the ecosystem gross photosynthetic CO₂ uptake (GPP) is necessary to reduce the uncertainties on continental vegetation response to climate change. As GPP cannot be directly measured at the ecosystem scale, different proxies of vegetation CO₂ uptake have emerged. These proxies are essential for land surface modelers to estimate GPP at large scale. Carbonyl sulfide (COS) shows many similarities with CO₂, following the same diffusional pathways through the leaves. However, unlike CO₂ that is also emitted by plants during respiration, COS is essentially only taken up by leaves and not re-emitted back to the atmosphere. Therefore, COS is a promising proxy of photosynthetic activity. In previous studies, fixed values of the leaf relative uptake (LRU) ratio of COS to CO₂ fluxes normalized by their respective concentration have typically been used to infer GPP for the different biomes. However, it has been shown that LRU ratio changes with varying Photosynthetically Active Radiation (PAR), which limits its accuracy to constrain photosynthetic activity. Therefore, we redefined the COS-based GPP estimation approach to better capture GPP response to changing environmental conditions, by implementing a mechanistic model of COS exchange by continental vegetation in the ORCHIDEE land surface model. We compared the modelled fluxes against field measurements at two sites and studied the model behavior and environmental drivers. Then, we ran global simulations and computed the annual COS vegetation uptake that was found in the middle range values of previous reported budgets (-490 to -1335 Gg S yr^-1), with -756 Gg S yr^-1. The simulated fluxes were transported, and COS concentrations were evaluated against measurements from the NOAA atmospheric stations. Our results show that the mechanistic approach is more appropriate when studying photosynthetic activity at high temporal resolution, but similar results in concentrations are obtained between the mechanistic and LRU approaches at the global scale. Accurate evaluation of the continental vegetation COS uptake is necessary as it is the main COS sink. However, COS can also be absorbed or emitted by soils, a flux that complicates the use of eddy covariance COS flux measurements or atmospheric COS measurements to derive information on GPP estimates. Therefore, the soil COS exchange should also be represented in land surface models. We implemented two soil COS exchange models in ORCHIDEE, a mechanistic model (based on Ogée et al. 2016) and a second model based on an empirical relationship with soil respiration (following Berry et al., 2013). We evaluated the two models at several sites against field measurements. We also performed global simulations to evaluate the spatial distribution of soil COS fluxes and their seasonal variations. Finally, we estimated the contributions of the combined impact of soil COS exchange and leaf COS uptake (both from the ORCHIDEE model) to the global COS budget and on the COS atmospheric concentration latitudinal gradient.