Relative uptake of carbonyl sulphide to carbon dioxide: insights from a coupled boundary layer – canopy inverse modelling framework

Abstract. Carbonyl sulphide (COS) is an atmospheric trace gas that has been suggested as a proxy to estimate carbon uptake by plants. To this end, the concept of leaf relative uptake (LRU), the ratio of deposition velocities of COS and CO2, has been introduced to obtain plant CO2 uptake fluxes from COS flux measurements. In our study we use a coupled soil – canopy – atmospheric mixed layer model to simulate CO2 and COS uptake by vegetation explicitly, and derive LRU. In this modelling framework, the exchange of COS is coupled to the exchange of H2O and CO2 via stomatal conductance. The latter is calculated using an assimilation–stomatal conductance (A–gs) photosynthesis model, accounting for separate exchange at sunlit and shaded leaves. Despite limited complexity, our coupled model includes most key processes involved in daytime land atmosphere exchange. The model is embedded in an inverse modelling framework, allowing for a structured model parameter estimation. We performed a parameter optimisation for a boreal forest in Finland (Hyytiälä), using observation data from July 2015. We took a holistic approach and aimed to obtain model parameters consistent with a large set of observations, including COS and CO2 molar fractions (measured in and above the canopy) and fluxes. By optimising parameters, we obtained a good fit to many observation types simultaneously. Analysing the corresponding modelled LRU, we found strong within-canopy variations at the leaf scale, with highest LRU values for shaded leaves near the bottom of the canopy. These variations can be explained to a large extent by differences in photosynthetically active radiation (PAR), vapour pressure and leaf temperature. Based on these findings, we propose a new parameterisation of canopy-scale LRU based on absorbed PAR and vapour-pressure deficit of sunlit leaves near the canopy top. We performed several additional optimisations, without re-optimising leaf exchange parameters: two for the same location, but for the months August and September, and two for a needleleaf forest in Austria (Mieming). We obtained a generally good fit with observations in all of these optimisations, suggesting transferability of model parameters to different months and locations. When testing the LRU parameterisation using Hyytiälä model data from August and September (data not used for deriving the parameterisation), the results of the physical model were well-approximated, although observations suggest somewhat lower LRU values for a large part of the day. For Mieming, the parameterisation also provided a satisfactory fit to the physical model. For both locations we found that the LRU of sunlit leaves near the top of the canopy provides a good approximation of the canopy-scale LRU. Our results provide insight in the behaviour of LRU in the canopy, and the new parameterisation, based on both absorbed PAR and vapour pressure deficit, can contribute to improving COS-based ecosystem plant carbon uptake estimates in needleleaf ecosystems, but further validation is needed.