Evaluation of modelled climatologies of O3, CO, water vapour and NOy in the upper troposphere–lower stratosphere using regular in situ observations by passenger aircraft

Evaluating global chemistry models in the upper troposphere–lower stratosphere (UTLS) is an important step toward an improved understanding of the chemical composition in this region. This composition is regularly sampled through in situ measurements based on passenger aircraft, in the framework of the In-service Aircraft for a Global Observing System (IAGOS) research infrastructure. This study focuses on the comparison of the IAGOS measurements in ozone, carbon monoxide (CO), nitrogen reactive species (NOy) and water vapour, with a 25-year simulation output from the LMDZ-OR-INCA chemistry–climate model. For this purpose, we present and apply an extension of the Interpol-IAGOS software that projects the IAGOS data onto any model grid, in order to derive a gridded IAGOS product and a masked (sub-sampled) model product that are directly comparable to one another. Climatologies are calculated in the upper troposphere (UT) and in the lower stratosphere (LS) separately but also in the UTLS as a whole, as a demonstration for the models that do not sort out the physical variables necessary to distinguish between the UT and the LS. In the northern extratropics, the comparison in the UTLS layer suggests that the geographical distribution in the tropopause height is well reproduced by the model. In the separated layers, the model simulates well the water vapour climatologies in the UT and the ozone climatologies in the LS. There are opposite biases in CO in both UT and LS, which suggests that the cross-tropopause transport is overestimated. The NOy observations highlight the difficulty of the model in parameterizing the lightning emissions. In the tropics, the upper-tropospheric climatologies are remarkably well simulated for water vapour. They also show realistic CO peaks due to biomass burning in the most convective systems, and the ozone latitudinal variations are well correlated between the observations and the model. Ozone is more sensitive to lightning emissions than to biomass burning emissions, whereas the CO sensitivity to biomass burning emissions strongly depends on location and season. The present study demonstrates that the Interpol-IAGOS software is a tool facilitating the assessment of global model simulations in the UTLS, which is potentially useful for any modelling experiment involving chemistry climate or chemistry transport models.