Simulation of tropospheric chemistry and aerosols with the climate model EC-Earth

We have integrated the atmospheric chemistry and transport model TM5 into the global climate model EC-Earth version 2.4. We present an overview of the TM5 model and the two-way data exchange between TM5 and the IFS model from the European Centre for Medium-Range Weather Forecasts (ECMWF), the atmospheric general circulation model of EC-Earth. In this paper we evaluate the simulation of tropospheric chemistry and aerosols in a oneway coupled configuration. We have carried out a decadal simulation for present-day conditions and calculated chemical budgets and climatologies of tracer concentrations and aerosol optical depth. For comparison we have also performed offline simulations driven by meteorological fields from ECMWF’s ERA-Interim reanalysis and output from the EC-Earth model itself. Compared to the offline simulations, the online-coupled system produces more efficient vertical mixing in the troposphere, which reflects an improvement of the treatment of cumulus convection. The chemistry in the EC-Earth simulations is affected by the fact that the current version of EC-Earth produces a cold bias with too dry air in large parts of the troposphere. Compared to the ERA-Interim driven simulation, the oxidizing capacity in EC-Earth is lower in the tropics and higher in the extratropics. The atmospheric lifetime of methane in EC-Earth is 9.4 years, which is 7 % longer than the lifetime obtained with ERA-Interim but remains well within the range reported in the literature. We further evaluate the model by comparing the simulated climatologies of surface radon-222 and carbon monoxide, tropospheric and surface ozone, and aerosol optical depth against observational data. The work presented in this study is the first step in the development of EC-Earth into an Earth system model with fully interactive atmospheric chemistry and aerosols.

The paper itself is well structured and well written. The rationale behind implementing chemistry and aerosols in Earth System models is established well. The experimental set up of the various present-day simulations carried out is described in sufficient detail. However, there are aspects of the gas-phase chemistry evaluation which could be more comprehensive and discussed quantitatively rather than qualitatively, and could include more comparisons with observations rather than solely relying on comparisons with the offline simulations. Details can be found under "Specific Comments". In relation to the aerosol evaluation, the focus is solely on aerosol optical depth and some recommendations for further evaluation are detailed under "Specific Comments". However, on balance, once the specific comments are adequately addressed, the paper will be a useful addition to the scientific community in documenting the development of the EC-Earth climate model and will be wholly suitable for publication in Geoscientific Model Development.
Specific Comments: 1. The introduction includes aspects on the role of stratospheric chemistry and stratospheric aerosols in the Earth system. Given that TM5 does not include these aspects, I suggest that they be removed or reduced due to lack of relevance for the current model description.
2. The version of TM5 being coupled to IFS includes aqueous-phase chemistry for the oxidation of dissolved SO2 by O3 and H2O2 but details of this chemistry haven't been included either in this paper or that of Huijnen et al. (2010). Please add sufficient details.
3. Can you provide some indication of the increase in computational cost of EC-Earth C459 when TM5 is included? In particular, it would be useful to know what additional cost comes from the OASIS coupler. 4. Although this paper isn't detailing aspects of the TM5 model, it would still be useful to include some information on deposition processes. Can you also include an explicit statement on whether there is any coupling between convective transport, for example, and wet deposition? 5. The implementation of emission heights has been altered in TM5 since the publication of Huijnen et al. (2010). Can you discuss the rationale behind these changes? What were they based on? Are the emission heights identical between the offline and online simulations? Further details would be useful here.
6. Given the importance of the specific humidity bias in EC-Earth on global mean OH and the oxidizing capacity of the atmosphere, can you include an equivalent plot to Figure 1 but for specific humidity? Some model physics changes (e.g. convection) can affect humidity without a corresponding change in temperature. It is also worth putting these biases in the context of other climate models.
7. The evaluation of 222Rn consisted primarily of comparisons with the offline simulations. It would greatly help if the comparisons could be extended to include observations. Despite the simplicity of the tracer experiment and the emissions used, it is still a useful tool for model assessment. Does the online simulation perform worse or better than the offline simulation relative to observations? The assessment could also usefully be extended to include that of 210Pb. 10. The evaluation of the aerosol component of TM5 in the online simulations has focussed solely on aerosol optical depth. No comparison of aerosol precursor gases (e.g. SO2) with observations is included. No comparison of component aerosol burdens (e.g. sulphate, dust, organic carbon) with observations is included. No comparison between aerosol budgets (e.g. primary production, secondary production, burdens, lifetimes etc..) between the offline and online simulations is presented. These would greatly extend the evaluation of the aersosol component of EC-Earth and would make a useful and valuable addition to the manuscript. It would also improve the balance of the paper between the gas-phase chemistry and the aerosol evaluation.
11. In a number of instances, there are differences between the offline and online simulations (e.g. CO lifetime, chemical destruction of CO, CO burden, as examples from Section 4.4). It would be useful to establish whether some of these differences (and those in other sections) are statistically significant and at what confidence interval.
12. Finally, there are a number of instances in the manuscript, where the comparisons between simulations or comparisons between simulations and observations could be made more quantitative. As an example, in Section 4.1 (pg 1952, line 11), cold and warm biases in EC-Earth are discussed but there is no detail in the text on how large these biases are and in which seasons they apply? The same is also applicable in Sections 4.3, 4.4, 4.5, and 4.6 -the inclusion of quantitative measures of skill in the manuscript will provide a useful benchmark against which successive model improvements can be assessed.