Coupling CLM3.5 with the COSMO regional model Edouard L. Davin, Reto Stöckli, Paulo J. C. Oliveira, Sonia I. Seneviratne Institute for Atmospheric and Climate Science, ETH Zürich The goal of this presentation is to give you a brief overview of our project to developp a new regional climate model, which will be a combination of the COSMO model as the atmospheric component and the CLM as the land surface componant. LMWG meeting, Breckenridge, June 2008 1

The Land-Climate Interactions group @ ETH Context The Land-Climate Interactions group @ ETH This is first a more general introduction about our group and what link us to the CLM. So the land-climate interaction group is led by SS and hosted by the IAC at ETH Zurich. The group exists only since 2007 and it has been growing fast since that time as you see on that picture. We currently have 2 projects involving CLM in our group: *one project is to use CLM in stand alone mode in order to better understand past change in hydrological cycle and also to explore possible change in the future. *the second project which is the topic of this presentation, rely on the coupling of the Community Land Model with the COSMO RCM. The implementation of the coupling has been already started by RS and since last month it is me who are in charge of continuing this project during my post-doc. 2 projects using CLM3.5 are starting in the group: PhD Paulo Oliveira: Past and future trends in the hydrological cycle (poster). Coupling of CLM3.5 with COSMO (started by Reto Stöckli) 2

The MAIOLICA project ETH-Funded project MAIOLICA (started april 2008). Goal: “explore the links between the atmosphere and the biosphere within the Earth system”. Activity 3a: “understanding of regional interactions of the biosphere with the climate at the scale of the European continent, for both present and future climate conditions“. The implementation of this coupling will be done within the framework of the MAIOLICA project. This is a project funded by ETH and which overall goal is to study biosphere-atmosphere interactions within the climate system. This will involve different spatial and temporal scales from tower measurements to global scale modeling. And more particularly we will be in charge of the regional climate modeling activity which focuses on biosphere-atmosphere interaction at the scale of the european continent. 3

The COSMO model Non-hydrostatic mesoscale model. LM 1.1 LM 2.19 LM 2.16 Climate-LM 1 Climate-LM 2 Climate-LM 2.x LM 3.x COSMO 4.0 Numerical weather prediction Regional climate simulations To achive this goal we will use the COSMO RCM. This shows a brief overview of this model. It is a non-H mesoscale model that has been historicaly developped by the german weather service for opperational forecasts. then from a latter version of this model, a climate version has been derived for regional climate simulations. The two branches have evolved in parralel in the last ten years. Finally the last released version is a unified model called COSMO that can be used both for weather forecast and for climate simulations. Non-hydrostatic mesoscale model. Community Model shared with several institutions in Europe: GKSS, ETH Zurich, JRC, DWD, MeteoSwiss, ... Resolution typically around 50km for climate simulations over Europe.

Land surface scheme The new COSMO-CLM RCM: COSMO CLM3.5 TERRA is currently integrated in COSMO as the land surface component. TERRA is mostly a “soil model” without explicit representation of vegetation processes. A more detailed land surface scheme is needed in order to represent biosphere-atmosphere interactions (coupling between photosynthesis and transpiration, root uptake, phenology, carbon cycle…). The new model represents an advance in the regional climate modelling community (e.g., EU-projects PRUDENCE & ENSEMBLES). The new COSMO-CLM RCM: COSMO Concerning the land surface parametrization, the COSMO model use a parametrization called TERRA. None of the RCMs involved in the PRUDENCE & ENS have this level of detail regarding vegetation processes. CLM3.5

Summer temperature interannual variability: COSMO minus CRU 20th century climate There is systematic biases in simulations of european climate by RCMs (e.g., RCMs within PRUDENCE). Could some of these biases be alleviated by a better representation of land surface processes? We will run a 20th century simulation with COSMO-CLM3.5 driven by ERA40 and compare it with existing simulations with COSMO and other RCMs (e.g. PRUDENCE, ENSEMBLES). Summer temperature interannual variability: COSMO minus CRU What can we expect from this new model. I think the first point is that we can expect that a better representation of land surface will help to improve the simulation of the 20th century climate. For instance there is still systematic biases in the simulation of the european climate by RCMs. This has been highlighted for example by the PRUDENCE project. (Beside these different aspects of the developpement of the model, I would like now to give some exemples of the questions we would like to adress in the future with this new model. First concerning the simulation of the 20th century climate it has been shown that current RCMs exhibit some systemtic biases. (Jaeger et al., 2008)

Future scenarios Identify land surface feedbacks and processes relevant for future climate projections over Europe. Atmosphere-sol moisture coupling. (Seneviratne et al, 2006) The ultimate goal motivating the implementation of this new tool, will be to be able to identify which processes and feedback may be particularly relevant for future climate projections over Europe. To explore this we will use the COSMO-CLM for simulations of 21st century. I noted here three processes/feedbacks that can be of particular interest for future climate change over Europe:

Future scenarios Identify land surface feedbacks and processes relevant for future climate projections over Europe. Simulations over the 21st century will be carried out with the new COSMO-CLM3.5 driven by a CGM. Atmosphere-sol moisture coupling. Change in vegetation functioning and structure (LAI, stomatal conductance..). Land cover change. The ultimate goal motivating the implementation of this new tool, will be to be able to identify which processes and feedback may be particularly relevant for future climate projections over Europe. To explore this we will use the COSMO-CLM for simulations of 21st century. I noted here three processes/feedbacks that can be of particular interest for future climate change over Europe:

Preliminary results Comparison of COSMO-CLM3.5 versus COSMO (1 year run without spinup) Annual mean latent heat (W/m2) Annual mean sensible heat (W/m2) Comparison with flux measurements at Tharandt (Germany) RS has already implemented an initial version of COSMO-CLM3.5 and this shows very preliminary results about how this version performs compared to the standard version. You can see first a comparison of the simulated mean fuxes of latent and sensible heat. There is a clear tendency of the new version to simulate lower evaporation and higher sensible heat. Now that we have seen that there is a clear difference in the simulated surface fluxes we may wonder if this represent an improvement with respect to observations. I have compared the simulated fluxes at the 50km grid cell with tower flux measurements.