Proposal for the intercomparison and systematic evaluation of global and regional chemistry transport models in the frame of AMMAMIP. Frédéric Hourdin, novembre 2006 Introduction This text presents a proposition of extension of the AMMAModel Intercomparison Project (AMMA MIP) to the intercomparison of chemistry transport models. The main goal of this proposal is to provide a framework which may help compare and validate the atmospheric composition together with an evaluation of the transport (both large scale and convective) as preformed already in the AMMAMIP project. In order to illustrate the ideas behind the proposal, we present some results of a study performed recently with the coupled climatechemistry model, LMDZINCA, endowed with additional idealized tracers. We also illustrate the potential interest of this approach based on recent observations of chemical compounds during the AMMA 2006 field campaign and comparison with modelling before giving the baseline of what the chemical part of this intercomparison program may be. AMMAMIP AMMAMIP is a light Model Intercomparison Project built in the frame of the AMMA program. In practice, contributing teams are asked to send subsets of outputs of their climate models, for specific years (2000, 2003 and soon 2005 and 2006). A series of sytematic diagnoistics, are generated and made accessible on a web site together with similar graphs issued from observations. The model outputs are also accessible directly for finer diagnostics. The intercomparison is organized in two sub sets. − AMMAcroos, the first part, is dedicated to the study of the mean latitudinal structure of the monsoon. It is motivated by the relative “zonallity” of the WAM. The crosssection consists in zonal averages of model outputs (10E10W). − AMMAmaps is dedicated to the study of synoptic dynamics (waves, onset, breaks, ...) and stationnary structures (orography, ocean/coninent, ...). It is based on 2D maps provided at a daily frequency. The idea is to compare results in a climate sense (mean state, sstatistics, ...). Simulations of a few days dedicated to case studies are not intended to be used here. So we ask modellers to provide a full year of results, or at least several months (at least june to octobre). A more detailed description can be found at http://ammamip.lmd.jussieu.fr LMDZINCA results. The results presented here are based on numerical simulations performed with two versions of the LMDZINCA model which only differ by the way deep convection is parameterized. The idea here is not to detail the description and evaluation of those two schemes but to use those two different runs to illustrate why we think the crosssection approach could be a good framework for model intercomparison and validation, and the interest of idealized tracers to isolate the role of transport from that of chemistry and emissions. er Fig 1 : Illustration of the difference Emanuel/Tiedtke in the AMMAcross framework. Top : mean meridional circulation (red) and mean zonal wind (contours, m/s, with a strong impact on the AEJ). Below, we show the relative humidity (%, colors) and paramterized convective heating rate (contours, K/day). Simulation performed by MarieAngèle Filiberti and Mai Pham. The model results are analysed adopting the AMMAcross framework in which fields are averaged in longitude over the region 10E10W. The differences between the Emanuel and Tiedtke simulations (Fig1) are clearly visible with a convective heating (lower panels, contours, K/day) peaking higher up in the atmosphere, with larger intensity, for the Emanuell scheme. The cooling (dashed contours) in the lower atmosphere, over Sahel, due to the reevaporation of convective rainfall, is also much stronger with Emanuel. Those modifications have a strong impact on the large scale dynamics : the African Easterly Jet is much better represented with the Emanuel scheme (the dashed contours in the upper panels which corresponds to the easterlies in m/s); the mean meridional circulation (red arrows) with a well marked ITCZ in the Emanuel case. The relative humidity (colors in the lower panels) show a signature of this different transport with, in particular, a local maximum stronger and higher in the ITCZ for Emanuel. The CO concentrations obtained with the chemistrytransport version LMDZINCA of the same models are also quite different (Fig 2). The local maximum of relative humidity in Fig 1 for Emanuel is associated with a local maximum of CO. The air comming from the Sahara is also much less loaded in CO in the case of Emanuel scheme. Fig 2: Cross section of CO concentrations (contours, ppmv) and idealized tracer 1 (colors, arbitrary units). Simulations performed by M.A. Filiberti and M. Pham. 3 1 2 Fig 3 : Regions chosen for the sources of idealized tracers Those differences are analysed further by using idealized tracers emited in the boundary layer. More precisely, the concentration of the tracers are forced to be larger than an hyperbolic tangent function of altitude, decreasing from 1 at the surface to 0 above, with a sharp transition at 850 hPa. The different tracers correspond to different regions of emission shown in Fig 3. The sink consists of a radioactive decay with a life time of a few days. The tracer (colors in Fig 2) emited over the African continent, south of 10N (emitted over the green region numbered 1 in Fig 3), behaves very similarly to the CO obtained in INCA (colors in Fig. 2), which is of course not a surprise for CO which is a long live tracer of the atmospheric circulation. However, the idealized tracer presents the avantage for model intercomparison of being independant of emissions and chemical scheme. Fig.4 illustrate further how those idealized tracers can be used to go into more insight in the understanding of atmospheric transport. Colors :Tracer cocentration with and without parametrized convection Contours : Heating rates (K/day) Effect of parametrized physics (withwitout) Fig 4: Separation between large scale and parametrized transport for tracer 1+2. The tracer transport is computed both with and without parametrized physics. In the case without parametrized transport, the transport is done by large scale winds only but those winds themselves are computed with parametrized convection Crosssection versus chemical observations: 15 August CO Fig 6 : track of the BAE transect flights (summer 2006, from Claire Reeves). BAE observations of summer 2006 may give a good argument to promote the idea of a crosssection approach framework for the intercomparison of chemistrytransport models. A number of flights were performed by the British BAE aircraft as illustrated in Fig 5. Those flight were intentionally concentrated along the AMMA transect and cover a period of 5 weeks. Fig 7 : CO observations (left) gathered for all the BAE flight and numerical results from the TOMCAT chemistry transport model (right). Simulations presented by Xin Yang et al. at the AMMA conference in Toulouse, 2006. In view of this long period, the dispersion of the measurements seems remarkably weak. Wehter it is something general or specific of this particular period is still to be analysed in details. Proposal So the idea would to extend the AMMAcross/AMMAmaps initiatives to the evalutation and intercomparison of chemistry transport models. Although the direct comparison of model results with insitu measurements along the flight track is a better approach for the validation of operational or chemistry transport models, the present approach seems a good alternative for the intercomparison and evaluation of the various models (including climate versions not forced by analysed winds) in a common framework, and focus on the link between convection, large scale transport and composition. With respect to the original AMMAcross part, the following specification could be made : − add the concentration of a number of chemical coumpounds, when avaliable. − add some idealized tracers to be defined. − the 10E10W transect is probably too wide and should be completed with a 1W3W transect. This second transect may also be adopted as an additional framework for the climate part. − the « total column » of some chemical compounds as well as other relevant variables may also be added in the AMMAmaps framework.