Date: Tue, 26 Mar 2002 09:41:50 -0700 (MST)
From: Jerry Meehl
To: CDKKL@NASAGISS.GISS.NASA.GOV, abeouchi@ccsr.u-tokyo.ac.jp, ach@dar.csiro.au, barthele@cerfacs.fr, cmglr@giss.nasa.gov, cubasch@dkrz.de, dinar@inm.ras.ru, esch@dkrz.de, fairhead@lmd.jussieu.fr, george.boer@ec.gc.ca, greg.flato@ec.gc.ca, hbg@dar.csiro.au, hellbach@dkrz.de, hogan@nrlmry.navy.mil, jbahn@pusan.ac.kr, jonathan.gregory@metoffice.com, jtkon@ncar.ucar.edu, kimoto@ccsr.u-tokyo.ac.jp, letreut@lmd.jussieu.fr, meagles@meto.gov.uk, meehl@ncar.ucar.edu, noda@mri-jma.go.jp, r.colman@bom.gov.au, rjs@gfdl.gov, roeckner@dkrz.de, sbp@bom.gov.au, tcjohns@meto.gov.uk, terray@cerfacs.fr, yyq@lasg.iap.ac.cn, zxh@lasg.iap.ac.cn
Subject: CMIP Coordinated Experiments
Cc: B.McAvaney@bom.gov.au, aclarke@netra1.wmo.ch, ahu@ncar.ucar.edu, covey@cirrus.llnl.gov, gates@cirrus.llnl.gov, jfbmitchell@meto.gov.uk, latif@dkrz.de, weaver@ocean.seos.uvic.ca, wmw@ncar.ucar.edu
Dear CMIP model group representatives,
A new activity, "Coordinated Coupled Model Experiments", was approved at the last Working Group on Coupled Models meeting at the Hadley Centre in February, 2002, and is being coordinated by CMIP. These are not intended to be standard CMIP integrations, but specific experiments conceived by an individual or group of individuals who wish to invite other modeling groups to participate on a voluntary basis. Thus, this activity is a collaboration among modeling groups rather than a centrally organized intercomparison project. It is expected that each group will mainly concentrate on analyzing results from its own model. However, collaborative exchange of results and conclusions, and shared participation in the writing of papers describing results, should help gain a better understanding of why the responses of the various models differ. We therefore welcome participation by the modeling group you represent.
So far the following groups have committed to participate in this initial coordinated experiment: CCCma, GFDL, Hadley Centre, UVic, NCAR. Concerning timetable and deadlines, the organizers view this effort as a process and not an event. The groups above are already running or are planning on running shortly the experiments described below. If your group wants to participate in the first paper, then you should be in the analysis phase by early 2003 when the organizers plan to compare the results and begin the process of publishing a paper.
If you or someone in your group is interested in joining this project, please contact Jonathan Gregory (jonathan.gregory@metoffice.com) and Ron Stouffer (rjs@gfdl.noaa.gov), with a copy to Jerry Meehl (meehl@ncar.ucar.edu) and Curt Covey (covey1@llnl.gov) as soon as possible. Specific questions should be addressed to Jonathan or Ron. Description of the experiment and updates on status will be posted on the CMIP web page (http://www-pcmdi.llnl.gov/cmip/). Suggested standard diagnostics to produce once the experiments are completed, as a starting point for discussion, are given below.
Thank you for your consideration.
Jerry Meehl (for the CMIP Panel)
-----------------------------------------------------
Coordinated Coupled Model Experiment: Standard sensitivity tests for the response of the Atlantic meridional overturning to surface forcing
The experiments are:
Starting from a control state, (a) CO2 increasing at 1% per year compounded for 140 years (up to 4xCO2), and (b) a corresponding control run. This is an extended CMIP2 experiment. Daily surface water fluxes will be saved from each run, and two further runs will be then done: (c) like (a), but with the daily water fluxes from (b) used instead of those generated by the model, and (d) like (b) with the daily water fluxes from (a). We refer to (c) and (d) as "partially coupled experiments". This design was developed by Dixon et al., 1999, "The influence of transient surface fluxes on North Atlantic overturning in a coupled GCM climate change experiment", Geophys. Res. Lett., 26, 2749-2752, and by Mikolajewicz and Voss, 2000, "The role of the individual air-sea flux components in CO2-induced changes of the ocean's circulation and climate", Clim. Dyn, 16, 627-642. As shown by these papers, the GFDL and ECHAM3/LSG models differ in the relative importance of changes in surface water fluxes and surface heat fluxes (caused by raising CO2), and it will be useful to extend this analysis to other models. The groups have made different choices regarding whether the surface water flux in sea-ice regions is taken to be at the surface of the ice (i.e. the P-E) or at the underside of the ice (in which case it also includes ice melting and freezing).
Diagnostics: A plot like Fig 1 of Dixon et al. or Fig 5 of Mikolajewicz and Voss.
A further aim is to find a diagnostic of the density field which can be used as a predictor of the overturning strength, such as the steric height gradient identified by Hughes and Weaver, 1994, "Multiple equilibria of an asymmetric two-basin model", J. Phys. Oceanogr., 24, 619-637. If possible it is recommended to include diagnostics for the rate of change of temperature and salinity in the ocean due to advection, diffusion and vertical mixing, at least on decadal timescales. These can then be used to relate changes in the density field back to the surface flux changes, as by Thorpe et al., 2001, "Mechanisms determining the Atlantic thermohaline circulation response to greenhouse gas forcing in a non-flux-adjusted coupled climate model", J. Climate, 14, 3102--3116.
Diagnostics: A plot like Fig 13 of Hughes and Weaver or Fig 3 of Thorpe et al. As a first step towards performing an analysis like that of Thorpe et al., it is necessary to find an appropriate diagnostic of the density field.
This experiment aims to establish a benchmark for the sensitivity to an imposed surface freshwater flux. The design is to apply a surface flux of 0.1 Sv in total, uniformly distributed over the Atlantic between 50N and 70N, for a period of 100 years, starting from a control state. This additional flux is a net addition of freshwater to the ocean; it is not compensated for by removal elsewhere. The "Atlantic" may include parts of the Labrador Sea and North Sea within the latitude band; groups have made different decisions about this. After 100 years, the imposed water flux will be switched off, and the experiment continued to run, so that any recovery can also be investigated. It could be valuable to try this run starting from an LGM state also, if one is available.
Diagnostics: Timeseries of maximum of the Atlantic overturning streamfunction, maps of surface air temperature differences and of differences in surface fluxes (heat, freshwater and windstress).
-----------------------------------------------------
Date: Tue, 26 Mar 2002 09:55:12 -0700
From: Jerry Meehl
To: Ron Stouffer
CC: weaver@ocean.seos.uvic.ca, gflato@ec.gc.ca, covey1@llnl.gov
Subject: Re: CMIP standard THC experiments
Dear Jonathan and Ron,
As you can see from the announcement I sent out this morning, I have added NCAR to the list of groups who have agreed to participate in the first Coordinated Coupled Model Experiment. I think at minimum we will use the PCM for this, though the brand new CCSM may also be used depending on various factors.
Note that I embedded some of Jonathan's introduction into the general announcement--hope you don't mind Jonathan. One question I had for experiment 2: the run is for 100 years with the N. Atl. fresh water forcing, but about how long do you think it should be run in the recovery phase? I presume this may be dependent to some degree on how the particular model in question recovers, but for computer allocation purposes (which we are going through right now), about what would you estimate?
Let me know if there are any more issues/problems to deal with.
Jerry
-----------------------------------------------------
Date: Tue, 26 Mar 2002 17:03:51 +0000
From: Jonathan Gregory
Subject: Re: CMIP standard THC experiments
To: Jerry Meehl
Cc: Ron Stouffer
Dear Jerry
Thanks for sending that out. It is good in that form, and I like the way the timescale is described, thanks to you and Ron.
I am glad you're going to join in too.
One question I had for experiment 2: the run is for 100 years with the N. Atl. fresh water forcing, but about how long do you think it should be run in the recovery phase? I presume this may be dependent to some degree on how the particular model in question recovers, but for computer allocation purposes (which we are going through right now), about what would you estimate?
I guess another 100 years is the maximum. In our previous experiments, we find the THC mainly adapts to the surface forcing within a few decades, so I would expect, if your model is similar, that after a few decades beyond 100 years it should have settled again into some steady state, whether it be a complete recovery, complete collapse, or something between - which might be the most interesting outcome! There must also be much longer timescales than 100 years over which the model will actually continue to drift slowly, but this experiment isn't investigating those in the first phase either.
Best wishes
Jonathan
-----------------------------------------------------
Date: Tue, 26 Mar 2002 09:31:16 -0800
Subject: Re: EMIC workshops @ EGS
From: Andrew Weaver
To: Martin Claussen
CC: Jonathan Gregory
Dear friends,
A new activity, "Coordinated Coupled Model Experiments", was approved at the last Working Group on Coupled Models meeting at the Hadley Centre in February, 2002, and is being coordinated by CMIP. These are not intended to be standard CMIP integrations, but specific experiments conceived by an individual or group of individuals who wish to invite other modeling groups to participate on a voluntary basis. Thus, this activity is a collaboration among modeling groups rather than a centrally organized intercomparison project. It is expected that each group will mainly concentrate on analyzing results from its own model. However, collaborative exchange of results and conclusions, and shared participation in the writing of papers describing results, should help gain a better understanding of why the responses of the various models differ. We therefore welcome participation by the modeling group you represent.
So far the following groups have committed to participate in this initial coordinated experiment: CCCma, GFDL, Hadley Centre, UVic, NCAR. Concerning timetable and deadlines, the organizers view this effort as a process and not an event. The groups above are already running or are planning on running shortly the experiments described below. If your group wants to participate in the first paper, then you should be in the analysis phase by early 2003 when the organizers plan to compare the results and begin the process of publishing a paper.
Members of the WCRP WGCM also felt that this intercomparison project would be enhanced with the inclusion of results from EMICS.
If you or someone in your group is interested in joining this project, please contact Andrew Weaver (weaver@uvic.ca), Jonathan Gregory (jonathan.gregory@metoffice.com) and Ron Stouffer (rjs@gfdl.noaa.gov), with a copy to Jerry Meehl (meehl@ncar.ucar.edu) and Curt Covey (covey1@llnl.gov) as soon as possible. Specific questions should be addressed to Andrew, Jonathan or Ron. Description of the experiment and updates on status will be posted on the CMIP web page (http://www-pcmdi.llnl.gov/cmip/). Suggested standard diagnostics to produce once the experiments are completed, as a starting point for discussion, are given below.
Andrew Weaver
School of Earth and Ocean Sciences
University of Victoria
3964 Gordon Head Road
Victoria, BC, V8N 3X3
Canada
email: weaver@uvic.ca
http://climate.uvic.ca/
Tel: (250) 472-4001
Fax: (250) 472-4004
-----------------------------------------------------
Date: Thu, 04 Jul 2002 14:03:10 +0100
From: Jonathan Gregory
Subject: runoff included in freshwater flux
To: rjs@gfdl.noaa.gov, weaver@ocean.seos.uvic.ca, smurakam@mri-jma.go.jp, meehl@meeker.ucar.edu, gflato@ec.gc.ca, latif@dkrz.de, hgs@astr.ucl.ac.be, fichefet@astr.ucl.ac.be, stefan@pik-potsdam.de, ganopolski@pik-potsdam.de
Cc: covey1@llnl.gov
Dear All
With regard to the standardised CMIP-EMIC experiments in which surface fresh- water fluxes from a 1% CO2 experiment are imposed on a control experiment and vice-versa: We did not make clear in the original explanation that the surface freshwater flux includes runoff (P-E+R) in the experimental design of Dixon et al. (1999), Geophys. Res. Lett., 26, 2749-2752. Runoff should be included in order to ensure freshwater conservation, as well as because of its local influence on salinity. Our apologies for that omission.
Yours sincerely
Jonathan