GUIDELINES: As a framework for the Ocean Carbon-Cycle Model Intercomparison Project (OCMIP) -------------------------------------------------------- %_ --------------------------------------------------------------------- %_ Revision Control System (RCS) used here to manage changes to this text %_ RCS output lines preceded by "%_ " %_ --------------------------------------------------------------------- %_ $Header: /home/geo/orr/ocmip/RCS/guidelines.ascii,v 1.4 1995/02/08 16:07:04 orr Exp orr $ %_ %_ $Log: guidelines.ocmip,v $ %_ Revision 1.6 1995/05/04 13:46:57 orr %_ Updated information concerning OCMIP library, file transfer, etc. %_ %_ Revision 1.4 1995/02/08 16:07:04 orr %_ Added changes from remarks of first e-mail distribution to %_ OCMIP participants. %_ %_ Revision 1.1 1995/01/25 15:55:09 orr %_ Initial revision %_ --------------------------------------------------------------------- The first OCMIP (GAIM/IGBP) planning workshop was held in Hamburg on January 20, 1995. Present were four research groups that use global 3-D ocean models to study the ocean's carbon cycle. All have agreed to participate in the OCMIP comparison. These groups and their representatives at the OCMIP workshop include GFDL/AOS: R. Toggweiler and J. Sarmiento Hadley: N. Taylor IPSL (LMCE-CFR, LODyC): J. Orr and P. Monfray MPIM: E. Maier-Reimer, M. Heimann, U. Mikolajewicz, K. Six, and A. Winguth. (Note: Expanded names and addresses for these laboratories will be found at the end of this document, see Abbreviations section) At this workshop, participants first developed a list of tracer simulations that would be desirable to include within the framework of this comparison. The list was then prioritized. Secondly, protocols were established to make simulations and analyses as comparable as possible. Thirdly, all agreed that it was well worth the effort early on to develop an on-line library of surface fields and algorithms, as necessary to compute standard surface boundary conditions for these simulations. Such a library would clearly improve internal consistency, allow for sensitivity tests, and encourage new participants to join in later on. Subsequently, participants discussed logistical constraints: (1) who will do what when and (2) potential funding support for this effort. The remainder of this document provides details on discussions held and decisions taken at this meeting. ------------------------------------------------------------------------ Planned Experiments including priorities: A = 1st priority B = 2nd priority C = for later study ------------------------------------------------------------------------ (1) Natural CO2: The pre-industrial steady-state ocean carbon cycle (a) Simulation using only the solubility pump (A) (b) Simulation using the solubility pump and the biological pump (B) (2) Perturbation CO2 (a) Anthropogenic CO2 - observed atmospheric record as forcing (A) (b) Impulse response function (B) (3) Natural C-14 (A) (4) Perturbation C-14 (a) Bomb C-14 (A) (b) Impulse response function to a doubling of 14C/C (B) (5) Other tracers (a) CFC's (C) (b) Ar-39 (C) --------------------------- Classification by Priority --------------------------- A. 1(a), 2(a), 3, 4(a) B. 1(b), 2(b), 4(b) C. 5(a), 5(b) ------------------------------------------------------------------------ Simulation Guidelines (At this time, details are complete only for simulations of Priority A) ------------------------------------------------------------------------ A. Initialization -------------- Natural C-14 - Atmospheric C-14 to be held at 0 permil - Simulations will be run until tracer fields reach "equilibrium" (definition of equilibrium may vary between models). Natural CO2: - Atmospheric CO2 will be held constant at 277.9514 ppm, the initial atmospheric pCO2 (at year 1765.0) used in the 1993 IPCC intercomparison, hereafter referred to as IPCC93 (Enting et al., 1994). Obviously, ice core measurements of pCO2 do not justify retaining 4 decimal places after the zero; however, simulations must use the above value to be entirely consistent with the IPCC93 record and protocols (and thus those for OCMIP). Statistically speaking, one should refer to the above preindustrial pCO2 concentration as 278 ppm. - Alkalinity field will be initialized by multiplying modeled salinity times the mean ocean ratio of Total Alkalinity to salinity (TALK/SAL) as determined from the GEOSECS data - Simulations will be run until tracer fields reach "equilibrium" (definition of equilibrium may vary between models). Perturbation Runs (Bomb C-14 and Anthropogenic CO2): - From the simulated natural equilibrium or by perturbation approach (A) - From the simulated natural equilibrium only (B) B. Atmospheric forcing ------------------- Natural CO2: 1(a) and 1(b) - Fixed atmospheric pCO2 = 277.9514 ppm Anthropogenic CO2: 2(a) Historical atm. pCO2 record + Future Scenarios Data from Enting et al. (1994), hereafter referred to as IPCC93 (i) 1765.0 - 1990.5 (IPCC93 file: splco2.dat) (ii) 1990.5 - 1995.5 (Scenario S650 using IPCC93 file: stab.dat) (iii) 1990.5 - 2300.5 (IPCC93 Scenarios S450 and S650, file: stab.dat) 2(b) Impulse response function (i) Instantaneous input (step function) of 10 GT C in atmosphere All modelers will use a ratio of 2.123 GT C / ppm CO2, thereby fixing the size of the atmospheric reservoir. In practice, this experiment consists of (1) allow model to reach equilibrium with atmospheric pCO2 = 277.9514 ppm, (2) subsequently, increase atmospheric pCO2 to 282.6617 ppm for just 1 timestep (clock set to time 0 at this point), (3) then allow atmospheric pCO2 to respond to ocean uptake throughout the rest of the simulation (ii) Simulation for 1000 model years. Natural C-14: 3. Fixed atmospheric C-14 at preindustrial value = 0 permil Bomb C-14 (including Suess effect): 4(a) Historical record (i) From 1765.0 to 1995.5 (ii) Atmospheric C-14 separated into 3 latitude bands - 90S-20S (IPCC93 file: c14sth.pak) - 20S-20N (IPCC93 file: c14equ.pak) - 20N-90N (IPCC93 file: c14nth.pak) (Note: the three files above contain records that begin in 1765.5; for consistency with CO2, C-14 runs will begin in 1765.0. For the first half year 1765.0-1765.5, atmospheric C-14 will be held at 0 permil). 4(b) Impulse response function (i) Instantaneous atmospheric input (step function) of +1000 permil (ii) Atmospheric CO2 will be fixed at 277.9514 ppm (consistent with the anthropogenic CO2 simulations) (iii) Simulation for 1000 model years. C. Gas exchange coefficient(* See Footnote at end of document) ------------------------ (i) Gas exchange vs. Wind algorithm from Wanninkhof (1992, eq. 8), adjusted to take into account the satellite wind distribution (ii) Schmidt numbers from polynomial fits of Wanninkhof (1992, Table A1) (iii) CO2 Solubility from Weiss (1974) (iv) Ocean surface temperature from Levitus (1982) (iv) Satellite Scatterometer derived wind fields (Etcheto et al., 1991) (v) To account for the impediment of gas exchange due to sea ice, multiply the gas exchange coefficient times (1 - Fice), where Fice is the fractional sea ice cover from the climatological data of Walsh (1978) and Zwally et al. (1983). By convention, all values of Fice below 0.2 must be reset to zero. (vi) Sensitivity tests will be run with at least one model including different gas exchange formulations (e.g., Constant; Liss and Merlivat, 1972; Broecker et al., 1985; Heimann and Monfray, 1989) and wind data sets (Esbensen and Kushnir, 1981; ECMWF) D. Carbonate chemistry(* See Footnote at tend of document) ------------------- (i) Participants will use the set of constants now being recommended by A. Dickson and F. Millero (See README and funcchem.f) (ii) Sensitivity tests to be run later E. Biology ------- (i) Current plans do not call for making sensitivity tests concerning how different ocean circulation patterns may produce different results with the same biological model (i.e., running the same biological model in different ocean models); hopefully however, that option can be explored in the near future. (ii) Likewise, studying the differences between several different biological models in one ocean circulation model is left to future efforts. F. Model Results ------------- - All participants must provide their output on a rectangular grid. Participants whose models use a rectangular grid (GFDL/Princeton and Hadley Centre) should retain their standard model grid. On the other hand, participants whose models do not use a rectangular grid (i.e., Hamburg and LODyC models) will thus be required to interpolate their results to the standard OCMIP 2 x 2 rectangular grid (see README). - Participants using a non-rectangular grid must also supply their non-rectangularly gridded output, i.e., as used when plotting their results. NOTE: Interpolation of non-rectangularly gridded data onto a rectangular mesh can in some cases result in loss of important details near land boundaries or it may distort model results. With the aid of some recently developed software, we are now able to treat most any non-rectangular grid in a sophisticated manner. Eventually we hope that OCMIP participants who employ non-rectangular grids will not have to bother with interpolating their results to the standard rectangular grid (2 x 2). Our goal is to be able make all interpolations at the analysis center in a consistent fashion while keeping loss of detail to a minimum. (i) Required fields (Priority A): 3-D scaler - tracer conc., T, and S 2-D - air-sea tracer flux, gas exchange, wind speed, sea ice, grid, area, landmask, and topography Other - Atmospheric tracer concentration for impulse response functions (ii) Desired Fields (Priority B): 3-D scaler - Rate of change of the tracer concentration due to advection, diffusion, and convection (i.e., typically eight 3-D scalar fields - 3 for advection, 3 for diffusion, 1 for convection, and 1 for the total change) plus radioactive decay. From these scalar fields, the analysis center will calculate tracer fluxes at grid box boundaries for advection, diffusion, and convection. (iii) Optional fields (Priority C): 3-D (vector) - advection (U,V,W) tracer fluxes at grid box boundaries due to advection, diffusion, and convection (iv) Frequency of Model Output (a) Equilibrium runs (experiments (1) and (3)) - Annual mean fields (once after model reaches equilibrium) - Monthly mean fields (for 1 yr after model attains equilibrium). (b) Transient runs (experiments 2(a) and 4(a)) - Annual mean fields for 1800, 1850, 1900, every year from 1950 on. (c) Impulse response functions (experiments 2(b) and 4(b)) - Atm. tracer concentration every year: 0.0, 1.0, 2.0, ..., 1000.0 - Annual mean 3-D tracer fields at years 1, 2, 5, 10, 20, 50, 100, 200, 500, and 1000. Notes: (1) Details concerning model output T, S, (as well as priority C vector fields of U, V, and W): - For off-line models, these five fields will be written out just once as the annual mean, and then for seasonal models an additional twelve times (once for each monthly mean). - For on-line models, these fields will be stored out at the time intervals specified in F(iv) above, but in a separate file from output for tracer concentrations, the air-sea flux, and Gas Exchange (Kg), Wind Speed, Sea Ice: - Stored once for annual mean; + 12 monthly means for seasonal models Topography, Latitude, Longitude, Depths, and Land Mask (1=Ocean;0=Land): - To be written out just once (separate file) - Latitude, Longitude, and Depths for both the center and edges of grid boxes - Surface areas of each tracer grid box (x-y plane) Format of output (essential for consistent analysis): - Participants must provide a standard Fortran subroutine that will provide a standardized list of model output arrays. An example of this subroutine is provided in the OCMIP library (see README). - To save space, participants should provide data files as binary, 32 bit (i.e., the IEEE standard). This is the standard used by almost all Unix machines (DEC excluded). Even from a Cray, one can easily produce IEEE 32 bit files, although by default, Cray binaries are 64 bit. For example, one way to produce a Cray 32-bit file named "myfile.dat" and associated with Fortran unit 61, issue the following command assign -a -N ieee -F f77 u:61 before running the Fortran program. Retain corresponding OPEN statements in the Fortran code as before. See routines "gridlab.f" and "trunlab.f" for another approach, entirely contained within the Fortran code itself. (2) An annual mean for a given year is defined as the average over the whole year. As in the real world, model years begin on 1 January. (3) Annual means are preferred to mid-year snapshots, even in annual mean models, such as that used at GFDL. For example, a time step taken at mid-year can occasionally fall on the mixing time step (in models using a leap-frog numerical scheme). Instantaneous fluxes are particularly sensitive to these numerics, thus their saved output can be noisy; annual means avoid such difficulties. ------------------------------------------------------------------------ LIBRARY ------------------------------------------------------------------------ A high initial priority has been given to developing an on-line database of the information necessary to make 3-D model simulations as specified under the guidelines given above. This library consists of a collection of data fields and algorithms necessary to compute surface boundary conditions. There will be an additional effort to include alternative fields in order to facilitate sensitivity tests. All fields will be given on a standard grid, probably 1 x 1 degree following the format of Levitus (1982). This library is accessible by ftp. We will also try to make it accessible by mail-server and perhaps WWW (restricted mode). Amongst other things, the library includes (i) Detailed experiment guidelines (ii) Standard fields of Gas Exchange and Sea Ice (iii) Atmospheric pCO2 and C-14 records (iv) Carbonate Chemistry Equilibrium Constants (v) Mean GEOSECS Alkalinity and Salinity (vi) Example of subroutine used to pass model output to analysis center For future efforts including sensitivty studies, efforts will be devoted to adding other components to the library: (i) Additional fields for gas exchange, sea ice, and wind (ii) Algorithms to compute gas exchange (as Fortran subroutines) (iii) Some results from this model intercomparison (iv) Documentation updates (v) Fields of delta pCO2 for the N. Atlantic (from two consolidation efforts: Takahashi et al. (in press) and Lefevre (1995) (vi) Other data for model validation (Ocean color and C-14) as it becomes available ------------------------------------------------------------------------ PERSONNEL ------------------------------------------------------------------------ Efficient intercomparison is only possible through common analysis of the results from the four different 3-D models. That is, one site where all analysis will be centralized is absolutely necessary, as agreed upon at the OCMIP planning meeting. To launch OCMIP, IGBP/GAIM has come up with some modest start-up funds. To keep OCMIP viable however further funding will certainly be necessary. It is planned that one post-doc would work full time under the supervision of J. Orr and P. Monfray at LMCE-CFR (Gif-sur-Yvette, France), where the OCMIP effort is now being coordinated. The first task to set-up the standard portion of the OCMIP on-line library is nearly complete, i.e., all standard fields and programs are now ready. The second phase, analysis of model results has just begun. ------------------------------------------------------------------------ TENTATIVE CALENDAR: ------------------------------------------------------------------------ May, 1995 *** PHASE A: - Available model output (Phase A) should be made available to LMCE-CFR (via ftp or surface mail--as exabyte tape; in tar or cpio format) - May 31, Deadline for all Phase A results to be sent to LMCE - Modify analytical tools as necessary to follow OCMIP protocols June, 1995: - Begin analysis of results from phase A experiments - Add fields for sensitivity tests to OCMIP library - Continue development of analytical tools July to September 1995: - Complete analysis of results from phase A experiments - Distribution of preliminary results to OCMIP participants - Discussion of early results - Preparation for the GAIM conference in Garmisch End September 1995: - Presentation of preliminary OCMIP results at the First GAIM Science Conference (25-29, September, Garmisch, Germany) - Second OCMIP meeting at Garmisch (30 September)???? October 1995: - Complete analysis of Phase A Results November 1995 *** PHASE B Begins: Spring 1996: - Third OCMIP meeting (3 days, LMCE, Gif): using workstations available at LMCE, substantial focus will be placed on simultaneous analysis of all models, side-by-side, including process-based investigations to distinguish differences as a function of advection, diffusion, and convection. 1996/1997: - Continue with Phase B - Begin Phase C ------------------------------------------------------------------------ ------------------------------------------------------------------------ REFERENCES: Broecker, W. S., T.-H. Peng, G. Ostlund, and M. Stuiver, 1985. The distribution of bomb radiocarbon in the ocean. J. Geophys. Res., 90, 6953-6970. Enting, I. G., T. M. L. Wigley, and M. Heimann, 1994. Assessment of CO2 projections, CSIRO Aust. Div. Atmos. Res. Tech. Pap. No. 31. Esbensen, S. K., and Y. Kushnir, 1981. The heat budget of the global ocean: An atlas based on estimates from surface marine observations, Rep. 29, Clim. Res. Inst., Oreg. State Univ., Corvallis. Etcheto, J., J. Boutin, and L. Merlivat, 1991. Seasonal variation of the CO2 exchange coefficient over the global ocean using satellite wind speed measurements, Tellus, 43B, 247-255. Heimann, M., and P. Monfray, 1989. Spatial and temporal variation of the gas exchange coefficient for CO2: 1. Data Analysis and Global Validation. Report No. 31, 29 pp. Max Planck Institut fur Meteorologie, Hamburg, Germany. Levitus, S., 1982. Climatological atlas of the world ocean. NOAA Prof. Paper 13, U.S. Government Printing Office, Washington, DC. Liss, P. S., and L. Merlivat, 1986. Air-sea gas exchange rates: Introduction and synthesis, in The Role of Air Sea Gas Exchange in Geochemical Cycling, ed. by P. Buat-Menard, 113-128, D. Reidel, Hingham, Mass. Walsh, J. 1978. A data set on northern hemisphere sea ice extent, 1953-1976. Glaciological Data, World Data Center for Glaciology (Snow and Ice), Report GD-2, 49-51. Wanninkhof, R., 1992. Relationship between wind speed and gas exchange over the ocean, J. Geophys. Res., 97, 7373-7382 Weiss, R. F., 1974. Carbon dioxide in water and seawater: the solubility of a non-ideal gas, Marine Chem., 2, 203-215. Zwally, H. J., J. Comiso, C. Parkinson, W. Campbell, F. Carsey, and P. Gloerson, 1983. Antarctic Sea Ice, 1973-1976: Satellite Passive Microwave Observations, NASA, 206 pp. ------------------------------------------------------------------------ ABBREVIATIONS: AOS - Program in Atmospheric and Oceanic Sciences, Princeton University P.O. Box CN710, Princeton, NJ 08544-0710 USA) CFR - Centre des Faibles Radioactivites, Laboratoire Mixte CNRS-CEA L'Orme, Bat. 709/LMCE, CE Saclay, 91191 Gif-sur-Yvette, FRANCE ECMWF - European Center for Medium-Range Weather Forecast GAIM - Global Analysis, Interpretation, and Modeling GFDL - Geophysical Fluid Dynamics Laboratory NOAA, P.O. Box 308, Princeton, NJ 08542, USA) Hadley - Hadley Center Meteorol. Office, London Rd., Bracknell, Berkshire RG12 2SY, ENGLAND IGBP - International Geosphere-Biosphere Program IPSL - Institute Pierre Simon Laplace, France LMCE - Laboratoire de Modelisation du Climat et de l'Environnement CEA/DSM, L'Orme, Bat. 709, CE Saclay, 91191 Gif-sur-Yvette, FRANCE LODyC - Laboratoire d'Oceanographie Dynamique et de Climatologie Universite Pierre et Marie Curie, Paris, FRANCE MPIM - Max Planck Institut fur Meteorologie Bundesstrasse 55, D-2000 Hamburg 13, GERMANY OCMIP - Ocean Carbon-Cycle Model Intercomparison Project ------------------------------------------------------------------------ FOOTNOTE (20 January 1995): * Groups that have already run comparable simulations but have not followed these guidelines explicitly are at this time encouraged to participate in OCMIP; however it is mandatory that new simulations follow these guidelines in order to be useful to the OCMIP effort.