[MITgcm-support] synchronous vs asynchronous integrations

David Wang dwang at ldeo.columbia.edu
Fri Jun 2 14:31:32 EDT 2006


Hi Samar,

Thanks for pointing this out to me. I think this is the thread that you 
mentioned:
http://mitgcm.org/pipermail/mitgcm-support/2005-January/002816.html

to be clearer, I need a quasi-equilibrium of the upper ocean, instead of 
a full equilibrium which would take much longer time. If the strategy 
that Martin suggested is implemented, when should I switch from 
asychronous mode to sychronous mode (i.e., how to know "the T/S-drift 
has become negligible" technically?); or, just safely stay with 
synchronous runs?

Thx, -d.

Samar Khatiwala wrote:
> David
> 
> Not sure why you expect the two to give similar results, especially  
> with seasonal forcing.
> (See papers by Danabasoglu and Danabasoglu & McWilliams.)
> My experience is that the extent to which they differ is (at the very  
> least) a function of the
> resolution, advection scheme, and the ratio of tracer to momentum  time 
> steps. Of course,
> you should only be comparing the final equilibrium solutions, and not  
> some intermediate
> transient solution. I take it you haven't run it long enough to spin  up 
> your model to equilibrium.
> There is a long exchange on this topic between myself and Martin  Losch 
> on the mitgcm-support
> list.
> 
> Samar
> 
> On Jun 2, 2006, at 11:40 AM, David Wang wrote:
> 
>> Hello,
>>
>> I ran two model runs with the same configuration, the same initial  
>> fields and the same surface forcing fields, but different timesteps  
>> (one is sychronous: deltaT = 3600s; another is asynchronous:  
>> deltaTmom = 900s, deltaTtracer = deltrTClock = 10800s). They give  
>> considerably different results, which is not what I expected. Of  
>> course I prefer the asychronous one since it takes much less  
>> integration time, but the differences make me to doubt the  integrity 
>> of these runs.
>>
>> The model domain is the tropical pacific. The initial fields are  
>> climatological potential temp and salinity. The model is forced by  
>> climatological monthly forcings. kpp, gmredi, rbcs packages are used.
>>
>> I'm attaching my data namelist file and a plot of momKE_ave  
>> timeseries over the first model year (which is to show how  different 
>> they are. blue is the sychronous run and red is the  asychronous run). 
>> The temperature field snapshots at the end of the  first model year 
>> also differ, esp. near the equator.
>>
>> If anyone would have any ideas on what might go wrong, plz let me  
>> know. Thanks very much.
>>
>> - d.
>> # ====================
>> # | Model parameters |
>> # ====================
>> #
>> # Continuous equation parameters
>>  &PARM01
>> tRef = 25.95, 25.85, 25.70, 25.41, 25.02,
>>        24.56, 24.00, 23.45, 22.84, 22.23,
>>        21.61, 20.99, 20.38, 19.77, 19.17,
>>        18.56, 17.86, 16.91, 15.74, 14.44,
>>        13.09, 11.74, 10.27,  8.90,  7.64,
>>         6.61,  5.82,  5.18,  4.63,  4.21,
>>         3.84,  3.51,  3.22,  2.91,  2.59,
>>         2.24,  1.95,  1.71,  1.54,  1.36,
>>         1.22,  1.10,  1.02,  0.95,  0.92,
>> #
>>  sRef = 45*35.,
>> #
>>  viscA4  = 1.E13,
>>  viscAz  = 1.E-4,
>>  diffKhT = 0.E3,
>>  diffKzT = 5.E-6,
>>  diffKhS = 0.E3,
>>  diffKzS = 5.E-6,
>> #
>>  rhoNil =   1025.,
>>  gravity =  9.81,
>>  eosType = 'LINEAR',
>>  tAlpha = 2.E-4,
>>  sBeta =  7.4E-4,
>> #
>>  rigidLid =            .FALSE.,
>>  implicitFreeSurface = .TRUE.,
>>  no_slip_bottom     =  .TRUE.,
>>  no_slip_sides      =  .TRUE.,
>> #
>> # ivdc_kappa=10.,
>>  implicitDiffusion=.TRUE.,
>>  implicitViscosity=.TRUE.,
>> #
>> # turn on looped cells
>>  hFacMin=0.05,
>>  readBinaryPrec=32,
>> #
>>  &
>>
>> # Elliptic solver parameters
>>  &PARM02
>>  cg2dMaxIters=1000,
>>  cg2dTargetResidual=1.E-13,
>>  &
>>
>> # Time stepping parameters
>>  &PARM03
>> #
>>  startTime  =   0.,
>>  endTime    =   31104000.,
>> #
>>  deltaT =         3600.,
>> # deltaTmom =      900.,
>> # deltaTtracer =   10800.,
>> # deltaTClock  =   10800.,
>> #
>>  abEps = 0.1,
>>  pChkptFreq =  0.,
>>  chkptFreq =   0.,
>>  monitorFreq = 86400.,
>>  dumpFreq = 0.,
>>  dumpInitAndLast = .TRUE.,
>> #
>> # cAdjFreq = -1.,
>> #
>> # 1 months restoring timescale for surface temperature
>> # 2 months restoring timescale for surface salinity
>> #
>>  tauSaltClimRelax =    5184000.,
>>  tauThetaClimRelax =   2592000.,
>>  periodicExternalForcing = .TRUE.,
>>  externForcingPeriod = 2592000.,
>>  externForcingCycle =  31104000.,
>>  &
>>
>> # Gridding parameters
>>  &PARM04
>>  usingCartesianGrid =      .FALSE.,
>>  usingSphericalPolarGrid = .TRUE.,
>> #
>>  delX = 160*1.,
>> #
>> delY =       1.0000000,  1.0000000,  1.0000000,  0.9998741,
>>  0.9956741,  0.9835052,  0.9675407,  0.9441032,  0.9175377,
>>  0.8843880,  0.8491249,  0.8084641,  0.7670174,  0.7215805,
>>  0.6768684,  0.6297283,  0.5849037,  0.5392418,  0.4974575,
>>  0.4563580,  0.4205704,  0.3867931,  0.3595371,  0.3353481,
>>  0.3185692,  0.3055687,  0.3004913,  0.2995090,  0.3004913,
>>  0.2995071,  0.3004932,  0.2995071,  0.3004932,  0.2995071,
>>  0.3004913,  0.2995090,  0.3004913,  0.2995090,  0.3004904,
>>  0.2995100,  0.3004894,  0.2995110,  0.3004894,  0.2995100,
>>  0.3004904,  0.2995090,  0.3004913,  0.2995090,  0.3004904,
>>  0.2995100,  0.3004899,  0.2995100,  0.3004899,  0.2995100,
>>  0.3004904,  0.2995095,  0.3004904,  0.2995095,  0.3004906,
>>  0.2995093,  0.3004907,  0.2995093,  0.3004907,  0.2995093,
>>  0.3004906,  0.2995095,  0.3004904,  0.2995095,  0.3004904,
>>  0.2995100,  0.3004899,  0.2995100,  0.3004899,  0.2995100,
>>  0.3004904,  0.2995090,  0.3004913,  0.2995090,  0.3004904,
>>  0.2995100,  0.3004894,  0.2995110,  0.3004894,  0.2995100,
>>  0.3004904,  0.2995090,  0.3004913,  0.2995090,  0.3004913,
>>  0.2995071,  0.3004932,  0.2995071,  0.3004932,  0.2995071,
>>  0.3004913,  0.2995090,  0.3004913,  0.3055687,  0.3185692,
>>  0.3353481,  0.3595371,  0.3867931,  0.4205704,  0.4563580,
>>  0.4974575,  0.5392418,  0.5849037,  0.6297283,  0.6768684,
>>  0.7215805,  0.7670174,  0.8084641,  0.8491249,  0.8843880,
>>  0.9175377,  0.9441032,  0.9675407,  0.9835052,  0.9956741,
>>  0.9998741,  1.0000000,  1.0000000,  1.0000000,  1.0000000,
>> #
>>  delZ = 10.00, 10.00, 10.00, 10.00, 10.00,
>>         10.00, 10.00, 10.00, 10.00, 10.00,
>>         10.00, 10.00, 10.00, 10.00, 10.00,
>>         11.13, 14.46, 19.82, 26.94, 35.48,
>>         44.99, 55.00, 65.01, 74.52, 83.06,
>>         90.18, 95.54, 98.87, 100.00, 100.00,
>>         100.00, 100.00, 111.42, 143.93, 192.60,
>>         250.00, 307.40, 356.07, 388.58, 400.00,
>>         400.00, 400.00, 400.00, 400.00, 400.00,
>> #
>>  phiMin = -29.,
>>  thetaMin = 130.,
>>  &
>>
>> # Input datasets
>>  &PARM05
>>  bathyFile=      '/data1/fats/dwang/myexps/trop_pacific/input/1x3rd/ 
>> bathymetry.bin',
>>  hydrogThetaFile='/data1/fats/dwang/myexps/trop_pacific/input/1x3rd/ 
>> woa01_init_theta.bin',
>>  hydrogSaltFile= '/data1/fats/dwang/myexps/trop_pacific/input/1x3rd/ 
>> woa01_init_salt.bin',
>>  zonalWindFile=  '/data1/fats/dwang/myexps/trop_pacific/input/1x3rd/ 
>> dasilva94_monthly_taux.bin',
>>  meridWindFile=  '/data1/fats/dwang/myexps/trop_pacific/input/1x3rd/ 
>> dasilva94_monthly_tauy.bin',
>>  thetaClimFile=  '/data1/fats/dwang/myexps/trop_pacific/input/1x3rd/ 
>> woa01_monthly_sst.bin',
>>  saltClimFile=   '/data1/fats/dwang/myexps/trop_pacific/input/1x3rd/ 
>> woa01_monthly_sss.bin',
>>  surfQnetFile=   '/data1/fats/dwang/myexps/trop_pacific/input/1x3rd/ 
>> dasilva94_monthly_qnet0.bin',
>>  the_run_name=   'trop_pacific.160x124x45',
>> #
>>  &
>> <momKE.png>
>> _______________________________________________
>> MITgcm-support mailing list
>> MITgcm-support at mitgcm.org
>> http://mitgcm.org/mailman/listinfo/mitgcm-support
> 
> 
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support



More information about the MITgcm-support mailing list