[MITgcm-support] Varying deltaTtracer in a barotropic model

Fri Oct 14 10:19:17 EDT 2016

Just two additional comments on this:

To clarify: If you set deltaTmom and deltaTtracer independently, the model uses tracer acceleration, following Bryan (1984), which distorts the physics. (It’s NOT just a split time stepping) If you aren’t familiar with this and confident that it won’t affect your results, you should not use it and only set deltaT.

In your specific case, I think what happens is that by default deltaTclock is set to deltaTtracer. So what you are doing when choosing a longer deltaTtracer is you accelerate the model time. That means in practice (since your physics only care about the time elapsed in the momentum equation) you are simply running your model for a much shorter time than you think. 

Cheers,
Malte

------------------------------------------------------
Malte F Jansen
Assistant Professor
Department of the Geophysical Sciences
The University of Chicago
5734 South Ellis Avenue
Chicago, IL 60637 USA

> On Oct 14, 2016, at 8:52 AM, Martin Losch <Martin.Losch at awi.de> wrote:
> 
> Hi Ryan,
> 
> have a look at your STDOUD (the stuff that mitgcmuv flushes to the screen if you don’t redirect it). There is a configuration summary that tells you what values the parameters really have. You’ll probably find that in your “synchronous setup”, deltaT is probably much smaller than in in the asynchronous set up. The cause is the way the parameters are used. Have a look at ini_parms.F starting around line 882, where deltaT etc are set. If you don’t set deltaT, then you should set deltaTclock, because that’s the “true” model time.
> 
> Martin
> 
>> On 14 Oct 2016, at 18:36, Patmore, Ryan D. <ryapat30 at bas.ac.uk> wrote:
>> 
>> Hi all,
>> 
>> I have picked up on some strange time stepping behaviour and I was wondering if someone can enlighten me. I am running a barotropic channel model with fixed temperature and salinity.  I have set
>> saltStepping=.FALSE.,
>> tempStepping=.FALSE.,
>> and I have also checked the monitoring output and as expected the temperature and salinity of the model do not change throughout the model run.
>> 
>> My problem is as follows. If I compare model runs with synchronous time stepping, i.e.
>> 
>> deltaTmom = deltaTfreesurf= deltaTtracer
>> 
>> and asynchronous time stepping, i.e.
>> 
>> deltaTmom = deltaTfreesurf = a * deltaTtracer,  where  a < 1,
>> 
>> the synchronous setup uses a considerable amount more walltime than the asynchronous setup. The results in both cases are also different. If I was running a baroclinic model then this would be expected, but my model is barotropic and thus I expected no tracer operations to occur. Does anyone know why changing deltaTtracer in a barotropic run has any effect on the results?
>> 
>> An example data file is below.
>> 
>> Any suggestions will be much appreciated,
>> Ryan D. Patmore
>> 
>> #
>> # Model parameters
>> # Continuous equation parameters
>> &PARM01
>> # ########################################
>> # boundaries
>> bottomDragQuadratic=2.5e-3,
>> no_slip_sides=.FALSE.,
>> no_slip_bottom=.FALSE.,
>> # #########################################
>> # viscosities
>> viscAz=1.E-2,
>> useFullLeith=.TRUE.,
>> viscC2Leith=2.0,
>> viscC4Leith=0.2,
>> # #########################################
>> # misc param, eq-state & gravity & f
>> selectCoriMap = 1,
>> f0 = -1.27393e-4,
>> beta = 1.1144e-11,
>> gravity=9.81,
>> readBinaryPrec=64,
>> useSingleCPUio = .TRUE.,
>> # #########################################
>> # misc flags
>> saltStepping=.FALSE.,
>> tempStepping=.FALSE.,
>> staggerTimeStep=.TRUE.,
>> implicitFreeSurface=.TRUE.,
>> implicitDiffusion=.TRUE.,
>> implicitViscosity=.TRUE.,
>> rigidLid=.FALSE.,
>> nonlinFreeSurf= 4,
>> exactConserv = .TRUE.,
>> metricTerms = .TRUE.,
>> # #########################################
>> # topog params
>> hFacMin=0.2,
>> hFacMinDz=20.0,
>> # #########################################
>> &
>> 
>> # Elliptic solver parameters
>> &PARM02
>> cg2dMaxIters=1000,
>> cg2dTargetResidual=1.E-13,
>> ##########################################
>> &
>> 
>> # Time stepping parameters
>> &PARM03
>> niter0=0,
>> endTime=1581120000.0,
>> deltaTmom=900.,
>> deltaTtracer=900.,
>> deltaTfreesurf=900.,
>> abEps=0.1,
>> dumpInitAndLast=.TRUE.,
>> dumpFreq=0,
>> pChkPtFreq=31622400.0,
>> chkPtFreq=2635200.0,
>> monitorFreq=2635200.0,
>> # pickupSuff='ckptA',
>> ##########################################
>> &
>> 
>> # Gridding parameters
>> &PARM04
>> usingCartesianGrid=.TRUE.,
>> usingSphericalPolarGrid=.FALSE.,
>> ygOrigin=-75.0,
>> delX= 288*25e3,
>> delY= 288*25e3,
>> delZ= 20.0,      21.11083314,   22.28336379,   23.52101874,   24.82741509,
>>  26.20637086,   27.66191612,   29.19830477,   30.820027  ,   32.53182237,
>>  34.33869368,   36.24592163,   38.25908017,   40.38405288,   42.62705009,
>>  44.99462708,   47.49370322,   50.13158219,   52.91597333,   55.85501416,
>>  58.9572942 ,   62.23188   ,   65.68834173,   69.33678107,   73.18786078,
>>  77.25283583,   81.54358634,   86.07265223,   90.85326995,   95.8994111 ,
>> 101.22582329,  106.84807324,  112.78259227,  119.04672431,  125.65877663,
>> 132.63807329,  140.00501165,  147.78112198,  155.98913035,  164.65302512,
>> 173.79812695,  183.45116289,  193.64034444,  204.39545001,  215.74791197,
>> 227.73090848,  240.37946047,  253.730534  ,  267.82314826,  282.69848968,
>> 
>> 
>> ##########################################
>> &
>> 
>> &PARM05
>> bathyFile='bathy.bin',
>> zonalWindFile='wind.bin',
>> &
>> ________________________________
>> This message (and any attachments) is for the recipient only. NERC is subject to the Freedom of Information Act 2000 and the contents of this email and any reply you make may be disclosed by NERC unless it is exempt from release under the Act. Any material supplied to NERC may be stored in an electronic records management system.
>> ________________________________
>> 
>> _______________________________________________
>> 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