[MITgcm-support] Correct estimation of Viscosity and Diffusion parameters.

[Martin Losch]

Harry,

selecting the viscosity parameters always involves a bit of tuning, as not only physical considerations are important, but also numerical stability. Please have a look at the header of pkg/mom_common/mom_calc_visc.F where there’s a short summary of how the horizontal viscosities are calculated. Also, have a look at the manual, in particular section “2.21 Nonlinear Viscosities for Large Eddy Simulation” <http://mitgcm.org/public/r2_manual/latest/online_documents/node85.html>. By default all viscosity parameters are set to 0. Using the unscaled parameters, e.g. viscA4, is resolution dependent.

implicitDiffusion = .TRUE., will only lead to an implicit solution technique for the vertical diffusion terms. There’s also the parameter implicitViscosity, that does the same for viscosity. For variable vertical viscosity/diffusion you need to turn on a mixed layer model (KPP, GGL90, …) or a convection scheme (e.g. by setting ivdc_kappa = some large value, e.g. 1).

Martin

On Jun 17, 2014, at 2:17 AM, HARRISON STEEL <hste3178 at berkeley.edu> wrote:

> Hello everyone, 
> I'm working on using MITGCM to simulate a slow moving (maximum velocities less than 1cm/second) ice covered lake. The lake is about  6 by 3 km, and at most  150 meters deep. I've set up a coordinate system using delX=50 meters in X and Y directions, and in the Z direction I have 20 intervals, starting from one meter thickness at the surface, and smoothly approaching 10 meters thickness at the bottom of the lake.
> 
> I am wondering, if the viscosity and diffusion parameters I have selected are going to function correctly for this task?
> 
> viscC2smag=2.2,
>  viscAz=0.001,
>  viscA4=300.0,
>  diffKhT=1.0e-7,
>  implicitDiffusion=.TRUE.,
> 
> viscC2smag has been chosen as this was similar to values recommended in the manual. 
> 
> viscAz has been chosen as 0.001 as studied in this paper: Sensitivity of MITgcm to different model parameters in application to Cayuga Lake, A. Dorostkar,  L. Boegman, P. DIamessis, A. Pollard
> 
> viscA4 has been chosen as 300 in keeping with that in the paper: Numerical experiments on subaqueous melting of Greenland tidewater glaciers in response to ocean warming and enhanced subglacial discharge, Y. Xu, E. RIgnot, D. Menemenlis, M. Koppes
> 
> diffKhT is again chosen as a value similar to the manual and example problem setups. 
> 
> implicitDiffusion was chosen as I believe this will take care of itself in terms of the vertical diffusion, if what I read in the manual is correct.
> 
> 
> I have read Adjoint sensitivities of sub-ice-shelf melt rates to ocean circulation under the Pine Island Ice Shelf, West Antarctica, P. Heimbach, M. Losch
> 
> 
> In which the authors talk about grid-scaled velocities being implemented, but I am unsure how exactly to go about doing this, but their vertical viscosity is the same as mine. 
> 
> 
> Thank you in advance for any wisdom you may have regarding how these parameters can be set to make my simulation as physically realistic as possible.
> 
> As an aside, does anyone know if and where the Icefront package outputs ice melt rates? In the paper on Subaqueous Greenland glacier melting they plotted melt rates, but I haven't found this in the output files...
> 
> Regards,
> Harry
> 
> 
> 
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support