[MITgcm-support] initial velocity field for staggered grid

[Martin Losch]

Hi there,

I am sorry for the confusion, I mixed up east and west. In more general terms, I would interpolate my boundary data to the velocity grid point ***inside*** of the tracer grid point. In your example you’d interpolate to 96E (if the tracer index OB_Iwest=1 is a boundary point) for the western boundary (left hand side of the domain) and to 99E for the eastern boundary (if the tracer index OB_Ieast=5=Nx is a boundary point). I think the model then copies theses fields to 96E (and 95E) and 99E (and 100E), respectively (where the positions in parentheses are the “less important” ones, less important, because these points are generally not used). See obcs_apply_uv.F for details.
There is the option OBCS_u1_adv_T/S to make the model use 1st order upwind at the boundaries. In this case you only need the “interior” velocity gridpoints (96E and 99E) and the exterior points (95E and 100E) are not used.

Martin

On Sep 24, 2014, at 10:19 AM, 钱钰坤 <qianyk at mail3.sysu.edu.cn> wrote:

> Hi Martin,
> 
> Thanks very much for your detailed explanation.  I'm still a little confused.
> 
> For example, my setup (x-direction) ranges from 95E (lon_west) to 100E (lon_east) with 1-deg resolution.
> Tracer (T, S) points are 95.5E, 96.5E, 97.5E, 98.5E, 99.5E, yielding 5 grids;
> velocity (u) points are 95E, 96E, 97E, 98E, 99E, yielding 5 grids;
> 
> I derive my IC and OBCS from SODA.  So I guess the west OBCS of tracer should be set by interpolating SODA onto 95.5E while that of u should be set by interpolating SODA onto 95E.  As for the east OBCS, tracer can be set by 99.5E-SODA while u can be set by 100E-SODA.
> 
> As you mentioned earlier, OBCS of u will be used at i=OB_least and, more importantly, at OB_Ieast+1.  That is, u will be used at 95E and 96E?  It seems a little strange for me because u are only derived from 95E SODA.
> 
> 
> ------------------
> Best regards 
> 
> Yu-Kun Qian (钱钰坤) 
> Center for Monsoon and Environment Research
> Department of Atmospheric Sciences
> School of Environmental Science and Engineering 
> Sun Yat-sen University 
> No. 135 Xingang West Road, Haizhu District 
> Guangzhou, 510275, P.R. China 
> Tel; 020-84115227 
> Email: qianyk at mail3.sysu.edu.cn
>  
>  
>  
>  
> ------------------ 原始邮件 ------------------
> 发件de人: "Martin Losch"<Martin.Losch at awi.de>;
> 发送时间: 2014年9月24日(星期三) 下午3:37
> 收件人: "MITgcm Support"<mitgcm-support at mitgcm.org>;
> 主题: Re: [MITgcm-support] initial velocity field for staggered grid
>  
> Hello,
> 
> the default boundary conditions for the MITgcm is double period, i.e. when you specify an initial u-velocity at i = 1 (which is a “lon_low”), then this is copied to u at i = Nx+1  (which is at “lon_high” = “lon_low” + Nx*dx, I am assuming a regular grid with constant dx for simplicity), analogous for the j-direction and v
> 
> If you don’t want periodic BC, then you need to have a wall (depth = 0) along i=1 or i = Nx (or both, but one is enough), or you need to use open boundary conditons. With the OBCS package you can specify u a i = 1 (more generally i=OB_Ieast) and i = Nx (OB_Iwest). These values overwrite the default periodic boundaries. Because of the staggered nature of the grid, the specified u will be used at i=OB_least and, more importantly, at OB_Ieast+1, and at (more important) OB_Iwest and OB_Iwest+1. The j-case is analogous. For T and S, the boudary values are at grid index OB_Ieast/west, but as you have found out, the position of the scalar points with these indices is between the velocity points OB_Ieast/west and OB_Ieast/west+1. If you want to be super accurate, you’ll have to take tha into account.
> 
> Martin
> 
> On Sep 24, 2014, at 7:19 AM, 钱钰坤 <qianyk at mail3.sysu.edu.cn> wrote:
> 
> > Hi all,
> > 
> > I setup a regional run using OBCS.  The initial conditions (ICs) includes U, V, T, and S.  I notice that the model is defined on C-grid (staggered grid).  Therefore, the ICs of T and S can be specified on cell grids (first longitude = lon_low + resolution / 2).  For zonal velocity U, however, they are defined on face grids  (first longitude = lon_low).  Do I need to prepare the face-grid IC for U which is different from cell-grid IC for T and S?  If yes, then I think that there should be Nx+1 U grids (one more than the cell grids Nx) but in fact IC of U contains only Nx grids (as example case shows).  Similar problem could be found for V (Ny).
> > 
> > I also notice that when specifying the OBCS, the west OBCS for T and S may be defined at lon_low + resolution / 2 while that for U should be defined at lon_low.  Has this to be taken into account when generating OBCS?  Does the OBCS at t0 exactly the same as IC? for different BCs? for different variables?
> > 
> > ------------------
> > Best regards 
> > 
> > Yu-Kun Qian (钱钰坤) 
> > Center for Monsoon and Environment Research
> > Department of Atmospheric Sciences
> > School of Environmental Science and Engineering 
> > Sun Yat-sen University 
> > No. 135 Xingang West Road, Haizhu District 
> > Guangzhou, 510275, P.R. China 
> > Tel; 020-84115227 
> > Email: qianyk at mail3.sysu.edu.cn
> >  
> >  
> > _______________________________________________
> > 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
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support