[MITgcm-support] The problem in the long run the model

[Martin Losch]

Hi,

I don’t know, if you will be successfull with starting with 5 layers and then later increase to 40 layers, because the vertical resolution is also important for watermasses in particular. I would not recommend it.

You can create “delx.bin” and “dely.bin” as you like (they should contain 1D arrays of dx and dy), but you should make sure that the dx and dy vary smoothly so that the numerical schemes do not become too inaccurate. There are examples of how to specify and create these files, in tutorial_plume_on_slope/input/data:
 delXfile='dx.bin’,
(and similarly delYfile = …)
 and in gendata.m dx.bin is created from 1D array dx like this in matlab:
fid=fopen('dx.bin','w’,’ieee-be'); fwrite(fid,dx,’real*8'); fclose(fid);
in python it would be similar:

import numpy as np
import sys
dx = np.array([10e3,10e3,12e3,15e3, ….])
if sys.byteorder == 'little': dx.byt
eswap(True)
fid = open(fname,"wb")
dx.tofile(fid)
fid.close()

using these files will not slow down the code, because they are read only once in the beginning. having a grid with variable dx.dy does not slow down the code, but reduces the accuracy of th schemes (you can look that up in any textbook about numerical differentiation).

Hope that helps


> On 2. Jun 2017, at 19:18, rahele shafiee <rahele_shafiee at yahoo.com> wrote:
> 
> Hello everyone
> I would like to trace the water mass from the Red Sea and the Persian Gulf to the the North of the Indian Ocean by MITgcm Model. Because the Basin is large (1016 * 904 with 40 vertical layers) so the run of model will be long. Can I choose the number of layers to 5 and then increase to 40 layers after the model stability?
> Is it possible making files of Dely.bin and Delx.bin in the gridding parameters (delx=’delx.bin’ and dely=’dely.bin’) gives different precisions to domain? How can I make these two files? will use of these two files slow the speed of modeling?
> 
> Thanks very much
> 
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support