Nesting Models

MOHID Water forum. Questions and discussion around MOHID Water
gonzalo.dizlois.palomares@xunta.es
MOHID Beginner
Posts: 5
Joined: Thu Jan 01, 1970 00:00

Nesting Models

Postby gonzalo.dizlois.palomares@xunta.es » Fri Jan 29, 2010 09:57

Hi!

I've been trying to figure out how to nest 2 models: father and son, and how to modify the Samples in Mohid to configure the nesting right.
I really don't understand it totally ... please correct me If I'm wrong:

1. First of all, we should have 2 Grid Data files, each corresponding to father and son grids. The father grid domain should include the whole son's grid. So I supose the nodes of the fathers' and son's grid should be the same in the boundaries between them, right? or the son's grid can be unattached to the father's grid?, what ratio is suitable for the cell reduction between both grids? 1:3, 1:5?)
2. Then I should set up both models independently, the son's model won't have boundary conditions, because it'll get them from the fathers' (apart from the case that a part of the boundary limits are in an open boundary (is that possible?).
3. In the son's model (=submodel) I should enter in module hydrodynamic:
SUBMODEL : 1
!Flather radiation scheme using the father model as a reference solution.
RADIATION : 2
LOCAL_SOLUTION : 2
!Activate the Flow relaxation scheme
DATA_ASSIMILATION : 1

(What does LOCAL_SOLUTION mean? is BRFORCE needed for these case?)

4. I should set up the assimilation.dat file using AssimilationCoefs to correctly configure relaxation times (for the subgrid, right?). There's no general rule to correctly configure these times, so I should make it up with a trial an error approach. Lesser relaxation times means the imposed solution (from the father's grid) is strictly taking into account by mohid, otherwise, the solution is relaxed and is not solved strictly. But the file input (AssimilationZones.dat) only asks for NUMBER_CELLS, the cells in the XX direction, to refer the relaxation times to. What about the other boundaries (you may have boundary with XX= constant, a vertical boundary).
5. The Assimilation.dat file should include the properties to be assimilated by the son's model (in my case u, v). These properties are read from a hdf5 file (wich is the father's grid output hdf5, right?) What do FILE_IN_TIME and TYPE_ZUV mean?. I know TYPE_ZUV relates to where the assimilation data (= father's model solution) are referred to (center cells (z), XX faces (V) or YY faces (V)), but why area they ='z' in the '<beginfield><endfield>' block and =u/v in the <begincoef><endcoef> block?
6. Both models should have the DT set up independently from each other, depending only on the size of the cells for each of the models (father and son).
7. If configuring a grandfather grid (2D) to avoid open boundary problems, then U, V should be 2D (DIMENSION : 2D)?

Thank you, I really appreciate the help,

Gonzalo

paulo.chambel@hidromod.com
MOHID Advanced user
Posts: 117
Joined: Thu Jan 01, 1970 00:00
Location: Lisbon
Contact:

Postby paulo.chambel@hidromod.com » Tue Feb 09, 2010 14:41

Hi Gonzalo,
First of all, we should have 2 Grid Data files, each corresponding to father and son grids. The father grid domain should include the whole son's grid. So I supose the nodes of the fathers' and son's grid should be the same in the boundaries between them, right?
No. They do not need to match.
or the son's grid can be unattached to the father's grid?, what ratio is suitable for the cell reduction between both grids? 1:3, 1:5?)
Yes. Maximum 1:5 ideal 1:3. But can be 1:3,5. You do not need to have rigid racios. The nesting is one-way. This means that you do not need to take care to much about how the nesting grid is matching the father grid.
Then I should set up both models independently,
Yes but is better to follow the strcture use by the GUI for a question of file organization.
the son's model won't have boundary conditions, because it'll get them from the fathers' (apart from the case that a part of the boundary limits are in an open boundary (is that possible?).
Not exactly . You need to specify boundary condition in the son model where you consider the father model has you reference solution.
3. In the son's model (=submodel) I should enter in module hydrodynamic:
SUBMODEL : 1
!Flather radiation scheme using the father model as a reference solution.
RADIATION : 2
LOCAL_SOLUTION : 2
!Activate the Flow relaxation scheme
DATA_ASSIMILATION : 1

That's right
(What does LOCAL_SOLUTION mean?
This is where you specified the reference solution of you boundary solution. The reference solution is the solution that use to :
1) radiate over the model solution;
2) relax (or nudge) the model solution to your reference solution
is BRFORCE needed for these case?)
Yes if you want to relax (or nudge) your model velocities to you reference velocities. This is a very powerfull way of impose a smooth transtion between the father and son solutions. The DATA_ASSIMILATION : 1 option only says that you want to relax (or nudge). After you need to say what you want to nudge.
BRFORCE : means you want to nudge the velocities
BRWATERLEVEL : means you want to nudge water levels (please do not use this option)
there are other hydrodynamic properties you can nudge but their are only use in schematic studies (coriolis force, advection, etc.).
4. I should set up the assimilation.dat file using AssimilationCoefs to correctly configure relaxation times (for the subgrid, right?). There's no general rule to correctly configure these times, so I should make it up with a trial an error approach. Lesser relaxation times means the imposed solution (from the father's grid) is strictly taking into account by mohid, otherwise, the solution is relaxed and is not solved strictly. But the file input (AssimilationZones.dat) only asks for NUMBER_CELLS, the cells in the XX direction, to refer the relaxation times to. What about the other boundaries (you may have boundary with XX= constant, a vertical boundary).
This program automaticly do the same for the other boundaries. It generates a spatial field similar to the frame of portrait.
I have a new Mohid.exe where this is done all automaticly and you do need to run in prepocessor this tool to generate the assimilation time scales. If you want you can send me an email and will send this new exe.

This will cotinue in a new post because I'm having problem when the post is too big.
The peace be with you

paulo.chambel@hidromod.com
MOHID Advanced user
Posts: 117
Joined: Thu Jan 01, 1970 00:00
Location: Lisbon
Contact:

Postby paulo.chambel@hidromod.com » Tue Feb 09, 2010 16:00

5. The Assimilation.dat file should include the properties to be assimilated by the son's model (in my case u, v). These properties are read from a hdf5 file (wich is the father's grid output hdf5, right?) What do FILE_IN_TIME and TYPE_ZUV mean?. I know TYPE_ZUV relates to where the assimilation data (= father's model solution) are referred to (center cells (z), XX faces (V) or YY faces (V)), but why area they ='z' in the '<beginfield><endfield>' block and =u/v in the <begincoef><endcoef> block?

No the son model in run time interpolates the father solution to the son grid. You only need to define the relax time scales.

The TYPE_ZUV is use to define the ASCII_FILES format and is the way to say if a property is defined in the Z poins (cell center), U point (X face) or V (Y face).


6. Both models should have the DT set up independently from each other, depending only on the size of the cells for each of the models (father and son).

Yes but to avoid interpolations in time we have introduce the rule of the DT father model must be a multiple of the DT son model.


7. If configuring a grandfather grid (2D) to avoid open boundary problems, then U, V should be 2D (DIMENSION : 2D)?

I do not understand this question. The mohid model is able to nest father models with son models with completly different vertical geometry (e.g. 2D father - 3D son, 3D father - 2D son, 3D cartesian father and 3D sigma son, etc.).
The peace be with you

gonzalo.dizlois.palomares@xunta.es
MOHID Beginner
Posts: 5
Joined: Thu Jan 01, 1970 00:00

Thank you.

Postby gonzalo.dizlois.palomares@xunta.es » Fri Feb 12, 2010 12:05

Thank you very much for your answers. I'll try to apply them and do some nesting around ... By the way, could you send me the mohid.exe with automatic calculation of relaxation times?
Thank you again

gonzalo.dizlois.palomares@xunta.es
MOHID Beginner
Posts: 5
Joined: Thu Jan 01, 1970 00:00

email

Postby gonzalo.dizlois.palomares@xunta.es » Fri Feb 12, 2010 12:09

Sorry, could you send it to gonza.dlp@gmail.com?

Thanx

administrator@.
Site Admin
Posts: 349
Joined: Thu Jan 01, 1970 00:00
Location: Instituto Superior Tecnico - Lisbon - Portugal
Contact:

Postby administrator@. » Mon Feb 15, 2010 08:55

Hi Gonzalo,

You can download the AssimilationCoefs executable from the download area in www.mohid.com.
Best regards,
Luis

adelcampo@pas.azti.es
MOHID Beginner
Posts: 4
Joined: Thu Jan 01, 1970 00:00

Nested models

Postby adelcampo@pas.azti.es » Wed Apr 14, 2010 16:40

Hello!!

I'm triying to nest 2 models and I cannot manage to do it.... :| I have followed the same structure we used in the last MOHID course in september for a small sample bay. In the course we didn't use the flather radiation condition at the open boundaries as you suggest in the Wiki, instead we impose a wave tide in the open boundary of the father model and compute the tide and the residual flow. So, we wrote the following in the Hydrodynamic file:

For the father model:

TIDE : 1
RESIDUAL : 1

And for the son model:

TIDE : 0
RESIDUAL : 1
SUBMODEL : 1

The results of the father model are ok, but when I run the soon model the run stops telling me \"FATAL; INTERNAL; ActualizeSubModelValues; Hydrodynamic. ERRO8\" I don't understand why is not working!!! I have checked everything :o (both grid data files, father and soon, are in the same coordinate system with a 1:3 ratio, I've estimated both DT using the Courant No, the rest of the models settings are the same except for the horizontal viscosity, smaller for the soon).

Please!!! Could you help me to see what I'm doing wrong?

adelcampo@pas.azti.es
MOHID Beginner
Posts: 4
Joined: Thu Jan 01, 1970 00:00

Re: Nested models

Postby adelcampo@pas.azti.es » Thu Apr 15, 2010 15:07

Problem is solved....I forgot to replace the missing values by zero...

MISSING_NULL: 1

Now the nesting is working ok! 8)

Bay!

luis_at_mohid.com
MOHID Developer
Posts: 553
Joined: Thu Jan 01, 1970 00:00
Location: Instituto Superior Técnico - Technical University of Lisbon
Contact:

Postby luis_at_mohid.com » Tue Jun 08, 2010 10:31

Hi Andrea,

Good. Cheers,
Luis


Return to “MOHID Water”

Who is online

Users browsing this forum: Bing [Bot] and 3 guests