about model iteration parameter : ocean.in & bio_fennel.in

General scientific issues regarding ROMS

Moderators: arango, robertson

Post Reply
Message
Author
jschoi
Posts: 11
Joined: Wed Apr 12, 2017 6:20 pm
Location: Chonnam National Univ.

about model iteration parameter : ocean.in & bio_fennel.in

#1 Unread post by jschoi »

Hello dear,

I have ran the PHY-BIO coupled model.
I faced the problem which is changed result whenever model-run although same condition.
Is it possible?
If possible, (1) I wonder why the gap is occurred (2) and how I will resolve it.

I asked a similar question previously and I guessed by non-linear term. Is it right?
Your definite answer is a great help to my argument.

For solution, I will try to change ERend and BioIter of parameter in ocean.in and bio_fennel.in respectively.
EX) ERstr=1, ERend=20, BioIter=20
If the above is correct, I have to know how the gap is decreased.


I'm sorry I have so many questions and thank to you and everyone in here.

p.s : added pictures are changed result although same condition. show to differ a isothermal line and current field.
Attachments
temp_2009_09_001.jpg
temp_2009_09_001.jpg

User avatar
kate
Posts: 4088
Joined: Wed Jul 02, 2003 5:29 pm
Location: CFOS/UAF, USA

Re: about model iteration parameter : ocean.in & bio_fennel.

#2 Unread post by kate »

You are running an ensemble of simulations and you're asking why different ensemble members are different? I don't know what ROMS is doing in this case, but generally you want enough spread in your ensemble members to represent the natural (nonlinear) variability in the system. What is it that you are trying to do? Change biological parameters and have the same physics solution in each run?

jschoi
Posts: 11
Joined: Wed Apr 12, 2017 6:20 pm
Location: Chonnam National Univ.

Re: about model iteration parameter : ocean.in & bio_fennel.

#3 Unread post by jschoi »

kate wrote:You are running an ensemble of simulations and you're asking why different ensemble members are different? I don't know what ROMS is doing in this case, but generally you want enough spread in your ensemble members to represent the natural (nonlinear) variability in the system. What is it that you are trying to do? Change biological parameters and have the same physics solution in each run?
One input have to one result in my think.

But whenever model-running, result is changed although only one input.
I know that same result don't be made by non-linear term.
I want to just decrease the gap at least for one input one result.
Therefore I want to convergence non-linear term.

I don't increased iteration number not yet.
and I don't changed any condition(initial, boundary, ATM forcing, parameter .. etc.)
and I don't use data assimilation.

Thank you always

User avatar
kate
Posts: 4088
Joined: Wed Jul 02, 2003 5:29 pm
Location: CFOS/UAF, USA

Re: about model iteration parameter : ocean.in & bio_fennel.

#4 Unread post by kate »

Like I said, I've never used the ROMS ensemble capability, so I don't know what it does. You should perhaps find out for yourself what it is doing to "inflate the solution space", i.e., not get the same answer for each member. If you really want the same solution, you can do that by running many individual runs.

jschoi
Posts: 11
Joined: Wed Apr 12, 2017 6:20 pm
Location: Chonnam National Univ.

Re: about model iteration parameter : ocean.in & bio_fennel.

#5 Unread post by jschoi »

kate wrote:Like I said, I've never used the ROMS ensemble capability, so I don't know what it does. You should perhaps find out for yourself what it is doing to "inflate the solution space", i.e., not get the same answer for each member. If you really want the same solution, you can do that by running many individual runs.
Thank you for replying despite my poor English.
Let me ask you the last question.
I will use the method below to get the ensemble mean effect(to convergence non-linear term).

ERstr=1
ERend=20 :::: ocean.in

BioIter=20 :::: bio_fennel.in

The number of iteration not enough, even though, please check this is the right way for the convergence effect.
I'm sorry that I ask you who have not used ensemble.

add question : If I set parameter ERstr=10, ERend=20, is the number of iteration 10? and is it differ "ERstr=1, ERend=10"?

Thanks kate always.

Post Reply