help regarding the time step

Report or discuss software problems and other woes

Moderators: arango, robertson

Post Reply
Message
Author
mashinde
Posts: 135
Joined: Mon Jun 22, 2009 3:46 pm
Location: Indian Institute of Tropical Meteorology, Pune, INDIA

help regarding the time step

#1 Post by mashinde » Thu Sep 09, 2010 6:11 pm

Hi frds,

I have doubts of time steping.

I created 1/16 deg resolution with 72 levels over Mediterranean sea. when I use time step say dt=180 and ndfast=30, the it blows up after few time steps. If I put very small time step say dt=10 and ndfast=5 then model works fine.

it is not good idea to keep such low time step for long simulation. it will take so much time to complete simulation.

here I am attaching .in,.h and .log file for more details.

any suggestion/comments?


Thanks !
Attachments
rmed16_grd_a.log
(67.18 KiB) Downloaded 132 times
rmed16.h
(4.54 KiB) Downloaded 142 times
rmed16_grd_a.in
(84.71 KiB) Downloaded 136 times

User avatar
kate
Posts: 3780
Joined: Wed Jul 02, 2003 5:29 pm
Location: CFOS/UAF, USA

Re: help regarding the time step

#2 Post by kate » Thu Sep 09, 2010 11:34 pm

Our experience is that going from N=42 to N=60 created a much stiffer timestep limit. What are you trying to resolve with N=72? Can you do as well with a smarter vertical stretching and not quite so many levels?

mathieu
Posts: 74
Joined: Fri Sep 17, 2004 2:22 pm
Location: Institut Rudjer Boskovic

Re: help regarding the time step

#3 Post by mathieu » Fri Sep 10, 2010 7:03 am

I think the value of rx1 at 14 is too high and that you need to smooth more your bathymetry. Your choice of vertical stratification Vtransform=1 and Vstretching=1 is no longer the common one. I would use Vtransform=2 and check if the vertical stratification is really the one that I need. Otherwise, it is sometimes a good idea to let the initial T/S stratification (if it comes from an interpolation procedure) get smoothed by the model by running ROMS with no forcing and small time step for, say, 1 day and then run it with larger time step and full forcing.

mashinde
Posts: 135
Joined: Mon Jun 22, 2009 3:46 pm
Location: Indian Institute of Tropical Meteorology, Pune, INDIA

Re: help regarding the time step

#4 Post by mashinde » Mon Oct 18, 2010 5:09 pm

Hi,

I am still not get satisfactory time step to run the model for long simulation. I need the time step of the order of 100 to 200 seconds.I am using ETOPO5. I am smoothing bathymetry according to smth_bath.m from the ROMS matlab package. I have following questions about the r factor.

(1)what should be the values or rx1 and rx0 in the Mediterranean sea ?(1/16 deg res. and N=72)
(2)How to determine these values while smoothing ? Is there any specific criteria to follow ?
(3)Is r factor depends on other parameters such as theta_s, theta_b, hc, N, Dx & Dy etc...?

Thanks

User avatar
arango
Site Admin
Posts: 1117
Joined: Wed Feb 26, 2003 4:41 pm
Location: IMCS, Rutgers University
Contact:

Re: help regarding the time step

#5 Post by arango » Mon Oct 18, 2010 7:52 pm

This is very simple to solve. My intuition tells me that you are running near the limit of stability. Then, any change in the strength of the external forcing (say, atmospheric forcing) is making the model to blow-up. Anyone who run long simulations are aware of this fact. You have to check your external forcing to see if this is the case.

User avatar
shchepet
Posts: 185
Joined: Fri Nov 14, 2003 4:57 pm

Re: help regarding the time step

#6 Post by shchepet » Tue Oct 19, 2010 4:21 am

First, it is a common experience with such a problems that they may be a bit
tricky to start up. How did you create initial conditions? Possibly some kind of
Matlab script. Because your kinetic energy starts with zero, it looks like you
have created temperature and salinity field for the initial state, while
velocities are set to zero. This is OK. But what is likely that the initial
density field is not well balanced resulting in strong excitation of internal
waves after the model starts. This leads to exceeding Courant number for vertical
advection. One way to overcome this problem is, indeed, to run the model with
riduculusly small time step for some duration (usually few days to a month) and
let the waves to calm down. Then obtain restart file and restart using normal
time step thereafter.

Unfortunately your does not supply the all needed information. What is your
barotropic Courant number?

One think caught my attention is setting of the coefficients in set_weights.F
Your log file says:

Power filter parameters, Fgamma, gamma = 0.28400 0.18933

While I do not recognize the exact meaning of these numbers, I thing I may
have some idea about where the 0.284 come from. Your setting may be a bit
too agressive. May be not. I do not know, but it is worth checking.

What are the coefficients Falpha, Fbeta, Fgamma (in Hernan terms) in
set_weight.F ?

You may be a bit too aggressive. Start with Falpha=2, Fbeta=4, Fgamma=0.15
or even less to see whether in changes anything. [Adjust it toward the ideal
value later when model runs.]

Post Reply