Blow up problem before start

General scientific issues regarding ROMS

Moderators: arango, robertson

Post Reply
Message
Author
changjw214
Posts: 25
Joined: Thu Feb 09, 2017 6:44 pm
Location: Inha university

Blow up problem before start

#1 Unread post by changjw214 »

Hi friends~~

I have same problem.
In my case, I wanna start with stable condition. So I set all values(u,v,ubar,vbar,zeta) as zero except salt(32 psu), temp(10 Celcious)in initial condition.
and set same condition in boundary file.

These values can occur error???
I attach my log file for model..

Thanks a lot!
Attachments
log_File.txt
(37.83 KiB) Downloaded 214 times

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

Re: Blow up problem before start

#2 Unread post by kate »

There are several scary values here:

Code: Select all

 Minimum X-grid spacing, DXmin =  2.27035279E-07 km
 Maximum X-grid spacing, DXmax =             Inf km
 Minimum Y-grid spacing, DYmin =  3.30415286E-07 km
 Maximum Y-grid spacing, DYmax =             Inf km
 Minimum Z-grid spacing, DZmin = -3.59640000E+02 m
 Maximum Z-grid spacing, DZmax =  2.26800000E+01 m

 Minimum barotropic Courant Number =  0.00000000E+00
 Maximum barotropic Courant Number =  9.08002235E+04
 Maximum Coriolis   Courant Number =  1.03698266E-02
Also here:

Code: Select all

      0     0 00:00:00           NaN           NaN           NaN           NaN
           (000,000,0)  0.000000E+00  0.000000E+00  0.000000E+00  0.000000E+00
The second set is doubtless due to the first set. DXmin and DYmin are really, really small. What do you think your grid spacing is? You'll need a much smaller timestep if that really is your grid spacing. You don't want to have DXmax and DYmax of Inf. You also don't want a negative DZmin unless you are playing with wetting and drying.

Here is your vertical coordinate:

Code: Select all

 level   S-coord     Cs-curve   Z   at hmin       at hc    half way     at hmax

     5   0.0000000   0.0000000        0.000       0.000       0.000       0.000
     4  -0.2000000  -0.0400000       39.960       0.000      18.720      -2.520
     3  -0.4000000  -0.1600000      159.840       0.000      74.880     -10.080
     2  -0.6000000  -0.3600000      359.640       0.000     168.480     -22.680
     1  -0.8000000  -0.6400000      639.360       0.000     299.520     -40.320
     0  -1.0000000  -1.0000000      999.000       0.000     468.000     -63.000
Is five levels enough for what you are trying to do?

changjw214
Posts: 25
Joined: Thu Feb 09, 2017 6:44 pm
Location: Inha university

Re: Blow up problem before start

#3 Unread post by changjw214 »

I always thank for your reply, Kate.

I think, DX(DZ) problems occur because my grid file didn't have domain length and get negative value of bathymetry data.
So I revised grid data and blows up in 2nd step.
this error has been seen many topics in forum. So I try to run the model in short time step, but didn't change.
(the file I attach is set as ntimes=720, dt=120)

Is it input file(tides or boundary)problem or others?
Attachments
log_8.txt
(26.8 KiB) Downloaded 197 times

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

Re: Blow up problem before start

#4 Unread post by kate »

Code: Select all

u                        1   Flather      Closed       Closed       Flather

 v                        1   Flather      Closed       Closed       Flather

 temp                     1   Radiation    Closed       Closed       Radiation

 salt                     1   Radiation    Closed       Closed       Radiation
Flather isn't an option for the 3-d velocities. I use RadNud for all four of these and provide boundary values from a larger model. Probably didn't cause a blowup that fast, though.

Code: Select all

 level   S-coord     Cs-curve   Z   at hmin       at hc    half way     at hmax

     5   0.0000000   0.0000000        0.000       0.000       0.000       0.000
     4  -0.2000000  -0.0400000        0.000       0.000      -1.260      -2.520
     3  -0.4000000  -0.1600000        0.000       0.000      -5.040     -10.080
     2  -0.6000000  -0.3600000        0.000       0.000     -11.340     -22.680
     1  -0.8000000  -0.6400000        0.000       0.000     -20.160     -40.320
     0  -1.0000000  -1.0000000        0.000       0.000     -31.500     -63.000
Without WET_DRY, you don't want a minimum depth of 0, pick something like 5 m instead.

Code: Select all

 Minimum X-grid spacing, DXmin =  6.28611591E-02 km
 Maximum X-grid spacing, DXmax =  1.40051290E+04 km
 Minimum Y-grid spacing, DYmin =  8.16982139E-02 km
 Maximum Y-grid spacing, DYmax =  1.40051877E+04 km
 Minimum Z-grid spacing, DZmin =  4.00000000E-18 m
 Maximum Z-grid spacing, DZmax =  2.26800000E+01 m
This is still a really wide range for DX and DY. What sort of grid is this? You want very smooth, gradual changes to both DX and DY.

Still, you need to develop some skills for looking at the model output, especially the netcdf files. It wrote all the fields into the restart file - where are they bad?

adundunis
Posts: 15
Joined: Mon Jul 04, 2016 1:31 pm
Location: Inha University

Re: Blow up problem before start

#5 Unread post by adundunis »

hi Kate.

My model keeps blowing up. The one who posted this topic is actually my friend and now his model runs. I compared my .in and header files with his, still my model blows up. I'm looking closer into my input files because I have a strong feeling that the problem is in them. It cannot calculate the Kinetic and Potential energies in the first step. (See below)

Code: Select all

   STEP   Day HH:MM:SS  KINETIC_ENRG   POTEN_ENRG    TOTAL_ENRG    NET_VOLUME
          C => (i,j,k)       Cu            Cv            Cw         Max Speed

      0     0 00:00:00  0.000000E+00  3.637960E+02  3.637960E+02  6.806820E+10
          (000,000,00)  0.000000E+00  0.000000E+00  0.000000E+00  0.000000E+00
      DEF_HIS     - creating  history      file, Grid 01: Results/nakdong_his_0001.nc
      WRT_HIS     - wrote history     fields (Index=1,1) in record = 0000001
      1     0 00:00:05           NaN           NaN           NaN           NaN
          (000,000,00)  0.000000E+00  0.000000E+00  0.000000E+00  0.000000E+00

 Blowing-up: Saving latest model state into  RESTART file



I'm trying to look for any aspect where the error could've possibly been caused. In my grid input file, I have negative h values because I have barrier islands in the middle of my grid. From what I understood, I need to activate Wetting and Drying in my header file if one has to have negative h values. Is my understanding correct?

As a result, I have the following log:

Code: Select all

 Minimum X-grid spacing, DXmin =  6.96509153E-02 km
 Maximum X-grid spacing, DXmax =  8.00000000E-01 km
 Minimum Y-grid spacing, DYmin =  8.04500000E-02 km
 Maximum Y-grid spacing, DYmax =  1.09658561E+00 km
 Minimum Z-grid spacing, DZmin = -1.90909095E-01 m
 Maximum Z-grid spacing, DZmax =  2.07639668E+01 m

 Minimum barotropic Courant Number =  1.38420216E-05
 Maximum barotropic Courant Number =  7.42650581E-02
 Maximum Coriolis   Courant Number =  4.19916072E-04
and

Code: Select all


 Basin information for Grid 01:

 Maximum grid stiffness ratios:  rx0 =            Inf (Beckmann and Haidvogel)
                                 rx1 =            Inf (Haney)

 Initial basin volumes: TotVolume =  6.8064832647E+10 m3
                        MinVolume = -2.4500359567E+03 m3
                        MaxVolume =  1.2914617809E+07 m3
                          Max/Min = -5.2711952140E+03
My grid stiffness ratios are Inf and I have negative DZmin. I also have a negative volume which is very unrealistic. Does that pose an error on my model? If yes, what's the best thing to do to the bathymetry of my barrier islands?

By the way the other input files that I have are river, tide, and initial conditions. I just set my initial conditions to zeros except for the salinity (all 33.29) and temp (all 8 degrees).

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

Re: Blow up problem before start

#6 Unread post by kate »

I have a domain with wetting and drying and my minimum dz is negative. My Beckmann and Haney numbers are not Inf, but rather 1e2 and 1e5, respectively. I can't tell from here why your job is blowing up. Can you tell from the restart record it saved which field is going NaN and where?

adundunis
Posts: 15
Joined: Mon Jul 04, 2016 1:31 pm
Location: Inha University

Re: Blow up problem before start

#7 Unread post by adundunis »

Thanks for the reply Kate. I realized the problem was in my "f" variable (grid file). The land cells all have NaN values as their f so I set them to 0, and the model ran. My Beckmann and Haney numbers are still Inf though.

Post Reply