Ocean Modeling Discussion

arango

I think that there is a confusion here with the forcing terminology. The surface and bottom heat ( Q ) and freshwater fluxes ( E-P ) are the boundary conditions to the vertical tracer (temperature and salinity, respectively) diffusion terms in the governing equations. We need to provide boundary con...

arango

The lake Jersey test case works for me. What you did is not LAKE_JERSEY :!: You changed the grid, land/sea mask, and added big and small islands. Stop calling it Lake Jersey because it is a different application and will confuse other users with misinformation! If doing nesting, it will require diff...

arango

Yes, thank you. It just need:

Code: Select all

#ifdef NESTING
      USE mod_nesting
#endif

Please, update.

arango

What do you mean by the same results? profiling time or state solution?

Usually, it is much efficient to do less partition in the I-direction (NtileI) and more in the J-direction (NtileJ) because of vectorization. Also, it depends on the computer architecture.

arango

Yes, thank you!

arango

Everybody has their own procedure for updating code from repositories. I have several research repositories, and I usually do the following steps: Copy the version that I want to update to a temporary one. So, I have a version of the code that works in case something goes wrong with the updating. I ...

arango

Well, you cannot pick and choose what routines you need to update. It doesn't work that way. Every change done to ROMS is well documented in the trac system, and the changes sometimes depend on previous changes and may be modified in future changes. The last major update to the Parallel I/O was done...

arango

I can believe that this is done :shock: We are not responsible for corrections to the code by a third party. Your solution is straightforward. Update your code from the official ROMS repositories. The agreement to receive a freely distributed version of the ROMS framework from us is that Users need ...

arango

Well, the error is telling your problem, if you think carefully about it. I assume that you previously defined SOLVE3D . Now, it is undefined , and you are solving the shallow-water equations of ROMS where you must have NDTFAST =1 for all nested grids, and you have NDTFAST = 30 :!: If you don't und...

arango

The varinfo.dat is an user-changeable file. In the beginning, we were thinking of writing omega in the restart NetCDF file as m 3 /s, and true vertical velocity in history files. But that's no longer the case. Nowadays, omega is recomputed internally. I rarely activate the switch for writing omega i...

arango

Well, the omega vertical velocity in ROMS is special and its orthogonality changes at each grid point because it depends on the bathymetry. It does not have a physical interpretation since it is only related to ROMS descrete, terrain-following coordinates, 3D continuity equation. Internally in ROMS,...

arango

I also have been using this capability for years and frequently and I never have any problems. Our group at Rutgers uses the same strategy on a daily basis in our operational system. Something is wrong in your files. When I do nesting, I only specify the forcing files for the first grid, and ROMS wi...

arango

The reason why nobody replayed to your post is that you didn't provide any concrete information about the ROMS or Fortran error. We are not magicians or have any potions to see the error written to your standard output. I have mentioned several times over the years in this forum to never never never...

arango

Are you trying to change ROMS equation of state? The definition of in situ density anomaly or potential density anomaly is that a 1000 kg m-3 value is subtracted to the density. The in situ density anomaly, as a field, is only used in ROMS in the pressure gradient algorithm where its value is not e...

arango

Yes, that's a small bug, thank you. I am running nowadays with the ROMS_STDOUT option that generates the output file log.roms instead of the standard Fortran output from unit 6. My bad, I forgot to run without the ROMS_STDOUT . Having this option activated is very convenient when running coupled sys...

arango

Oh my goodness, no. We cannot use WET_DRY in any of the adjoint-based algorithms including any of the 4D-Var drivers. That's the reason that you will find WET_DRY_NOT_YET in the code. Every time that you see a multiplication of a state variable with a mask array, we will need another term, which is ...

arango

Yes, gfortran is a stricter compiler. Other compilers relax the Fortran standard rule that we cannot have a module, subroutine, or variable with the same name. I missed this one. Thank you for reporting this issue. Please, update your code.

arango

In our experience, it is challenging for either the tangent linear ( TLM ) or adjoint models ( ADM ) to blow-up in the first iteration of the inner loops. Something must be wrong in your step-up. Both models are linearized from the nonlinear model ( NLM ) state trajectory (prior). The TLM and ADM st...

arango

Yes, thank you. Please update your code.

arango

Update your code again!

arango

I actually removing that parameter in an update that I am releasing today. Remove the couple of ocurrences of Ltracing in read_phypar.F. I have several reseach repositories and sometimes unwanted features leaks into the trunk. Sorry about that.

arango

Great work, Stefan. I concur with john about your telescoping approach into meter scales. It will take me some time to digest the changes that you make to the code. If you provide us the modifications that you have to do for the shallow-water nested applications to work, we can incorporate such chan...

arango

Okay, update your repository. It is now fixed. Here is the

trac ticket

arango

Of course, you need the observation error covariance (diagonal matrix) in the ROMS 4D-Var algorithm! The NetCDF variable is obs_error , which is read into ROMS variable ObsErr . You are searching for the wrong variable. Check ROMS/Modules/mod_fourdvar.F for the description of all variables.

arango

Yes, it will be easy to fix. I will update the repository tonight. Thank you for reporting this issue.

arango

Hi Koushik, great study, and thank you for the research that you continue doing about ROMS parallel I/O :!: Let me digest the information and come up with a strategy for more efficient parallel I/O. Are you aware that ROMS has the CPP option NO_WRITE_GRID , which suppresses the definition and writin...

arango

Yes, good catch :!: Thank you. I have been looking for a bug in the HSIMT scheme on and off. We fixed the issue that it needs three ghost points a few days ago, but forgot to make changes to the range of oHz . We need have an additional switch, Lhsimt : logical :: LapplySrc, Lhsimt, Lmpdata ... Lhsi...

arango

Yes, thank you for reporting this typo in the documentation. However, why do you want to set a different tracer advection scheme for temperature and salinity across nested grids? I will never do such a thing in any of my nesting applications. It doesn't make sense to me. You need to understand how t...

arango

Yes, I already knew about this one and included in a repository update that will be released today.

arango

Hmmm, I noticed that you activated both BULK_FLUXES and CCSM_FLUXES . I am not familiar with that version of ROMS. However, my understanding is that you should use a single formulation. You need to choose one and not both. If the sea ice model is activated, we need BOTH freshwater (E-P) and saltwate...

arango

I revised this logic recently. My assumption is that ndefAVG is an exact multiple (integer factor) of ntimes . If it is not the case, the file creation definition is changed. What values are you using that gives you OutFiles=0 ? The time averaged files are special because of its delayed creation. Th...

arango

I don't know if this behaviour will be related, but I would turn off SPLINES_VDIFF and SPLINES_VVISC and see what happens. I also noticed that you are using a different version of ROMS that has sea ice modeling. This will complicate your analysis.

arango

Well, u2dbc_im.F and v2dbc_im.F need to be modified to include code relevant to 2D shallow-water nesting applications. I guess that we never accounted for a User needing nesting in a 2D application. It will need an equation to impose the transport from the donor coarser grid. We will have to figure ...

arango

Notice that the indices told and tnew are local (dummy) time indices: ! ! Set time snapshot indices for the donor grid data. ! told=3-RollingIndex(cr) tnew=RollingIndex(cr) However, if you are running a 2D shallow-water application ( SOLVE3D is off), why are you dealing with state variables ( DU_avg...

arango

Hmm. I will have to think about it but maybe it will be needed. It will need to be tested. There are a set of those indices for RHO-, PSI-, U-, and V-points. It will be only relevant it coastlines are present in the contact points.

arango

Yes, Great catch

Thank you for reporting these two bugs. I updated the

repositories.

arango

It is all a matter of the assumed background error covariance hypothesis and its parameters. What do you trust more the model or the observations? HF radar velocity observations are very tricky since it is a very noisy data, and depends on the processing: averaging, detiding, super-observations, err...

arango

You need to make sure that the variables in your forcing NetCDF files have the _FillValue = 1.e+35 attribute or whatever other value you used. Otherwise, ROMS doesn't know how to process such data without the _FillValue that have land/sea masks. However, the fact the you have Min = 1.0E+35 and Max =...

arango

Check your standard input file, roms.in or your standard output log file, which contains lots of information. Notice that in roms.in we have: ! Logical switches (TRUE/FALSE) to nudge the desired climatology field(s). ! If not analytical climatology fields, users need to turn ON the logical ! switche...

arango

The nesting configuration is tricky since we support various types of grid connectivity. We only provide a Matlab interface to generate the nesting connectivity NetCDF file needed by ROMS. Unfortunately, we don't officially have any other software to create such a connectivity NetCDF file nor have t...

arango

The frozen versions of ROMS are available only on SVN. Usually, one checkout the tags SVN repository: svn checkout https://www.myroms.org/svn/src/tags tags svn checkout https://www.myroms.org/svn/src/tags/roms-3.7 roms-3.7 The GIT repository works in a different way. If you need the version that was...

arango

Yes, your first correction is fine. Thank you :!: I updated the repository. We didn't catch that one before because all our grids always have the ( XI , ETA ) coordinates even if the applications grid is spherical. It turns out that in curvilinear grids is better to do the interpolation in ( XI , ET...

arango

This is not a bug. It always has to be equal or greater than one by design. If no input forcing files are needed, ROMS will ignore those values on standard input. I don't understand what it is the Fuzz about it. There are a lot of parameters in ROMS standard input that will be only relevant for a sp...

arango

Look at Nonlinear/ini_fields.F instead. It is very clear how it is computed there. It is how the inital ubar anf vbar is computed in ROMS.

arango

I learned how to program and plot with the NCARG graphical library. I spent so much time reading the manuals and learning how to use that library. It was very satisfying plotting the results with computer-generated contours. I became a pro using that wonderful library. I still used it today. I wrote...

arango

Indeed, I took that class from John Klinck when I was a graduate student at Texas A&M. I learned a lot about coding and ocean numerical modeling in that first class. It was a great introduction to a world that it is now my professional life. I recall that we, the students, and John, enjoyed that cla...

arango

Hi Frank, great catch. Thank you! I updated the repository. See the following

trac ticket for more information.

arango

An application of size 899x629x40 is TOO small to be run in 1350 PEs :!: No wonder why is not scaling after 240 PEs. You are killing scalability with excessive MPI collective communications and exchanges. Your 20x20 tile size is too small, and it is killing vectorization. The good news is that you f...

arango

Great study. We are missing important information about your simulations. We need the grid size and the parallel domain decomposition. I have mentioned frequently in this forum that ROMS scalability is a function of the application and depends on the domain size and parallel partition. There is alwa...

arango

The version of the code that we distribute doesn't have a native sea ice model yet. I am sure that it will be included in the future. It has to be located in a subdirectory inside of the Nonlinear directory, like the biology and sediment models. The reason for it is quite logical and straightforward...

arango

The default upwelling example in ROMS test repository has UV_QDRAG activated in the build script. If you notice, there are various options in build_roms.sh : # can be used to write time-averaged fields. Notice that you can have as # many definitions as you want by appending values. setenv MY_CPP_FLA...

arango

ROMS/TOMS 3.7 Released The latest version of the :arrow: ROMS/TOMS svn repository , revision 974 , was frozen and tagged as ROMS 3.7 . We haven't tagged a frozen version of ROMS in several years. This version is quite stable, and it is the one to be used at the upcoming 2019 ROMS 4D-Var Workshop to...

arango

Indeed, thank you! I didn't test with LIMIT_STFLX_COOLING activated.

arango

Indeed. Thank you for reporting this issue. Please update.

arango

The ROMS basic numerical kernel has not changed much in a few years. However, configuration, CPP options, and parameters in the standard input have been modified and enhanced. You need to be sure that your header file ( application.h ) and standard input script ( roms.in ) is equivalent between the ...

arango

ROMS has nothing to do with the time convention and elapsed time. The user selects how the time is managed in an application. There is nothing that we can do about it. It is up to the person(s) that set-up that application to use a consistent time zone in all the input NetCDF files (initial conditio...

arango

Your problem is that you haven't configured the build_roms.sh and build_roms.bash correctly. For example, notice that in the build_roms.sh script we have: #-------------------------------------------------------------------------- # If applicable, use my specified library paths. #-------------------...

arango

Segmentation violations (SIGSEGV) are usually a sign of memory problems on the computer and available stack for the run. Perhaps, your computer doesn't have too much memory for your application, and you are deploying too many floats for it. I will try to run without floats to see if it works. Then a...

arango

You didn't understand at all what I was saying to you, or what WET_DRY does. The wetting and drying option ( WET_DRY ) is modeled in ROMS by changing the associated land/sea masking arrays at every barotropic time-step in step2d , which affect the ocean state variables. The wet/dry mask is a new arr...

arango

The WET_DRY option is not adjointed because it not fully differentiable, and because of every time-step of the forward trajectory need to be saved to have a consistent linearization of the tangent linear (TLM) and adjoint (ADM) models. You cannot use such time-depending land/sea masking in any ROMS ...

arango

My current strategy is to activate LONGWAVE_OUT and provide downward longwave radiation so bulk_flux will subtract the outgoing IR by using ROMS surface temperature ( K ), infrared emissivity, and Stefan-Boltzmann constant. I also use downward shortwave radiation and I activate COOL_SKIN to correct ...

arango

ROMS has nowadays a very robust calendar. See the routine ROMS/Utility/dateclock.F for more details. It is consistently internally and can handle different reference date ( epoch ) for each forcing field from NetCDF files regardless of what we choose in the ROMS input parameter TIME_REF . I have to ...

arango

I agree with John. This part is not desirable. We and others have been using this standard for years. There are lots of datasets out there that follow this nomenclature. I am sympathetic with the issue of the links but some of the changes suggested are radical.

arango

Indeed, I believe that unsavory piece of code was in the original COARE code. It is quite clever if you ask me. I recall putting enough comments in the code when I did some research about this threshold value many years ago. There is not a standardized unit for specific humidity in the atmospheric d...

arango

Thank you. The ROMS metadata file varinfo.dat is a user file and can be edited in any way that the user feels is necessary. You can even change the language. Some users in France wanted some of the attribute values to be in French. I haven't read the documentation about the UDUNITS-2 package. I look...

arango

Finally, I have time to browse the literature, Googling, and write a Matlab script to check the issue mentioned above. Indeed, the 15.0 factor used to compute the zenith is incorrect :!: We need to have instead: zenith=cff1+cff2*COS(Hangle-lonr(i,j)*deg2rad) In my Matlab script, I compared against...

arango

I updated the code to correct the bug in regrid.F. Check the following trac ticket

src

808 for more details. The parallel bug was corrected. Good luck.

arango

It doesn't make sense to me. MyXout is a state, tiled variable and it is allocated as the others, and the pointer is passed correctly. It is the only way that this can be done. I bet that the problem is not in regrid . It seems like a memory leakage somewhere else. Yes, you can put the application i...

arango

I took a look in the debugger on our US East Coast application and I cannot find anything wrong. I activated both ATM_PRESS and PRESS_COMPENSATE . I am also using BULK_FLUXES , which also need Pair . I don't see a parallel bug. Pair.png shflux.png The Pair is kind of jagged but it is because of the ...

arango

Yes, the Dmem is a diagnostic quantity to estimate the memory requirement for an application. It has nothing to do with the numerical kernel. It is in my TODO list to look at your problem in the debugger. The issues that you are talking about sound like a parallel bug. However, the regrid subroutine...

arango

Both ana_srflux.h and ana_specir.h were coded longtime ago when we were testing the EcoSim bio-optical model in the late 90's. It was like twenty years ago. So I don't remember the details. At that time, both routines were intended for idealized toy problems for debugging purposes. The EcoSim model ...

arango

Thank you, Mark. Your logic is correct. We don't need to add processing logic since it needs ATM_PRESS anyway. I loaded your request ticket to the repository.

arango

Just do it in Matlab.

arango

It has been in ROMS plotting package that uses the NCAR's GKS library since the beginning. Check the file src/Utility/svel.F in the plot repository. It uses the Che and Millero (1976) formula. Chen, C.T. and F.J. Millero, 1976: re-evaluation of Willson sound-speed measurements in pure water. J. Acou...

arango

The oceanography community and the world lost one of its most renowned scientists. Walter Munk was 101. I met Walter in the mid-'80s when I was a graduate student at Texas A&M University. He visited and gave us an excellent Seminar/Lecture. My thesis advisor Robert O. Reid introduced me to him. It w...

arango

:shock: Well, that's your problem. You are using biharmonic diffusivity (m4/s), which scales as 1/12 * ABS(U) * (dx ** 3) . A value for TS_DIF4 of 2 m4/s is too small :!: You need values of the order of 2.0d+8 m4/s . That is, you are using a value that is eight orders of magnitude too small :!: No w...

arango

I guess that the more fundamental question is what type of horizontal advection are you using? Are you specifying explicit diffusion and viscosity? What do you specify for vertical diffusion/viscosity parameterization closure? You should compute your Reynolds Number (U dx/nu), where U is the maximum...

arango

Hi Mark, Are you ready to submit the changes due to the PRESS_COMPENSATE?

Thank you.

arango

Indeed, good catch

Thank you for reporting this bug. I will update the code soon.

arango

Thank you for reporting this problem. I forgot to take care of the roundoff issues with the interpolation weights when I implemented the single precision ( SINGLE_PRECISION option) simulations. Your suggested solutions are incorrect :!: We cannot take the ABS value when computing the interpolation w...

arango

I haven't updated the code yet. I think that Ramsey Harcourt is correct. The shape functions used in mostly all vertical mixing parameterization should be smooth and continuous. We should stay away from sharp discontinuities that lead to eye bullet values in vertical diffusion and vertical viscosity...

arango

Thank you for looking this carefully and for providing a Matlab script to show the difference between the formulation currently coded in ROMS and the correction to the limiter (green curve). Obviously, the coded formulation has a discontinuity spike just before it remains constant. It is clearly wro...

arango

Again, Nope :!: The governing equations are timestep from n (right-hand-side terms) to n+1 (left-hand-side term: time rate of change). The vertically-integrated equations are integrated with smaller timestep to resolve fast dynamics due to gravity wave phenomena. For 3D total momentum coupling, the ...

arango

Nope , there are actually M* barotropic timesteps :!: The reason for it is that the cosine-square shape filter needs to provide time-averaged values of ubar , vbar , and zeta centered at the n+1 baroclinic timestep, which gives us second-order temporal accuracy for time-averaged barotropic motions....

arango

Done. Thank you for reporting this typo.

arango

Nope. What I said above is correct :!: If the last argument to the mp_bcastf routine is single- precision it will use mp_bcast_1d , and if it is double-precision it will use mp_bcastf_1dp . Just look the subroutines in distribute.F :idea: If you don't understard, you need to check a Fortran 1990, 19...

arango

You are doing something wrong or something is weird with the compiler. You need to read carefully the following :arrow: trac ticket . Notice that even if SINGLE_PRECISION is activated, there are several variables that are still in double precision ( dp 64-bit kind parameter) and the r8 kind paramete...

arango

You don't need to do that. One of the great things about NetCDF is that it is self-described. I have mentioned in the past that we need to follow CF convections to guarantee the functionality of metadata and avoid missinterpretation issues in ROMS. I highly recommend users to add the units , time , ...

arango

I am going to try again for the last time. Read carefully :arrow: trac ticket 747 . In the older version of the code, we choose either lower- or higher-level MPI function for exchanges. We no longer do that in the newer versions, you need to experiment and select which options are more efficient in ...

arango

I think that you need to read the following :arrow: trac ticket and choose the MPI communication options that are more efficient in the computer environment that you are running. You should check the profiling information that ROMS reports to standard output to see in what region of the code are slo...

arango

Hmmm, I haven't had the time to examine the issue that you are having. I haven't modified this part of the code in years. I use the split of history files frequently and I have never observed the behavior that you are reporting. I wonder if it is a matter of configuration. Are you changing DSTART ev...

arango

Yes, your problem is the stack size per CPU and it seems to be associated with the automatic arrays used in distributed-memory for I/O operations. This is not a direct ROMS problem, but a computer problem because of not enough memory to handle automatic arrays that are either allocated on stack or h...

arango

It is very simple as I have mentioned several times before. You just need to edit your login script and add one of the lines below:

Code: Select all

.cshrc, .tcshrc, etc.

limit stacksize unlimited

or  .bashrc

ulimit -s unlimited

I wrote lots of information in previous

trac ticket.

arango

Nowadays, severe segmentation errors are associated with stack size, which is used for allocating automatic arrays. They are allocated on stack or heap according to you choice of compiler options. I mentioned this in the last trac ticket.

arango

Nope. That's not what I have in the repository for def_diags.F : # ifdef DIAGNOSTICS_BIO # if defined BIO_FENNEL || defined HYPOXIA_SRM DO ivar=1,NDbio2d ifield=iDbio2(ivar) IF (TRIM(var_name(i)).eq.TRIM(Vname(1,ifield))) THEN got_var(ifield)=.TRUE. DIA(ng)%Vid(ifield)=var_id(i) END IF END DO # endi...

arango

Indeed, thank you. I have more stuff to update in the computation of ROMS memory requirements. I am missing arrays to account for. I am trying to come up with the best estimate.

arango

Yes, we need to update the documentation.

arango

The ECMWF data has a minimal interval of 3 hours; you can easily create annual files instead monthly. It will be more efficient in ROMS to avoid inquiring and open so many NetCDF files. The time units for forcing file is usually days since the reference date ( epoch ). However, you can have any time...

arango

Okay, that will do it. I put safeguards to the code when processing standard input file ocean.in . I made a couple of updates recently for it. It looks that your grid is not for a regional application, but it is a global grid. Is that correct? If that's the case, you have a problem with the design o...

arango

It is a weird error if you didn't repeat the NtileJ parameter in ocean.in , which triggers the allocation of several modules. In yesterday update, I put safeguards for this to never happen. I don't know what to make of your case. I don't think that a corrupted ocean.in should trigger to process the ...

arango

It is weird that you are not activating periodic boundary conditions in the south and northern boundaries :!: It is the perfect and appropriate case for it because your grid is an annulus :roll: You can set the desired flow in the initial condition instead of using Flather boundary conditions. If yo...

arango

That can be also a problem. However, the main issue here is that there are one or more grid cells with a value of 1E+40 1/km for both inverse metric factors pm and pn . We cannot multiply those variables with the mask because of division by zero in ROMS. The grid spacing ( DX=1/pm and DY=1/pn ) is a...

arango

The information is in the front of you. If you don't understand that information, you have homework to do before running ROMS. Look the large and a small numbers. Check your grid files. ( DXmin , DYmin ) cannot be zero!!! It is a trillion times smaller than the scales of electrons... I am sure that ...

arango

I usually put lots of information in trac when I update the code. We have noticed that this information is generally ignored and our emails are treated as spam. There is nothing that we can do about it, and we cannot force users to update and or read the detailed information provided. We have thousa...

arango

PARALLEL_IO works for me. You need to use a recent version of the code. Also, make sure that all your input NetCDF files are NetCDF-4 format. You can check which format you have using ncdump : % ncdump -k filename.nc netCDF-4 To convert from NetCDF-3 to Netcdf-4 try: % nccopy -k netCDF-4 nc3name.nc...

arango

First, I don't think that ROMS will run with the PNETCDF library. It will need special function calls in ROMS. I never tried that library, it is a third party library. If you use PARALLEL_IO, you need to compile with the NetCDF-4 library with parallel MPI I/O and HDF5 support and your computer need ...

arango

Hi Marc and John,

I removed the restriction of having the noncoincident grid points in the indirect connectivity between grids for telescoping applications. See the following

trac ticket for more information.

Please update. Your particular set-up should work now.

arango

Yes, it is Darwin configuration that needs it. As Kate said, just comment the line in the Linux-ifort configuration file. Linux will take the unlimited value from the environmental variable. It is the loader LD that it is having the problem. It depends on what you have in your computer.

arango

By the way, you don't need to modify that routine. ROMS prints out the NetCDF error code at the end of the standard output file when it closes all the files. Notice that there are NetCDF-3 and NetCDF-4 input files for this application. You may be using the wrong one for the compiled code.

arango

You need to provide more information. The new inquiry.F was modified to provide the correct transition between annual files.

arango

I haven't noticed this in my applications that have run stable for years. Are you using those three refinement grids? If so, one of the refined grids is touching the boundary of the other. That will affect the order of the contact points between them. It is the chicken and egg question, what came fi...

arango

It cannot be done during initialization because the CPP include syntax <*****.h> allows to use the header files from user's directories other than ROMS/Functionals . Notice that in analytical.F , we have # ifdef SOLVE3D # if defined ANA_BIOLOGY && defined BIOLOGY # include <ana_biology.h> # endif # ...

arango

There is a reason for it that I need to remember exactly. It is not a bug, but why does it matter? I think that it has to do with with the global attributes of output NetCDF files and the fact that the user can overwrite the location of the code header files instead of using the ones distributed by ...

arango

Yes, you need to select a better option for MPI communications as I mentioned yesterday. Check the information on the svn trac ticket. If you are using too many processes for this application, the data exchanges due to nesting is a bottleneck. See the number for point data gathering . It explains th...

arango

Maybe your application is not big enough to justify 144 nodes, and your application is slowing down because of excessive MPI communications. It is a critical issue that some users ignore always. There is always an optimal parallel partition for each ROMS application. Now if you are using a recent ve...

arango

Do you have the cycle_length attribute in the time variable? For the forward time-stepping, we can also have the following logic in get_cycle.F: IF (job.gt.0) THEN ! forward: Tval(i) =< mday =< Tval(i+1) IF (Linside) THEN tstart=Tmin DO i=2,ntime IF ((tstart.le.mday).and.(mday.le.Tval(i))) THEN Tstr...

arango

Something is weird here. It doesn't make sense to me. Notice that the logic in get_cycle.F that set the switches is: ! ! Is the model time inside the data time range? If not, check if the ! data just has the LOWER- or the UPPER-snapshot interpolant. ! IF ((Tmin.le.mday).and.(mday.le.Tmax)) THEN Lins...

arango

Smaller values of NDIA are possible but its value needs to be an exact factor of NRST, say NDIA=NRST/2 should give you an integer. The time window (in seconds) should be exactly divided by DT.

arango

Well, this logic has worked for us for years. Your problem is that NAVG and NDIA are inconsistent with NRST . If you have AVERAGES and DIAGNOSTICS_TS and/or DIAGNOSTICS_UV activated, you need to set NAVG=NDIA . It doesn't make any sense to have different values because if you are going to compute ba...

arango

The PARALLEL_IO is slower than serial I/O, and it won't help you to improve I/O performance. The last time that I checked that option, it was not working well, and I haven't checked if the NetCDF4 library has made progress on it. I can fix the issue with nf_read2d.F but it will not help you much. Th...

arango

You cannot change those values capriciously. They have to be exactly the same values used by the person that generated the initial conditions, boundary conditions, atmospheric forcing, and observations for data assimilation. The time clock has to be the same in all those files for consistency. The t...

arango

There is a weird set-up of the ROMS time clock here. You need to ask the person that set-up this application. First, the input parameters: 13295.000 dstart Time-stamp assigned to model initialization (days). 19000101.00 time_ref Reference time for units attribute (yyyymmdd.dd) It implies that the RO...

arango

Provide more information from the log file, not just the end. If you are using the latest version of ROMS, there is detailed information in the log file with source code and line number where the error occurred.

arango

Yes, the primal formulation I4DVAR is very expensive because the minimization is done in the physical space (i.e. full application grid for the state vector, 10^5 to 10^7 points) so we need to do a lot of I/O manipulations in cgradient to compute and orthogonalize the Lanczos vectors. We need to rea...

arango

What do you mean by the Lanczos algorithm? Are you talking about the minimization routine, rpcg_lanzos (RPCG)? The Lanczos vectors are stored in a NetCDF and use the adjoint model iterations over all outer and inner loops. I recently profiled WC13 with W4DPSAS : Nouter=2 and Ninner=50 on 4 CPUs on m...

arango

Postdoctoral Researcher Position in Continental Shelf Currents and Their Surface Wave Interactions Departments of Atmospheric and Oceanic Sciences University of California, Los Angeles (UCLA) We are seeking a researcher to work on problems of oceanic circulation dynamics using the Regional Oceanic ...

arango

Fixed the bug. It was not a Y2K bug. I coded datevec incorrectly in module dateclock.F . I was able to reproduce your problem with a simple test. Now, I get: 1 2000-02-28 00:02:30.00 9.193018E-09 1.954312E+04 1.954312E+04 1.659839E+17 (001,32,30) 1.799562E-05 8.159871E-10 0.000000E+00 4.667957E-03 2...

arango

Weird. What calendar are you using? The behavior of your time reporting indicates to me that you are using the Proleptic Gregorian Calendar ( time_ref=0 or time_ref>0 ). The datenum and datevec routines in ROMS starts in in March 1, 0000 to facilitate the management of leap years: MyMonth=MOD(month+...

arango

Weird, I compiled with ifort , gfortran , and pgi and didn't get those errors. It turns out that I missed to transfer the updated files gasdev.F , nrutil.F , and ran1.F when I committed the ticket in another computer. We need these files to have i8b instead i4b . This is the reason why you are havin...

arango

Yes, thank you for reporting this one. The weird thing is that I ran and debugged ( TotalView ) this version several times with the ifort compiler and didn't report any errors or warnings. It seems to be happy with the dot syntax. Of course, the correct syntax in Fortran is to use Rclock % . This ha...

arango

What are you reporting does not make sense to me. The NL_iDataPenalty and NL_fDataPenalty is computed and any of the 4D-Var drivers; w4dpsas_ocean.h in your case. In none of those drivers the VERIFICATION option is used. You just need to use the grep command to confirn this. The definition of these ...

arango

Well, it seems that there is a problem with the latest version of OpenMPI . I wonder what did they change. This kind of problem is difficult to track and frustrating. Compilers and libraries are known to have bugs. One thing to check is if the OpenMPI library was compiled correctly with the same com...

arango

Good catch, thank you

I updated the repository. Check the following

trac ticket for more information.

arango

Yes, thank you very much for reporting this problem with great detail and providing a solution :!: I know that this kind of problems are kind of frustrating and takes a lot of patience to find them. It is refreshing when users get deep into the code and try to understand and find solutions to the pr...

arango

Yes, thank you for reporting this. It is difficult to guess the length of that string. I set it to 4096. See the following

trac ticket for more information.

arango

Yes, thank you. It is always a good strategy to initialize all the switches and variables. I updated the repository. For more information check the following

trac ticket.

arango

Yes, I can see why some compilers may have problems with this type of syntax. Thank you for reporting this issue. This is now fixed. See following

trac ticket.

arango

2016 ROMS/TOMS Asia-Pacific Workshop Institute for Marine and Antarctic Studies Hobart Waterfront, Tasmania, Australia October 17 - 20, 2016 Andrew M. Moore will be giving a half day tutorial on ROMS 4D-Var. We are posting the slides for this presentation :arrow: here ahead of time so attendees can...

arango

2016 ROMS/TOMS Asia-Pacific Workshop Institute for Marine and Antarctic Studies Hobart Waterfront, Tasmania, Australia October 17 - 20, 2016 We have had several cancellations, making it necessary to rework the agenda. If you are presenting , please check :arrow: the agenda again as your presentatio...

arango

This is our design for the ROMS T-shirt for the upcoming Asia-Pacific Workshop. The front is the an artistic rendition of the Hobart skyline with Mt. Wellington as a backdrop.

: front.png (197.79 KiB) Viewed 5308 times

: back.png (179.06 KiB) Viewed 5308 times

: sleeve.png (135.73 KiB) Viewed 5308 times

: front_detail.png (17.64 KiB) Viewed 5308 times

: sleeve_detail.png (2.24 KiB) Viewed 5308 times

arango

2016 ROMS/TOMS Asia-Pacific Workshop Institute for Marine and Antarctic Studies Hobart Waterfront, Tasmania, Australia October 17 - 20, 2016 :!: NOTE: :!: Due to scheduling conflicts with other meetings the workshop has been compressed into 4 days. We apologize for the late notice on this. Please r...

arango

2016 ROMS/TOMS Asia-Pacific Workshop Institute for Marine and Antarctic Studies Hobart Waterfront, Tasmania, Australia October 17 - 21, 2016 This is a reminder that the abstract deadline has passed. Please check the :arrow: participants list to confirm that your abstract has been submitted. If your...

arango

Yes, please update. The error is due to the lack of the QCKNAME keyword in ocean.in . Notice that in :arrow: ticket 706 the field for NetCDF file ID, XXX%ncid=-1 , is initialized to close state in all the I/O structure arrays. This will allow to run newest ROMS version with old input ocean.in scripts.

arango

2016 ROMS/TOMS Asia-Pacific Workshop Institute for Marine and Antarctic Studies Hobart Waterfront, Tasmania, Australia October 17 - 21, 2016 The extended registration deadline for the 2016 ROMS/TOMS Asia-Pacific Workshop is less than a week away. This is a reminder that registration deadline was ex...

arango

Nope. It is not a vertical sum divided by the number of levels. It is a matter of semantics here. Perhaps, depth-averaged will be more appropriate. The shallow water equations are obtained by vertically integrating momentum and continuity equations in flux form. In curvilinear coordinates (xi, eta) ...

arango

Nope, this cannot be done in varinfo.dat . I have thought about this in the past but we have so many variables. However, I can see how this is important for velocity components and third-party potting tools. It is very easy to add the standard_name to the variables in def_avg.F and def_his.F . We ju...

arango

2016 ROMS/TOMS Asia-Pacific Workshop Institute for Marine and Antarctic Studies Hobart Waterfront, Tasmania, Australia October 17 - 21, 2016 The registration deadline for the 2016 ROMS/TOMS Asia-Pacific Workshop is approaching. This is a reminder that registration is open until June 30, 2016 . The ...

arango

Thank you. I have to think about this. I need to reproduce your problem so I can look in the debugger.

arango

The MY2.5 scheme is coded correctly in ROMS and has several corrections that are not available in the original published formulation of the Mellor-Yamada 2.5. There are several papers in the literature about this. Also some people argue that the MY2.5 was not originally formulated correctly and the ...

arango

2016 ROMS/TOMS Asia-Pacific Workshop Institute for Marine and Antarctic Studies Hobart Waterfront, Tasmania, Australia October 17 - 21, 2016 We are inviting you to participate at the 2016 ROMS Asia-Pacific Workshop to be held at the Institute for Marine and Antarctic Studies building on the Hobart ...

arango

This is not related to earlier posts in this forum thread. Anyway, this is weird. If you added biology, only mp_exchange4d is used in just two routines: pre_step3d.F: # ifdef DISTRIBUTE CALL mp_exchange4d (ng, tile, iNLM, 1, & & LBi, UBi, LBj, UBj, 1, N(ng), 1, NT(ng), & & NghostPoints, & & EWperiod...

arango

That's good that you guys use the full stencil at the interface to compute the mass fluxes entering the refined grid. I haven't observed smoothness problems in my testing applications with one-way nesting and 1:5 ratios. Anyway, we should always use two-way nesting since the one-way interaction has ...

arango

We need to take into account that ROMS_AGRIF has a reduced and simpler barotropic stepping engine ( step2d.F ), so it is more efficient. Therefore, we are not comparing apples with apples. Rutgers version of step2d.F includes additional terms so it is slower. In ROMS_AGRIF, all of those terms are in...

arango

What is the size of your grids? Also, what are the partitions that you are using? I want to be sure that you are not using excessive parallel nodes. There is always an optimal partition in ROMS. This is computer engineering and we should abstain to put a lot of processes just because they are availa...

arango

Absolutely not. Why do you want to do such a thing? ROMS has over a million lines of code. There are ways to run simple and small Fortran programs in Matlab but the capability is very limited and you need to write an interface.

arango

Please update. I fixed the problem that you were having. See the following

trac ticket for details. Thank you for reporting this problem

arango

Yes, we are aware of this and it is in our TO-DO list. The issue here is that in two-way nesting (default) we are currently making global gathering of data (full 2D and 3D arrays) between all the MPI nodes to compute easily the fine-to-coarse averaging. This causes a bottle neck because of the exces...

arango

I have a new version of the Matlab scripts that I haven't released yet. I am still thinking about them. They are experimental and I will send them to your e-mail latter today. It provides alternatives to computing the metrics via interpolation instead of the great circle formula. Please sent me an e...

arango

Thanks Jamie for the detailed information. I have to think about this one. Obviously, the spherical grid is doing something to it. It makes me suspicious of the area/volume conservation. I wonder if the grid spacing (area) between coarse and fine grids is conserved in the sherical case. I assume tha...

arango

I will have to reproduce this problem with a test case to see what it is going on. Your fix to main3d.F is not correct.

arango

Refined grid by a factor of 4 is illegal. I don't know why contact.m allowed you to work by such a factor. You should get an error warning and stop. You only can have refimnement ratios of 3 , 5 , 7 , 9 , etc in ROMS because it is discritized on a C-grid . As a matter of dynamics and because of nume...

arango

Perhaps, you should show us a diagram (plot) of all grids and their refinement ratio with respect the donor grid to check if you are setting the grids correctly.

arango

Nope, this is not a good idea! because of the internal manipulations in several routine inside module nesting . It seems to me that you didn't specified the parameters correctly. The order of the grids are crucial. You need to think about the logical order when generating the contact NetCDF file wi...

arango

Yes, I did moved this code after I tested. Thank you for report this. I updated the repository. Nope, the values of BOUNDARY_CONTACT(:,:)%Mflux(:) are only used during debugging and testing to write several diagnostics to Fortran fort.100 , fort.200 , and fort.300 output files. Yes, they are loaded ...

arango

Yep, good catch. Thank yTo be precise we need instead:

Code: Select all

# if defined MASKING || defined WET_DRY
      PRIVATE :: mask_hweights
# endif

I will update the code today.

arango

There are two ways to construct refined nested grids: (1) coarse to fine grid refinement or (2) fine to coarse grid extraction. I prefer to do option (2) since all the quantities needed in the contact area are already available in the donor grid. If you don't see this, think about refinement and com...

arango

Yes, it needs that information from the global attributes. Otherwise, contact.m will not work. These scripts are complex and there is a lot of internal details. They need to be used as a full package. I won't guarantee that a mixture of third party scripts will work together. There is a lot of subtl...

arango

That's your problem! You need to use Lmask=false when generating the contact points with the Matlab script. Please read the instructions! Notice that the default value is false . :idea: I have said several times in this forum that if we don't have experience in nesting applications for a particular ...

arango

This is weird because close_inp is used in the adjoint-based algorithms to facilitate the iterations in the propagators and 4D-Var algorithms. This means that these files were already processed when reading input forcing fields. In the example that you provide above, which is exactly what I wrote in...

arango

I don't know what are you talking about. I checked the code and it already have that correction! I think that you need to update. As you see below, ad_v2dbc_im.F already have the change as shown on

trac for Revision 767:

arango

I did a major overhaul of the sponge areas in ROMS sometime ago. We can have now visc_factor and diff_factor in the Grid NetCDF file. This is the way all application (simple or complex) that require areas of enhanced viscosity and diffusion. See the following :arrow: trac ticket for details. Notice ...

arango

The nesting in ROMS (one- or two-way) is designed for all tracers active and passive. The nesting interface is generic for all tracers. The only constraint is that the bio-geochemical or sediment model needs to be activated in all grids in order to compute the contact point data between all grids (d...

arango

The routines in the white noise module white_noise.F are specific to the adjoint algorithms and we pass the Imin , Imax , Jmin , Jmax dummy variables as arguments. These are tiled parallel routines and need to be used by passing the appropriate values for these I- and J-range parameters. Using the a...

arango

Yes, thank you.

arango

The function zlevs is not part of our official distribution code and we are not responsible for third party functions coded by others. In the ROMS Matlab repository we offer: set_depth.m . This is the official Matlab version of such a function for computing depths at any C-grid location, if you use ...

arango

Users can use the following thread to discuss the

Parallel Global Reduction and Volume Conservation post from the ROMS Messages forum.

arango

WARNING: You cannot replay or post comments in this forum thread. If you have any comments, use the following :arrow: post instead. In ROMS, we have the option to impose a volume conservation in applications with open boundaries if tidal forcing is not anabled: ! Set lateral open boundary edge volu...

arango

I already fixed this problem in contact.m . I pretty much rewrote this logic so it is more robust. However, I am not ready to release it yet. There are other changes that we need to think carefully about and discuss between us before we can release the updated scripts. I don't have the time to check...

arango

Yes, the ROMS restart file is always, by default, in double precision since all the computations in ROMS are in double precision. The option RST_SINGLE was added because some users were complaining that the restart file was getting too big in very large applications. The restart file has only two ro...

arango

The error covariance balanced operator is kind of tricky. It is based on Weaver et al. (2005) method. It imposes a multivariate constraint on the background error covariance such that the unobserved variable(s) information is extracted from the observed (assimilated) data using T-S empirical relatio...

arango

I don't know what are the units. This is used in the downwelling spectral irradiance routine ana_specir.h . This was adapted from a routine provided by Paul Bissett for the EcoSim model. That routine says that it is based on Gregg and Carder (1990) model and few units were provided. It is used to co...

arango

In the ROMS Matlab repository you can find the function depths.m or set_depth.m . More importantly, the above plot was produced with scoord.m , which outputs the quantity that you seek: [z,s,C]=scoord(h, x, y, Vtransform, Vstretching, theta_s, theta_b, ... hc, N, kgrid, column, index, plt, Zzoom) Yo...

arango

You need to start paying more attention to the updates to the code, specially if you are using nesting. You need to read carefully the following :arrow: post . It contains information about the quadratic interpolation. It is not ready yet :!: Also, the information about the wet/dry update was provid...

arango

I hope that you are plotting the cost function to see if 4D-Var is converging. It is not clear to me if you are using 5 or 30 inner loops. If you are using 5 inner loops, I doubt that the 4D-Var algorithm converged and this may explain your blow-up when running the nonlinear model. If your prior SSH...

arango

I pretty much doubt that you will be able to do this. These coastlines files are very special and extremely difficult to reproduce. We need closed polygons and the order of values matters to distinguish between continents, islands, and lakes. The filling algorithm in the plotting package requires th...

arango

Yes, that's correct. There are several coastline files at various resolutions including global datasets: global_low.cst, global_crude.cst, and global_inter.cst. Western longitudes and southern latitudes are negative with ranges -180 to 180 and -90 to 90, respectively.

arango

Although I have not tried passive tracers yet in nesting, the algorithm is designed to work on all tracers (active and passive). We operate on the full 5D tracer array. If works on temperature and salinity, it will work also on passive tracers. As designed, you need passive tracers (like biology) in...

arango

Nope, what I am saying is that in a model like BIO_FENNEL with so many options for additional biological components, you need to know what the order of biological tracer is in idbio(:) and that's the order that you need follow in LBC(isTvar) . If you don't know what that order is, you need to look a...

arango

Okay, I need to track this problem in contact.m . I was able to reproduce you error. I will look at this soon to see what is the solution for this out-of-range values when computing the interpolation weights. I also have in my to do list to have a realistic application with composite grids. However,...

arango

Your first panel is completely illegal and a numerical absurdity; the western boundary is periodic and the eastern boundary closed for zooplankton and others. You better understand what periodicity is. Periodic boundary conditions works in pairs: west/east boundaries and south/north boundaries. ROMS...

arango

Yes, that fixed the problem.

arango

The only Matlab scripts that we support for processing our version of ROMS are located in the Matlab svn repository: https://www.myroms.org/svn/src/matlab :!: These are the official version of these scripts containing the correct functions to process vertical coordinate schemes, nesting, land/sea ma...

arango

I changed the TS_MPDATA to TS_MPDATA_NOT_YET in several TLM , RPM , and ADM routines. It think that this will solve your compilation problem. See :arrow: trac ticket for more information. You need to update. By the way, your header file angola_3km_4dvar.h needs some cleaning, specially when you use ...

arango

Yes, it needs to be SOURCES(ng)%Dsrc(is) , thank you :!: That's was easy an obvious. The ARPACK library is needed in the propagators drivers for Generalized Stability Analysis that uses the tangent linear and adjoint models. What are the undefined references in ad_step3d_t ? We don't need the ARPACK...

arango

I have checked the new released codes carefully for the last two months (countless hours in the debugger) with idealized and a realistic applications in the South China Sea and the US East Coast and they work fine. The one that you were using before have several serious bugs. I don't know what is wr...

arango

The option TS_MPDATA is not adjointed. It is highly nonlinear. You can still use the TS_MPDATA for the forward nonlinear model and any other option for the adjoint and tangent linear models. In cppdefs.h , we have: ** The flags below are optional. By default, the same options chosen ** ** for the no...

arango

Recommendations: :idea: This post has lots of detailed information. Please carefully read and study the provided information before loading this forum with lots of questions that are probably answered below. The nesting algorithms are complex and require experience and patience. I highly recommend ...

arango

This is very simple. You are gaining volume and this is due to the open boundaries. You may plot the domain volume for your solution as a function of time. It seems that using climatology at the boundary is not working for you. Nowadays, we get boundary condition from global models. Imposing climato...

arango

I changed the logic how is this done. Please be patient. An upgrade to the nesting algorithms will be released in the next two days. I am working on the documentation.

arango

One person reported much better performance from the ROMS_Agrif code over both the Rutgers ROMS and the COAWST ROMS when using the nesting. You might try that. Actually, this has a good explanation that mostly everybody is not aware. The barotropic engine in either the UCLA ROMS or ROMS_Agrid (I be...

Search found 1131 matches