Ocean Modeling Discussion

m.hadfield

I suspect it's all a matter of time. ROMS has a variable called time (usually appearing in the code as time(ng)) representing time in seconds relative to a reference. (The variable TIME_REF can then be used to specify the calendar date-time of this reference.) It also has a variable tdays(ng) equal ...

m.hadfield

> One thing I have noticed is all examples in the/Test/ folder that employ grid coordinates in meters instead of lat-lon... Hello CJ. Greetings from my home in sunny Welllington. I suggest you try specifying float locations in grid coordinates and see if that works. The other thing to look at is the ...

m.hadfield

You've probably misunderstood something I wrote: I'm not smart enough to manually assign MPI processes.

Good luck, Stefan. I too work with grids with ridiculously large fractions (~50%) of tiles masked. In principle one could avoid this with composed grids.

m.hadfield

The guidelines if you're running the Conda package manager: You can use PIP to install modules into your Conda environment, but do this only when you have to. You don't have to for the netCDF library or any of its dependencies. The reason for avoiding PIP with Conda is that you are more likely to ...

m.hadfield

Are you uysing Anaconda? are you able to? The latest instructions assume that that's what you are using. Anaconda handles all the netCDF dependencies automatically?

m.hadfield

For the libgridgen problem, take a look at this issue of GitHub: https://github.com/ESMG/pyroms/issues/26 The original poster on that issue found a solution that involved installing gridgen and its dependencies from the Git repositories. Alternatively, you can install these packages with Conda, but ...

m.hadfield

For anyone who's interested, I am interested in contributing to the development of PyROMS . I have opened a couple of issues: Issue 23 : Improve installation and documentation thereof Issue 24 : Improve documentation and example code As the README says, there are a few rough edges in PyROMS. The ...

m.hadfield

Non-linear feedbacks are not included in the code.

The PRESS_COMPENSATE option should be used in the case you describe where the model providing the boundary conditions does not allow for the inverse barometer effect.

m.hadfield

Thank you. I did not know that lpsolve55 is available via Conda.

m.hadfield

Hi John I've just created a ticket for PRESS_COMPENSATE on the Trac system: https://www.myroms.org/projects/src/ticket/805 I hadn't noticed your follow-up to my comment, sorry, so didn't take it into account. My work weeks are rather fragmented these days! A result of my semi-retired status, an ...

m.hadfield

The UPWELLING test case works OK when either STATIONS or BIO_FENNEL is defined, but fails when both are defined. It gives the message below . I haven't found a solution yet, so I'm just reporting what I've found. It may well make more sense to Hernan. This is the message (Linux-gfortran in serial ...

m.hadfield

Sure. In the next few days.

m.hadfield

Thank you very much for that Sasha.

For a while now I have handed my bathymetry interpolation duties to the GMT surface utility:

http://gmt.soest.hawaii.edu/doc/latest/surface.html

I'll have a good look at what you've done and compare them.

m.hadfield

The RadNud boundary condition nudges the boundary value towards the specified value. This will only affect the interior values when the radiative scheme diagnoses inflow. So: Have you set a nudging time scale for dye at the southern boundary? Have you waited long enough to see an effect in the ...

m.hadfield

If you use git svn to clone the trunk, as Kate has talked about from time to time, you will get a copy of the entire history of the repository and can check out any revision you want.

m.hadfield

Another PyROMS problem; might as well use the same thread... I have more-or-less successfully (?) built and installed PyROMS with the following commands # The system is CentOS 7 with Miniconda (single-user install). # Python 3.6, numpy, netcdf, etc are installed in a custom environment called ...

m.hadfield

Sure do! I submitted the one and only pull request I've ever prepared to you a few weeks ago.

m.hadfield

Python 3.6.6 |Anaconda, Inc.| (default, Jun 28 2018, 17:14:51) [GCC 7.2.0] on linux Type "help", "copyright", "credits" or "license" for more information. >>> import pyroms_toolbox Traceback (most recent call last): File "<stdin>", line 1, in <module> File "/home/hadfield/.local/lib/python3.6/site ...

m.hadfield

I am running "many rivers scenario" without specifying temperature (only salinity) of the river sources and all is working fine

If you're lucky it will continue to do so.

m.hadfield

I don't know if this is relevant here, but even with a well-specified river, ROMS tends to generate an estuarine flow (outflow at the surface, inflow at the bottom) at the grid cell the river enters. The advection scheme can then generate hypersaline water at the bottom, which tends to form a plume ...

m.hadfield

Near the end of your build.bash file you should be able to see lines like the following # Go to the users source directory to compile. The options set above will # pick up the application-specific code from the appropriate place. cd ${MY_ROMS_SRC ...

m.hadfield

kate wrote:Well, needs evolve. The gang wants to build tools that work with more than one model, supporting ROMS, MOM6, Mitgcm in an object-oriented way.

Too much for a bear of little brain like me, I suspect.

So, will pyroms development cease?

m.hadfield

I'm getting emails about pyroms suddenly, right when the guys want to scrap it and start over. I want to develop my own library of Python software for ROMS. It seems to me that the key components will be conda, python3, xarray & dask, esmf and matplotlib. The missing component is the vertical ...

m.hadfield

On Ubuntu, I installed netcdf-fortran with apt instead of with conda: sudo apt install libnetcdff-dev Be sure to use nf-config to find the fortran-netcdf because you'll still have the nc-config in conda. When tried to link to the system netCDF I got errors relating to libgfortran.so version ...

m.hadfield

I get a similar error building pyroms (python3 branch): https://github.com/ESMG/pyroms/tree/python3 The system is Python 3.6, Miniconda, CentOS 7. The warning is Warning: distutils distribution has been initialized, it may be too late to add an extension _interp Warning: distutils distribution has ...

m.hadfield

I suggest you delete *everything* and start over again, and use a configure command like this

Code: Select all

CPRDEF=INTEL CC=icc FC=ifort MPIFC=mpif90 ./configure

m.hadfield

See

https://www.myroms.org/svn/src/trunk/RO ... et_tides.F

and search for "ramp"

m.hadfield

wilkin wrote:We'd be interested to know more about your PRESS_COMPENSATE.

See

https://github.com/hadfieldnz/roms-trun ... u2dbc_im.F
https://github.com/hadfieldnz/roms-trun ... v2dbc_im.F

but replace "&" with "&&".

m.hadfield

Thanks. PRESS_COMPENSATE is something I added to ATM_PRESS. I think it lets you use lateral boundary zeta data from a model without ATM_PRESS (the inverse barometer effect) for a model that does have it. I haven't used ATM_PRESS for a while, but if you're interested in storm surge it's a must-have ...

m.hadfield

kate wrote:Though I don't bother with RAMP_TIDES.

An aside: If you have shallow water in your domain and wetting and drying is not enabled, RAMP_TIDES can be useful to avoid transients that drop the water level below the bathymetry, hence crashing the model.

m.hadfield

arango wrote:So to report all the header files together, I decided to write its usage at the end when there are visible instead of the middle of time-stepping. So when you see weird things like that in ROMS, there is also a reason for it I think that I mentioned this before in the documentation.

m.hadfield

I'm putting this in ROMS Bugs, but it's really an oddity, not a bug. In output from a ROMS non-linear model, there is a list of directories and files near the start of the output, just before the grid and tiling information. Eg. from one of my runs: Local Root : /home/hadfield/ROMS/roms-trunk-mgh ...

m.hadfield

Like many ROMS users--I suspect--I have my own lightly hacked version of the ROMS source code and a collection of roms-related scripts. I have just put a selection of these on GitHub.com for the edification and amusement of the ROMS community. There are 3 repositories: roms-trunk-mgh : There are two ...

m.hadfield

What is the advantage on moving the code from python 2.7 to python 3? I'm approaching this as someone who has used Python (version 2) for many years as a scripting language but is looking to adopt it as my main language for data analysis. I have looked at various discussions like this... https ...

m.hadfield

kate wrote:What are you planning to do exactly?

General stuff: analysis of model output; generating input files. With an emphasis on nested grids.

Does PyCNAL have a Python 3 version?

m.hadfield

I have created a python3 branch of the pyroms repository on github. This is much like the previous PyCNAL repository, but the guys want to start fresh with all new PyCNAL code. So, if I want to develop my own ROMS Python 3 code (using existing packages like xarray) but also want to build on (and ...

m.hadfield

OK, so an all-Git ROMS would be better for people like us. And for everyone after the transition period.

m.hadfield

kate wrote:Honestly, I've had mixed to bad luck with "git svn" on Linux. I gave up and do that through the Mac.

I've been using Git and Git-SVN on Cygwin, but have now switched to Linux, with good results so far. I'll keep an eye out for problems. I should have though Git was very platform-agnostic.

m.hadfield

kate wrote:Hernan hates my suggestion of switching to git.

The good new is that (as you know, Kate) all the users can ignore Hernan's choice of repository format and use git svn to clone their own copy.

m.hadfield

Hi Jamie

Good to hear.

One quibble: if USE_NETCDF4 is "on" then you shouldn't need NETCDF_INCDIR; if it's off then setting NF_CONFIG should have no effect.

Mark

m.hadfield

has-f03 is "yes" so it should work with ROMS.

m.hadfield

According to this https://www.unidata.ucar.edu/software/netcdf/netcdf-4/newdocs/netcdf-install/Binaries.html "The easiest way to get netCDF is through a package management program, such as rpm, yum, adept, and others. NetCDF is available from many different repositories, including the default Red ...

m.hadfield

Here, for completeness, is Linux-ftn-gnu.mk. Build time with the Gnu programming environment is much less than for the others; I conjecture this might be due to a lack of inter-procedural optimizations at link time. I expect run-time performance to be poor, but the Gnu option might be useful when ...

m.hadfield

Oops, forgot compiler options needed for OpenMP on the Intel variant. Corrected file attached.

m.hadfield

Here in New Zealand we have access to a new Cray XC50 (whoopee!) and I have been working out how to build ROMS on it. As discussed in this thread https://www.myroms.org/forum/viewtopic.php?f=31&t=4086&p=15704&hilit=Linux+ftn#p15704 it has 3 programming environments, PrgEnv-cray, PrgEnv-intel, PrgEnv ...

m.hadfield

> Did anybody have similar issue? Yes, a week or two ago. ROMS seemed to object to one forcing file in a sequence, the third or fourth as I recall, but the file was similar to all the others as far as I could see. I resolved to look into this at some stage, but haven't got around to it. I have ...

m.hadfield

To answer Mark Hadfield's remark that coastal modelers have been content to use zeta-only boundary conditions for many years I might cynically suggest that modelers have been happy to accept a lot of things that aren't necessarily particularly good or in any way optimal. No argument from me! :wink:

m.hadfield

I'm afraid there is no manual. But I will add some extra documentation (installation, examples) to both the IDL-Motley and IDL-ROMS packages in the next few days and update the repository.

IDL-ROMS requires IDL-Motley. Have you managed to install IDL-Motley?

m.hadfield

I know there are very few ROMS modellers using IDL out there, but those who are might be interested in my IDL-ROMS package, which is now hosted on GitHub along with my IDL-Motley package, which it requires: https://github.com/hadfieldnz/idl-roms https://github.com/hadfieldnz/idl-motley IDL-ROMS does ...

m.hadfield

+1 to what Kate said. Keeping your local modifications in a Subversion working copy is skating on thin ice. The modifications have no back-up and no history. Personally I use Mercurial with the hgsubversion extension, but Git is more widely supported and Kate has written instructions for using it ...

m.hadfield

See this thread

viewtopic.php?f=19&t=4521&p=17727&hilit=y2k#p17727

and update your ROMS

m.hadfield

I think I've found a Y2K leap year bug in the new string-based date-time output. I have a simulation that runs from mid-1999 to mid-2000. Below are some excerpts from the log file's time-step output. Shortly after 2000-02-28 12:30, the date string switches from 2000-02-28 (which is correct) to 1999 ...

m.hadfield

With Cygwin these days you should install the pre-built Cygwin netCDF libraries rather than building from source.

There are a few Cygwin netCDF packages. I think the one you really need is netcdf-devel, but the others may come in useful too.

m.hadfield

Possibly related issue: Since I updated to 3.7 I have also occasionally had segfaults that don't reproduce themselves. That is, it will segfault, I resubmit the job without even recompiling, and then it doesn't segfault. I understand segfaults that show up inconsistently are often related to ...

m.hadfield

The oceanS executable gets the location of the varinfo.dat file from the following line in ocean_upwelling.in ! Input variable information file name. This file needs to be processed ! first so all information arrays can be initialized properly. VARNAME = ROMS/External/varinfo.dat Your output log ...

m.hadfield

Oops, that advice in the tutorial is about a decade out of date!

m.hadfield

I have a problem which I think is ultimately due to a broken netCDF binary package on CentOS Linux, but before taking it up with the package maintainers I want to raise it on the ROMS forum. I've also raised it on the netCDF group... http://www.unidata.ucar.edu/mailing_lists/archives/netcdfgroup ...

m.hadfield

I second John's suggestion. Don't build netCDF on Cygwin yourself; use the Cygwin packages.

Also, don't try adding extra command-line switches like -lnetcdff; make sure build.bash sets USE_NETCDF4 and let nc-config sort it out.

m.hadfield

This is not specific to ROMS, but please bear with me. There are quite a few groups running ROMS routinely, in configurations that might variously be called operational, quasi-operational and not-really-operational. At the ROMS Asia-Pacific Workshop in Hobart I talked to a few people about a tool ...

m.hadfield

The quick-save output facility introduced in ticket 704 had a bug that was fixed in ticket 708:

https://www.myroms.org/projects/src/ticket/704
https://www.myroms.org/projects/src/ticket/708

m.hadfield

That was quick!

Thanks for the info.

m.hadfield

This is arguably not a bug, rather than an opportunity for user error. When I updated to the latest version of the code, with the new, quick-save file functionality and ran an existing case I got NETCDF_CLOSE - error during closing of file, ncid = 0 file: call from: close_out, close_io.F To cut a ...

m.hadfield

Forcing files require a time variable, but that can be given the attribute "calendar" with the value "none", in which case only a single value is required and will be applied to all times. I assume Kate's albedo files work the same way, and if they don't it wouldn't be hard to fix the code so they ...

m.hadfield

Not only do you know that the crash occurs in check_multifile, you know that it involves a variable called check_multifile_$LCHECK (which I think is a local or module level variable called lcheck in the scope of check_multifile) being used at line 74 without a value being set. Time to employ the ...

m.hadfield

FLOAT_VWALK enables a vertical random walk for the floats. VWALK_FORWARD has an effect only if FLOAT_VWALK is activated. With VWALK_FORWARD, the float vertical movement is implemented using a forward-only time-stepping scheme, otherwise it uses the same predictor-corrector method as the rest of the ...

m.hadfield

You should try (and you should be able to) install all the software components you need as Cygwin packages. This includes svn, gfortran, openmpi, netcdf and the miscellanous development tools: make, perl, etc. There are various things that suggest you haven't done this... If you used Cygwin svn to ...

m.hadfield

In an earlier post you said The program stops at the same point as for this user, although it does give me a line number I believe Does it still give you a line number? If so, what is it? Is there any other output to stderr that you haven't shown us? Once you have a source code line number you can ...

m.hadfield

kate wrote:I see you have zero depth at the coast. Try deepening that just a bit.

A note for novices: without wetting and drying enabled, ROMS will blow up if zeta drops below -h at any grid point.

m.hadfield

You can add non-tidal boundary values to your tidal boundary conditions with the options ADD_FSOBC and ADD_M2OBC. See ROMS/Include/cppdefs.h. The relevant comments in that file say, "use to add tidal elevation/current to processed OBC data". Actually, I think it would more accurate to say "use to ...

m.hadfield

I have never successfully set up a point source detached from the land mask using the standard ROMS facilities. (If someone can tell me how, I would love to hear it.) For point sources of this type I add code to step3d_t.F. This represents the source of sediment mass, but not the accompanying water ...

m.hadfield

How are you implementing your point source and how do you know it stops after 24 hours?

m.hadfield

It doesn't really matter whether you modify the master file system table (/etc/fstab) or the user table (/etc/fstab.d/hadfield in my case). However putting things in the user table is cleaner, in my opinion. As you say, the user table does not exist by default: if you want one, you need to create it ...

m.hadfield

You appear to have a directory name with spaces (/cygdrive/c/Program Files/AMD APP/bin/x86) on your path. Bad idea. The cure is to add an entry like this to your mount table: C:/Program Files on /prog type ntfs (binary,user) That way, Cygwin will convert the above path to /prog/AMD APP/bin/x86 at ...

m.hadfield

What detraceyb says. You did download using Cygwin tools, didn't you? Also, you don't say what netCDF packages you installed, but these days you should use the ones available through Cygwin setup. If you're using Cygwin, you can save yourself a lot of grief by using Cygwin all the way. That, plus ...

m.hadfield

Or if the data are supposed to repeat year on year, you could add a cycle_length attribute to the zeta_west variable.

m.hadfield

I think the names you mention are commands (or maybe packages) that were needed during the transition from GCC 3.x to GCC 4.x, to select version 4 instead of version 3. GCC 4.x is the default now. If you install Cygwin and ensure the following packages are selected (they are currently at version 4.9 ...

m.hadfield

What Hernan said, but do remember that

ROMS levels are numbered from the bottom to the top; and
The ROMS s coordinate is -1 at z=-h and 0 at z-zeta.

I periodically confuse myself by forgetting that s is negative!

m.hadfield

I suggest you get into read_fltpar with a debugger or (my choice) add some WRITE statements. It would be helpful if your compiler told you the line number at which the error occurred. Most compilers can be persuaded to do this. For Gfortran (on most platforms) you just need to set the USE_DEBUG make ...

m.hadfield

Based in Wellington, New Zealand Permanent, full-time NIWA is a dynamic research organisation whose purpose is to enhance the economic value and sustainable management of New Zealand's aquatic resources and environments, to provide understanding of climate and the atmosphere and increase resilience ...

m.hadfield

The code to set the Lnudging switch for each grid in inp_par.F (lines 853-863) is #ifdef SOLVE3D IF (LnudgeM3CLM(ng).or. & & LnudgeM3CLM(ng).or. & & ANY(LnudgeTCLM(:,ng))) THEN Lnudging(ng)=.TRUE. END IF #else IF (LnudgeM2CLM(ng)) THEN Lnudging(ng)=.TRUE. END IF #endif At line 854, shouldn't that be ...

m.hadfield

Hang on, the fix can be simpler than that, just copy the value of the boundary structure's acquire field along with the others: #ifdef SOLVE3D ! ! If processing tracers and last standard input entry (Icont=0), set ! unspecified tracer values to the last tracer entry. ! IF ((iTrcStr.gt.0).and ...

m.hadfield

In a post back in April... https://www.myroms.org/forum/viewtopic.php?f=19&t=3326&hilit=compact ...I reported a problem when radiative+nudging boundary conditions were specified for Fennel biological tracers using the compact specification, for example: LBC(isTvar) == RadNud RadNud RadNud RadNud ...

m.hadfield

I agree. I noticed this a month ago and intended to report it, but never got around to it.

m.hadfield

You don't need to edit the makefile. You can override the NC_CONFIG make variable by setting an environment variable with the same name before running make, eg:

Code: Select all

export NC_CONFIG=nf-config

Perhaps the make variable should now be renamed NF_CONFIG and point to nf-config by default.

m.hadfield

Hi Mark and Bene, Unless things have changed, the NC_CONFIG macro is not very smart, and doesn't actually search for installed versions of NetCDF...it just assumes the netCDF files are in the location associated with the NC_CONFIG macro. Thus, if you have multiple netCDF installations, or have just ...

m.hadfield

Good

, but you shouldn't need to define NETCDF_INCDIR if NC_CONFIG is set correctly.

m.hadfield

AlexisEspinosa wrote:If a particle leaves the partition domain to go into another, then the particle information would be sent into the receiving partition.

I think this is the tricky bit.

m.hadfield

Is OpenMP an option for you? In my experience, this is much better for parallel runs with floats than MPI. I presume this is because the overhead that Kate describes is absent or greatly reduced.

m.hadfield

I think it's crashing at some code in distribute.f90 (is 2674 the line number??), which is called from some code in wrt_gst.f90 (line 186??). I would attack this by inserting WRITE statements at the relevant locations to print out relevant variables, rebuilding and rerunning (repeat as necessary). I ...

m.hadfield

When you set LtracerSrc to false, what is your TS advection scheme? Sometimes TS_U3HADVECTION, sometimes TS_MPDATA, but I may have to back down on my reassurances about LtracerSrc = F. As I said, I've used this setting before now with no problems I have been aware of, but right I'm now looking at a ...

m.hadfield

You have to set those LtracerSrc to True and provide tracer values for the incoming river flow. Otherwise you have a "downwind" advection scheme, known to be unstable. The only time it makes sense to turn LtracerSrc to False is on outflow from the domain (which could argue for LtracerSrc to be on ...

m.hadfield

Surface tracer fluxes in ROMS are stored in FORCES(ng)%stflx, with the convention that this is positive down , i.e. that a positive flux causes the amount of the tracer in the water column to increase. (Incidentally, the same is true of momentum fluxes: a positive value of sustr, say, tends to ...

m.hadfield

My voice may not count for much, but I intend shortly to get into serious use of the nested ROMS (having been forced to abandon such an attempt about a year ago because of a looming project deadlines and the threat to my mental health--I am not entirely joking). I would very much like to see John ...

m.hadfield

I recently set up a simulation with the simple, 7-component Fennel model and--being an adventurous but trusting soul--specified Rad+Nud boundary conditions all round with the compact form, thus: LBC(isTvar) == RadNud RadNud RadNud RadNud \ ! idbio(*) Hmmm. I wonder if the trailing backslash on that ...

m.hadfield

These days, boundary conditions for biological tracers (Fennel model) are set in a file like ROMS/External/bio_Fennel.in, which has the following comment ! NOTICE: It is possible to specify the lateral boundary conditions for ! ====== all biological tracers in a compact form with a single entry ...

m.hadfield

Thank you very much for your reply,kunlv! The log file include all the info ROMS output.I am confused of the model blowing up. So am I, but can we please be clear about the status of the error you originally pointed out? On my reading of the log file you have asked ROMS to read temperature and ...

m.hadfield

This isn't actually a ROMS bug at all (or even a picture of one) but for 6 hours on Wednesday I thought it was. I ran into a problem where ROMS was not reading the units attribute of a time variable in an input netCDF file. In fact it wasn't correctly reading any string attributes at all. The files ...

m.hadfield

Thanks, Hernan!

m.hadfield

ATLANTIS is a spatially compartmentalised ecosystem model (Link et al 2010; van Putten et al. 2013) that requires data on time-varying horizontal and vertical fluxes between the compartments from a hydrodynamic model like ROMS or HYCOM. The preparation of the fluxes is conceptually simple (provided ...

m.hadfield

See also the discussion on this thread:

viewtopic.php?t=3265

m.hadfield

Yes, MPI does work and is the best choice for most situations. OpenMP is generally not used as much and is therefore less well-tested. However OpenMP is much faster than MPI for floats simulations, in my experience. I don't think this bug will be too hard to fix, but don't have time to look at it ...

m.hadfield

It's a bug in the current ROMS when restarting under OpenMP, not specific to any specific compiler or platform. See this thread https://www.myroms.org/forum/viewtopic.php?f=19&t=3260 AS far as I am aware, there has been no solution posted. You may be able to work around it by reverting to an earlier ...

m.hadfield

antoinemarie wrote:For instance, I make a source, and I want this source has the same temperature has a particular point of my model.

Is the particular point the point at which the source is being applied? If so, you can just set LtraceSrc(1) to F in the input file.

m.hadfield

A couple of extra data points: The problem I reported also occurs in the UPWELLING test case. It's not specific to float simulations (though float simulations are where I tend to use OpenMP). Contrary to what I said earlier, it does occur with Linux/Gfortran, as well as Cygwin/Gfortran and AIX/xlf.

m.hadfield

Attached are output files from two consecutive runs of the FLT_TEST case. In both cases ROMS (latest source) is run under OpenMP with 1x2 tiles, using Gfortran under Cygwin. The first run (rom001.log) proceeds as it should: the model is initialised analytically (NRREC=0) and runs to 1152 time steps ...

m.hadfield

wangzc wrote:Anyone know Why?

Not I. I suggest you build ROMS with the USE_DEBUG variable set to on (in build.bash) and try again. This *might* give you some useful extra information.

m.hadfield

In a 4-thread OpenMP run of the UPWELLING case I'm getting the header line for the time stepping (the one that begins with "STEP") printed four times. The first occurrence is where it should be and the remaining 3 occur at irregular intervals, eg. STEP Day HH:MM:SS KINETIC_ENRG POTEN_ENRG TOTAL_ENRG ...

m.hadfield

Indeed but various compilers behave not the same in different computers. Therefore, I initialize it like this because debuggers will add junk characters if it is not initialized. This is primarily done as a safeguard to insure that file names are not padded with non blank characters. I have very ...

m.hadfield

Sorry to revive an old thread, but it seems relevant. I want a reasonably high resolution (0.5 deg) surface stress dataset with global or Pacific Ocean coverage continuing to the present, more or less. (Aren't we all?) CCMP, as suggested by saulo above, does seem to be a good product, but currently ...

m.hadfield

In ROMS/Modules/mod_ncparam.F at line 487, ANANAME is declared with dimension 39 character (len=256), dimension(39) :: ANANAME but at lines 715-719 only the first 38 elements are initialised: DO i=1,38 DO j=1,LEN(ANANAME(1)) ANANAME(i)(j:j)=' ' END DO END DO This can lead to platform-dependent ...

m.hadfield

A minor remaining issue: looking at this...

https://www.myroms.org/svn/src/trunk/RO ... ultifile.F

...the variable lstr is still undeclared, and this is not picked up by the compiler because there is still no "implicit none".

m.hadfield

With the multifile-checking code introduced a couple of days ago I'm getting compilation errors in check_multifile when the model is built in a configuration that requires the reading of forcing files: check_multifile.f90:68.33: ncname=BRY(ng)%files(Fcount) 1 Warning: Legacy Extension: REAL array ...

m.hadfield

The "n" in "build.bash -j n" has nothing to do with the number of processes in which you intend to run the model. It merely sets the number of processes (threads) that Gnu make can call on when it is compiling and building the model. There are two quite different concepts being referred to here ...

m.hadfield

Just backing up Kate: I have run ROMS+MPI+Gfortran using MPICH2 and OpenMPI at different times, no worries. Using mpif90 is definitely recommended: the mpif90 script ensures that environment variables & libraries are correctly set up for compiling and linking. It doesn't handle run-time environment ...

m.hadfield

Thanks for all your effort on this, Hernan. It's impressive looking stuff, but quite complicated, obviously.

m.hadfield

The dogbone composite-grid case appears to work OK.

m.hadfield

I have updated my code with this fix, but the dogbone-refined case is still not working properly. I have attached a couple of videos (gzipped AVIs) showing zeta in the outer and inner domains. In the outer domain, the positive-zeta wave moves in from the LH boundary and approaches the constriction ...

m.hadfield

Well, I guess the obvious thing to check is whether LD_LIBRARY_PATH is set correctly for ROMS when it is running inside the shell script. I haven't used PBS, but my impression is that batch scripting environments generally don't pass environment variables from the caller unless you specifically tell ...

m.hadfield

I don't know about that, but I do know that when I run the dogbone-refined case, the water level in the refined grid trends steadily downwards in a completely unphysical way. Perhaps these two problems are related.

m.hadfield

I trust you are running with bound checking, as I suggested a few posts ago.

When adding a line of code changes the results, it usually means you are doing something a bit dodgy, and writing or reading outside the bounds of an array is definitely dodgy.

m.hadfield

It looks like a bit of digging into the particle initialisation code might not go amiss!

m.hadfield

How can I check if usr/local/bin subdirectory is on my path? It looks like this directory is on your path, as you can call nc-config and nf-config without including the full name (which is what you want to happen, and the reason why I asked about the path). So, as Kate says, something else is ...

m.hadfield

Hi PedroMarques The configure command you gave several posts back specified an installation prefix of /usr/local. Is that where you installed netCDF and netCDF-Fortran? If so, why is ROMS looking for netcdf.mod under /opt? (I myself don't have a Cygwin /opt directory. All user-installed software ...

m.hadfield

I'm stumped. I suspect configure is failing the first time it tries to link Fortran and C code together. But I don't know why that would happen. You're on Cygwin, right? On Linux you probably wouldn't want "--disable-shared". You could sort out what's happening by trying to reproduce the command ...

m.hadfield

Reduce Lm in the input file to a smaller number. As Kate said, the minimum is probably 3; whatever, 1 is not permitted. Whatever value of Lm you choose, the UPWELLING simulation is homogeneous in the xi direction, so you can do your analyses on any slice

m.hadfield

An rx1 of 20 is too high. Somewhere in the vicinity of 5 to 8 should be OK. In my experience, the failure mode for large rx1 is that a large patch of the bottom layer decides, for no obvious reason, to move rapidly in one direction or other. This produces large velocities at the coastline, which ...

m.hadfield

I believe you can get ROMS to output the surface stresses, which in your case should have been calculated by BULK_FLUXES. Do these look reasonable?

m.hadfield

Hi robjenkinsiii You could begin by reading all the posts in this thread. The linker is failing to find a netCDF Fortran routine, probably because the linker command is lacking a "-lnetcdff", and the most likely reason for that is that you have netcdf-4.x installed but have failed to define USE ...

m.hadfield

I have not and I noticed this in your earlier posts. This might seem like a silly question (I'm very new to all of this), but does this require that I uninstall cygwin and reinstall with the appropriate packages or is there a simpler way to handle this? No you should not need to uninstall ...

m.hadfield

Have you installed HDF5? It's available via the Cygwin setup utility. You need the hdf5 and libhdf5-devel packages, and also maybe libhdf5_7 (but I think that gets installed automatically with the other two.)

m.hadfield

Hi jande023

Did you get this sorted out? I missed it back in November 2012.

Mark

m.hadfield

Hi Leo That's all as I expected. See the string that is returned by "nc-config --flibs" or "nf-config --flibs": -L/usr/local/lib -lnetcdff -lnetcdf -L/usr/local/lib -lnetcdf -lhdf5_hl -lhdf5 -lm -lz -lcurl The ROMS build process needs to incorporate that string in the command used to link the ROMS ...

m.hadfield

I have another potential use for a 4-D forcing field: to represent density of farmed mussels in order to calculate their effect on phytoplankton, nutrients, etc. The mussel density is a 3D (x,y,z) field that can vary in time.

m.hadfield

I think Mark's suggestion about adding the -lnetcdff library might have gotten lost in the noise. You have to do that. You can let nc-config(*) do that. You just need to ensure that nc-config is on the path and that USE_NETCDF4 is defined. Then the following lines (from the current CYGWIN_gfortran ...

m.hadfield

Hello Mark Yes I´m using the netcdf-4.2.1.1 version and the netcdf-fortran-4.2 version. Anyway I guess my mistake is in here, I configured netcdf-4.2.1.1 with this: ./configure --prefix=/usr/local --disable-shared --enable-netcdf-4 --enable-dap --disable-dap-remote-tests and configured netcdf- ...

m.hadfield

Personally, I would lay out the grid for an area like yours on a domain that's rectangular on a suitable (possibly rotated) map projection, with maybe some stretching towards the boundaries. I would do the whole thing in IDL (similar to Matlab) code. To my mind the benefit you're getting from ...

m.hadfield

Yes, Kate is on the right track. The command that triggers the error is /usr/bin/gfortran-4 -frepack-arrays -O3 -ffast-math /home/Usuario/roms/Projects/Upwelling/Build/esmf_roms.o /home/Usuario/roms/Projects/Upwelling/Build/master.o /home/Usuario/roms/Projects/Upwelling/Build/ocean_control.o /home ...

m.hadfield

There are holes (white patches) in the plot of barotropic time step limit. How do they arise?

m.hadfield

Thanks for the follow-up, Leo

m.hadfield

OK. If you look at ROMS/Functionals/ana_grid.h, you'll see a block of code that looks like the following (with a lot of lines in the middle deleted) #if defined BASIN Xsize=3600.0E+03_r8 Esize=2800.0E+03_r8 depth=5000.0_r8 f0=1.0E-04_r8 beta=2.0E-11_r8 ... #elif defined WINDBASIN Xsize=2000.0_r8 ...

m.hadfield

A maximum barotropic Courant number of 0.62 is fine. (I find 0.8 is usually optimal for robustness and performance.) I can't see anything wrong with those boundary conditions, with temperatures in the range 0-27 and salinities in the range 0-36. I suspect the 0 values shouldn't be in there, but I ...

m.hadfield

I have no trouble compiling INLET_TEST with an up-to-date copy of the trunk code.

Did you clear out all the files generated during your first attempt before re-trying?

m.hadfield

A quick Google search on "bad reloc address" indicates that this message is usually preceded by other error messages, eg undefined reference, that give you a better indication of the problem. Please, if you can, reply on the forum attaching the *entire* output of your ROMS build, or at least the ...

m.hadfield

Hello Everyone I´m installing ROMS via CYGWIN. I´ve followed the instructions of the Tutorial for Windows and I´ve got to the Compile ROMS process in order to get the executable oceanS file; when I do so, I get this message: ./build.bash: line 40: $'\r': command not found ./build.bash: line 43: $ ...

m.hadfield

Rebuild with bounds-checking enabled, maybe try a few different compilers, to establish the statement and array reference that triggers the bounds violation. Then use your preferred debugging technique (mine is WRITE statements inserted into the code) to establish why this is happening. This being ...

m.hadfield

There's just been a long discussion on installing netCDF on Cygwin on this thread: https://www.myroms.org/forum/viewtopic.php?f=31&t=2766 I have a couple of comments. The first is that it is impossible to determine how a make check failure occurred without the full output. Perhaps you could attach ...

m.hadfield

Thanks, Leo.

It failed during an OpenDAP test with a "Cannot locate test server" error. However it had already run a series of OpenDAP tests. This is probably just an isolated connectivity problem. It is safe to ignore it and install netCDF.

m.hadfield

Thanks for those warnings, Chris.

Remember that the ROMS makefile uses a variable called NC_CONFIG to find nc-config. The default is "nc-config", but you can override it via an environment variable.

m.hadfield

Incidentally, both Kate and I have said on this thread that if you are installing the netCDF C and Fortran libraries separately (which is how they're packaged nowadays) then the command that the ROMS make file needs to call to determine the location of libraries and include files is nf-config, not ...

m.hadfield

Hi Leo

"make check" worked fine for me once I'd worked around the #elifdef bug.

I can't tell the cause of the error with the output you've shown. Could you please include the entire make check output as an attachment, or email it to me separately.

Mark

m.hadfield

Hi Leo The compilation problem occurs because the file in question, oc/ocinternal.c, contains the preprocessor directive "elifdef", which is not valid. Here is the offending block of code /* Note: TMPPATH must end in '/' */ #ifdef __CYGWIN__ #define TMPPATH1 "/cygdrive/c/temp/" #define TMPPATH2 ...

m.hadfield

Hi Leo

I get the same error with netdf-4.2.1.1. It also occurs with 4.2.1. I suggest you install netcdf-4.2. I will check this problem out further and report it to the netCDF developers.

"nice" is a utility that runs executables at a reduced priority.

Mark

m.hadfield

When I run the UPWELLING case with ECOSIM defined and with bounds-checking enabled, I get the following error 0: Subscript out of range for array idttav (mod_ncparam.f90: 1938) subscript=3, lower bound=1, upper bound=2, dimension=1 The error arises in this section of code: SELECT CASE (Vinfo(1 ...

m.hadfield

Hi lalvarez Please don't hesitate to ask for further help if you need it. A few messages back you referred to an error you got from the configure command for the netCDF-Fortran package (because you misunderstood a comment by Kate and substituted nf-config for nc-config). But you should have only ...

m.hadfield

...Anyway "$(nc-config --libs)" is not the correct syntax. It is saying that whole expression is a variable of some kind due to the "$". The configure command that I suggested, namely CC=gcc FC=gfortran LIBS="$(nc-config --libs)" CPPFLAGS="-I$(nc-config --includedir)" nice ./configure --disable ...

m.hadfield

Thanks.

m.hadfield

Hi Kate

Can you comment on the relationship between the Github link you gave

https://github.com/TRACMASS/

and this one

http://tracmass.org/

which has an SVN repository

http://svn.tracmass.org/trunk

Mark

m.hadfield

What Kate says. But what may not be immediately clear from my last post is that the files for the C and Fortran libraries are unpacked into different directories, because the tar archives contain subdirectories that match the archive names (without the .tar.gz bit). This is why the 4th command in ...

m.hadfield

OK, you install the C library then the Fortran library. The C library optionally depends on several other prerequisite packages. If you want to be able to produce files in netCDF-4 format then you need HDF5. If you want to use OpenDAP then you need Curl. Also, in recent versions of netCDF, building ...

m.hadfield

I suggest you download the netCDF source code from the Unidata FTP site with your Web browser. You will need the C and Fortran libraries: ftp://ftp.unidata.ucar.edu/pub/netcdf/netcdf-4.2.1.1.tar.gz ftp://ftp.unidata.ucar.edu/pub/netcdf/netcdf-fortran-4.2.tar.gz If you want any help regarding ...

m.hadfield

Hi ming

I have replied on the other thread relating to this issue.

Mark

m.hadfield

With my up-to-date installation of Cygwin, The make version is 3.82.90. The version-testing code in the make file is not very clever: it doesn't find "3.82.90" in the list of acceptable make versions, so it aborts. In fact, this version has no problems in building ROMS. Suggested solutions: either ...

m.hadfield

My advice is to follow Kate's first suggestion: run your model in segments; use a shell or Python script to generate the input files and scripts required.

m.hadfield

Um, I think you're right! And setting Pwave_bot to 8s suppresses the oscillation.

Thank you.

m.hadfield

You only need Pwave_bot if SEDIMENT is enabled. The original case where I saw this behaviour did have sediment, but I stripped it out in order to simplify things.

m.hadfield

Also attached FYI is the input file (which includes the GLS model coefficients)

m.hadfield

Hi s091334 I think your problem is with the -fno-second-underscore option. In the past I have recommended using this for building ROMS. (It controls symbol name-mangling. Try a Google search on "g95 -fno-second-underscore Hadfield" for some history.) However this recommendation comes with the ...

m.hadfield

I was inspired by your question to take another look at the status of MPI on my system (PC with dual-core Intel, Windows 7, Cygwin). I have MPICH2 1.4.1p1 installed. I did this by building from source. The commands were CC=gcc FC=gfortran ./configure --prefix=/usr/local nice make nice make install I ...

m.hadfield

This is more a MPICH2 on Windows problem than a ROMS problem. There might be an MPICH2 forum somewhere wher people could help you. That said, I have run MPI on Cygwin successfully in the past (though for the past year or two it seems to have been broken) so I can offer a few pieces of advice. The ...

m.hadfield

Hello colleagues Preface: I don't know how big a problem the following really is in practice, but I've noticed some curious & spurious behaviour, so I thought I'd let the community know about it. Short version: A toy case with oscillating tidal flow shows pronounced, rapid oscillations in bottom ...

m.hadfield

A couple more comments: If the initial dye field is simple, you can set it with ana_passive.h. Just #define ANA_PASSIVE in the header file, set DSTART in the input file to the time at which you want the dye introduced and set NRREC to 0. You will not get a perfect restart in this case: the model ...

m.hadfield

Hi again Hernan At the moment, values of 1.E37 are written to lon_rho/x_rho, lat_rho/y_rho, h and zeta for points in the land mask. Are you saying that this should not happen? I am inclined to agree. However if it is happening, then it is appropriate to mark them as fill values via that variables ...

m.hadfield

OK, it's a bit of a stretch to call this a bug report, more a suggestion for a minor improvement. When a station is inside the land mask (or just outside it in the case of velocity variables) values of 1.0E37 get written to the station file. The netCDF variables in which these values are written ...

m.hadfield

Hi Hernan The problem with ana_psource isn't a big issue for me and I know you are busy on other things, but I just can't let these things go unsolved! I think the issue is with the BARRIER directives inside ana_psource.h. According to IBM, the rules are Rules A BARRIER directive binds to the ...

m.hadfield

Yes, that works: it runs to completion with AIX/xlf+OpenMP, and gives identical results to a serial run. Thank you!

m.hadfield

I have done a little more investigation and now think that The use of the OpenMP SINGLE directive in ROMS is incorrect. The xlf compiler on AIX is correct in recognising this. The PGI and Gfortran compilers are incorrect in failing to do so. Consider the following OpenMP test program: module openmp ...

m.hadfield

So far, my conclusions for the floats vertical-walk code are the same as the one drews reported for the river source: There are no obvious violations of the OpenMP rules The simulation runs and appears to give correct results if you remove the PARALLEL directives in main3d Perhaps it's an AIX ...

m.hadfield

Thanks, Hernan. I'll look into it.

m.hadfield

Incidentally, I also get the same error message when I run the RIVERPLUME1 test case in OpenMP mode on the IBM. (I checked this out because the problem reported in the previous thread was related to ana_psource.) As before, the simulation runs to completion with Linux/gfortran/OpenMP.

m.hadfield

Did you know that a search for "1587-114 Incorrect nesting of OpenMP directives" on Google returns only one match, namely the following thread on the ROMS Forum https://www.myroms.org/forum/viewtopic.php?t=1714 Shortly there should be a second one! I have run into the same error message, this time ...

m.hadfield

You have added the -m64 flag to C and Fortran compiler options when building netCDF, but I see no sign that it has been used when building ROMS. That might be your problem.

m.hadfield

Hi Liz You say you have netCDF libraries here? NETCDF_INCDIR ?= /opt/intelsoft/netcdf4/include NETCDF_LIBDIR ?= /opt/intelsoft/netcdf4/lib Then nc-config should be in /opt/intelsoft/netcdf4/bin. To help the ROMS make system find it, you can put that directory on the path or specify the following in ...

m.hadfield

Yes. It can be adjointed. But we have several modules in ROMS/Nonlinear that are used in the ROMS/Utility directory. This includes the bc2d.F , bc3d.F , exchange_2d.F , exchage_3d.F , scale_omega , and so on. I don't understand why uv_rotate.F has to be different. This cross-directory references ...

m.hadfield

+1 to changing your own makefile. Probably better than moving source files.

It's getting off topic, I know, but I have had good results with using Mercurial to manage my copies of ROMS. However I haven't made a video about it.

m.hadfield

Hi simion123... I've just noticed this thread, which deals with the same issue https://www.myroms.org/forum/viewtopic.php?f=31&t=2530 There Hernan answers the questions I posted in my previous message on the current thread. ...we have several modules in ROMS/Nonlinear that are used in the ROMS ...

m.hadfield

I just ran into the problem reported by Entrapmen (with the Gfortran linker). I fixed it by moving uv_rotate.F from ROMS/Nonlinear to ROMS/Utility. Then I read this thread :roll: Is there any reason why uv_rotate.F should be in ROMS/Nonlinear and not in ROMS/Utility? Could the other circular ...

m.hadfield

With the recent addition of rotated velocity I/O, a couple of minor declaration bugs were introduced into wrt_his.F when SOLVE3D is undefined: The following symbols are undeclared:

The integer tile
The subroutine uv_rotate2d

A corrected version of the file is attached.

m.hadfield

A potentially attractive alternative would be to specify "dt" and fast-time step "dtfast", and let the model figure out the mode-splitting ratio within the code (along with a slight adjustment to "dtfast", so ndtfast*dtfast=dt made by the model via a simple algorithm described above). +1 to that. I ...

m.hadfield

shchepet wrote:Counters should always be of an integer type.

Quite right. The quick fix should do for now. If it doesn't (I have some runs with a much larger number of time steps) I'll send Hernan some code for all the different drivers.

m.hadfield

I have a simulation with the non-linear, forward model driven by the M2 tide only. Accordingly I have made the simulation length a multiple of the M2 period, which is 44714.1 s. The simulation ends by writing a record to the restart file. Parameters are: NTIMES = 23040 DT == 7.7628819D0 NDTFAST == 1 ...

m.hadfield

Make sure you're looking at line 1187 of read_phypar.f90 (the preprocessed file) not read_phypar.F.

m.hadfield

In the absence of any direction to the contrary, netCDF will have been built for the Gfortran compiler. To tell ROMS to use the right configuration, set the following environment variables (Bash style) export FORT=gfortran export USE_NETCDF4=on export NC_CONFIG=/usr/local/bin/nc-config

m.hadfield

gfortran

m.hadfield

Mark, I had the same problem until I restricted the inflow to the top bin or few bins (I forget just now). I would be interested to know the outcome if you were to try this (top bin only). But what if inflow is at or near the bottom, like the brine from a desal plant. Has someone sorted this out? I ...

m.hadfield

So what is wrong with Q_PSOURCE ? This is not in my list of ROMS problems. This type of source is complicated because involves the continuity equations of ROMS. I am always uneasy messing around with the 2D and 3D continuity equations. It may explains why this is broken. This is very delicate stuff ...

m.hadfield

The reason I use both PERFECT_RESTART and RST_SINGLE is because I am trying to emulate a very large, real-life ROMS application. I that application, we have to continue the ROMS simulations by breaking them in to segments (as we only get 4-hours of time on the supercomputer at a time) and the ...

m.hadfield

Mark, All of those scenarios would best be represented - in ROMS - as a rho-point-centered source of tracer. Not as a horizontal flux at the vertical face of a grid cell in the middle of the water column. Several of those scenarios could very sensibly be represented as surface or bottom fluxes of ...

m.hadfield

But even this seems ill conceived - what sort of physical source to you envision acts this way in the middle of the water column? Usually sources like this are intended to represent the footprint on the model-resolved scale of some subgrid process. A wastewater pipe? A cloud of sediment stirred up ...

m.hadfield

Perhaps the word "wish" isn't quite what we need here! :twisted: Here's a wish, in the sense of something I have thought about and may well implement but would appreciate feedback on: an interior tracer point-source option (M_SOURCE?), each source to be defined in terms of its position, vertical ...

m.hadfield

When I run the floats with my modified test problem without the PERFECT_RESTART and RST_SINGLE CPP options, the serial and parallel (MPI) outputs are the same. However, when I include these two CPP options, the parallel ROMS outputs has zero X, Y values upon carrying out a restart and the problems ...

Search found 524 matches