Generating model bathymetry from alternative datasets

edoddridge · 12 January 2026 05:16

Hi everyone,

Summary: I’m trying to apply the bathymetry pipeline to new datasets and not succeeding.
If that interests you, or you think you might be able to help, please read on!

Gory details

I’m trying to build on the topography workflow described here to create an ensemble of PanAntarctic runs using different bathymetries.

I’ve modified the gen_topo.sh script to take a command line argument for the bathymetry dataset, and verified that it works for me when I use GEBCO as the input. Unfortunately, I can’t make it work with other datasets yet.

From the script gen_topo.f90 in bathymetry-tools, I can see that the data variable needs to be called ‘height’ or ‘elevation’, so I’ve modified the bathymetry files accordingly.

When I call gen_topo.sh, it falls over with an error from gen_topo.f90. This is my gen_topo.sh.e* file

Loading conda/analysis3-25.11
  Loading requirement: singularity
+ set -e
+ cp -L --preserve=timestamps /g/data/vk83/prerelease/configurations/inputs/access-om3/mom/grids/mosaic/global.025deg/2025.01.30/ocean_hgrid.nc ./ocean_hgrid.nc
+ cp -L --preserve=timestamps /g/data/vk83/prerelease/configurations/inputs/access-om3/mom/grids/vertical/global.025deg/2025.01.30/ocean_vgrid.nc ./ocean_vgrid.nc
+ ln -sf /g/data/jk72/ed7737/datasets/bathymetry/ML_topos/DTU_DKL_6_Ker50_pre_processed.nc ./BATHY.nc
+ ./bathymetry-tools/bin/float_vgrid --vgrid ocean_vgrid.nc --vgrid_type mom6
+ ./bathymetry-tools/bin/topogtools gen_topo -i BATHY.nc -o topog_new.nc --hgrid ocean_hgrid.nc --tripolar --longitude-offset -100
FATAL ERROR: NetCDF: Index exceeds dimension bound

The error comes from line 364 of gen_topo.f90, where it is trying to get data from the supplied bathymetry file.

The output file is

-- FindNetCDF defines targets:
--   - NetCDF_VERSION [4.7.3]
--   - NetCDF_PARALLEL [FALSE]
--   - NetCDF_C_CONFIG_EXECUTABLE [/apps/netcdf/4.7.3/bin/nc-config]
--   - NetCDF::NetCDF_C [SHARED] [Root: /apps/netcdf/4.7.3] Lib: /apps/netcdf/4.7.3/lib/libnetcdf.so
--   - NetCDF_Fortran_CONFIG_EXECUTABLE [/apps/netcdf/4.7.3/bin/nf-config]
--   - NetCDF::NetCDF_Fortran [SHARED] [Root: /apps/netcdf/4.7.3/include] Lib: /apps/netcdf/4.7.3/lib/Intel/libnetcdff.so
-- Configuring done (0.2s)
-- Generating done (0.2s)
-- Build files have been written to: /home/552/ed7737/coding-and-projects/panant_bathy_ensemble/make_OM3_panant_topo/bathymetry-tools/build
[  8%] Building Fortran object CMakeFiles/M_CLI2.dir/external/M_CLI2/src/M_CLI2.F90.o
[ 16%] Linking Fortran static library libM_CLI2.a
[ 16%] Built target M_CLI2
[ 25%] Building Fortran object CMakeFiles/topography.dir/src/kdtree2.f90.o
[ 33%] Building Fortran object CMakeFiles/topography.dir/src/utils.f90.o
[ 41%] Building Fortran object CMakeFiles/topography.dir/src/gen_topo.f90.o
[ 50%] Building Fortran object CMakeFiles/topography.dir/src/vgrid.f90.o
[ 58%] Building Fortran object CMakeFiles/topography.dir/src/topography.f90.o
[ 66%] Linking Fortran static library libtopography.a
[ 66%] Built target topography
[ 75%] Building Fortran object CMakeFiles/topogtools.dir/src/topogtools.f90.o
[ 83%] Linking Fortran executable topogtools
[ 83%] Built target topogtools
[ 91%] Building Fortran object CMakeFiles/float_vgrid.dir/src/float_vgrid.f90.o
[100%] Linking Fortran executable float_vgrid
[100%] Built target float_vgrid
-- Install configuration: "Release"
-- Installing: /home/552/ed7737/coding-and-projects/panant_bathy_ensemble/make_OM3_panant_topo/bathymetry-tools/./bin/topogtools
-- Installing: /home/552/ed7737/coding-and-projects/panant_bathy_ensemble/make_OM3_panant_topo/bathymetry-tools/./bin/float_vgrid
Reading vgrid from file 'ocean_vgrid.nc'
Writing vgrid to file 'ocean_vgrid.nc'
Reading supergrid info
  j_lat located at          903 latitude    65.0000000000000
 Mosaic grid starts at   -80.9675013444044       topography jstart =
           1  lat =   -79.9916666666667
 Mosaic grid starts at   -80.9675013444044       topography jstart =
         -58  lat =   0.000000000000000E+000
 Shifted topography longitude grid by   -100.0000     (ishift =        -6000 )
 jmosaic =           1  -80.9675013444044       -74.4608728325249
         -58         332
 imosaic =           1  -280.000000000000       -255.000000000000
           1        1500
 ncid =       131072 hid =            3 itopo =
 listart =        15601 jstart =          -58
 icount =         1500 jcount =          391

The bathymetry files I have tried span ~80S to ~80N, while GEBCO is pole-to-pole, so I wondered if the Index exceeds dimension bound error was due to the negative value for jstart (topography jstart =-58) being used in the nf90_get_var call.

So, I tried expanding the bathymetry arrays by padding them with zeros. Unfortunately, this made gen_topo.f90 error out earlier.

my error file:

Loading conda/analysis3-25.11
  Loading requirement: singularity
+ set -e
+ cp -L --preserve=timestamps /g/data/vk83/prerelease/configurations/inputs/access-om3/mom/grids/mosaic/global.025deg/2025.01.30/ocean_hgrid.nc ./ocean_hgrid.nc
+ cp -L --preserve=timestamps /g/data/vk83/prerelease/configurations/inputs/access-om3/mom/grids/vertical/global.025deg/2025.01.30/ocean_vgrid.nc ./ocean_vgrid.nc
+ ln -sf /g/data/jk72/ed7737/datasets/bathymetry/ML_topos/DTU_DKL_6_Ker50_pre_processed.nc ./BATHY.nc
+ ./bathymetry-tools/bin/float_vgrid --vgrid ocean_vgrid.nc --vgrid_type mom6
+ ./bathymetry-tools/bin/topogtools gen_topo -i BATHY.nc -o topog_new.nc --hgrid ocean_hgrid.nc --tripolar --longitude-offset -100
FATAL ERROR: NetCDF: Start+count exceeds dimension bound

This time the error comes from line 221 of gen_topo.f90, where it is trying to put the longitude variable in to the output file.

And the output file, gen_topo.sh.o*

-- Found NetCDF: /apps/netcdf/4.7.3/include;/apps/netcdf/4.7.3/include/Intel (found version "4.7.3") found components: Fortran
-- FindNetCDF defines targets:
--   - NetCDF_VERSION [4.7.3]
--   - NetCDF_PARALLEL [FALSE]
--   - NetCDF_C_CONFIG_EXECUTABLE [/apps/netcdf/4.7.3/bin/nc-config]
--   - NetCDF::NetCDF_C [SHARED] [Root: /apps/netcdf/4.7.3] Lib: /apps/netcdf/4.7.3/lib/libnetcdf.so
--   - NetCDF_Fortran_CONFIG_EXECUTABLE [/apps/netcdf/4.7.3/bin/nf-config]
--   - NetCDF::NetCDF_Fortran [SHARED] [Root: /apps/netcdf/4.7.3/include] Lib: /apps/netcdf/4.7.3/lib/Intel/libnetcdff.so
-- Configuring done (0.2s)
-- Generating done (0.2s)
-- Build files have been written to: /home/552/ed7737/coding-and-projects/panant_bathy_ensemble/make_OM3_panant_topo/bathymetry-tools/build
[  8%] Building Fortran object CMakeFiles/M_CLI2.dir/external/M_CLI2/src/M_CLI2.F90.o
[ 16%] Linking Fortran static library libM_CLI2.a
[ 16%] Built target M_CLI2
[ 25%] Building Fortran object CMakeFiles/topography.dir/src/kdtree2.f90.o
[ 33%] Building Fortran object CMakeFiles/topography.dir/src/utils.f90.o
[ 41%] Building Fortran object CMakeFiles/topography.dir/src/gen_topo.f90.o
[ 50%] Building Fortran object CMakeFiles/topography.dir/src/vgrid.f90.o
[ 58%] Building Fortran object CMakeFiles/topography.dir/src/topography.f90.o
[ 66%] Linking Fortran static library libtopography.a
[ 66%] Built target topography
[ 75%] Building Fortran object CMakeFiles/topogtools.dir/src/topogtools.f90.o
[ 83%] Linking Fortran executable topogtools
[ 83%] Built target topogtools
[ 91%] Building Fortran object CMakeFiles/float_vgrid.dir/src/float_vgrid.f90.o
[100%] Linking Fortran executable float_vgrid
[100%] Built target float_vgrid
-- Install configuration: "Release"
-- Installing: /home/552/ed7737/coding-and-projects/panant_bathy_ensemble/make_OM3_panant_topo/bathymetry-tools/./bin/topogtools
-- Installing: /home/552/ed7737/coding-and-projects/panant_bathy_ensemble/make_OM3_panant_topo/bathymetry-tools/./bin/float_vgrid
Reading vgrid from file 'ocean_vgrid.nc'
Writing vgrid to file 'ocean_vgrid.nc'
Reading supergrid info
  j_lat located at          903 latitude    65.0000000000000
 Mosaic grid starts at   -80.9675013444044       topography jstart =
         543  lat =   -80.9583333333324
 Mosaic grid starts at   -80.9675013444044       topography jstart =
         542  lat =   -80.9749999999991
 Shifted topography longitude grid by   -100.0000     (ishift =        -6000 )

I’m not sure what to look at next. I have a vague worry that perhaps the ordering of dimensions is getting mucked up, but I don’t have anything to back that up at the moment.

Help greatly appreciated!

anton · 12 January 2026 21:45

Yeah - this seems likely, it looks like there is hidden assumption that the topography source file is larger than the model grid:

github.com/COSIMA/bathymetry-tools

src/gen_topo.f90

a29d07a28


      
          do j = 1, ylen
            jstart = jstart + 1
            if (ytopo(jstart) >= y_c(1, 1)) exit
          end do
          
          write(output_unit,*) 'Mosaic grid starts at ', y_c(1, 1), ' topography jstart = ', jstart,' lat = ', ytopo(jstart)
          
          yt_delta = (ytopo(ylen) - ytopo(1))/(ylen - 1)
          xt_delta = (xtopo(xlen) - xtopo(1))/(xlen - 1)
          
          jstart = nint((y_c(1, 1) - ytopo(1))/yt_delta) + 1
          write(output_unit,*) 'Mosaic grid starts at ', y_c(1, 1), ' topography jstart = ', jstart,' lat = ', ytopo(jstart)
          
          ! Shift topography grid if requested
          ishift = nint(offset/xt_delta)
          if (ishift /= 0) then
            xtopo = xtopo + offset
            write(output_unit,*) "Shifted topography longitude grid by ", offset, "(ishift = ", ishift, ")"
          end if
          
          !

This doesn’t make sense, this line shouldn’t by impacted by changing the topo file. It’s just writing the model grid coordinates to the new output file.

github.com/COSIMA/bathymetry-tools

src/gen_topo.f90

a29d07a28


      
          call handle_error(nf90_def_var(ncid_topo, 'geolon_t', nf90_float, dids, geolon_id, deflate_level=1, shuffle=.true.))
          call handle_error(nf90_put_att(ncid_topo, geolon_id, 'long_name', 'tracer longitude'))
          call handle_error(nf90_put_att(ncid_topo, geolon_id, 'units', 'degrees_E'))
          call handle_error(nf90_put_var(ncid_topo, geolon_id, x_t))

I think line 218 might have a bug - it should use dids_topo not dids , although how it ever works is a fluke/mystery!

edoddridge · 12 January 2026 22:53

Nice catch! If I change dids to dids_topo in both L218 and L224

github.com/COSIMA/bathymetry-tools

src/gen_topo.f90

a29d07a28


      
          call handle_error(nf90_def_var(ncid_topo, 'geolon_t', nf90_float, dids, geolon_id, deflate_level=1, shuffle=.true.))
          call handle_error(nf90_put_att(ncid_topo, geolon_id, 'long_name', 'tracer longitude'))
          call handle_error(nf90_put_att(ncid_topo, geolon_id, 'units', 'degrees_E'))
          call handle_error(nf90_put_var(ncid_topo, geolon_id, x_t))
          
          call handle_error(nf90_def_var(ncid_topo, 'geolat_t', nf90_float, dids, geolat_id, deflate_level=1, shuffle=.true.))
          call handle_error(nf90_put_att(ncid_topo, geolat_id, 'long_name', 'tracer latitude'))

the script makes it through that section. It then manages to execute the imosaic loop a couple of times and before throwing an error here

github.com/COSIMA/bathymetry-tools

src/gen_topo.f90

a29d07a28


      
          if (.not. (1 <= listart .and. listart < xlen .and. 1 < liend .and. liend <= xlen) .or. &
            (listart < liend .and. liend - listart + 1 /= icount)) then
            write(error_unit,'(a)') "ERROR: some of the required points for the interpolation are not available in the topography file."
            error stop
          end if

Last part of output file:

...
Reading supergrid info
  j_lat located at          903 latitude    65.0000000000000
 Mosaic grid starts at   -80.9675013444044       topography jstart =
         543  lat =   -80.9583333333324
 Mosaic grid starts at   -80.9675013444044       topography jstart =
         542  lat =   -80.9749999999991
 Shifted topography longitude grid by   -100.0000     (ishift =        -6000 )
 jmosaic =           1  -80.9675013444044       -74.4608728325249
         543         932
 imosaic =           1  -280.000000000000       -255.000000000000
           1        1500
 ncid =       131072 hid =            1 itopo =
 listart =        15601 jstart =          543
 icount =         1500 jcount =          390
         100
 imosaic =         101  -255.000000000000       -230.000000000000
        1501        3000
 ncid =       131072 hid =            1 itopo =
 listart =        17101 jstart =          543
 icount =         1500 jcount =          390
         100
 imosaic =         201  -230.000000000000       -205.000000000000
        3001        4500
 ncid =       131072 hid =            1 itopo =
 listart =        18601 jstart =          543
 icount =         1500 jcount =          390
         100
 imosaic =         301  -205.000000000000       -180.000000000000
        4501        6000

The error happening with longitude of -180 makes me think it is possibly a dateline issue (but I don’t understand the purpose of the longitude shift, so might well be missing something).

anton · 12 January 2026 23:27

There seems to be a column missing in the source data?

Where is 179.991667°E ?

edoddridge · 13 January 2026 00:37

That was the problem.

With an extra column and the dids_topo bug fix, the script is running all the way through without errors. I’ll submit a PR for the bathymetry-tools repo.

Thanks @anton!

anton · 13 January 2026 00:44

No worries - a couple of thoughts:

It’s worth checking that:

the data at -180 looks ok the final output? I am not sure if that code knows to wrap from 180 to -180.
that netcdf _FillValue value (xarray tends to show these as NaN) is used for land in the output (as its just 0 in the source data)

Topic		Replies	Views
Bathymetry for ocean model at any resolution COSIMA inputs	12	689	8 June 2023
Adapting GFDL CM2.1 coupled model for deep time paleoclimate experiments Earth System experiments , user-story	0	332	1 June 2023
ACCESS-coupled N48 issue #1: change topography, bathymetry and land-mask Earth System	3	389	21 March 2023
Generating the exchange grid for MOM6 Ocean	3	277	3 February 2023
ACCESS-coupled N48 for deep paleo Earth System palaeogeography	44	919	25 August 2023

Generating model bathymetry from alternative datasets

Related topics