neg. dp (m) problem near the southern open boundary

[Jeongin Kim]

Hi, I made a new domain for north atlantic ocean, and I'm getting neg. dp (m) in about 5-7 days of run when I turn on the nesting option. 

Without the nesting option, it runs pretty well for a few months (I haven't runned a longer period), but whenever I turn on the nesting option, the outcome gives neg. dp(m) near the southern open boundary. 

It's really close to the western edge of the southern open boundary, near the relaxation zone.

Should I modify e-folding dates (Currently 0.1-10 days within 10 grid points, similar to the GOMb0.08 example), or baclin and batrop, etc.?

I've tried several methods from the solution in previous threads, but those didn't work.

And it is okay when I run without nesting, so I guess the problem is related to nesting.

Could you please share your ideas?

My running bathymetry: Grid size 0.08, from GEBCO 2025 (From the GEBCO_2025I2021.nc file you've uploaded recently)
My nesting bathymetry: Grid size 0.08, from GLBb0.08 (09m11, generated similarly with GOM example)

I attached a powerpoint slides for plots of my topologies, nest_rmu_invday.a, ports, etc., for your information.

Many thanks,

Jeongin

[Alan Wallcraft]

I see that for GOMb0.08 we used 0.1 to 9 day e-folding time even for monthly nesting files, but this is more appropriate for daily nesting files.

When relaxing to monthly climatology we would likely extend the rmu region to at least 20 grid points with an e-folding time of 5-50 days (say).

So I suggest trying 5-50 day e-folding time over 10 grid points for monthly nesting.

Alan.

[Jeongin Kim]

I appreciate your answer!

Okay, I'll revise the e-folding days.

And just to confirm, what is the difference between nest_rmu.a and relax_rmu.a?

Becaise I found that nest linear rmu.csh file generates a nest_rmu.a file, but not the relax_rmu.a. 

Should I copy and rename nest_rmu.a to use it as a relax_rmu.a?

Many thanks,

Jeongin

[Alan Wallcraft]

nest_rmu is for relaxing to nesting archives, relax_rmu is for relaxing to monthly climatology.

The latter is controlled by relax:

   0      'relax ' = activate lateral boundary nudging    (0=F,1=T)

Typically you would not use nesting and relax at the same time.  

Note that relax is only for T,S,p and nudging is usually for T,S,p,u,v

Alan.

[Jeongin Kim]

Hi Alan, 

I revised the e-folding days to 5-50 days and generated new relax files from the new topography files.

Now it doesn't blow up for a few days, so I really want to appreciate you for this huge progress!

However, it still shows neg. dp (m) on early March(63 days of run).

Should I just landmask the southern boundary? 

i don't think this would be the problem of batrop and baclin, because baclin for my case is relatively smaller than other examples.

baclin = 240, batrop = 12 (cfl minimum is around 18)

For your information, I've attached my topo maps and csh codes for generating merged topography and nest_rmu files.

I ran topo_landfill after generating depth_..._30.[ab] file to generate depth_..._31.[ab] ., then merged with 09m11.[ab] (From GLBb0.08) for nesting.

Now I suspect I made a mistake in merging topography data, but not pretty sure.

Let me know if you have any other files or info to determine where the error happens.

Sincerely,

Jeongin

2026년 1월 23일 금요일 AM 11시 16분 7초 UTC-5에 Alan Wallcraft님이 작성:

[Jeongin Kim]

I saw some of the near-global case, people modified dp00 and solved the problem.

Would that apply to my case as well?

Best,

Jeongin

2026년 1월 23일 금요일 AM 11시 16분 7초 UTC-5에 Alan Wallcraft님이 작성:

nest_rmu is for relaxing to nesting archives, relax_rmu is for relaxing to monthly climatology.

[Jeongin Kim]

Hi, 

I decreased the baclin to 180 (originally 240) and the batrop to 10 (originally 12), and I could run it for 169 days (Early June).

However, now neg. dp (m) appears at the Northern Open Boundary!

This is the zoomed map where neg. dp (m) occurs.

Although this is the quiver plot of the surface velocity (k=1), it looks like there are some weird velocity swirls in here.

Should I landmask this area?

I'm also curious if landmasking the problematic area will resolve all neg. dp (m) errors, or I am missing some important physical parameters, like baroclinic time step, etc.

If you could give an idea or a related forum thread, it would be really helpful for me.

Thanks,

Jeongin

2026년 1월 26일 월요일 PM 4시 3분 35초 UTC-5에 Jeongin Kim님이 작성:

[Alan Wallcraft]

What lat,lon is creating the neg. dp's?  In a directory containing regional.grid.[ab], issue the command:

hycom_ij2lonlat 562 951

Check the bathymetry by plotting or printing the difference between the model bathymetry and the nesting bathmetry.  The difference should be zero where rmu is non-zero, and small close to that region.

Another possibility is that the rmu (1/s) is weak at the open boundary and strong at the sea-ward edge of the relaxation zone.  I normally produce and plot the e-folding time (1/rmu) since it is easier to interpret (the e-folding time should be shortest at the open boundary).

Use hycom_print:

narwhal10 91> hycom_print
 Usage: hycom_print file.a idm jdm k cfmt [if il [id] jf jl [jd]]

C                 cfmt     - output format, e.g. "(f6.2)" or "(*)" or
C                            "(2i5,...)" to print i j value (skipping land)
C                         or "(2i6,...)" to print i j value (skipping land)

hycom_print rmu.a 1476 954 1 '(2I5,e20.6)' 562 562 921 954

You can use an f format instead of e if you know the likely range of the field.

If you want to try modifying the bathymetry, I suggest instead making the relaxation zone larger (20 grid points) which implies re-merging the bathymetries and regenerating the relax climatology files.  The nesting files should be using the outer model bathymetry already and so are ok.  You can also use hycom_sea_ok to check the other forcing files to make sure the coastline is consistent with them.  Over 20 grid points, I would hold the e-folding time at 5 days for 5 grid points.

Alan.

[Jeongin Kim]

Lon Lat of i = 562, j = 951 is 53.120W  54.878N.

And I found the answer to the error.

I actually ran landfill.csh to fill single-width inlets and 1-point seas AFTER THE MERGE.CSH!!!

That's why I got different coastlines in the nesting bathymetry and the running bathymetry... I was so stupid :(

I really really appreciate you for pointing that out.

The direction of nest_rmu_invday.[ab] is fine. I've double-checked that.

But still, can you please explain more in detail about the last paragraph? 

I set the relaxation zone of the Northern and Southern edges as 20 grid points, but for the Eastern edge, I made a large relaxation zone because I thought that the bathymetry there is quite complex.  

e-folding days are 5-50 days. (10 grids: 5-15 days, 10 grids: 15-50 days)

For your information, the script below is how I set the e-folding days at the Northern edge, near the problematic zone.

 490      'if    ' = first i point of sub-region (<=0 to end)
1140      'il    ' = last  i point of sub-region
 944      'jf    ' = first j point of sub-region
 953      'jl    ' = last  j point of sub-region
  15.0    'efoldA' = bottom left  e-folding time in days
  15.0    'efoldB' = bottom right e-folding time in days
   5.0    'efoldC' = top    right e-folding time in days
   5.0    'efoldD' = top    left  e-folding time in days
 490      'if    ' = first i point of sub-region (<=0 to end)
1140      'il    ' = last  i point of sub-region
 934      'jf    ' = first j point of sub-region
 944      'jl    ' = last  j point of sub-region
  50.0    'efoldA' = bottom left  e-folding time in days
  50.0    'efoldB' = bottom right e-folding time in days
  15.0    'efoldC' = top    right e-folding time in days
  15.0    'efoldD' = top    left  e-folding time in days

I want to learn more about this sentence, "Over 20 grid points, I would hold the e-folding time at 5 days for 5 grid points."

If you have time, could you please review my nest_rmu_linear.csh file?

Meanwhile, I'll revert my topography file before the landfill.csh.

This is the map.log of nesting bathymetry (09m11)

and the map.log of running bathymetry (31)

This is the nest_rmu_linear.log

2026년 1월 28일 수요일 AM 7시 7분 1초 UTC-5에 Alan Wallcraft님이 작성:

[Alan Wallcraft]

  You have:

2      'jf    ' = first j point of sub-region

  10      'jl    ' = last  j point of sub-region

   5.0    'efoldA' = bottom left  e-folding time in days

   5.0    'efoldB' = bottom right e-folding time in days
  15.0    creasinglinearly in 'efoldC' = top    right e-folding time in days

  15.0    'efoldD' = top    left  e-folding time in days

I suggest holding the e-folding time at 5 days from 2 to 5 and from 5 to 15 days from 5 to 10.

For the Med. Sea, I suggest only including the Alboran  Sea, setting the rest to land.

I would compile HYCOM using:

setenv OCN_MISC "-DMASSLESS_1MM -DRDNEST_MASK -DLATBDT_NPLINE3 -DMOMTUM_CFL"

The CFL option may  reduce the number of negative dp's.   The NEST and LINES options are for open boundaries.

Alan.

[Jeongin Kim]

I realized that I was interpolating the nesting file to the running bathymetry, not to depth_ATLb0.08_09m11s.a.
In most cases, we interpolate forcing or relaxing files to the running bathymetry, so I changed VU in nest.csh to be 31, 43, 50, 60, or whatever version of the bathymetry I've generated.
So I turned it back to the 09m11s, and now it works. I could run the model for 2 years.
I appreciate your help.

I'm writing down this for anyone having similar issues like me.

Thanks,

Jeongin

2026년 1월 29일 목요일 PM 2시 39분 48초 UTC-5에 Alan Wallcraft님이 작성: