surface velocities and surface transports that are significantly larger than observed values

[hao ming]

Dear Alan,

I hope this email finds you well. I'm writing to seek your advice on some issues I'm encountering with my HYCOM model configuration.

Main Issues: The model is producing surface velocities and surface transports that are significantly larger than observed values. Strangely, I'm implementing nesting, the model appears to reach stability almost immediately at the beginning of the run. Is this normal? When I use relax relaxation, it typically takes 5-10 years to stabilize.And I must enable both relax relaxation and nesting relaxation simultaneously to maintain stable runs; otherwise, I get errors with dp < -10.

Current Configuration:

• Daily nesting with a frequency of once per day

I suspect the open boundary conditions might be causing these issues. My current boundary settings are:

Nest zones :

IF = 499, 497, 495, 493, 491, 132, 132, 132, 132, 132, 301, 301, 301, 301, 301, IL = 500, 498, 496, 494, 492, 500, 500, 500, 500, 500, 500, 500, 500, 500, 500, JF = 189, 189, 189, 189, 189, 1, 3, 5, 7, 9, 499, 497, 495, 493, 491, JL = 500, 500, 500, 500, 500, 2, 4, 6, 8, 10, 500, 498, 496, 494, 492, EFOLD = 1.0, 5.8,10.5,15.2,20.0, 1.0, 5.8,10.5,15.2,20.0, 1.0, 5.8,10.5,15.2,20.0,

ports.input:

• North boundary (port 1): i=382-500, j=501
• West boundaries (ports 2&4): Two sections along western edge
• South boundary (port 3): i=1-369, j=1
• East boundary (port 5): i=501, j=189-500
• Port e-folding time: 5.0 days

Questions:

1. Could the open boundary configuration be causing the excessive surface velocities?
2. Is the immediate stability with nesting unusual, or does this indicate over-constraining?
3. What modifications would you recommend to address these issues?

Any guidance you could provide would be greatly appreciated. Thank you for your time and expertise.

Best regards,

Minglei Zhong

[Alan Wallcraft]

What do you mean by "stabilize"?

I suggest, runing for one year (say):

a) From climatology with closed boundaries.

b) From climatology with relaxation near "open" boundaries boundaries (e-folding time of 5 to 90 days)

c) From climatology with nesting (e-folding time of 0.1 to 9 days)

d) From the outer model initial state with nesting

Your target in most cases is (d), but all the above should work.  

I think you are saying that (b) does not work.  If so, then check that (a) has the same problem and if it does then either your time steps are wrong or your atmospheric forcing is wrong.

Alan.

[hao ming]

Dear Alan,

Thank you for your prompt response and suggestions.

To clarify what I mean by "stabilize": When I use closed boundaries with relaxation and run a 10-year climatological simulation, if I plot the transport at a specific point over time, the transport typically remains lower during the first few years and only reaches a stable state after year 5-6. However, when I use nesting, the transport remains nearly constant from year 1 through year 10 - it appears to be stable from the very beginning.

This immediate stability with nesting seems unusual to me compared to the spin-up period I observe with relaxation alone. Is this behavior expected when using nesting, or could it indicate that the model is being over-constrained by the boundary conditions?

I will follow your suggestions to systematically test configurations (a) through (d) to better diagnose the issue. I'll start with the closed boundary case to check if the excessive surface velocities persist without any open boundary influence.

Thank you again for your guidance.

Best regards,

Minglei Zhong

[Alan Wallcraft]

Are you starting the nesting cases from an outer model initial state?

Typically, cases that start from climatology "spin-up" over a few years.  Region-wide kinetic energy is usually close to its maximum after about a year but the equatorial Pacific takes about 3 years to form a realistic undercurrent and a western boundary currents can take 5-10 year to develop fully.  On the other hand, cases that start from a data-assimilative outer model (e.g. from GOFS) typically spin-down slightly because their realistic currents are stronger than the "climatology" of the ocean model.  This spin-down is relatively quick, within a few months.

As example, we would run an Atlantic basin model for 10 years from climatology but only 5 years from GOFS, taking results from the the last 3 years in both cases.  In the case of the Gulf of Mexico (Gulf of America) we never now start from climatology, always from GOFS, but because Loop Current eddies have about an annual time scale we take statistics over at least 10 years.

Alan.

[hao ming]

Dear Alan,

Thank you for your detailed explanation and the insightful examples. They really help clarify the differences between spin-up and spin-down behaviors.

To answer your question: No, I am starting the nesting cases from a climatology initial state (cold start), not from an outer model initial state. It's essentially a climatology setup combined with nesting.

Given that, I'm still puzzled by the lack of an apparent spin-up period in my nesting runs, where the transport appears stable from the very beginning, unlike the 5-6 year spin-up I see in the relaxation-only cases. Based on your description, I would have expected a similar spin-up process here.

I will proceed with the systematic tests you suggested (a) through (d) to diagnose this further and check for any underlying issues with the configuration.

Thank you again for your guidance.

[Alan Wallcraft]

It is possible that the stable velocities from nesting is because your sample locations are in the nest relaxation zone.  Climo relax is only for T&S but nest relax also includes velocities.

If you think you are sampling away from the nest relax zones, check the rmu file.

Alan.