Add tidal potential forcing for 8 major constituents #371

sbrus89 · 2019-09-18T14:23:47Z

Includes diurnal constituents: K1, O1, P1, Q1 and semi-diurnal constituents: K2, N2, S2, M2. A tanh ramp in time is used to gradually apply the forcing (similar to time-varying atmospheric forcing). It has been implemented for both split_explicit and RK4 time-stepping.

Also, there are several updates to the Hurricane Sandy test case:

Increase maximum depth to 6000m (from 2000m)
Decrease minimum depth to 10m (from 30m)
Use 100layerE3SMv1 vertical grid (was 60layerPHC)
Decrease forcing ramps to 10 days and shorten simulation duration to 10/10-11/3 (was 10/1-11/8)

Any feedback on the implementation is welcome.

sbrus89 · 2019-09-18T14:33:46Z

Here are some results compared to our previous best solution without tides.

These runs use a mesh with 2km resolution in Delaware Bay and 5km in the Chesapeake Bay and Long Island regions.
Finer resolution is going to be required to improve these results, especially inside the bays.
Also, these runs do not use wetting and drying, which should also help improve the comparison to observed data.

src/core_ocean/Registry.xml

src/core_ocean/mode_forward/mpas_ocn_time_integration_split.F

sbrus89 · 2019-09-30T14:44:31Z

Results of the finer mesh USDEQU60at15cr5rr1 compared with USDEQU120at30cr10rr2. The pre-storm tidal amplitudes are too high, but the phase errors are better and the surge is also improved.

pwolfram · 2019-10-15T21:11:09Z

Will want to add simple SAL in before we merge this week in conversations with Brian Arbic and Nairita Pal.

pwolfram · 2019-10-15T21:12:52Z

@vanroekel and @xylar, if you'd like to review this week could you please take a look over the next day or so? In conversations with @mark-petersen we need to get this in this week. @proteanplanet
has offered to take a look in a G case with this merged branch. Thanks!

xylar · 2019-10-15T21:17:18Z

@pwolfram, I wasn't aware a review had been requested. I'll take a look at this tomorrow.

vanroekel · 2019-10-15T21:20:40Z

I don't feel qualified to review these changes. I'll remove my name from the list

xylar · 2019-10-16T06:42:48Z

@sbrus89 or @pwolfram, could you point to the design doc here if it's publicly available?

xylar · 2019-10-16T06:53:06Z

If I have found the correct design doc via confluence, I see the following:

For instance, in a shallow water model, we replace the term $\nabla \eta$ in the momentum equation with a gradient of $\eta$ referenced to the equilibrium tide and self-attraction and loading terms...In essence, the equilibrium tide and self-attraction and loading terms reset the reference against which pressure gradients are computed.

Is this the approach that is being taken here? If so, I'm concerned that just resetting the SSH will have adverse consequences such as lack of mass conservation in a fully 3D model. In other words, the SSH is an important dynamical variable beyond its influence on the pressure gradient. Instead, it seems to me that tides (and eventually changes in equilibrium sea level) would need to be incorporated as change the reference gravitational potential, which is currently assumed to correspond to z=0.

See Ch. 7 and 8 of the MOM5 manual for details on how this would work: https://mom-ocean.github.io/assets/pdfs/MOM5_manual.pdf

@sbrus89 and @pwolfram (and maybe @mark-petersen), could we discuss this over the phone sometime today?

xylar

A few small comments on the code, but my main one is my concern about the overall approach above.

src/core_ocean/mode_forward/mpas_ocn_time_integration_split.F

pwolfram · 2019-10-16T14:38:08Z

@xylar, the design doc is at https://www.overleaf.com/read/dxgprfyjnnnw Is this the one you were looking at for your review?

xylar · 2019-10-16T15:02:34Z

@pwolfram, yes, that's the document I was looking at. However, I don't think it's a good idea to share an editable link to it on a public forum (though that ship has sailed).

pwolfram · 2019-10-16T17:05:26Z

Thanks @xylar, please let me know if think isn't "view only" because that is what was selected in overleaf. Please email me if there is another editable link out there and we'll remove it.

pwolfram · 2019-10-16T17:24:56Z

@xylar, can you please be more specific about your concern? We are effectively using 8.2 from the MOM link (with SAL needing to be added as noted above). But, we are just going to use the very simple scalar approximation for SAL for this time with a more advanced SAL added later. Are you concerned about using this form itself?

Maybe this is the issue as implied in design doc that we replace ssh (we don't):

∇η→∇(η−η_EQ−η_SAL) (3.4)

We are not modifying the free surface as this implies but rather adding on the gravitational potential terms to account for the tide (reflected in the momentum equation via the gradient of the potentials to represent the tidal forcing). This results in the free surface as an emergent property. Thus, η ≠ η−η_EQ−η_SAL, but we are adding the following forcing terms to the momentum equation RHS: -∇(η_EQ+η_SAL). These terms aren't there currently but need to be added as tendencies to have the tidal solution computed as an emergent property of the free surface.

xylar · 2019-10-16T18:41:06Z

Thanks @xylar, please let me know if think isn't "view only" because that is what was selected in overleaf. Please email me if there is another editable link out there and we'll remove it.

I'm definitely able to edit. Maybe you gave me special permission?

src/core_ocean/shared/mpas_ocn_tendency.F

sbrus89 · 2019-10-17T16:44:28Z

@xylar, thanks so much for reviewing so carefully! I really appreciate it. In light of your comments, I made some changes which (I hope) help clarify what is being done and make things a bit cleaner. Thanks again.

src/core_ocean/mode_forward/mpas_ocn_time_integration_split.F

- Equilibrium tidal potential for K1, O1, P1, Q1, K2, N2, S2, M2 - Simple scalar approximation for self attraction and loading - Incorporation into Hurricane Sandy test cases

pwolfram · 2019-10-18T16:51:21Z

@sbrus89, had to update the design doc part to include the other tex files and remove the pdf.

pwolfram · 2019-10-18T16:53:07Z

Just to confirm no damage was done:

design_docs pwolfram$ git diff 2aae280 34698ba --stat
 docs/ocean/design_docs/tides.pdf             |  Bin 204589 -> 0 bytes
 docs/ocean/design_docs/tides/ametsoc.bst     | 1741 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 docs/ocean/design_docs/tides/tides.bib       |   35 +++++
 docs/ocean/design_docs/{ => tides}/tides.tex |    0
 4 files changed, 1776 insertions(+)

pwolfram · 2019-10-18T16:53:26Z

@mark-petersen, can you please approve so that I can merge?

sbrus89 · 2019-10-18T16:57:54Z

Oops. Sorry I forgot to include the other tex files

pwolfram · 2019-10-18T17:00:55Z

@sbrus89, it is all good. Please enjoy your day off and many thanks for the help!

mark-petersen

Merged locally into ocean/develop. Tested with gnu and intel debug on grizzly. Passes bfb against previous.

Adds tidal forcing from 8 major constituents and simple self-attraction and loading. Includes diurnal constituents: K1, O1, P1, Q1 and semi-diurnal constituents: K2, N2, S2, M2. A tanh ramp in time is used to gradually apply the forcing (similar to time-varying atmospheric forcing). It has been implemented for both split_explicit and RK4 time-stepping. It also includes a simple self attraction and loading (0.09*eta) parameterization. Also, there are several updates to the Hurricane Sandy test case to provide verification that tides is working propertly: * Increase maximum depth to 6000m (from 2000m) * Decrease minimum depth to 10m (from 30m) * Use 100layerE3SMv1 vertical grid (was 60layerPHC) * Decrease forcing ramps to 10 days and shorten simulation duration to 10/10-11/3 (was 10/1-11/8)

pwolfram · 2019-10-18T18:15:33Z

Thanks everyone!

This PR is an accumulation of PRs into the ocean/coastal branch. We are bringing them in at once for efficiency: #285, #289, #284, #295, #310, #311, #312, #335, #354, #356, #358, #359, #365, #371

This PR is an accumulation of PRs into the ocean/coastal branch. We are bringing them in at once for efficiency. #285, #289, #284, #295, #310, #311, #312, #335, #354, #356, #358, #359, #365, #371

pwolfram · 2020-01-17T00:15:41Z

src/core_ocean/shared/mpas_ocn_tendency.F

+
+        do iCell = 1, nCells
+          do k = 1, maxLevelCell(iCell)
+            tidalPotentialZMid(k,iCell) = zMid(k,iCell) - tidalPotentialEta(iCell) &


@sbrus89, potential issue may be here with #366 including tidal forcing. zMid is a strong negative with the spurious datum shift we are employing in #366 until the vertical datum issue is resolved.

pwolfram · 2020-01-17T00:19:53Z

src/core_ocean/shared/mpas_ocn_tendency.F

+        do iCell = 1, nCells
+          do k = 1, maxLevelCell(iCell)
+            tidalPotentialZMid(k,iCell) = zMid(k,iCell) - tidalPotentialEta(iCell) &
+                                        - config_self_attraction_and_loading_beta * zMid(k,iCell)


Will test config_self_attraction_and_loading_beta=0 and report back because offset at zMid is likely cause of problem with offset.

That fixed the issue, so having a water column shifted away from 0 for zMid will cause a pretty spurious artificial tidal flow.

Fixed means code was stable, but results are not correct with tides on (but are ok with tides off), which means we still have a bug here.

SE no wetting / drying

RK4 no wetting / drying

no tides with wetting / drying

tides with wetting / drying but no SAL

These are all for USDEQU120at30cr10rr2* meshes.

Using the approach of adding back out the vertical shift produces:

The tidal signal is contained in the output now, but the result is not as good as say just the standard RK4 without wetting / drying, so something is still not completely right.

Del4 differed between the runs, am reruning with matching del4 values.

Note, hard-coded code change was tidalPotentialZMid(k,iCell) = (zMid(k,iCell) + 977.502_RKIND) - tidalPotentialEta(iCell) &...

Station 8519483 looks better though, especially compared with the previous results.

Possible correction

@sbrus89, if we allow for zMid offset from zero, this means that this
- config_self_attraction_and_loading_beta * zMid(k,iCell) may not be correct.
Shouldn't it be something more like this?
- config_self_attraction_and_loading_beta * (ssh(iCell)-ssh_dry(iCell))

Given it is a small correction I'm wondering if we shouldn't just turn it off.

Given the approximation really is only deep water as the text implies, maybe we should do something like this:
- config_self_attraction_and_loading_beta * (bottomDepth > 100)*ssh(iCell)

So, we wouldn't do any SAL for shallow waters with less than 100m depth.

Reference

https://www.sciencedirect.com/science/article/pii/S0967064504002048

pwolfram · 2020-01-24T15:15:12Z

src/core_ocean/shared/mpas_ocn_tidal_potential_forcing.F

+      real (kind=RKIND), dimension(:,:), pointer :: latitudeFunction
+      integer, dimension(:), pointer :: constituentType
+      real (kind=RKIND), dimension(:), pointer :: eta
+      real (kind=RKIND), dimension(:,:), pointer :: zMid, tidalPotentialZMid


We don't need these here, do we?

cc @sbrus89

Correct, the zMid and tidalPotentialZMid could be removed. At one point I was using them here, but moved those calculations elsewhere and forgot to get rid of the declarations. Thanks for catching this.

Adds tidal forcing from 8 major constituents and simple self-attraction and loading. Includes diurnal constituents: K1, O1, P1, Q1 and semi-diurnal constituents: K2, N2, S2, M2. A tanh ramp in time is used to gradually apply the forcing (similar to time-varying atmospheric forcing). It has been implemented for both split_explicit and RK4 time-stepping. It also includes a simple self attraction and loading (0.09*eta) parameterization. Also, there are several updates to the Hurricane Sandy test case to provide verification that tides is working propertly: * Increase maximum depth to 6000m (from 2000m) * Decrease minimum depth to 10m (from 30m) * Use 100layerE3SMv1 vertical grid (was 60layerPHC) * Decrease forcing ramps to 10 days and shorten simulation duration to 10/10-11/3 (was 10/1-11/8)

This PR is an accumulation of PRs into the ocean/coastal branch. We are bringing them in at once for efficiency: MPAS-Dev#285, MPAS-Dev#289, MPAS-Dev#284, MPAS-Dev#295, MPAS-Dev#310, MPAS-Dev#311, MPAS-Dev#312, MPAS-Dev#335, MPAS-Dev#354, MPAS-Dev#356, MPAS-Dev#358, MPAS-Dev#359, MPAS-Dev#365, MPAS-Dev#371

sbrus89 added feature coastal labels Sep 18, 2019

sbrus89 requested review from mark-petersen, pwolfram, vanroekel and xylar September 18, 2019 14:23

sbrus89 assigned pwolfram Sep 18, 2019

mark-petersen reviewed Sep 18, 2019

View reviewed changes

src/core_ocean/Registry.xml Outdated Show resolved Hide resolved

mark-petersen reviewed Sep 18, 2019

View reviewed changes

src/core_ocean/mode_forward/mpas_ocn_time_integration_split.F Outdated Show resolved Hide resolved

vanroekel removed their request for review October 15, 2019 21:20

xylar reviewed Oct 16, 2019

View reviewed changes

src/core_ocean/mode_forward/mpas_ocn_time_integration_split.F Outdated Show resolved Hide resolved

src/core_ocean/mode_forward/mpas_ocn_time_integration_split.F Outdated Show resolved Hide resolved

pwolfram reviewed Oct 16, 2019

View reviewed changes

src/core_ocean/mode_forward/mpas_ocn_time_integration_split.F Outdated Show resolved Hide resolved

xylar reviewed Oct 16, 2019

View reviewed changes

src/core_ocean/shared/mpas_ocn_tendency.F Outdated Show resolved Hide resolved

xylar reviewed Oct 16, 2019

View reviewed changes

src/core_ocean/shared/mpas_ocn_tendency.F Outdated Show resolved Hide resolved

xylar reviewed Oct 17, 2019

View reviewed changes

src/core_ocean/mode_forward/mpas_ocn_time_integration_split.F Outdated Show resolved Hide resolved

sbrus89 mentioned this pull request Oct 17, 2019

Adds Delaware flooding case #366

Closed

sbrus89 force-pushed the tides branch from 57ba568 to 34698ba Compare October 18, 2019 16:16

Add tidal forcing

2aae280

- Equilibrium tidal potential for K1, O1, P1, Q1, K2, N2, S2, M2 - Simple scalar approximation for self attraction and loading - Incorporation into Hurricane Sandy test cases

pwolfram force-pushed the tides branch from 34698ba to 2aae280 Compare October 18, 2019 16:50

pwolfram approved these changes Oct 18, 2019

View reviewed changes

mark-petersen self-requested a review October 18, 2019 17:47

mark-petersen approved these changes Oct 18, 2019

View reviewed changes

pwolfram merged commit 2aae280 into MPAS-Dev:ocean/coastal Oct 18, 2019

pwolfram deleted the tides branch October 18, 2019 18:15

mark-petersen mentioned this pull request Oct 18, 2019

Merge coastal features into ocean/develop #384

Merged

pwolfram reviewed Jan 17, 2020

View reviewed changes

pwolfram reviewed Jan 24, 2020

View reviewed changes

pwolfram mentioned this pull request Feb 27, 2020

Verify SAL for RK4 (wetting / drying cases) #458

Open

matthewhoffman added coastal feature labels Mar 17, 2021

Add tidal potential forcing for 8 major constituents #371

Add tidal potential forcing for 8 major constituents #371

Uh oh!

Conversation

sbrus89 commented Sep 18, 2019

Uh oh!

sbrus89 commented Sep 18, 2019

Uh oh!

Uh oh!

Uh oh!

sbrus89 commented Sep 30, 2019

Uh oh!

pwolfram commented Oct 15, 2019

Uh oh!

pwolfram commented Oct 15, 2019

Uh oh!

xylar commented Oct 15, 2019

Uh oh!

vanroekel commented Oct 15, 2019

Uh oh!

xylar commented Oct 16, 2019

Uh oh!

xylar commented Oct 16, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

xylar left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

pwolfram commented Oct 16, 2019

Uh oh!

xylar commented Oct 16, 2019

Uh oh!

pwolfram commented Oct 16, 2019

Uh oh!

pwolfram commented Oct 16, 2019

Uh oh!

xylar commented Oct 16, 2019

Uh oh!

Uh oh!

Uh oh!

sbrus89 commented Oct 17, 2019

Uh oh!

Uh oh!

pwolfram commented Oct 18, 2019

Uh oh!

pwolfram commented Oct 18, 2019

Uh oh!

pwolfram commented Oct 18, 2019

Uh oh!

sbrus89 commented Oct 18, 2019

Uh oh!

pwolfram commented Oct 18, 2019

Uh oh!

mark-petersen left a comment

Choose a reason for hiding this comment

Uh oh!

pwolfram commented Oct 18, 2019

Uh oh!

Choose a reason for hiding this comment

Uh oh!

pwolfram Jan 17, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

SE no wetting / drying

RK4 no wetting / drying

no tides with wetting / drying

tides with wetting / drying but no SAL

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

xylar commented Oct 16, 2019 •

edited

Loading

pwolfram Jan 17, 2020 •

edited

Loading