Wind AND tidal forcing in the Inlet test

[drews]

How would you modify the Inlet test to add onshore wind forcing?

The Inlet test uses the northern domain boundary to generate tidal oscillations that propagate south through a narrow strait and into an inlet. Here is the tidal forcing in ana_fsobc.h within the Coupled sub-directory:

#if defined INLET_TEST
IF (NORTHERN_EDGE) THEN
cff=-1.0_r8*sin(2.0_r8*pi*time(ng)/(12.0_r8*3600.0_r8))
DO i=IstrR,IendR
BOUNDARY(ng)%zeta_north(i)=cff
END DO
END IF

This code generates a nice sinusoidal oscillation in the free-surface zeta. The peak amplitude of the tides is 1.0 meter and the mean is 0.0 meters (sea level). In a few places of the domain the zeta tides have a slightly greater amplitude, but the mean water level is always zero, as expected.

I increased the value of NTIMES from 720 to 7200 in ocean_inlet_test.in in order to run the model for a few days.

I added the following code to ana_smflux.h to simulate a wind blowing from the north:

#if defined INLET_TEST
DO j=JstrR,JendR
DO i=IstrR,IendR
svstr(i,j)= -2.0_r8 / rho0
END DO
END DO

That's about 28 m/s. Now the southern edge of the domain shows a sinusoidal tide with a peak amplitude of about 1 meter and a mean water level of about 0.5 meters above sea level. I expected this result; the wind from the north is causing storm surge at the southern edge of the domain, and we have tides imposed on top of storm surge, just like in real life.

The problem is at the northern domain boundary. The tidal peak amplitude there is 1, but the mean water level is relentless 0.0 meters. I want the water level to drop on the upwind side of the domain (wind setdown), as I have seen many times in ROMS without tidal forcing. The northern domain boundary should oscillate in zeta with the tidal forcing, but the mean water level should also float free with the wind forcing. Somehow.

I might be overlooking a simple setting. How can the Inlet test unlock the tidally forced domain boundary from its mean value of 0.0 meters?

[jcwarner]

you might have to try a few things, but to start how about adjusting

cff=-1.0_r8*sin(2.0_r8*pi*time(ng)/(12.0_r8*3600.0_r8))
DO i=IstrR,IendR
BOUNDARY(ng)%zeta_north(i)=cff
END DO

to have some setdown condition, such as
cff=-1.0_r8*sin(2.0_r8*pi*time(ng)/(12.0_r8*3600.0_r8))-0.20_r8
or some other value?

-j

[drews]

Since storm surge and wind setdown develop over many hours, I took John's suggestion and made the lowering of the water level at the upwind/northern boundary dependent on time:

#if defined INLET_TEST
IF (NORTHERN_EDGE) THEN
cff=-1.0_r8*sin(2.0_r8*pi*time(ng)/(12.0_r8*3600.0_r8))
cff = cff - 0.10_r8 * (2.0_r8*pi*time(ng)/(12.0_r8*3600.0_r8))
DO i=IstrR,IendR
BOUNDARY(ng)%zeta_north(i)=cff
END DO
END IF

The Inlet test domain is 14 kilometers from south to north, so this expression will produce far more vertical displacement in the water level over time than is realistic. As intended - I want to watch what's happening as the tides and wind work together. Here is a plot of the free surface zeta in two places:

Inlet test: wind and tide

InletTestWindAndTide.png (20.42 KiB) Viewed 3147 times

The lower red line is the northern (tidally forced) boundary, and the upper white line is the southern (free) boundary. We have a tidal signal superimposed over a drop in the water level. I can see that since the two lines are gradually diverging, the wind from the north is pushing the water mass southward and producing storm surge on the southern shore. So far, so good.

At this point I could run the Inlet test without tides, record the wind setdown at the northern domain boundary, and use those values as my baseline for the northern boundary. I would either formulate an analytical expression to match the wind-only model results, or use a forcing file. Then I would add in the tidal signal and run the model a second time with wind and tides. Storm surge and tides are not strictly additive, but this two-step approach would be better than simply guessing what the wind setdown should be.

When a typhoon approaches Inlet test and I'm trying to issue real-time forecasts for the people who live there, running the model twice would pose an operational problem.

Or maybe I should not try to make something realistic happen at the domain boundary, since a free-floating wall in the middle of the ocean is not realistic anyway? The important thing is to get accurate surge and tidal height on the southern coast of Inlet test.

Any further comments?