Oscillation in bottom stress for GLS_MIXING plus SSW_BBL

[m.hadfield]

Hello colleagues

Preface:

I don't know how big a problem the following really is in practice, but I've noticed some curious & spurious behaviour, so I thought I'd let the community know about it.

Short version:

A toy case with oscillating tidal flow shows pronounced, rapid oscillations in bottom stress (also near-bottom speed and TKE) when SSW_BBL and GLS_MIXING are enabled. These oscillations occur as the tidal current speed drops away from its maximum value. They can be avoided by substituting LMD_MIXING for GLS_MIXING or UV_LOGDRAG from SSW_BBL, or by reducing the prescribed bottom orbital velocity.

Long version:

I have attached the necessary header files and the input file, also a time series plot of bottom stress (bustr) and a snapshot of velocity (u).

The toy case involves an oscillating tidal flow (period 12 hours) in a 20m deep, periodic domain. The expected behaviour is that velocity oscillates between about -0.4m/s and +0.4m/s, with a more-or-less logarithmic drop-off towards the bottom. When SSW_BBL and GLS_MIXING are enabled and a reasonable large bottom orbital velocity is imposed (here 0.5m/s) there are rapid oscillations in the near-bottom velocity, bottom stress and TKE when the tidal current is reducing in amplitude. These oscillations develop and die away rapidly. In the velocity snapshot shown here, the bottom velocity at a maximum; at the next time step it is less than 0.

Reducing the bottom orbital velocity to 0.3m/s essentially eliminates the oscillation.

Question:

Has anyone else seen similar behaviour?

Mark

[m.hadfield]

Also attached FYI is the input file (which includes the GLS model coefficients)

[jcwarner]

ok. still need to know how to force this thing?
roms_frc.nc?
was there an ana_smflux?

[jcwarner]

you still need to set a Pwave_bot value.
i did not see one set in the ana_wwave. try 8 secs or something and see if that helps.
-j

[m.hadfield]

You only need Pwave_bot if SEDIMENT is enabled. The original case where I saw this behaviour did have sediment, but I stripped it out in order to simplify things.

[jcwarner]

um, i think you need pwave no matter if sed is defined or not.
if you activate ssw_bbl, then check the Build/bbl.f90 and see what it needs. if you specify ana_wwave (not sure why there are 2 w's there, wind wave?), then it needs an amplitude and a period. so i still think you need a period. the bbl routines were coded to work with sed on or off.
-j

[m.hadfield]

Um, I think you're right! And setting Pwave_bot to 8s suppresses the oscillation.

Thank you.