We are trying to use ROMS to predict tides in the East China Sea. The model-predicted phase compares well with the observation. But the amplitude is smaller then the observation by about 10-20 cm over the whole domain, especially along the coast. We further checked the model-predicted value (output) at the open boundary and found that it is also smaller than the given value by about 20percent (10cm). Why does this happen? Whether is it due to the boundary condition or any other issue?
I also checked the results of the Chesapeake Bay model which uses the same boundary flag. The difference between the model-predicted value and the given value is really small, only about 1cm.
We use TPXO6.0 as the open boundary condition and the following flag for tides (2D).
#define SSH_TIDES
#define UV_TIDES
/* Open boundary conditions */
#define EAST_FSCHAPMAN
#define EAST_M2FLATHER
#define SOUTH_FSCHAPMAN
#define SOUTH_M2FLATHER
#define WESTERN_WALL
#define NORTHERN_WALL
Has anyone encountered this question or any comments?
Thanks for any help.
why does model underestimate the tidal amplitude???
Once I fix all my Northeast Pacific issues, I have a tidal issue to look into as well. I've got a 2D domain covering the Bering Sea at 4 km resolution. It also covers Cook inlet, where the tides are OK over the southern part of the domain, far too small in the northern part. For this 2D run, the horizontal viscosity was used to tune the model to some extent. It is currently larger than I am comfortable with - the model runs with it set to zero, but gives too large amplitudes over much of the Bering Sea shelf.
-
m.hadfield
- Posts: 521
- Joined: Tue Jul 01, 2003 4:12 am
- Location: NIWA
When you say the tide amplitude is small compared to observations, do you mean direct in situ observations or the TPXO analysis? If it's the former, have you compared your observations to TPXO itself?
If TPXO is low also, then possibly the boundary condition values are low - ROMS has little scope to compensate for this.
If there is a progressive loss of amplitude of the ROMS solution compared to the TPXO solution in the direction of propagation, perhaps you have overestimated friction. As I recall from some old papers with co-phase diagrams for the region, the cyclonic advance of tidal phase around the East China Sea means there is quite a long propagation path to observation sites close to Shanghai.
If TPXO is low also, then possibly the boundary condition values are low - ROMS has little scope to compensate for this.
If there is a progressive loss of amplitude of the ROMS solution compared to the TPXO solution in the direction of propagation, perhaps you have overestimated friction. As I recall from some old papers with co-phase diagrams for the region, the cyclonic advance of tidal phase around the East China Sea means there is quite a long propagation path to observation sites close to Shanghai.
John Wilkin: DMCS Rutgers University
71 Dudley Rd, New Brunswick, NJ 08901-8521, USA. ph: 609-630-0559 jwilkin@rutgers.edu
71 Dudley Rd, New Brunswick, NJ 08901-8521, USA. ph: 609-630-0559 jwilkin@rutgers.edu
I don’t know whether it’s due to the progressive loss of amplitude, since the model value at the boundary is also about 20 percent smaller than the given value (TPXO).
Also the model value reduced to the smaller value very quickly (half M2 cycle) before it reaching the coast. So I think it may not due to the adjustment of the inside.
I also try the CLAMPED boundary, the model value is exactly the value given at the boundary, but no much change inside the domain and blocks occur at the boundary.
Also the model value reduced to the smaller value very quickly (half M2 cycle) before it reaching the coast. So I think it may not due to the adjustment of the inside.
I also try the CLAMPED boundary, the model value is exactly the value given at the boundary, but no much change inside the domain and blocks occur at the boundary.
-
m.hadfield
- Posts: 521
- Joined: Tue Jul 01, 2003 4:12 am
- Location: NIWA
I'm still inclined to suspect the boundary velocities. Could there be a mistake in converting the outer model tidal velocity data into the form required for ROMS forcing files? How well do the velocities simulated by ROMS at the boundary match the outer model velocities?
You're right: FSCLAMPED is not going to work well.
You're right: FSCLAMPED is not going to work well.
You are right!
The TPXO velocity at the boundary is too small compared to the model output (don't know whether due to my fault or the TPXO data yet, checking). It seems that the Flather boundary tends to transform potential energy to kinematic energy, right?
I use SSH_Tides & FSOBC_REDUCED, it works fine now.
I note that FSOBC_REDUCED works together with *_M2FLATHER or *_M2 REDUCED. What's the difference between these two? When using together with *_M2FLATHER, the reduced physics only applies to tides, but when using together with *_M2REDUCED, the reduced physics applies to both tides and the processed OBC data? Is my understanding right?
The TPXO velocity at the boundary is too small compared to the model output (don't know whether due to my fault or the TPXO data yet, checking). It seems that the Flather boundary tends to transform potential energy to kinematic energy, right?
I use SSH_Tides & FSOBC_REDUCED, it works fine now.
I note that FSOBC_REDUCED works together with *_M2FLATHER or *_M2 REDUCED. What's the difference between these two? When using together with *_M2FLATHER, the reduced physics only applies to tides, but when using together with *_M2REDUCED, the reduced physics applies to both tides and the processed OBC data? Is my understanding right?