Problem with the river discharge

[dandanouc]

Hello everyone,
The model domain of my work is the East China Sea. I consider 8 rivers in the model.
I have checked the information in this forum; usually the river discharged water cannot be spread away. However, in my case, the fresh water spread far away compare with the observed one, especially for the Changjiang River. The figures are the comparison of the observed sea surface salinity with model result. I have checked the volume of the Changjiang River which has no problem.
What will be the reason that the fresh water spread far away? Thank you very much.

Here are my CPP options:

/* general */

#define CURVGRID
#define MASKING
#define SALINITY
#define NONLIN_EOS
#define SOLVE3D
#ifdef SOLVE3D
# undef SPLINES
#endif

/* advection, dissipation, pressure grad, etc. */

#ifdef SOLVE3D
# define DJ_GRADPS
#endif

#define UV_ADV
#define UV_COR
#define UV_QDRAG

#ifdef SOLVE3D
# undef MIX_GEO_TS
# define MIX_S_TS
# define TS_DIF2
#endif

#define MIX_S_UV
#define UV_VIS2

#ifdef SOLVE3D
# define TS_MPDATA
# ifdef TS_MPDATA
# undef TS_U3HADVECTION
# undef TS_C4VADVECTION
# else
# define TS_U3HADVECTION
# define TS_C4VADVECTION
# endif
#endif

/* vertical mixing */

#ifdef SOLVE3D
# define MY25_MIXING
# undef GLS_MIXING
#endif

#if defined MY25_MIXING || defined GLS_MIXING
# define KANTHA_CLAYSON
# define Canuto A
# define N2S2_HORAVG
#endif

/* surface and side corrections */

#define SCORRECTION
#define QCORRECTION

/* point source (rivers, line sources) */

#define UV_PSOURCE
#define TS_PSOURCE


/* roms quirks */

#define ANA_BTFLUX
#define ANA_BSFLUX

/* tides */

#define SSH_TIDES
#ifdef SSH_TIDES
# define ADD_FSOBC
#endif

#define UV_TIDES
#ifdef UV_TIDES
# define ADD_M2OBC
#else
# define ANA_M2OBC
#endif

/* Boundary conditions */

# define RADIATION_2D

# define WESTERN_WALL

# ifndef NORTHERN_WALL
# define NORTH_FSCHAPMAN
# define NORTH_M2FLATHER
# define NORTH_TRADIATION
# define NORTH_TNUDGING
# define NORTH_M3RADIATION
# define NORTH_M3NUDGING
# endif

# ifndef WESTERN_WALL
# define WEST_FSCHAPMAN
# define WEST_M2FLATHER
# define WEST_TRADIATION
# define WEST_TNUDGING
# define WEST_M3RADIATION
# define WEST_M3NUDGING
# endif

# ifndef SOUTHERN_WALL
# define SOUTH_FSCHAPMAN
# define SOUTH_M2FLATHER
# define SOUTH_TRADIATION
# define SOUTH_TNUDGING
# define SOUTH_M3RADIATION
# define SOUTH_M3NUDGING
# endif

# ifndef EASTERN_WALL
# define EAST_FSCHAPMAN
# define EAST_M2FLATHER
# define EAST_TRADIATION
# define EAST_TNUDGING
# define EAST_M3RADIATION
# define EAST_M3NUDGING
# endif

/* output stuff */

#define PERFECT_RESTART
#ifdef PERFECT_RESTART
# define AVERAGES
# undef DIAGNOSTICS_BIO
# undef DIAGNOSTICS_TS
# undef DIAGNOSTICS_UV
# define RST_SINGLE
#endif

[kate]

You've got insufficient vertical mixing at the surface. The fresh water comes in at the surface, making things more stable, and the mixing scheme has no way to break down that stable fresh layer. We've seen this before off Alaska and our current "fix" is to supply the river water as an additional rainfall term. We'll have to do better if one of our proposals gets funded...

[rduran]

I'm wondering if changing the vertical profile of the river discharge would help?

[hetland]

There are a number of reasons for this discrepancy. First of all, what does the data represent? A climatology? A synoptic cruise? At what level were the salinities measured (usually a few meters depth for standard hydrography)?

As for errors in the model, the differences you show are too far away for the source to have much of an impact. I would guess there might be errors in the specification of wind stress of evaporation. There might also be differences based on different closure schemes -- particularly the stability functional (e.g., CANUTO_A or _B).

There is no easy fix for a problem like this. You need to understand all the things that go into your model, and how each factor might affect the solution.

[susonic]

I believe that you have used the climatological monthly windstress data.
Why don't you use daily mean or 6 hourly wind data?
I'm sure that you will get a better result.
-JH

[dandanouc]

I am very grateful for all the kindly replies and I have solved the problem according to the suggestion of replacing the climatology wind stress with the daily data.
However, I encountered another problem that has puzzled me for a period of time. The attachments are the sea surface current from my model results and the current distribution quoted from Guo et al. (2006). It is obviously that both the current from the Taiwan Strait and the branches of Kuroshio west of Ryukyu in my model are weaker than Guo’s. It looks like both the Taiwan warm current volume and Kuroshio volume exit East China Sea from the Tokara Strait without the Tsushima Strait.
Then, what contributes to the weaker of the Taiwan warm current and the branches of Kuroshio towards Tsushima Strait? Thank you!

[LiuZHQ]

Branching of the KC to Tsushima Strait is, nowadays, one of the major scientific questions related to interactions between the ECS shelves and the KC. It is not easy to answer that question: why Tsushima Warm Current and Taiwan Warm Current getting smaller during your modelling time.

The way I will check this problem is to calculate the volume transport from your boundary file to find out whether you give the right amount of waters from Taiwan Strait and get them out of your domain from that part of the eastern and northern boundaries in Japan Sea. That value, in fact, is critically related to the model you want to nest with. If they are wrong, you can not be right, since there is not algorithm to correct it.

I will also check the model output in Tsushima, Taiwan and the boundaries in Japan Sea to find out whether they give some reasonable value. I recommand that you refer to the review paper of Isobe [2008] for those values.

I suppose you included the tides, which are crticial for the right extending direction of Changjiang river plume, and I think you should double check your wind forcing, since spreading of those buoyant waters is gennearlly trapped in the Ekman layers, I think.

BTW, you may also want to check your topography, before you do all of those checking. That is the first-ordered important parameter.

Happy modelling,

ZQ