abnormal salinity with river point source

[tony1230]

Hi ROMS friends

These days i have been running a real case with river point source, including T,S and Nutrients. I'd like to test firstly if the river location are right to inflow.

My test was:
1. with tides in the open boundary;
2. with rivers;
3. no more other forcing;
4. constant values for T(25) and S(30) horizontally and vertically in the initial field;

In order to avoid numerically unstable, the vulues given to T and S (river transports)were gradually decrease (increase) to freshwater value like 0.5 (e.g, S in the rivers are 30,29,28,...,1,0.5,0.5,....0.5).

Ruselt: please refer to the attached figure (snapshot). Near the river points, there minus values appear, sometime they were here and over a time they switched to other river nearby. Besides, there many narrowband, with larger values than initial ones, rounded the less saline regions formed by freshwater. I have changed many times the locations of the river, all failed.

Attached are also the .in and .h file for easy diagnostic.

Any guidance/suggest will be warmly appreciated.

Thanks in advanced.

Steve W. Shou

[kate]

I'm afraid many of the ROMS horizontal advection schemes are capable of this sort of over- and under-shoot of tracer values. MPDATA is designed to be better for this - have you tried it?

[wilkin]

Looking at your parameters in bohai.in you have a turbulent Prandtl number (the ratio of eddy viscosity VISC2 to eddy diffusivity TNU2) of Pr = 2.5e-03. The Reynolds analogy for turbulent flow argues that Pr ~ o(1) while empirically it tends to be slightly (but only slightly) < 1.

It is possible that your viscosity is so weak that short wavelengths in the velocity field are contributing to overshooting of the tracer field. Try adopting more realistic values of VISC2 and TNU2 appropriate to the resolution of your grid.

Also, I would caution against using tracer mixing strictly along s-coordinate surfaces (you have #define MIX_S_TS) when there is sloping bathymetry. #define MIX_GEO_TS will reduce spurious diapycnal mixing where density surfaces cross s-coordinate surfaces.

[rduran]

Do your vshapes add up to one?

I got very weird salinities at a river mouth with a vshape that did not.

Another thing to check in your trial and error list.