Salinity larger than specified at the boundary

I have an EFDC model where I have two open boundaries and three river boundaries. On one open boundary, I have specified 34 ppt and on another open boundary I have 28 ppt. However, within the domain during the simulation period, I found that some cells have salinity of 35 ppt. Is is possible to have salinity greater than the maximum value of salinity specified at the boundary in the model grids ? If it is not possible what might be the cause of having larger salinity ?


Regarding this issue, it is possible your model is experiencing overshoot. To correct this you should enable the anti-diffusion correction, which is accessed from Active Modules Tab | Sub-model Computational Options | Anti-Diffusion Correction drop down | Flux Limitation drop down. Though this is computationally more intensive, it reduces overshoot. You may also want to reduce your time step to help avoid this problem.


I have already tried anti-diffusion correction and flux limitation in my drop down and the time step in my model is 10 second which is not a whole lot. I have run the model with 5 second time step to see if I get any improved results.
Do you have any other suggestions ?

I believe when you have high bottom elevation gradients, these types of error might occur. What do you think ? Would you please tell how does initial bathymetry smoothing work ?

It is possible that at the open boundaries there can be a flux problem if the bottom elevations of the two cells that make up the actual boundary are not the same ie if you have a north boundary then the north cell and the one directly to the south should have the same bottom elevation. If these cells have different bottom elevations it can lead to numerical errors in the model.

Thank you EE_Development_Team for your kind reply. I didn’t know about that. It is not written anywhere that you should use the same bottom elevation for the cell just upstream where you have specified the open boundary. It seems to greatly reduce that numerical error.