Answers from Matt Harrison:
The tangential velocity is difficult to control . You can try "OBLIQUE/OBLIQUE_TAN or ORLANSKI/ORLANSKI_TAN which do so via radiation. The TRACER_RESERVOIR_LENGTH_SCALE_(IN/OUT) parameters are meant to control the "memory" along the boundaries of the reservoirs, which serve as an outer b.c. for the computation of normal tracer and density gradients and hence the geostrophically balanced tangential flow at the boundaries. I recommend trying these parameters as a starting point before resorting to interior damping. In the development of the regional code, we were determined to go as far as we could to avoid modifications to the interior values. This is a philosophy which we want to try to continue to adhere to.
That said, you can enable the interior tracer nudging using the parameters described in the code here .
I have been experimenting with using enhanced Laplacian viscosity near the boundaries via a spatially-varying background . This seems to help with model stability and using it may improve the interior solution as well. This is something I am continuing to investigate.
Other discussions are also welcome.
The tangential velocity is difficult to control . You can try "OBLIQUE/OBLIQUE_TAN or ORLANSKI/ORLANSKI_TAN which do so via radiation. The TRACER_RESERVOIR_LENGTH_SCALE_(IN/OUT) parameters are meant to control the "memory" along the boundaries of the reservoirs, which serve as an outer b.c. for the computation of normal tracer and density gradients and hence the geostrophically balanced tangential flow at the boundaries. I recommend trying these parameters as a starting point before resorting to interior damping. In the development of the regional code, we were determined to go as far as we could to avoid modifications to the interior values. This is a philosophy which we want to try to continue to adhere to.
That said, you can enable the interior tracer nudging using the parameters described in the code here .
I have been experimenting with using enhanced Laplacian viscosity near the boundaries via a spatially-varying background . This seems to help with model stability and using it may improve the interior solution as well. This is something I am continuing to investigate.
Other discussions are also welcome.
kshedstrom@alaska_edu
Member
You can also have NUDGED_TAN to go along with your radiated ORLANSKI_TAN or OBLIQUE_TAN. Or you can pick SIMPLE_TAN to clamp to known values.