Hydrodynamic model troubleshooting - gher-uliege/Documentation GitHub Wiki

If a hydrodynamical model blows up, here is a list what can be done to investigate the problem:

• visualize the model solution several time steps before the blow-up
• verify model stability conditions
• check if the model is stable with a reduced time step (by a factor 2, 5, 10, 20, ...). If the model blows up at the same physical time, have special look at the forcing fields
• check for aberrant values in initial condition and forcing fields
• an increased viscosity and diffusivity can help to stabilize but one should try to avoid to mask an other problem
• check minimum depth of the bathymetry (unless your model supports "dry" grid cells)
• check the steepness of the bathymetry for a terrain following coordinate model ; if using z-partial step coordinates, check the minimum partial step height
• use a model spin-up to dissipate surface gravity waves. For example use a very small time step (and possibly an enhanced viscosity) and use the time-averaged simulation or the last time step as your new initial condition (and then use normal values for your time step and viscosity.
• if an unrealistic current renders the model unstable, check if the vertical profile is barotropic or baroclinic

Open boundary condition

If you have an open boundary condition:

• Verify that the preparation for the boundary conditions file is correct
• Parameters/settings that can be adjusted:
  • Mathematical formulation of the boundary condition
  • Relaxation time scales
  • Viscosity and diffusivity near the boundary
  • Position of the open boundary (a open boundary near steep topography can be problematic)
• A useful diagnostic parameter can be the vertical velocity. Strong vertical currents can be a sign of a problem.

(Please add additional ideas for troubleshooting an ocean model)