Towards Dynamically Consistent Boundary forcing
Jeroen Molemaker (UCLA)Evan Mason (ULPGC)
Sasha Shchepetkin (UCLA)Francois Colat (UCLA)
One way nesting
• Obvious limitations:– One way will never be two way!
The round peg and the square hole
Testing lab: Canary Current system
Forcing at side boundaries
• Forcing of the outermost grid.– Something ROMS
• Forcing of (off line) one way nested grids– ROMS ROMS
Forcing the outermost grid
• Output from other (global) models
• Observations (such as World Ocean Atlas)
Observation based forcing
• World Ocean Atlas 2005 T, S monthly climatology
• Absolute SSH (Rio, 2005)– Now, annual mean, but we should include at
least monthly averaged perturbations
Our ‘truth standard’
Drifter data SSH variance
World Ocean Atlas 2005• Using level of no motion (1300 m)
World Ocean Atlas + absolute SSH
World Ocean Atlas + absolute SSH
‘crude’ Ekman layer transport
Assessing large scale, slow dynamics
-Subtract geostrophic flow
- Scale vertical profiles with mixed layer depth, f and wind stress vector:
z’ = z/Hbl, (u’) = (u Hbl f)/
Roms’ Ekman spiral
World Ocean Atlas + SSH ‘spiral’ Ekman transport
Impervious to baroclinic structure?
Patrick style Ekman KPP spiral Ekman
How well did we do?
Data:
Model:
Drifters SSH variance
ROMS ROMS
• Expected consistency much higher– Only regime transition is unavoidable
• Starting point:– Methods as existing in ROMS tools (Pierrick Penven, Patrick Marchesiello…. Many others)
ROMS ROMS
• Sigma z-levels Sigma coordinate
• No boundary mass flux correction
Or not so reasonable?
Horizontal-vertical interpolation
Matching boundary mass flux with parent grid
Matching grids at the boundary
Summary
• One way nesting can be good when:
– Solutions have consistent dynamics• roms roms• No enormous jumps in resolution
– Interpolation does not destroy said consistency