HOW AESOP GOT A SUNTANA fractured fairy tale

(with apologies to the producers of the Rocky and Bullwinkle show)

The cast of this episode:

Oliver Fringer and Bob StreetEnvironmental Fluid Mechanics Laboratory at Stanford University

Synopsis:

•The context and collaborators

•The tools and some pretty pictures

•Some thoughts on our work plan

8 March 2005

The context: we are participating in AESOP and in NLIWI.

Our overall goals include collaboration with UCLA to support the continuing development of our models, SUNTANS and ROMS; their coupling; and the scientific issues to be set forth by the ONR NLIWI and AESOP DRI experiments.

Our collaborators:Jim McWilliams, Sasha Shchepetkin, Yulia Kanarska,

UCLA

Our Tools

• SUNTANS - a free-surface NS simulation code for the coastal ocean. Currently implemented for Monterey Bay applications.

• LES - tools for the large-eddy simulation of the flows, including subfilter models that parameterize the unresolved motions.

SUNTANS Overview

• SUNTANS:– Stanford– Unstructured– Nonhydrostatic– Terrain-following– Adaptive (not yet)– Navier-Stokes– Simulator

• Parallel computing • Large-eddy simulation

High-resolution simulations must be nonhydrostatic

Doman size: 0.8 m by 0.1 m (grid: 400 by 100)

Hydrostatic

Nonhydrostatic

Simulation results using 500 m grid + MY2.5

north-south velocity contours. Max velocity = 5 cm/s

Along-canyon generation sites

Comparison of results to measurementsOf Petruncio, et al. (1998,2002)

Cross-canyon generation sites

Comparison of results to measurementsOf Lien and Gregg (2001)

LES Overview

• Work based on decomposition of flow into resolved and subfilter motions by spatial filtering. Subfilter scales are further separated into – resolved subfilter scales [computational grid size is at least a

factor of 2 smaller than spatial filter size]. Highly accurate representation possible; facilitates energy transfer to and from large scales.

– subgrid scales. These must be modeled.

• These ideas have been proven and are being introduced in to SUNTANS. SUNTANS currently employs a RANS approach with, e.g., Mellor-Yamada 2.5 closure.

RSFS and SGS scale partitioning

• Resolved scales– Well-resolved

• Resolved subfilter scales (RSFS)

– Can be reconstructed

• Subgrid scales (SGS)– Must be modeled

• Numerical Errors (NE)– Limit reconstruction

Improvements near the wall as applied to a neutral boundary layer in the atmosphere.

Dynamic reconstruction + Near-wall model

Smagorinsky

Log law

Dynamic eddy viscosity+ Near-wall model

Thoughts on contributions to DRI• High-resolution nonhydrostatic modeling of Monterey Bay (up to 10

m resolution) can be made available, i.e., simulation of scales from regional to those at which mixing takes place.– Simulated cruise tracks

• Prediction and refinement via direct comparison to actual cruise tracks

– Basis for evaluation of coarser resolution simulations and their parameterizations.

– Boundary forcing for domains of simulators working on small scales.

• Evaluation of RANS and LES turbulence models [ours and those of others] at submesoscale [100 m and up resolutions].

• Nesting of SUNTANS into ROMS and follow-up simulations to assess value of high resolution in specific areas and how that impacts submesoscale parameterizations.