Transcript Report to WGOMD on GFDL and Co. ocean modelling activities

Report to WGOMD on GFDL Ocean
Modelling Activities 2004-2005
Stephen Griffies
NOAA/GFDL (and CSIRO)
•IPCC AR4 activities
•Model developments
IPCC activities
• Completed development of AR4 coupled
climate model in 2004, and submitted
simulations to PCMDI 2004/2005.
• ~1 degree ocean (mom4) with 50 levels and 1/3
degree at equator. Described at previous
meetings.
• Numerous studies now being conducted to
document the model’s design and simulation
characteristics. Will take years to fully evaluate.
GFDL Coupled Model results
References
Ocean Model
Plans for GFDL ocean model development
A briefing to WGOMD
• Inform WGOMD of plans for MOM4 over next 6 months
•Speculate on 3-5 years research/development goals and
applications involving ocean models.
MOM4 as of November 2005
• Four public releases
MOM4p0a Jan2004
MOM4p0b Mar2004
MOM4p0c Oct2004
MOM4p0d May2005
• Roughly 300 registered users from 30 countries using
~35 computational platforms. They represent about
1200 scientists, engineers, and programmers using the
code and simulation results for research and
development.
MOM4 user statistics
Vertical lines
are intermediate
code releases:
mom4p0a
mom4p0b
mom4p0c
mom4p0d
GFDL Applications of MOM4
• IPCC Global climate change modelling:
– Ocean component to the GFDL AR4 climate change models.
– Developed largely for global climate modeling applications.
– ~50 GFDL scientists directly involved with this research and
development.
• Earth system modelling:
– interactive land, atmosphere, ocean biogeochemistry and
ecosystems
– ~30 scientists at GFDL and Princeton University
• Global and regional process studies:
–
–
–
–
–
paleo-oceanography
idealized climate change simulations
thermohaline shutdown
physical process studies
~20 scientists, visiting researchers, post-docs, graduate
students
Gravity current-entrainment CPT
Eddy mixed layer interactions CPT
MOM4p1: vertical coordinate features
free surface z-model: mom4p0
•Partial step topography
•Trivial pressure gradient
errors
•Decades of experience
•Well known limitations
•Irregular and variable
computational domain
(i.e., land/sea masks and
vanishing surface layer)
  ( z   ) /( H   )
•Terrain following σ-model
•Smooth topography
•Regular computational
domain (no land/sea
masks)
•Time independent
computational domain
(-1 < sigma < 0)
•Pressure gradient errors:
requires topography filters
•Difficult neutral physics
implementation: not
commonly done in sigmamodels
z*  H ( z   ) /( H   )
•Irregular computational domain
(i.e., land/sea masks needed)
•Time independent computational
domain (-H < z* < 0): no vanishing
layers.
•Negligible pressure gradient errors
since isosurfaces are quasihorizontal. Correspondingly, can
use the same neutral physics
technology as in z-models.
Evolution of GFDL ocean codes
Evolution is in response to many inputs
– New applications:
• Refined resolution climate models
• Biogeochemistry and ecosystem applications
• Earth system modeling
• Coastal impacts of climate change
• Non-hydrostatic processes at very refined resolutions
– Enhanced features:
• physical parameterizations (e.g., mixed layers, mesoscale eddies)
• algorithm fundamentals (e.g., time stepping, vertical coordinates)
• better understanding of the ocean (e.g., equation formulations)
– Computational efficiency and platform portability
– Input from the international user communities (HIM, MITgcm, MOM4)
Main developers: Alistair Adcroft, Bob Hallberg, Steve Griffies
Evolution Path
• MOM4p1: ~March 2006 with rudimentary generalized vertical
coordinate features to expand mom4 applications.
• HIM-Fortran: Hallberg Isopycnal Model, publicly supported
within GFDL Flexible Modeling System (FMS); GFDL
development now aimed at coupled simulations to compare w/
mom4-based coupled model.
• Research: Merge three fundamental perspectives
– non-hydrostatic z-modeling from MIT (Adcroft)
– hydrostatic isopycnal modeling from HIM (Hallberg)
– Global ocean climate modeling from MOM4 (Griffies)
• Key NOAA application: climate impacts on coasts
– Global “BackBone Model” ~10 km with nest to ~1 km
– Tides, wave breaking, storm surge, sediment transport, etc.
• ~2008-2010 for first public code release
Horizontal grids: nesting and cubed sphere
• Multiple 2-way nested regions
• Mass and tracer conservation: Most nesting implementations in ocean and
atmospheric models are non-conservative
• Time sub-cycling: coarse region not constrained by time step used in fine
region. Essential for economical global models with nests.
• Envision applications in areas such as
– global climate models: boundaries, choke points, etc.
– Regional modeling with nests to estuary scale
– Coastal biogeochemistry and ecosystems
• Present development
– general grid description
– tools for parallel computing and coupled modeling
– analysis/visualization tools
– shallow water test cases
• Cubed Sphere
– technology from MITgcm
– Also envisioned for finite volume atmosphere model
Main Applications
• Coastal impacts of climate change
• Earth System Modelling w/ eddying
simulations (~1/3 to 1/4 degree mercator with
twoway nesting in selected critical regions)
• NOAA “BackBone” model, with ~1/10 global
to be nested with finer grids in certain coastal
areas. For use by many projects within NOAA.
• nonHydrostatic process studies and very refined
coastal and estuary simulations
• University PI and student research and
education
HIM + MIT + MOM = ???
~20 m
~100 m
~1 km
NON-HYDROSTATIC
~10 km
~100 km
~1000 km
HYDROSTATIC
Unified GFDL Ocean Code
• Bring together our understanding of the ocean and how
to simulate a wide range of scales.
• Various algorithms with stepwise evolution involving
suites of applications to test methods and flesh out
favourable approaches.
• This effort is a major research and development project,
presently in its early stages at GFDL. Much research
remains to determine particulars of algorithms.
• Various efforts (e.g., HOME) to develop a US
community model have failed to garner funds. GFDL is
committed to this project using in-house resources, and
will involve outside collaborators as best as possible.