Ocean Modeling Requirements for Decadal-to-Centennial Climate

1
Ocean Modeling Requirementsfor
Decadal-to-Centennial Climate

• Robert Hallberg
• NOAA/GFDL

2
Ocean Modeling Considerationsfor
Decadal-Centennial Climate

• Dec-Cen Climate is Operational, but not in Real
  Time.
• Ocean models are fully global and must include
  sea-ice.
• Models run for O(1000-year) timescales.
• Wall-clock model speeds are 103-104 times real
  time.
• Ocean biases after a spinup of hundreds of years
  must be acceptably small.
• Data assimilation during the run or unphysical
  damping are unacceptable in Dec-Cen climate
  models.
• Precise initial values are often relatively
  unimportant.
• The long timescales mean that a wide range of
  physical processes must be represented as
  credibly as possible.

3
Ocean Climate Models must be Conservative.

• Non-conservation leads to drift and uncertainty.
• Tolerances for non-conservation
• Mass small compared to sea-level rise.
• 20th Century rise 10 cm.
• Err ltlt 10-3 m cen-1 MOM, GOLD 10-7 m cen-1
• Heat small compared to anthropogenic forcing
• Anomalous Radiative Forcing 4 W m-2 at CO2
  doubling
• Err ltlt 0.1 W m-2 MOM, GOLD 3x10-6 W m-2
• Total Salt small compared to sea-level rise
  dilution?
• Dilution of Salinity 10-3 PSU cen-1
• Err ltlt 10-4 PSU cen-1 MOM, GOLD 3x10-9 PSU
  cen-1
• With care, these tolerances can be achieved in
  all classes of ocean models. (MOM Z-coord., GOLD
  r-coord.)

4
Key Metrics of a Global Ocean Climate Model

• SST biases at equilibrium
• ENSO statistics (Amplitude Frequency) at
  equilibrium
• Tropical ocean circulation and watermass
  structure at equilibrium.
• Stability and strength of meridional overturning
  circulation and gyre circulations.
• (Important for meridional heat transport
  sea-ice distribution)
• Equilibrium watermass properties, rates and
  processes of formation destruction.
• (Important for storage of heat, carbon, etc.)
• Spurious diapycnal mixing ltlt physical Kd 10-5
  to 10-6 m2 s-1
• Overflows entraining gravity currents
• Mode-water formation processes

5
100-year-mean SST Biases in GFDLs CM2.1 Coupled
Climate Model
6
Equatorial Pacific Velocities and Temperatures
after 500 years in CORE simulations with 7 models
Drifts accumulate over 50 years. Short runs may
not be indicative of equilibrium climate.
(Figs. from Griffies et al., Ocean Mod., Sub.)
7
Challenges in Global Ocean Climate Modeling

• Increasing resolution to admit mesoscale eddies

8
Role of eddies in Southern Ocean Dynamics

• Eddies alter the sensitivity to forcing changes
  of the Southern Ocean overturning circulation and
  the resultant ventilation of the interior ocean.

Hallberg Gnanadesikan, JPO 2006
9
Challenges in Global Ocean Climate Modeling

• Increasing resolution to admit mesoscale eddies
• Dramatic increase in model cost, reduction in
  speed
• Changes in required parameterizations
• Numerical constraints (e.g. on spurious diapycnal
  mixing) become harder to satisfy.
• Regional climate impacts

10
SST in a Prototype 1/8 Global Ocean Model
Upwelling zone
California Current
11 Box Contemporary Climate Model Resolution
11
Challenges in Global Ocean Climate Modeling

• Increasing resolution to admit mesoscale eddies
• Dramatic increase in model cost, reduction in
  speed
• Changes in required parameterizations
• Numerical constraints (e.g. on spurious diapycnal
  mixing) become harder to satisfy.
• Regional climate impacts
• Dominant scales of many ecosystems much smaller
  than well-resolved by global physical climate
  models.
• Two-way nesting is probably needed.

12
The Existing GFDL Suite of Ocean Models

• GFDL/Princeton has leading developers of each
  widely used class of large-scale ocean models.
• MOM B-grid Z-coordinate model
• MITgcm C-grid Z-coordinate model with
  nonhydrostatic capabilities
• HIM C-grid isopycnal coordinate model
• POM Princeton AOS programs C-grid sigma
  coordinate model
• The Generalized Ocean Layer Dynamics (GOLD)
  ocean model unites the GFDL efforts.
• GOLD is an ocean modeling system that combines
  the capabilities of GFDLs MOM and HIM and the
  MITgcm ocean models into a single flexible
  code-base.
• GOLD numerics will be suitable for studying
  climate, but with demonstrated proficiency for a
  variety of other applications (tides,
  nonhydrostatic mixing, etc.)
• GOLD will include significant nesting and data
  assimilation capabilities.
• GFDL ocean models focus on integrity for climate
  applications.

Developers at GFDL/Princeton
13
GOLD and HYCOM

• GOLD is structurally compatible with HYCOM.
• The GOLD dynamic core overall structure are
  derived from the isopycnal coordinate model HIM.
  Similar issues must be addressed as for HYCOM.
• GOLD has necessary qualities for long-term
  climate studies.
• Conservation-to-roundoff of heat, salt, mass, and
  tracers.
• Accurate with a fully nonlinear equation of
  state.
• Robust parameterizations of small-scale
  processes.
• Rotated diffusion tensor minimized spurious
  diapycnal mixing.
• An NSF-funded project, including GOLD, HYCOM and
  ROMS, is exploring standardization across ocean
  models.
• Collaborations are welcomed, provided they do not
  disrupt GFDLs primary mission in long-term
  climate studies.

GOLD
?
z
z/z/p/p
HIM
Poseidon
HyCOM
MOM
MITgcm
POP
14
Potential for One NOAA Ocean Model Addressing
Real-time Operational Global Dec-Cen Climate

• What is a model?
• A single model configuration.
• Unlikely, due to mismatch in timescales, model
  speed, emphasis on initial conditions vs. bias.
• A shared model code-base.
• Repository of best theories/techniques for both
  real-time operations and climate. Natural
  synergies could emerge.
• Will take considerable work and resources.
• Important not to disrupt the on-going operational
  requirements for real-time forecasts or climate.
• Establishing common model interfaces would be a
  natural first step.
• Some ocean forecasting applications (e.g.,
  tsunami warnings) are so different that they are
  unlikely ever to use the same model.

15
(No Transcript)
16
High-Resolution Ocean Modeling the Ocean
Research Priorities Plan

• Numerical ocean models are recognized as crucial
  tools for a wide range of questions (ORPP, p.
  47).
• High resolution global ocean modeling is central
  to the ORPPs Theme 4 The Oceans Role in
  Climate and the near-term priority Assessing
  Meridional Overturning Circulation Variability
• Resolving the critical processes is crucial for
  reducing the uncertainty in projections.
• The insight gained from the Global Ocean
  Observing System is maximized when the data is
  interpreted in the context of numerical models.
• High resolution global ocean simulations are
  valuable in support of the other 5 Themes.
• Provide consistent boundary conditions for
  ultrafine regional models.
• Provide estimates of variability in the
  conditions faced by ecosystems, marine
  operations, or processes affecting human health.
• NOAA/GFDL is taking the lead in addressing the
  ORPP call for a coherent, comprehensive global
  modeling capability, addressing Theme 4 in
  particular.

17
Mean Salinity Drifts in 7 CORE Forced Candidate
Ocean Climate models (Griffies et al., 2008)
Notes Sea-ice growth is the sole cause of
salinity drift in some models. Some
models balance salinity restoring, others do not.
18
Frontal Dynamics and Resolution

• Upwelling jets fronts require higher
  resolutions than current ocean climate models.
• With steady forcing all variability is due to
  ocean dynamics.

Hallberg Gnanadesikan, JPO 2006
19
Strengths and Weaknesses of Terrain-following
Coordinate Models

• (Issues for global climate application addressed
  in detail by G. Danabasoglu later.)
• Strengths
• Topography is represented very simply and
  accurately
• Easy to enhance resolution near surface.
• Lots of experience with atmospheric modeling to
  draw upon.
• Traditional Weaknesses
• Pressure gradient errors are a persistent
  problem.
• Errors are reduced with better numerics (e.g.,
  Shchepetkin McWilliams, 2003)
• Gentle slopes (smoothed topography) must be used
  for consistency
• Traditional requirement for stability (Beckman
  Haidvogel, 1993)
• ROMS requirement (Shchepetkin, pers. comm)
• Spurious diapycnal mixing due to advection may be
  very large. (Same issue as Z-coord.)
• Diffusion tensors may be especially difficult to
  rotate into the neutral direction.
• Strongly slopes require larger vertical stencil
  for the isoneutral-diffusion operator.

20
Resolution requirements for avoiding numerical
entrainment in descending gravity currents.

• Z-coordinate
• Require that
• AND
• to avoid numerical entrainment.
• (Winton, et al., JPO 1998)
• Many suggested solutions for Z-coordinate models
  
• "Plumbing" parameterization of downslope flow
• Beckman Doscher (JPO 1997), Campin Goose
  (Tellus 1999).
• Adding a separate, resolved, terrain-following
  boundary layer
• Gnanadesikan (1998), Killworth Edwards (JPO
  1999), Song Chao (JAOT 2000).
• Add a nested high-resolution model in key
  locations?
• Sigma-coordinate Avoiding entrainment requires
  that
• But hydrostatic consistency requires
• Isopycnal-coordinate Numerical entrainment is
  not an issue - BUT
• If resolution is inadequate, no entrainment can
  occur. Need

21
(No Transcript)
22
Horizontal Resolution (in km) Required to Permit
50m Vertical Resolution at Bottom
23
Horizontal Resolution (in km) Required to Permit
50m Vertical Resolution at Bottom
24
Horizontal Resolution (in km) Required to Permit
50m Vertical Resolution at Bottom