CLIVARCCSM Tropical Biases Workshop May 2830, 2003, GFDL

1
CLIVAR/CCSM Tropical Biases Workshop(May 28-30,
2003, GFDL)

• Chris Bretherton
• University of Washington
• ( Thanks to Ping Chang, Leo Donner, Jim Hack,
  Meghan Cronin, Peter Gent , Ken Sperber, and
  others)

2
Workshop Themes

• Status of biases in major coupled models
• Double ITCZ and E Pacific
• ENSO and seasonal variability
• Are observational uncertainties pacing progress?
• Are idealized and reduced models helping?

3
Participating Centers and Models

• NCAR CCSM
• GFDL CM2
• CSIRO
• HadGEM1
• MPI
• BMRC

• INGV
• CCSR, U. Tokyo
• FRSGC, Earth Simulator
• NSIPP (NASA Goddard)
• COLA
• IPRC, U. Hawaii

Observations, Process Modeling, Diagnostics
Obs Cronin, Kessler, Raymond, Mapes, L. Yu,
CRM/Idealized models Grabowski, Khairoutdinov,
Sobel Diagnostics NOAA/CDC, etc.
4
Observations
Coupled Models

• Double ITCZ bias
• in ALL models!

Sperber et al. (2003)
5
AGCMs forced by observed SST do not consistently
have double ITCZ
6
Coupled Model SST biases too cold tongue warm
subtropical NE, SE Pac, SE Atl
7
Status of Biases in Major Coupled Models

• Recent development cycles of major models have
  not reduced double ITCZ and too-cold tongue
  biases.
• E Pacific biases have reduced in models with
  realistic stratocumulus cloud forcing.
• However, warm biases off the S American and S
  African coast in coupled models persist.
• No smoking gun in model physics.

8
Sensitivity studies Results model-dependent

• AGCM parameterization sensitivities
  model-dependent
• cumulus momentum transport
• convective rainfall evaporation
• super-parameterization
• OGCM sensitivities also not smoking guns
• Dependence of penetrating solar radiation on
  upwelling reduces too-cold tongue but not
  enough.
• Artificially changing surface winds ameliorates
  too-cold tongue in some models, but AGCMs dont
  make consistent wind-stress errors.
• Effects of higher resolution also mixed
• Does not remove major systematic biases.

9
(No Transcript)
10
Possible causes of biases

• Deep convective feedbacks with PBL, SST wrong--gt
  double ITCZ
• Bad coastal winds (continental effect,
  resolution,) --gt bad coastal upwelling --gt
  stratus region too warm --gt double ITCZ
• Solar and longwave cloud forcing wrong --gt
  stratus region too warm --gt double ITCZ
• Ocean mixing wrong --gt cold tongue too cold --gt
  double ITCZ
• Air-sea interaction wrong --gt PBL not properly
  stabilized over cold tongue --gt trades
  upwelling too strong --gt cold tongue too cold
• Biology not included --gt penetrative radiation
  too large --gt cold tongue too cold

(Thanks, Meghan)
11
What does EPIC have to contribute?

• EPIC-inspired idea for warm bias in SE/NE Pacific
  SSTs GCMs have stratofogulus in the EPIC
  region so maybe high surface relative humidity
  leads to insufficient ocean cooling from latent
  heat fluxes.

Mean profiles at 20S 85 W 16-21 Oct 2001 (obs,
NCEP, ECMWF)
Oct AGCM climatology (CAM2, AM2)
12
CAM2 surface energy budget compared with WHOI
stratus buoy
Excessive SW heating
Too little heat into ocean (yet CCSM2 has warm
SST bias!)
Excessive LW cooling
SHFLHF surprisingly good
13
EPIC 95 W region does not show systematic
atmospheric, SST biases in coupled models

• This means coupled modelers probably wont come
  running for the 95W datasets (unless their
  subsurface oceanography looks awful).
• Nevertheless, there are worthy goals for us to
  aim for
• - a well-prepared TAO-EPIC dataset
• - a nice 95W composite atmospheric cross-section
  
• - useful, easily tested constraints on ITCZ
  convection
• Combined with applicable insights into component
  physical processes, these may allow EPIC to
  improve climate models so as to reduce these
  biases.