Tropical Biases in GFDL atmospheric and coupled models
Where are we?
How did we get there?
Where are we going?
(GAMDT/LMDT/OMDT/CMDT)
• Where are we? (cm2a11o2) Atmosphere (in AMIP mode)
• Mean
• Regressions against ENSO Coupled
• Mean
• ENSO Variability
AM2/LM2:comparison to other models
Differences in annual mean precipitation from CMAP (Xie-Arkin)
MAM precip (mm/day)
Low cloud amount in JJA
Observations
Nino3 regressions in am2p11 AMIP integrations
Zonal stress – all seasons Precip- all seasons
ECMWF
am2p11
difference
Regression of Nino3 vs Z200(m) DJF: 10 runs 1950-2000
NH SH
Cm2a11o2 SST bias
Coupled model cm2a11o2Annual precip MAM precip
Wavelet analysis of Nino3 SST anomalies
El Nino variability in AM2 coupled to OM2
Nino3 regressions in am2p11 AMIP and cm2
Zonal stress – all seasons Precip- all seasons
ECMWF
am2p11
cm2
How did we get here?
AM2p2 AM2p
6
AM2p8 AM2p1
0
AM2p12
OM2– MOM4 (fully integrated into FMS)– Tripolar grid; – 2 deg Mercator south of 60N outside of
equatorial zone– 50 vertical levels – 10m vertical resolution near surface- 2/3 degree meridional resolution at equator– Explicit free surface– Uniform GM thickness diffusion– Prescribed, spatially varying “color” (solar radiation penetration depth)
AM2p11•B-grid Dynamical Core for am2p11 (Wyman):
–2.5° lon X 2.0° lat X 18 vertical levels
–top at ~ 30 km
–Split time stepping: 200, 600, and 1800 seconds for gravity waves, advection, and all
physics except radiation
–Piecewise-parabolic finite volume vertical advection of tracers (S. J. Lin)
–Finite volume form of pressure gradient force calculation (S. J. Lin)
T42
N90 = 1 degree
Tropical SST bias
cm2_a11_o2(1K )
cm2_a10_o2old pressure gradient
(1K)
T : 11 –10(0.5K)
am2p11 (5% contour) am2p10 (5% contour)
Effect of pressure gradient form on % Low cloud
am2p11 –am2p102% contour
Changes in oceanic heating due to pressure gradient
4 w/m2 contour
total
sensible + evap radiation
p11 – p10
AM2p11
• Prognostic cloud scheme (Klein)– 3 prognostic cloud tracers which are
advected and diffused: cloud liquid, cloud ice and cloud fraction
– Cloud fraction parameterization from Tiedtke (1993) as is used in ECMWF model
– Cloud microphysics from Rotstayn (1997) as is used in CSIRO model
– Precipitation macrophysics (large-scale rain and snow areas) from Jakob and Klein (2000)
AM2p11• Relaxed Arakawa Schubert (RAS) Convection
(Moorthi/Suarez) (Sirutis)– Ensemble of cumulus updrafts – no downdrafts– Specified precipitation efficiencies as a function of
the depth of the updrafts. Non-precipitated fraction is a source of condensate for cloud scheme
– Closure: relax cloud work function to a threshold value with a timescale dependent upon cloud type
– Simple diffusive cumulus momentum transport (Held)
– For deep convection, a minimum bound on lateral entrainment rates is imposed (Tokioka modification)
cumulus momentum transport
• Parameterize cumulus momentum transport (CMT) as a simple vertical diffusion of horizontal momentum where convection occurs
• Km ~ Mcz ~ acwupz
AM2p11: without CMT
AM2p11: with CMT
El Nino variability in zonal wind stress in AMIP integrations
AM2p11: with CMT
AM2p11: without CMT
zonal wind stress regressed on NINO3 in AMIP integrations
AM2p11AM2p11: w/o CMT AM2p11:
w/o cmt and w/o Tokioka
Wavelet analysis of Nino3 SST anomalies
AM2p11 without CMT
biannual peakPhase locked to seasons
Moist static energy budget changes in evaporation more important than
changes in radiative fluxes(clouds)
In ENSO regressions, changes in stress larger than changes in precip diffusion directly affects vorticity budget
Dominance of baroclinic mode precip increases strength of
low level damping
precip(divergence)
vorticity evaporation
Dominant feedback determining response to cumulus momentum transport?
cmt
AM2p11
• Planetary Boundary Layer– Mellor-Yamada (1982) 2.5 order dry
parameterization with prognostic turbulent kinetic energy
– “Gustiness” enhancement to wind speed used in surface flux calculations (Beljaars 1995)
– Oceanic roughness lengths enhanced at low wind speed (Beljaars 1995)
• Gravity Wave Drag (Stern)– Orographic drag from Stern and Pierrehumbert
In Development:
• New boundary layer turbulence parameteriziation based upon UK Meteorological Office PBL (Klein)– Stability based upon moist thermodynamics– K-profile mixing for surface driven and cloud top
radiatively driven mixing– Explicit entrainment parameterization based
upon Large-eddy simulations and observations– Enhanced momentum drag in regions of
variable orography (“orographic roughness”)– Enhanced mixing in very stable conditions– 6 more vertical levels in the PBL – 9 levels
beneath 1500 m
Trade inversion height (annual mean)
L18 – Mellor-Yamada
L24 – Mellor-Yamada
L24 – UKMO PBL
meters
Low cloud amount in JJA
Observations
New PBL parameterization
latent heat fluxes (w/m2, colors)
2 meter relative humidity
(%, contours)
Changes in latent heat flux and relative humidity(annual mean) due to new PBL
Changes in precipitation (annual mean)
In Development: AM3
• New convection scheme to replace RAS (Donner et al. 2002)– Cloud microphysics in an ensemble of
updrafts with prognostic vertical velocity– Parameterized heating from a mesoscale
anvil based upon Leary and Houze (1980) observations
– Radiative impact of convective towers and mesoscale anvils included
– Convective and mesoscale downdrafts
In Development: AM3
• Enhanced stratosphere (Wilson)– Raise the model top and add 5 to 10 more
vertical levels– Replace Pierrehumbert-Stern orographic
gravity wave drag with anisotropic gravity wave drag parameterization from Garner
– Add convectively generated gravity waves from the parameterization of Alexander and Dunkerton (NWRA)
Thanks to:
Steve Klein
Paul Kushner
Tony Rosati
Andrew Wittenberg