Convectively Coupled Kelvin Waves
and the MJO in a Hierarchy of GCMs
DARGAN M. W. FRIERSON
UNIVERSITY OF WASHINGTON,
DEPARTMENT OF ATMOSPHERIC SCIENCES
COLLABORATORS: MARSHALL STONER,
DAEHYUN KIM, JIALIN LIN, IN-SIK KANG,
MYONG-IN LEE, ADAM SOBEL, ERIC
MALONEY, GILLES BELLON
Outline
 What sets speed/structure of convectively coupled
equatorial waves?


In a simplified GCM
Modeling work with SNU group
 What is required to generate a MJO-like structure?
 AM2 model work w/ Sobel, Maloney & Bellon
 Master’s thesis of Marshall Stoner
Convectively Coupled Equatorial Waves
 What sets speed?
 Moist 1st baroclinic mode? (gross moist stability: Neelin,
Emanuel, etc)
 Dry 2nd baroclinic mode? (Mapes, Majda, etc)
 Observations show clear 2nd baroclinic structure
(Kiladis et al 2009)
CCKWs in a Simplified GCM
 Convectively coupled Kelvin waves
(CCKWs)dominate tropical variability in a simplified
GCM
Unfiltered Hovmoller diagram of
precipitation at the equator
In this model, gross moist stability
controls the speed of these waves
Model of Frierson, Held & Zurita-Gotor (2006)
Plot from Frierson (2007)
Convectively coupled Kelvin waves
 GMS reduction leads to slower convectively coupled
waves:
GMS = 6.9 K
GMS = 3.9 K
GMS = 3.0 K
Ratio of grid-scale to convective (simplified Betts-Miller) precipitation sets the GMS
See Frierson (2007) for more detail
Simplified Moist GCM CCKWs
 These CCKWs are powered by evaporation-wind
feedback

Likely not true in reality in Indian Ocean…
 Vertical structure is purely first-baroclinic mode
 Unrealistic…
Composited pressure velocity
See Frierson (2007b) for more detail
Longitude
Equatorial Waves in a Full GCM
 Experiments with SNU atmospheric GCM
 Run over observed SSTs, realistic geography
 Simplified Arakawa-Schubert (SAS) and Kuo convection
schemes
 Varying strength of convective trigger:
Tokioka entrainment limiter for SAS
 Higher Tokioka parameter => least entraining plumes are
eliminated
 Moisture threshold for Kuo
 From always triggering convection to 95% RH required

See Lin, Lee, Kim, Kang and Fri. (2008, J Clim) & Fri. et al (submitted) for more
Moist Static Energy
 Vertical profile of MSE in the North West Pacific
ITCZ for SAS simulations:
Tokioka values:

Higher entrainment => harder to warm upper
troposphere
 Stronger

trigger => more unstable
GMS also reduced
Equatorial Waves in a Full GCM
 Phase speeds in SAS simulations:
 In Kuo simulations:
• Wavespeed decreases with stronger moisture trigger
• Simulated equivalent depths scale with gross moist stability
See Lin, Lee, Kim, Kang and Fri. (2008, J Clim) & Fri. et al (submitted) for more
CCKW Vertical Structures
 In full GCM, many cases show 2nd baroclinic mode
structures (unlike in simplified GCM)
Shallow -> deep ->
stratiform
Gradual moistening of
boundary layer/midtroposphere
Warm over cold temperature
anomalies
See Lin et al (2008) and Frierson et al (submitted) for more detail
CCKW Vertical Structures
 Depends on convection scheme though!
Kuo simulations
never show
tilted omega or
humidity.
Only most
inhibited case
shows realistic
temperature
perturbations
Least inhibited SAS case =>
No tilt in omega
(but OK temperature)
Most inhibited Kuo case =>
No tilt in omega, q
(but OK temperature)
Phase Speed Determination?
 Estimated equivalent depths versus GMS:
Circled cases have clear 2nd baroclinic structure


1st baroclinic mode seems to explain phase speed
Presence/absence of 2nd baroclinic mode doesn’t appear to
have effect
Phase Speed Determination?
 2nd baroclinic mode and cloud-radiative forcing effects
on GMS
Stratiform phase =>
higher GMS
Shallow phase =>
lower GMS
Mode structure effect on GMS
averages to zero, and are small near
center of the wave
CRF changes have small effect
everywhere
Open Questions
 Reasons for second baroclinic mode structure
 And why seen in some fields more easily than others?
 Applicability to other models?
 Need for thorough comparisons of composites
 Relation to changes in mean precipitation?
MJO in GCMs
 Work with Sobel, Maloney, & Bellon using GFDL
AM2 model w/ realistic geography
 First crank up Tokioka “entrainment limiter” to get a
better MJO simulation:
Obs (NCEP)
Modified GFDL model
Unmodified GFDL model
See SMBF (Nature Geoscience 2008; J. Adv. Modeling Earth Systems in press)
MJO in GFDL AM2 Model
 Ratio of variance in eastward/westward
intraseasonal bands: 2.6 for modified GFDL model

Less than the observed value of 3.5, but larger than nearly all
models in Zhang et al (2006) comparison
 Higher entrainment in convection scheme => more
sensitivity to midtropospheric moisture
 Next test role of evaporation-wind feedbacks in
driving the modeled MJO

Set windspeed dependence in drag law formulation to globally
averaged constant value
See SMBF (Nature Geoscience 2008; J. Adv. Modeling Earth Systems in press)
Evap-Wind Feedback in Modeled MJO
 MJO greatly weakened when evaporation-wind
feedback (EWF) is turned off!
With EWF
Without EWF
See SMBF (Nature Geoscience 2008; J. Adv. Modeling Earth Systems 2009)
MJO in Aquaplanet AM2
 What is required to have a MJO-like structure in a
model?



Land-sea contrast?
Zonal asymmetry/Walker cell?
Evaporation-wind feedback?
 Experiments with Neale & Hoskins aquaplanet AMIP
boundary conditions


“QOBS” & “Flat”
GFDL AM2 model with Tokioka modification
M.S. thesis work of Marshall Stoner (2010)
Zonally Symmetric Results
 Log(variance) spectra: QOBS (left) and “Flat” (right)
Enhanced power in eastward
intraseasonal band
Connected to moist Kelvin wave?
More clear dominance of east over west
Less connected to Kelvin wave?
M.S. thesis work of Marshall Stoner (2010)
Intraseasonal Composites
 Composites of structure:
QOBS
Connected to midlatitude wave trains,
smaller scale

Flat
More similar to observed MJO?
When WISHE is suppressed, QOBS ISV (left) remains, while
Flat ISV (right) disappears
M.S. thesis work of Marshall Stoner (2010)
Mean States
 Mean states (solid = QOBS, dashed = flat):
 Flat has weaker easterlies, and a double ITCZ
 Standard WISHE likely drives the waves
M.S. thesis work of Marshall Stoner (2010)
How about Flat + a Walker cell?
Surface winds
Now mean westerlies over much of the tropics
Will WISHE still be important?
(standard theory assumes mean easterlies)
M.S. thesis work of Marshall Stoner (2010)
Walker Cell Case
 MJO-like variability still exists (although weaker)
 Again it disappears if WISHE is suppressed
Log(variance)
Variance avoids
surface westerly
region?
Surface winds
M.S. thesis work of Marshall Stoner (2010)
WISHEful Thinking
 Evaporation composites for Flat (zonally symmetric)
and Flat + Walker
Flat + Walker cell
Flat
Both essentially have evaporation leading the wave
Open Questions
 What sets scale, speed of the MJO-like
phenomenon?


Related to Kelvin wave at all, or a moisture mode?
Advection of dry air by WWBs & Rossby cyclones appears to be
important in setting speed as well as WISHE
 Comparisons with other models (including CRMs)
 Similar mechanisms acting? (mechanism denial experiments
in a range of models)
 Compare composites as well as spectra
 Understanding of how/when different mechanisms
can power waves can help our interpretation of
observations
Conclusions
 Convectively coupled waves in simple and full GCM
are affected by “gross moist stability”

Full GCM shows second baroclinic mode characteristics
 MJO-like structures can exist in aquaplanet model
 Zonally symmetric or with Walker cell
 More realistic ISV is powered by WISHE in mostly traditional
manner