Simple tropical models and their relationship to GCMs

1
Simple tropical models and their relationship to
GCMs

• Adam Sobel, Columbia
• Chris Bretherton, U. Washington

2

• All aspects of tropical climate depend on tightly
  coupled feedbacks between convection, clouds,
  radiation, SST, surface fluxes
• In characterizing feedbacks, useful to look at
  local relationships between different physical
  variables (scatter plots) takes out geography
• Observed relationships can be used to constrain
  models of all sorts
• In GCMs, relationships can be diagnosed and
  compared to obs, and to other GCMs
• In simple models, relationships can be put in via
  bulk parameters, and sensitivity relatively
  easily understood may help to interpret GCM
  biases?

3
E.g. water vapor path vs. precip,from satellite
microwave
4
Normalizing to saturated WVP (mass-weighted
column mean RH) gives better fit
5
Sensitive test for GCMsCCM3.6
too much rain at low wvp (shallow cloud regions?)
6
Longwave cloud forcing (TOA),CCM3.6
7
Shortwave not as good -gtcoupled model errors
8
Simple models

• Neelin-Zeng Quasi-equilibrium tropical
  circulation model (QTCM)
• Single baroclinic mode for temperature, moisture,
  wind, barotropic mode for wind
• Betts-Miller type convection
• Underlying construction (single mode etc.)
  similar to models by Raymond, Emanuel, others
• QTCM is good bridge between GCMs, simple models
• We work with further reductions of QTCM (assumed
  symmetries etc.)

9
Idealized Walker circulation (Bretherton Sobel
2002)

• 1-dimensional (longitude) domain, sinusoidal SST,
  no rotation
• Temperature assumed constant in x, but unknown
• QTCM vertical structures
• Convective scheme hard adjustment or strict
  quasi-equilibrium
• Convective greenhouse feedback on radiative
  cooling RRclr rP, r0.2
• Fixed surface wind speed exchange coeff.

10
Equations

• u is baroclinic, has sign of upper trop. wind

In convective regions, qT. Gross moist
stability MMs Mq. Aq is constant coef. P
diagnosed from moist static energy plus (1).
11
System is very sensitive to cloud-radiative
feedback parameter
12
And also gross moist stability (Neelin Held
1987 Yu et al. 1998 Raymond 2000). r modifies
effective GMS
vertical motions efficiently
inefficiently export/import energy Radiative
cooling larger smaller
13
Same system with ocean mixed layer (Sobel 2003
Peters Bretherton 2004)

• set SWCFLWCF (both linear in precip as before)
• Finite conv adj time P(q-T)/?
• Forcing is background sfc rad ocean heat xport
  divergence, linear in x
• Now size of convective region much less dependent
  on r, because SWCF counteracts LWCF by cooling
  the surface but other parts of the solution,
  such as SST in convective region, are sensitive
  to r
• For large r can get instability to coupled
  oscillations on intraseasonal/subannual time
  scale (Sobel Gildor 2004)

14
Insights

• In these models, moisture is key horizontally
  varying fields. Controls precip, and through
  that, CRF.
• Key parameters are gross moist stability, and the
  constants relating WVP and CRF to precip

15
How is this relevant to biases in GCMs?

• Sensitivity to parameters such as r, GMS,
  convective time scale (WVP vs. precip) in simple
  models is relatively straightforward to
  understand. These same parameters can be
  diagnosed from GCMs. We can thus place the GCM
  in the parameter space of the simple model, which
  might if the simple model has enough of the
  right stuff in it - give us some insight into why
  the GCM behaves as it does. Would be especially
  interesting to compare these parameters in
  several different GCMs.