TOWARD IMPROVED PREDICTION OF CONVECTIVE PRECIPITATION Mitchell W. Moncrieff Cloud Systems Group NCAR/MMM

1
TOWARD IMPROVED PREDICTION OF CONVECTIVE
PRECIPITATION Mitchell W. Moncrieff
Cloud Systems Group NCAR/MMM



NSF Briefing, March 6
2003
2
What are the strategic needs?

• To realize skillful NWP models, global and
  regional
• To understand the water and energy cycles
• To narrow uncertainties in climate models that
  involve moist physics

3
CRM Multiscale approach

• Modern cloud-resolving models (CRM) have the
  dynamic range to resolve convection and the
  multiscale organization of convection, this is
  demonstrated by realizations of
• warm-season precipitation over the US continent
  (Moncrieff and Liu 2003).
• convectively-coupled tropical waves (Grabowski
  Moncrieff 2001, 2002).
• MJO-like systems using CRMs in the
  cloud-resolving-convection (super)
  parameterization approach (Grabowski 2001
  Khairoutdinov Randall 2002).

4
Organized convection
Is organized convection important from the
large- scale perspective?

• Organization product of coupled physical
  processes and dynamics.
• Climate models parameterizations not designed
  to represent organization.
• NWP models grid-scale convection is an
  unphysical form of organization.

5
Parameterization methodology
Prediction models
Measurements
Convection cloud parameterization
Cloud- resolving models
Resolved convection super-parameterization pathw
ay
Dynamical models, stochastic models
6
Large-scale forcing

• 2-D CRMs planetary-scale computational domains
  enable convection and advective forcing to be
  interactive and physically consistent (Grabowski
  Moncrieff 2001, 2002).
• 3-D CRM computational domains are presently
  smaller than the Rossby radius of deformation,
  requiring that large-scale forcing be specified.
• Accurate forcing not obtainable from standard
  measurements alone a role for field campaigns
  e.g., GATE, TOGA COARE, IHOP_2002, NAME Tier 1,
  among others.

7
Warm-season precipitation over US-continent

• Large-scale forcing specified from nested MM5
  using NCEP global analysis as a first-guess in
  the outer domain.
• 10-day composite forcing applied to each day of
  CRM simulation -- an ensemble approach.
• Questions
• Do CRMs represent sequences of
  precipitation?
• Improve on parameterized-convection
  results?
• Implications for convective
  parameterization?

8
Computational domains
Cloud-resolving domain ( )
9
Environmental shear
Travel speed of observed sequences (14 m/s)
Same deep shear, increased low-level shear
10
Sequences of precipitation
Kain-Fritsch
Grell
Carbone et al. (2002)
Betts-Miller
Betts-Miller
2-D CRM
16 m/s
10 m/s
14 m/s
10 days
16 m/s
14 m/s
14 m/s
6 m/s
14 m/s
5 m/s
5 m/s
Grell
CRM precipitation travels at observed speed (14
m/s) shear is the important quantity
11
Total condensate
18 UTC 12 MDT
00 UTC 18 MDT
06 UTC 00 MDT
12 UTC 06 MDT
12
System-relative zonal flow (CRM)
Midnight CST

• Steering-level is 7 km (400 mb)

13
Dynamical model
c
Dynamical model steering level 7 km (400 mb),
agrees with CRM and observations
c
c
Steering level

Traditional steering level (700 mb 3 km),
lies outside the solution range
CAPE and shear
Moncrieff and Green (1972)
Compressibility
14
Parameterization
Organized deep convection
Ordinary deep convection

Perhaps many grid volumes
Single grid volume

• Organized transport
• Shear effects
• Propagation effects
• Open system
• Dynamically consistent far-field

• Entraining plume
• No shear
• No propagation
• Mass balance within grid volume
• Far-field effects minimal

Ordinary and organized convection can occur
15
Conclusions

• In CRM, sequences of precipitation over the US
  continent travel at the observed speed (14 m/s)
  wind shear the key quantity.
• Dynamical models of organized convection provide
  a theoretical foundation.
• Failure of conventional parameterization
  attributed to wind shear not being represented.
• 2-D CRMs embedded in GCMs (super-parameterization
  ) is a new way forward.
• Dynamical/stochastic models possible ways to
  improve existing parameterization.