Date post: | 16-Jan-2016 |
Category: |
Documents |
Upload: | randolph-wilcox |
View: | 215 times |
Download: | 0 times |
SRNWP 27-29 Oct., 2003, Bad Orb
Masaki Satoh and Tomoe NasunoMasaki Satoh and Tomoe Nasuno
Frontier System Research for Global Change/Frontier System Research for Global Change/Saitama Inst. Tech.Saitama Inst. Tech.
Radiative-convective equilibrium calculations Radiative-convective equilibrium calculations with cloud resolving models:with cloud resolving models:
A standard experiment and parameter studyA standard experiment and parameter study
Fifth International SRNWP-workshop on nonhydrostatic modelling27-29 Oct. 2003, Bad Orb, Germany
SRNWP 27-29 Oct., 2003, Bad Orb
OutlineOutline
MotivationA global cloud resolving model
Investigation of physics
Model formulationNonhydrostatic core
Radiative-convective equilibrium experimentsSetup
Parameter study
Summary
SRNWP 27-29 Oct., 2003, Bad Orb
MotivationMotivation
Development of a global cloud resolving model Nonhydrostatic ICosahedral Atmospheric Model (NICAM)
⇒ Δx = 3.5km on the Earth Simulator
2hours for one day simulation
if 320 nodes are used (half of ES)
Climate study
⇒ direct calculation of cloud-radiation interaction
⇒ Radiative-convective equilibrium
NICAMNICAM
By H.Tomita
SRNWP 27-29 Oct., 2003, Bad Orb
RegionalRegional Nonhydrostatic modelNonhydrostatic model
Regional Nonhydrostatic model (Satoh, 2002,2003,MWR)A subset of the global cloud resolving model (NICAM)• Cartesian coordinates
• The same dynamical framework as NICAM except for the metrics
• Model hierarchy: can be used as 1D-vertical, 2D-slice, and 3D-regional models
> Development of new dynamical schemes: Dynamical framework and advection scheme
> Study of physics: cloud-radiation interaction
SRNWP 27-29 Oct., 2003, Bad Orb
CharacteristicCharacteristic s s of the nonhydrostatic modelof the nonhydrostatic model
Fully compressible non-hydrostatic equations Horizontally explicit and vertically implicit time integration with time splittin
g The Helmholtz equation is formulated for vertical velocity not for pressure:
> a switch for a hydrostatic/non-hydrostatic option can be introduced.
Conservation of the domain integrals (Satoh 2002, 2003,MWR) The finite volume method using flux form equationsDensity, momentum, and total energy are conserved.Conservation of total energy including TKE budget
Tracer advecionThird order upwind, or UTOPIAConsistency with Continuity
Exact treatment of moist thermodynamics (Ooyama 1990, 2001).Dependency of latent heat on temperature and specific heats of water substanceTransports of water, momentum, and energy due to rain.
An accurate transport scheme for rain (Xiao et al 2003,MWR) Conservative Semi-Lagrangian scheme with 3rd order
SRNWP 27-29 Oct., 2003, Bad Orb
Dry formulationDry formulation
Conservative flux form equations for density R, momentum V, and total energy E+K+G:
where
ETKE
Wj
zj
Vj
yj
Uj
xj
GChGKEt
Gx
wRgPz
Wt
Gx
vPy
Vt
Gx
uPx
Ut
Rt
vFv
V
V
V
V
V
2
0
2
gzGKpR
CTCeE
RpPwvuWVU
d
VV
in
,2
,
' ,' ,,,,,2v
V
SRNWP 27-29 Oct., 2003, Bad Orb
Governing equations (Governing equations (Ooyama, 1990,2000Ooyama, 1990,2000 ))
Transports due to rain
Release of potential energy of rain
SRNWP 27-29 Oct., 2003, Bad Orb
CharacteristicCharacteristic s s of the nonhydrostatic model (2)of the nonhydrostatic model (2)
PhysicsCloud physics: Choice of ice process for the global model is an issue.
Warm rain (bulk method)
Ice process: Grabowski(1998; simple 3 categories) (courtesy of W.G.)
planned: Lin et al.(1983); Grabowski (1999: 5 categories)
Bin or Spectral expansion method (K.Suzuki)
Turbulence: 1.5TKE (Deardorff)
or Mellor and Yamada Level 2, 2.5
Surface flux: Louis (1982)
Radiation: MSTRN-X (Nakajima et al, 2000, courtesy of CCSR)
SRNWP 27-29 Oct., 2003, Bad Orb
Radiative-convective equilibrium studiesRadiative-convective equilibrium studies
Small domain experimentsInvestigation of many parameters: physics and external parameters
Comparison between different models
Feasible on many computers: • 100km x 100km , Δx=2km (Tompkins and Craig 1998)
Can be used as a standard test
Large domain experiments1000km x 100km (Tompkins 2001)
3D domain : 1000 km x 1000 km , Δx=2km
Equatorial belt 2D or 3D : 40000km x 100km
Global experiments on ESAqua planet with uniform SST (Sumi; Grabowsky 2003)
Aqua planet with prescribed SST distribution (Hayashi and Sumi; APE)
AMIP ,… : Realizable climate condition is an equilibrium state of fully interactive radiative and convective processes.
SRNWP 27-29 Oct., 2003, Bad Orb
Large domain experimentsLarge domain experiments 3D large domain experiment: fo
llowing Tompkins (2001)1000 km x 100 km x 21 kmΔx= 2 kmUniform radiative cooling (-2K/day)Tropical SST (302K)Long-time simulation (56 days)No large scale forcing: pure RCE exp.Use of MRI/NPD-NHM
(Courtesy of Dr. T.Kato)
SRNWP 27-29 Oct., 2003, Bad Orb
Large-scaleorganization
Looselyorganized(small scale)
10 days
1000 km
Rainwater (z=35m)
y-averaged(100 km)
SRNWP 27-29 Oct., 2003, Bad Orb
Issues of radiative-convective equilibrium experimentsIssues of radiative-convective equilibrium experiments
Strong dependency on artificial parameters
Surface flux with bulk formula:
Depends on minimum surface velocity: Umin
Control of the shear:
Mean winds develop internally.
Strong interaction with radiation
Domain size, resolution, numerical diffusion…
Model dependency
⇒ Requires a suitable standard setup
To understand parameter dependency
To know model characteristics
Shie et al. (2003)
SRNWP 27-29 Oct., 2003, Bad Orb
Small domain experimentsSmall domain experiments
Basically follows Tompkins & Craig (1998)Dimension: 3D or 2D100km × 100km × 25km 200km×200km(3D); 1000km, 5000km(2D)Δx=Δy=2km 4, 10kmLowest level: 20m, 54 layers depend on number of vertical layers?Periodic boundary conditionFixed sea surface temperature with 300K or 302KRadiation: interactive with clouds and humidity
prescribed radiative cooling: 2K/day (z<9km) decreases to zero at z=12km or 1K/day, or interactive radiation (require solar flux and ozone profile)
Surface flux: minimum velocity for the bulk coefficient: Umin=4m/s or 1, 7m/sNo large-scale forcing: no momentum source
or nudging to prescribed zonal wind (0m/s)No Coriolis forcing: f=0Total integration time: 60(spin up)+40days Initial condition: uniform temperature(250K)
or TOGA-COARE, Marshall islands
SRNWP 27-29 Oct., 2003, Bad Orb
Control experimentControl experiment•100km × 100km, Δx = 2km•Warm rain•Prescribed cooling: -2K/day•TKE•Bulk method Umin=4m/s•Uniform initial cond. T=250K
Precipitation
Relative humidity
temperature
SRNWP 27-29 Oct., 2003, Bad Orb
Mass weighted mean temperature & precipitable waterMass weighted mean temperature & precipitable water
CTL
Courtesy of W.K. Tao
SRNWP 27-29 Oct., 2003, Bad Orb
Problems and further experimentsProblems and further experiments
Problems of the control experimentToo cold and too dry
Too moist in the upper troposphere
Domain size & grid intervals: Are they sufficiently large and fine?
If not, in what sense?
Bulk coefficient and minimum velocity
Statistic of maximum of the vertical velocity
Ice phase
SRNWP 27-29 Oct., 2003, Bad Orb
Bulk coefficient and minimum velocityBulk coefficient and minimum velocity
CTL : control case : bulk formula, Umin=4m/s Umin=1, 7 m/s: minimum wind for bulk coeff.=1, 7m/s CD=0.001, 0.01: constant bulk coefficient
SRNWP 27-29 Oct., 2003, Bad Orb
)()1(
))]()(([
)]()([
*
**
00
spD
sspD
sspD
TqrLTCVC
TTrqTqLTCVC
qqLTTCVC
EvapShQdzF
56 175
Surface temperature jumpSurface temperature jump
Radiative cooling
Ts: surface temperautre, T0 : atmospheric bottom temperature
qs: surface humidity, q0 : atmospheric bottom humidity
q*: saturation humidity, r: relative humidity
CDV: bulk coefficient x surface velocity
F: Total radiative cooling
Sh: sensible heat flux, Evap: evaporation flux
SRNWP 27-29 Oct., 2003, Bad Orb
pDVCCFT /max
)(8.0 *0 sTqq
TTTTqq s 00*
0 ),(8.0
pDs VCCqqLFT /)( 0
Possible range of
Bulk coefficient
Bulk CoefficientBulk Coefficient
SRNWP 27-29 Oct., 2003, Bad Orb
Saturation in the upper troposphereSaturation in the upper troposphere
CTL : control case with Kessler:Autoconversion rate:
g/kg 0.1
sec10
)(31
cr
auto
crcauto
q
K
qqKAUTO
Cloud water [kg/kg]
Relative humidity
SRNWP 27-29 Oct., 2003, Bad Orb
Relative humidityRelative humidity
G98Berry
G03 RE9
SRNWP 27-29 Oct., 2003, Bad Orb
Autoconversion rateAutoconversion rate
CTL : control case with Kessler
Berry
g/kg 0.1
sec10
)(31
cr
auto
crcauto
q
K
qqKAUTO
Grabowsky(2003):
Robe and Emanuel(1996):
Grabowsky(1998): simple ice (3categories) temperature dependent snow/rain
g/kg 0.0
sec10 2
crq
C)02(T g/kg 0.0
C)0(T g/kg 0.1
sec10 3
cr
cr
q
q
1
2
c
c
q
qAUTO
SRNWP 27-29 Oct., 2003, Bad Orb
Domain size, grid interval, and 3D vs 2DDomain size, grid interval, and 3D vs 2D
3D 100km CTL 100km x 100km Δx=Δy=2km 3D 200km 200km x 200km Δx=Δy=2km 3D 200km,dx=4km 200km x 200km Δx=Δy=4km 3D 500km,dx=10km 500km x 500km Δx=Δy=10km 2D 1000km 1000km Δx=2km 2D 5000km 5000km Δx=2km
SRNWP 27-29 Oct., 2003, Bad Orb
PDF of maximum vertical velocityPDF of maximum vertical velocity
100km x 100km, x=2km⊿
200km x 200km, x=2km⊿
200km x 200km, x=4km⊿
SRNWP 27-29 Oct., 2003, Bad Orb
Summary(1)Summary(1)
A new regional non-hydrostatic model using a conservative scheme.
Conservation of mass and total energy.
Accurate formulation of moist process.
A subset of a global nonhydrostatic model with icosahedral grid (NICAM)
Radiative-convective equilibrium experimentsProposal of a standard experiment
To be used for investigation of physics and parameters
SRNWP 27-29 Oct., 2003, Bad Orb
Summary(2)Summary(2)
Parameter dependency Large dependency on surface flux
Cloud physics: conventional warm rain scheme is inappropriate;
require ice physics
Domain size and resolutionAs the grid interval becomes coarser• Colder mean temperature and less precipitable water
• Larger CAPE
At the same resolution (Δx=2km),• Mean temperature and precipitable water take closer values.
• Statistics (PDF) of Wmax depend on domain size.
• 200km x 200km is preferable rather than 100km x 100km.