A new physics package for the next version of MIROC

A new physics package for the next version of MIROC

Masahiro WatanabeCCSR, University of Tokyohiro@ccsr.u-tokyo.ac.jp

May 27, 2008

Team “MIROC-physics” in KAKUSHN projectS. Watanabe1, T. Takemura2, M. Chikira1, T. Ogura3, T. Mochizuki1, K. Sudo4, T. Nishimura1, M. Watanabe5, S. Emori3, and M. Kimoto5

1: FRCGC/JAMSTEC, 2: RIAM/Kyushu Univ, 3: NIES, 4: Nagoya Univ, 5: CCSR/Univ of Tokyo

Climate change simulation by MIROC3.2 @ AR4

Climate change simulation by MIROC3.2 @ AR4

Global mean SAT – change from the end of 18th centuryGlobal mean SAT – change from the end of 18th century

Year

Full forcing (Natual + Anthropogenic)

Year

Anthropogenic forcingOnly

Year

Natural forcing Only(Solar + Volcano)

Glo

bal

mea

n S

AT

an

om

aly

(oC

)

Year

No forcing

Glo

bal

mea

n S

AT

an

om

aly

(oC

)

Observation Model ensemble mean

Further development of MIROCFurther development of MIROC

MIROC3.2 has presented as good ability as other MIROC3.2 has presented as good ability as other state-of-the-art CGCMs in simulating climate and itstate-of-the-art CGCMs in simulating climate and its variabilitys variability

Why we need to update it?Why we need to update it? We We knowknow the model still contains large uncertainty the model still contains large uncertainty

(model is (model is tunabletunable even if it generates realistic clim even if it generates realistic climate)ate)

Forthcoming CGCMs must be more robust as highForthcoming CGCMs must be more robust as higher accuracy will be required in AR5er accuracy will be required in AR5

CloudsClouds might be the key might be the key

Atmospheric component of MIROCAtmospheric component of MIROC

MIROC3.2 MIROC4.1Dynamical core Spectral + semi-Lagrangian sc

heme (Lin & Rood 1996)Spectral+ semi-Lagrangian scheme (Lin & Rood 1996)

Coordinate Eta (hybrid sigma) Eta (hybrid sigma)

Cloud Diagnostic (LuTreut & Li 1991) + Simple water/ice partition

Prognostic PDF (Watanabe et al. 2008) + Ice microphysics (Wilson & Ballard 1999)

Turbulence Level 2.0 (Mellor & Yamada 1982)

Level 2.5 (Nakanishi & Niino 2004)

Convection Prognostic A-S + critical RH(Pan & Randall 1998, Emori et al. 2001)

Prognostic A-S + critical RHwith water/ice partition

Radiation 2-stream DOM 37ch (Nakajima et al. 1986)

2-stream DOM 111ch (Sekiguchi et al. 2008?)

Aerosols simplified SPRINTARS(Takemura et al. 2002)

full SPRINTARS + prognostic CCN

Land submodel MATSIRO MATSIRO mosaic

Hybrid prognostic cloud (HPC) scheme Hybrid prognostic cloud (HPC) scheme

Large-scale condensation (LSC) Assume a subgrid-scale distribution of qt’ or s=aL(qt’-LTl’) ? Predict condensate amount and cloud?

Tompkins (2005) Prognostic equations for PDF variance & skewness Quasi-reversible operator between grid quantities & PDF

HPC-DU HPC-ST

Basis PDFs (varying skewness)

cloud water [g/kg]

clou

d fr

acti

on

C-qc relationship

cloud water [g/kg]

clou

d fr

acti

on

C-qc relationship

( ) , ( ) ( ) , ( , )c c

c c c l t s lQ Q

C G s ds q Q s G s ds Q a q q T p

Similar approach:Tompkins (2002, JAS)Wilson & Gregory (2003, QJ)

Single column model testSingle column model test

A-S, prognostic cloud scheme + simple cloud physics 12hr integration from Weisman & Klemp (1982) profile

Cloud mass flux

qc & qi Precipitation rate

Variance & skewness

McCf

V

S > 0

convectivestratiform

cumulus detrainment ⇒ V, S+ precipitation/snowfall ⇒ S-

0 1 2 3 4 [hr]

ice cloud

anvil

Snapshot of qc at hour 96 in NICAMCloud water at z=835m

Collaboration with NIES

How can we verify predicted PDF moments?

How can we verify predicted PDF moments?

Comparison w/ GCRM: 1-week integration from Dec. 25, 2006• GCRM (named NICAM) w/ 3.5km grid, realistic topography • MIROC atmosphere w/ T42

6400 points on avg. in a T42 grid diagnosis for the PDF moments

GCRMPDF variance

AGCM

8.3km

8.3km835m

835m

PDF variance

Improvement with HPCImprovement with HPC

ISCCP AGCM HPC AGCM HPC-ORG

ORG HPC

Annual-mean cloud water & cloud fraction along 10°S

* Low-cloud is yet insufficient over continents * Better representation of low clouds over the cold tongue

Annual-mean low cloud

Watanabe et al. (2008)

Higher-order turbulence closureHigher-order turbulence closurework done by M. Chikira (FRCGC)work done by M. Chikira (FRCGC)

From Level 2.0 (Mellor-Yamada 1982) to Level 2.5/3.0 (Nakanishi-Niino 2004)

• Evaluation of MLS locally (changing in space and in time)• Advection of TKE and other turbulent variables• Coupling with cloud scheme

Zonal and annual mean master length scales

Lev2.0 Lev2.5

Annual mean PBL height [m]

Lev2.0

Diff.

Lev2.5

Higher-order turbulence closureHigher-order turbulence closure

850hPa specific humidity

Annual mean clim. Bias

work done by M. Chikira (FRCGC)work done by M. Chikira (FRCGC)

Lev2.0

Lev2.5

ERA40

• MIROC3.2 has suffered from a low-level dry bias <- insufficient mixing

• Predicting TKE significantly improves the boundary layer structure -> reduced the dry bias

22

turb.

2 2

2

2 2

t l

l

L

Lt

Lq

t t

q q

t

a

Vt

V

• Turbulent variance/covariance can directly be used for predicting subgrid-scale PDF variance -> tighter coupling between turbulence & cloud processes

Sophisticated ice-cloud microphysicsSophisticated ice-cloud microphysicswork done by T. Ogura (NIES)work done by T. Ogura (NIES)

Ice

Cloud microphysics in MIROC3.2

Wilson and Ballard (1999)

Cloud microphysics in MIROC4.1

Liquid

Cloud liquid/ice fraction

ΔT2x=6.3K

Rotstayn et al. (2000)

Airborne measurements

Cloud liquid/ice fraction

ΔT2x=4.0K

• In MIROC3.2 climate sensitivity has largely been affected by a parameter for cloud liquid/ice partition

Coupling HPC-ice microphysics with cumuli

Coupling HPC-ice microphysics with cumuli

＊＊＊＊ ＊＊

● ●●＊ ＊ ＊

● ●●●

melting layer

cloud ice

cloud liquid

mixed-phase ice/liquid

MSE & total water budgets in A-S -> vapor, liquid and ice partitioned inside the cumulus with reference to temperature and saturation deficit in the cumulus tower

ice nucleation/deposition/fallout

change in the PDF variance and skewness

any type of cloud fractionis then calculated with HPC

/ / ( )cch t M h z D h h

/ / ( )ct c t t tq t M q z D q q

Preliminary model performancePreliminary model performance

T42L20 Atmos. + 0.5x1.4deg Ocean (corresponding to MIROC-mid)

qv 850hPa Dec-Feb clim., MIROC4.1 qv 850hPa Dec-Feb clim. bias NEWNEW

ORGORG

[kg/kg]

Preliminary model performancePreliminary model performance

T42L20 Atmos. + 0.5x1.4deg Ocean (corresponding to MIROC-mid)

Annual-mean precipitation

NEWNEW ORGORGObsObs

Preliminary model performancePreliminary model performance

T42L20 Atmos. + 0.5x1.4deg Ocean (corresponding to MIROC-mid)

SST Dec-Feb climatology, MIROC4.1 SST Dec-Feb climatology bias NEWNEW

ORGORG

[K]

SST interannual variability

Preliminary model performancePreliminary model performance

T42L20 Atmos. + 0.5x1.4deg Ocean (corresponding to MIROC-mid)

Annual-mean x & subsurface T

100E 60W

Obs

ORGNEW

100E 60W

SummarySummary

From diagnostic to prognostic schemesFrom diagnostic to prognostic schemes

Stronger coupling between subgrid-scale Stronger coupling between subgrid-scale processes processes

Better representation of climatology, Better representation of climatology, variability and climate sensitivity? variability and climate sensitivity?

Yes, we do hope so!Yes, we do hope so!

Major part of the atmospheric physics schemes was renewed in MIROC4.1

Concerns: characteristic timescale and difficulty in deriving diagnostic equations

Concerns: physical consistency, but errors in one scheme may be distributed