ccam simulations for cordex south asia

CSIRO Marine and Atmospheric Research 1

CCAM simulations for CORDEX South Asia

John McGregor, Vidya Veldore, Marcus Thatcher, Peter Hoffmann, Jack Katzfey and Kim Nguyen

CSIRO Marine and Atmospheric ResearchAspendale, Melbourne

CORDEX WorkshopKathmandu

28 August 2013

CSIRO Marine and Atmospheric Research

Outline

• Introduction to the downscaling approach

• GCM selection

• SST bias correction

• CCAM model features

• Behaviour of the simulations


Downscaling with CCAM

CCAM (~50 km)

CCAM (~14 km)

Bias correction

GCM (~200 km)

GCM SST/Sea-ice


Quasi-uniform C192 CCAM grid with resolution about 50 km, showing every 4th grid point

Stretched C96 grid with resolution about 14 km over Nepal, showing every 2nd grid point

• The 50 km run is then downscaled to 10 km by running CCAM with a stretched grid, but applying a digital filter every 6 h to preserve large-scale patterns of the 50 km run

• A separate 100 km global CCAM run is also used to drive RegCM4.2 at its boundaries for 20 km RCM runs

CCAM downscaling methodology

• Coupled GCMs have coarse resolution, but also possess Sea Surface Temperature (SST) biases such as the equatorial “cold tongue”

• We first run a quasi-uniform 50 km global CCAM run driven by the bias-corrected SSTs


Indonesia 14 km

Some previous CCAM downscaling projects

Pacific Islands 60 km and 8 km

South Africa

Australia20 km – 60 km

Tasmania8 km – 14 km

CSIRO Marine and Atmospheric ResearchGCM Selection | Peter

Hoffmann

GCM Selection


GCM Selection Requirements

• Good performance in present climate• Simulation of rainfall, air temperature etc.

• Reproduce observed trends

• Good SSTs• ENSO pattern/frequency

• SST distribution

• Good spread of climate change signals

GCM Selection | Peter Hoffmann


GCM Selection Evaluation studies

• 24 CMIP5 models

• > 20 evaluation studies

• 6 publications with rankings + evaluation used within the Vietnam project

• Peer-reviewed or submitted


ACCESS1.0ACCESS1.3CanESM2

CCSM4CNRM-CMS

CSIRO-Mk3-6-0FGOALS-g2FGOALS-s2GFDL-CM3

GFDL-ESM2MGISS-E2-HHadCM3

HadGEM2-CCHadGEM2-ES

inmcm4IPSL-CM5A-LRIPSL-CM5A-MR

MIROC4hMIROC5

MIROC-ESM MIROC-ESM-CHEM

MPI-ESM-LRMRI-CGCM3NorESM1-M


GCM Selection Example: performance in current climate over Indochina


ACCESS1-0ACCESS1-3CanESM2CCSM4CNRM-CM5CSIRO-Mk3-6-0FGOALS-g2FGOALS-s2GFDL-CM3GFDL-ESM2MGISS-E2-HHadCM3HadGM2-CCHadGM2-ESinmcm4IPSL-CM5A-LRIPSL-CM5A-MRMIROC4hMIROC5MIROC-ESM-CHEMMIROC-ESMMPI-ESM MRI-CGCM3NorESM1-M

RMS Error (mm/day)

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

Pat

tern

Co

rrel

atio

n

0.5

0.6

0.7

0.8

0.9

1.0

PR ANNUAL (Jan.Dec.)

Evaluation region Results annual rainfall


GCM Selection - Rankings

Bhend (pers. communication) Suppiah (2012, HRD VN)

Watterson et al. (Aust)

Watterson et al. (Kont) Grose et al. (2012, submitted) Kim and Yu (2012, GRL)

Kug et al. (2012, ERL)

GCMsZ-score Temp

trendRMSE Temp

RMSE Prec PC Prec M-Score M-Score No. ENSO

RMSE N3.4 Corr N3.4 Std N3.4

Cor EP ENSO EOF1

Cor CP ENSO EOF1 Cor N3 N4

ACCESS1.0 4 8 19 20 3 2 7 1 2 7 12ACCESS1.3 1 6 22 21 8 7 2 11 5 8 CanESM2 17 11 7 11 12 12 7 5 2 11 3 5 18

CCSM4 22 3 1 1 7 5 6 20 5 13 2 1 2CNRM-CMS 21 17 2 2 2 1 8 8 2 9 3 2 1

CSIRO-Mk3-6-0 10 10 14 4 15 16 6 14 11 1 11 8 6FGOALS-g2 12 23 9 10 14 13 2 4 3 4 13FGOALS-s2 23 16 10 12 19 20 9 21 4 14 GFDL-CM3 9 22 6 6 5 9 9 19 5 12 4

GFDL-ESM2M 2 15 11 9 11 14 4 22 12 15 5 7 3GISS-E2-H 20 12 24 24 20 17 9 7 4 10 9 1 HadCM3 15 13 23 19 3 1 1 3

HadGEM2-CC 9 15 18 6 6 2 13 9 4 7 3 7HadGEM2-ES 14 4 18 16 4 4 6 10 6 2 6 7 15

Inmcm4 8 24 13 17 13 15 4 16 8 11 10 4 5IPSL-CM5A-LR 16 21 17 15 22 22 1 6 3 4 4 4 16

IPSL-CM5A-MR 7 5 16 14 21 21 5 9 4 2 2 3 9MIROC4h 18 2 8 13 2 3 5 5 MIROC5 5 7 3 7 16 10 8 23 5 15 6 5 10

MIROC-ESM 19 19 20 22 18 19 3 17 10 14

MIROC-ESM-CHEM 3 20 21 23 17 18 3 15 7 13 14MPI-ESM-LR 11 1 5 5 1 3 7 18 7 6 1 4 17MRI-CGCM3 13 14 12 3 10 11 9 12 3 11 6 6 11NorESM1-M 6 18 4 8 9 8 5 2 4 6 8 1 8


GCM Selection Final ranking


Rank GCM Average Score1 CNRM-CM5 0.312 CCSM4 0.343 ACCESS1.3 0.354 NorESM1-M 0.355 ACCESS1.0 0.396 MPI-ESM-LR 0.417 GFDL-CM3 0.428 HadGEM2-CC 0.449 MIROC4h 0.46

10 MIROC5 0.4711 GFDL-ESM2M 0.4812 MRI-CGCM3 0.5113 HadCM3 0.5314 IPSL-CM5A-MR 0.5315 HadGEM2-ES 0.5416 FGOALS-g2 0.5717 CSIRO-Mk3.6.0 0.5718 inmcm4 0.6119 CanESM2 0.6120 MIROC-ESM-CHEM 0.6921 GISS-ES-H 0.7022 IPSL-CM5A-LR 0.7123 FGOALS-s2 0.8024 MIROC-ESM 0.84

The rankings of the 6 individual studies are averaged to yield a final ranking of the models.


GCM SelectionClimate change signal JJA - good spread

X

XX


SST correction


• Observations• daily optimum interpolation SST & SIC (Reynolds et al.,

2007)

• 1/4° resolution for 1982-2011

• Method

adjust variance adjust mean

OBS

GCM

SST

freq

uenc

y


SST bias correction Results: SST BIAS ACCESS1.0

JAN JUL

original

after correction

(K)


Results: SST variance ACCESS1.0 (January)

ACCESS1.0 ObservedBias & Variance

corrected

Mean SSTs

SST Stdev


The conformal-cubic atmospheric model

• CCAM is formulated on the conformal-cubic grid

• Orthogonal• Isotropic

Example of quasi-uniform C48 grid with resolution about 200 km


Variable-resolution conformal-cubic grid The C-C grid is moved to locate panel 1 over the region of interestThe Schmidt (1975) transformation is applied

- it preserves the orthogonality and isotropy of the grid- same primitive equations, but with modified values of map

factor

C48 grid (with resolution about 20 km over Vietnam


CCAM dynamics

• atmospheric GCM with variable resolution (using the Schmidt transformation)

• 2-time level semi-Lagrangian, semi-implicit• total-variation-diminishing vertical advection• reversible staggering

- produces good dispersion properties• a posteriori conservation of mass and moisture


CCAM physics• Cumulus convection:scheme for

simulating rainfall processes

• Detailed modelling of water vapour, liquid and ice to determine cloud patterns





• Parameterization of turbulent boundary layer (near Earth’s surface)






• Modelling of vegetation and using 6 layers for soil temperatures and moisture

• CABLE canopy scheme


CCAM physics

• Cumulus convection:scheme for simulating rainfall processes



• Modelling of vegetation and using 6 layers for soil temperatures and moisture. 3 layers for snow

• CABLE canopy scheme

• GFDL parameterization of radiation (incoming from sun, outgoing from surface and the atmosphere)


Cumulus parameterization• In each convecting grid square there is an upward

mass flux within a saturated aggregated plume• There is compensating subsidence of environmental

air in each grid square• As for Arakawa schemes, the formulation is in terms

of the dry static energy

sk = cpTk + gzk

and the moist static energy

hk = sk + Lqk


Above cloud base

plume

detrainment

downdraft

subsidence


Enhancements for Maritime Continent

The Maritime Continent has many islands with land or sea breeze effects, and extra SST variability

a) enhance sub-grid cloud-base moisture if diurnal increase of SSTs, or

b) enhance sub-grid cloud-base moisture if upwards vertical motion

Both (a) and (b) are beneficial over Indonesia, Australia, Vietnam, China – (b) slightly better

(b) seems less suitable over India

(a) still fine over India


Cloud microphysics scheme (Rotstayn)CCAM carries and advects mixing ratios of

water vapour (qg), cloud liquid water (ql) and cloud ice water (qi)


Latest GFDL radiation scheme

• Provides direct and diffuse components

• Interactive cloud distributions are determined by the liquid- and ice-water scheme of Rotstayn (1997). The simulations also include the scheme of Rotstayn and Lohmann (2002) for the direct and indirect effects of sulphate aerosol

• Short wave (has H2O, CO2, O3, O2, aerosols, clouds, fewer bands)

• Long wave (H2O, CO2, O 3, N 2O, CH4, halocarbons, aerosols, clouds)


A recent AMIP run 1979-1989

CCAM100 km

Obs

Tuning/selecting physics options:• In CCAM, usually done with 100 km or 200 km AMIP runs, especially

paying attention to Australian monsoon, Asian monsoon, Amazon region

• No special tuning for stretched runs

DJF JJA


CORDEX runs using CCAM• We are performing global runs at 50 km,

providing outputs for 4 CORDEX domains: Africa, Australia,

SE Asia, S Asia.

• RCP 4.5 and 8.5 emissions scenarios

• So far have downscaled 6 of the CMIP5 GCMs at 50 km/ L27 resolution (as part of large Vietnam project). Output now available.

• Doing more runs, and more at 100 km.

• Performing the runs at CSIRO, CSIR_South_Africa, and Queensland_CCCE


a 100 km

b 50 km – ACCESSOthers quite similar

a 14 km

a 50 km ERA-ITRMM JJAS

GPCP JJAS


Rainfall change by 2080 (mm/d)JJAS RCP 8.5

32

CCAM_MPI CCAM_GFDL

CCAM_CNRM CCAM_ACCESS


CCAM_MPI CCAM_GFDL

CCAM_CNRM CCAM_ACCESS

% rainfall change by 2080 (mm/d)JJAS RCP 8.5


Convection in 50 km runs included vertical velocity enhancement (b)

TRMM-3B43GPCP

CCAM-100kmCCAM-14km

CCAM-Coupled

CCAM-BVC_SST

Over land and sea

TRMM

100 & 14 km

50 kmruns

coupled GPCP


35

CCAM100 km

Obs

DJF JJA

CCAM14 km over N Indiastretched


CMAPCMAP CCAMCCAMMAMDJF

JJA SON

100 km AMIP runs vs CMAP

1979-1989 C96 100 km AMIP run

Generally good rainfall. Fresh 50 km CORDEX runs are underway


AphroditeAphrodite CCAMCCAMMAMDJF

JJA SON

14 km runs vs Aphrodite


14 km runs vs IMD obs


DHMobs

CCAM14 km

JJAS present-day rainfall over NEPAL


CCAM coupled model - 14 km over Asia

Quite acceptable rainfall


14 km coupled runs – 3 daysMSLP, wind vectors, mixed layer depth > 50 m


14 km coupled runs – 3 daysSSTs

ccam simulations for cordex south asia

Documents