ccam simulations for cordex south asia
DESCRIPTION
CCAM simulations for CORDEX South Asia. John McGregor, Vidya Veldore, Marcus Thatcher, Peter Hoffmann, Jack Katzfey and Kim Nguyen CSIRO Marine and Atmospheric Research Aspendale, Melbourne CORDEX Workshop Kathmandu 28 August 2013. Introduction to the downscaling approach GCM selection - PowerPoint PPT PresentationTRANSCRIPT
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
CSIRO Marine and Atmospheric Research
Downscaling with CCAM
CCAM (~50 km)
CCAM (~14 km)
Bias correction
GCM (~200 km)
GCM SST/Sea-ice
CSIRO Marine and Atmospheric Research
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
CSIRO Marine and Atmospheric Research
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
CSIRO Marine and Atmospheric Research
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
CSIRO Marine and Atmospheric Research
GCM Selection Evaluation studies
• 24 CMIP5 models
• > 20 evaluation studies
• 6 publications with rankings + evaluation used within the Vietnam project
• Peer-reviewed or submitted
GCM Selection | Peter Hoffmann
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
CSIRO Marine and Atmospheric Research
GCM Selection Example: performance in current climate over Indochina
GCM Selection | Peter Hoffmann
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
CSIRO Marine and Atmospheric Research
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
CSIRO Marine and Atmospheric Research
GCM Selection Final ranking
GCM Selection | Peter Hoffmann
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.
CSIRO Marine and Atmospheric Research
GCM SelectionClimate change signal JJA - good spread
X
XX
CSIRO Marine and Atmospheric Research
SST correction
CSIRO Marine and Atmospheric Research
• 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
CSIRO Marine and Atmospheric Research
SST bias correction Results: SST BIAS ACCESS1.0
JAN JUL
original
after correction
(K)
CSIRO Marine and Atmospheric Research
Results: SST variance ACCESS1.0 (January)
ACCESS1.0 ObservedBias & Variance
corrected
Mean SSTs
SST Stdev
CSIRO Marine and Atmospheric Research
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
CSIRO Marine and Atmospheric Research
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
CSIRO Marine and Atmospheric Research
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
CSIRO Marine and Atmospheric Research
CCAM physics• Cumulus convection:scheme for
simulating rainfall processes
• Detailed modelling of water vapour, liquid and ice to determine cloud patterns
CSIRO Marine and Atmospheric Research
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)
CSIRO Marine and Atmospheric Research
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
CSIRO Marine and Atmospheric Research
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. 3 layers for snow
• CABLE canopy scheme
• GFDL parameterization of radiation (incoming from sun, outgoing from surface and the atmosphere)
CSIRO Marine and Atmospheric Research
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
CSIRO Marine and Atmospheric Research
Above cloud base
plume
detrainment
downdraft
subsidence
CSIRO Marine and Atmospheric Research
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
CSIRO Marine and Atmospheric Research
Cloud microphysics scheme (Rotstayn)CCAM carries and advects mixing ratios of
water vapour (qg), cloud liquid water (ql) and cloud ice water (qi)
CSIRO Marine and Atmospheric Research
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)
CSIRO Marine and Atmospheric Research 29
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
CSIRO Marine and Atmospheric Research
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
CSIRO Marine and Atmospheric Research
a 100 km
b 50 km – ACCESSOthers quite similar
a 14 km
a 50 km ERA-ITRMM JJAS
GPCP JJAS
CSIRO Marine and Atmospheric Research
Rainfall change by 2080 (mm/d)JJAS RCP 8.5
32
CCAM_MPI CCAM_GFDL
CCAM_CNRM CCAM_ACCESS
CSIRO Marine and Atmospheric Research 33
CCAM_MPI CCAM_GFDL
CCAM_CNRM CCAM_ACCESS
% rainfall change by 2080 (mm/d)JJAS RCP 8.5
CSIRO Marine and Atmospheric Research
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
CSIRO Marine and Atmospheric Research
35
CCAM100 km
Obs
DJF JJA
CCAM14 km over N Indiastretched
CSIRO Marine and Atmospheric Research
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
CSIRO Marine and Atmospheric Research
AphroditeAphrodite CCAMCCAMMAMDJF
JJA SON
14 km runs vs Aphrodite
CSIRO Marine and Atmospheric Research
14 km runs vs IMD obs
CSIRO Marine and Atmospheric Research
DHMobs
CCAM14 km
JJAS present-day rainfall over NEPAL
CSIRO Marine and Atmospheric Research
CCAM coupled model - 14 km over Asia
Quite acceptable rainfall
CSIRO Marine and Atmospheric Research 41
14 km coupled runs – 3 daysMSLP, wind vectors, mixed layer depth > 50 m
CSIRO Marine and Atmospheric Research 42
14 km coupled runs – 3 daysSSTs