1. 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

2. Outline
Introduction to the downscaling approach
GCM selection
SST bias correction
CCAM model features
Behaviour of the simulations

3. Downscaling with CCAM Bias correction CCAM (~50 km) GCM (~200 km)
GCM SST/Sea-ice

4. 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
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
Stretched C96 grid with resolution about 14 km over Nepal, showing every 2nd grid point
Quasi-uniform C192 CCAM grid with resolution about 50 km, showing every 4th grid point

5. Some previous CCAM downscaling projects
Indonesia 14 km
Pacific Islands 60 km and 8 km
South Africa
Australia
20 km – 60 km
Tasmania
8 km – 14 km

6. GCM Selection
GCM Selection | Peter Hoffmann

7. 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

8. GCM Selection Evaluation studies
ACCESS1.0
ACCESS1.3
CanESM2
CCSM4
CNRM-CMS
CSIRO-Mk3-6-0
FGOALS-g2
FGOALS-s2
GFDL-CM3
GFDL-ESM2M
GISS-E2-H
HadCM3
HadGEM2-CC
HadGEM2-ES
inmcm4
IPSL-CM5A-LR
IPSL-CM5A-MR
MIROC4h
MIROC5
MIROC-ESM
MIROC-ESM-CHEM
MPI-ESM-LR
MRI-CGCM3
NorESM1-M
24 CMIP5 models
> 20 evaluation studies
6 publications with rankings + evaluation used within the Vietnam project
Peer-reviewed or submitted
GCM Selection | Peter Hoffmann

9. GCM Selection Example: performance in current climate over Indochina
Evaluation region
Results annual rainfall
GCM Selection | Peter Hoffmann

10. 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)
GCMs
Z-score Temp trend
RMSE Temp
RMSE Prec
PC Prec
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 12
ACCESS1.3 6 22 21 11 5
CanESM2 17 18
CCSM4 13
CNRM-CMS 9
CSIRO-Mk3-6-0 10 14 15 16
FGOALS-g2 23
FGOALS-s2
GFDL-CM3
GFDL-ESM2M
GISS-E2-H 24
HadCM3
HadGEM2-CC
HadGEM2-ES
Inmcm4
IPSL-CM5A-LR
IPSL-CM5A-MR
MIROC4h
MIROC5
MIROC-ESM
MIROC-ESM-CHEM
MPI-ESM-LR
MRI-CGCM3
NorESM1-M

11. GCM Selection Final ranking
Average Score 1
CNRM-CM5
0.31 2
CCSM4
0.34 3
ACCESS1.3
0.35 4
NorESM1-M 5
ACCESS1.0
0.39 6
MPI-ESM-LR
0.41 7
GFDL-CM3
0.42 8
HadGEM2-CC
0.44 9
MIROC4h
0.46 10
MIROC5
0.47 11
GFDL-ESM2M
0.48 12
MRI-CGCM3
0.51 13
HadCM3
0.53 14
IPSL-CM5A-MR 15
HadGEM2-ES
0.54 16
FGOALS-g2
0.57 17
CSIRO-Mk3.6.0 18
inmcm4
0.61 19
CanESM2 20
MIROC-ESM-CHEM
0.69 21
GISS-ES-H
0.70 22
IPSL-CM5A-LR
0.71 23
FGOALS-s2
0.80 24
MIROC-ESM
0.84
GCM Selection Final ranking
The rankings of the 6 individual studies are averaged to yield a final ranking of the models.
GCM Selection | Peter Hoffmann

12. GCM Selection Climate change signal JJA - good spreadX X X

13. SST correction

14. Observations
daily optimum interpolation SST & SIC (Reynolds et al., 2007)
1/4° resolution for
Method
adjust variance
adjust mean
OBS
GCM
SST
frequency

15. SST bias correction Results: SST BIAS ACCESS1.0
JAN
JUL
original
after correction
(K)

16. Results: SST variance ACCESS1.0 (January)
Bias & Variance corrected
ACCESS1.0
Observed
Mean SSTs
SST Stdev

17. 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

18. Variable-resolution conformal-cubic grid
The C-C grid is moved to locate panel 1 over the region of interest
The 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

19. 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

20. CCAM physics
Cumulus convection:scheme for simulating rainfall processes
Detailed modelling of water vapour, liquid and ice to determine cloud patterns

21. 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)

22. 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

23. 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)

24. 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

25. Above cloud base
subsidence
detrainment
plume
downdraft

26. 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

27. Cloud microphysics scheme (Rotstayn)
CCAM carries and advects mixing ratios of
water vapour (qg), cloud liquid water (ql) and cloud ice water (qi)
Lots of processes to be included.

28. 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)

29. Tuning/selecting physics options:
A recent AMIP run
DJF
JJA
Obs
CCAM
100 km
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

30. 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

31. TRMM JJAS
a 100 km
a 50 km ERA-I
b 50 km – ACCESS
Others quite similar
a 14 km
GPCP JJAS

32. Rainfall change by 2080 (mm/d) JJAS RCP 8.5
CCAM_MPI
CCAM_GFDL
CCAM_CNRM
CCAM_ACCESS

33. % rainfall change by 2080 (mm/d) JJAS RCP 8.5
CCAM_MPI
CCAM_GFDL
CCAM_CNRM
CCAM_ACCESS

34. Convection in 50 km runs included vertical velocity enhancement (b)
TRMM-3B43
GPCP
TRMM
CCAM-100km
CCAM-14km
CCAM-Coupled
100 & 14 km
CCAM-BVC_SST
GPCP
coupled
Over land and sea
50 km
runs
Convection in 50 km runs included vertical velocity enhancement (b)

35. DJF
JJA
Obs
CCAM
100 km
CCAM
14 km
over
N India
stretched

36. Generally good rainfall. Fresh 50 km CORDEX runs are underway
100 km AMIP runs vs CMAP
CMAP
CCAM
CCAM
CMAP
DJF
MAM
JJA
SON
C km AMIP run
Generally good rainfall. Fresh 50 km CORDEX runs are underway

37. 14 km runs vs Aphrodite
Aphrodite
CCAM
CCAM
Aphrodite
DJF
MAM
JJA
SON

38. 14 km runs vs IMD obs

39. JJAS present-day rainfall over NEPAL
DHM
obs
CCAM
14 km

40. CCAM coupled model - 14 km over Asia
Quite acceptable rainfall

41. MSLP, wind vectors, mixed layer depth > 50 m
14 km coupled runs – 3 days
MSLP, wind vectors, mixed layer depth > 50 m

42. 14 km coupled runs – 3 days
SSTs