The Australian Community Climate and Earth System Simulator

1
The Australian Community Climate and Earth System
Simulator

• Michael Naughton
• ACCESS Group
• CAWCR
• Bureau of Meteorology
• Melbourne
• talk prepared by
• Kamal Puri
• ACCESS Group Leader

2
Centre for Australian Weather and Climate
Research, CAWCR

• Joint initiative of the Bureau of Meteorology and
  CSIRO
• CAWCR commenced in September 2007
• CAWCR is arranged in two Programmes
• Weather, Ocean and Environment
• Earth System Modelling and Climate Science
• CAWCR has a Director and two deputy Directors
• Director Search for director under way Chris
  Mitchell is Foundation director
• Deputy Directors Tom Keenan and Helen Cleugh

3
CAWCR
Weather, Ocean and Environment
ESM and Climate Science
Seasonal and Inter-annual Prediction
Weather and Environment
Ocean Prediction
Climate change
Atmosphere Land Observation and Assessment
ACCESS Modelling, Assimilation
Ocean Observation and Assessment
4
ACCESSObjectives

• Develop a national approach to climate and
  weather prediction model development
• Focus on the needs of a wide range of
  stakeholders
• Providing the best possible services
• Analysing climate impacts and adaptation
• Linkages with relevant University research
• Meeting policy needs in natural resource
  management

5
Current situation with modelling

Global Model Reg. Model Data Assim RCM
Bureau Spectral Finite Diff. GenSI 1DVAR No
CSIRO Spectral C-CAM Finite Diff. C-CAM No C-CAM
6
Project PlanKey Recommendations

• Blueprint for ACCESS, K. Puri, June 2005
• Project Plan for ACCESS, K. Puri, September
  2005
• ACCESS should import the MetOffice atmospheric
  model HadGAM1 to provide the initial atmospheric
  model for ACCESS
• The MetOffice 4DVAR scheme should be imported to
  form the atmospheric data assimilation module in
  ACCESS

7
UM Atmospheric Model

• The atmospheric models used at the Met Office for
  operational NWP and climate modelling (including
  Regional Climate Modelling) are all derived from
  the Unified Model (UM) System

8
Domains used
Vertical layers
9
UM Atmospheric ModelKey Features

• Non-hydrostatic form of governing equations are
  used
• A semi-implicit/semi-Lagrangian scheme is used to
  solve the governing equations
• It is designed to conserve mass, mass-weighted
  potential temperature and moisture, and angular
  momentum

10
UM Atmospheric ModelKey Features

• A number of alternative formulations for the
  physical parametrisations are available at each
  release of the UM
• The parametrisations include
• clouds, radiation, boundary layer, precipitation,
  convection, land surface interactions and gravity
  wave drag

11
Data Assimilation

• Atmospheric data assimilation is performed using
  a variational scheme (3DVAR and 4DVAR) designed
  to be used by both global and limited area models
• The 4DVAR system includes perturbation forecast
  and adjoint models to provide explicit
  representation of the time dimension with a
  6-hour data window

12
Incremental Analysis update system with 3D-Var
13
Incremental 4D-Var
14
ACCESS modules based on Project Plan
recommendations
OPS
NWP
VAR
LPJ
UM
Assimilation (?)
OASIS
UKCA
CABLE
CICE
AusCOM (MOM4)
CSIRO OC
BODAS
OBS
15
ACCESS Teams(as recommended in the Project Plan)

• Atmospheric Modelling
• Data Assimilation
• Land Surface / Carbon cycle Modelling
• Ocean and Coupled Modelling
• Model Evaluation
• Modelling System

16
ACCESSKey Timelines

• NWP
• 3Q 2006 UM and VAR implemented
• 1Q 2007 Preliminary runs in NWP mode
• 3Q 2007 Consideration of model resolutions
  and limited area domains and resolutions
• 4Q 2008 Operational implementation
• Risk Data handling and access, particularly
  satellite data

17
ACCESSKey Timelines

• Climate Change simulation
• 2006/2007 Port component models to common
  computing environment and test
• 2007/2008 Construct coupled system
• 2009 Test and balance coupled system
• 2010 Perform and submit IPCC AR5 runs
• Risks Not meeting AR5 timeline
• Deficiencies in performance of coupled
  system
• Insufficient supercomputer time

18
ACCESS NWP Configuration
NWP component of ACCESS has been fully
implemented This represents a significant
milestone
M e t O f f i c e
OPS
VAR
Atmosphere
Coupler
MOSES
19
Met Office
ACCESS
BoM Obs.
UKMO Obs
Observations
ObservationData-Base
MetDB
ODB (ECMWF)
Observation
OPS
OPS
Variational Assimilation
VAR
VAR
UM
UM
Unified Model
Met. Archive Retrieval
MARS
MetDB
20
ACCESS Coupling AusCOM to UM
Atmosphere
Currently being worked on MetOffice have already
done this
Coupler
AusCOM Sea ice (CICE4 code)
AusCOM Ocean (MOM4p1 code)
Ocean Carbon cycle
BODAS
Dynamic Ocean Primary Prod.
OBS
21
ACCESSCurrent status

• AusCOM ocean and sea-ice models have been coupled
  using OASIS coupler
• Initial results are encouraging
• Coupling of AusCOM to the UM has commenced a
  clear way forward has been developed
• Expect to obtain version (UM7.0) with coupling
  code from MetOffice within the next few weeks
• Expect to have technical coupling achieved by the
  end of year (probably earlier)

22
ACCESS Coupling CABLE to UM
Single Column Model (1D UM)
Coupler
CABLE
1D UM is a subset of the 3D UM so transition to
the full UM should be technically straight-forward
23
ACCESS Coupling CABLE to UM Current status

• The technical coupling of CABLE to the UM is
  challenging - a clear way forward has been
  developed
• Work to achieve this coupling has made good
  progress
• Coupling is initially being carried in the 1D UM
  currently 90 done
• Only one land point with two tiles used so far
  tropical forest and bare ground - more testing
  for different climates and tiles is being
  performed
• Expect to have technical coupling achieved by the
  end of year (probably earlier)

24
ACCESSFully coupled ESM

• Expect to have technically fully coupled system
  available by end of year
• The necessary testing and balancing of the fully
  coupled system will then be able to commence

25
ACCESS NWP suitesCurrent status

• Global NWP suite running routinely with
  encouraging results
• Regional 37.5km NWP suite running routinely
• Successful runs have been conducted with the
  mesoscale Australian domain 12km suite
• VicTas 5km forecast suite running daily
• Detailed testing has commenced in preparation for
  operational implementation
• Last legacy systems upgrade was made in December
  all work has since shifted to ACCESS

26
UM, OPS, VAR

• In summary the NWP component of the ACCESS Earth
  System Model has now been implemented and
  detailed testing leading to operational
  implementation is about to commence.
• This represents a major milestone for ACCESS

27
ACCESS Current status NWP data used in
assimilation
AIREPS
SYNOPS
SONDE
AMV
ATOVS
SCAT
AIRS
28
Experimental NWP - Observations received
29
Experimental NWP - Observations accepted
30
Accepted surface pressure observations
31
Rejected surface pressure observations
32
Accepted surface pressure observations
33
Rejected surface pressure observations
34
Example of daily NWP runs
35
Example of daily NWP runs
36
Example of daily NWP runs
37
Skill score verifications over Australian region
MSLP
Z850
Z500
Z200
Met Office IC UM
GASP IC GASP
GASP IC UM
38
Skill score verifications over Australian region
MSLP
Z850
ACCESS IC ACCESS UM
GASP IC GASP
Met Office
39
MSLP verifications over Australian region
ACCESS IC ACCESS UM
GASP IC GASP
Met Office
Met Office IC ACCESS UM
40
MSLP verifications over S.Hem annulus
ACCESS IC ACCESS UM
GASP IC GASP
Met Office
Met Office IC ACCESS UM
41
MSLP verifications over N.Hem annulus
ACCESS IC ACCESS UM
GASP IC GASP
Met Office
Met Office IC ACCESS UM
42
Z500 verifications over S. Hem annulus
ACCESS IC ACCESS UM
GASP IC GASP
Met Office
Met Office IC ACCESS UM
43
Improved analysis and prediction by the ACCESS
NWP system
LAPS
ECMWF
GASP
AusLAM
44
AusLAM
Verifying NMOC analysis
LAPS
45
24 h forecast temperature bias andRMSE for
RTO-LAPS and AusLAM Units 0.1K Verification
against own analyses
46
(a)
(b)
(c)
(d)
47
(No Transcript)
48
(No Transcript)
49
ACCESS Ensemble PredictionPlans

• Implementation of Met Office Global and regional
  EPS (MOGREPS) is under way
• Test global and regions runs have been executed
  successfully

50
(No Transcript)
51
(No Transcript)
52
(No Transcript)
53
(No Transcript)
54
(No Transcript)
55
(No Transcript)
56
Example of experimental ACCESS/MOGREPS run
57
Example of experimental MOGREPS run
58
Towards operational implementation
59
Current Domains
60
Towards operational implementationNWP Expert
Group recommendations

• It is proposed that development of the NWP model
  suite occur in two phases.
• The primary goal of phase one is to make the
  transition from our previous LAPS/GASP systems,
  to a suite built on the ACCESS framework.
• The timelime for operational implementation of
  the ACCESS systems is quite tight, and simply
  implementing the ACCESS systems over the domains
  as they stand will require a great deal of
  effort.
• Even without a change in resolution, significant
  improvements in accuracy are expected to be
  achieved, primarily due to the use of more data,
  inside the 4D-VAR framework. Current experiments
  with the global ACCESS model appear to confirm
  this.

61
Towards operational implementation NWP Expert
Group recommendations

• Phase One Suite
• Global system at 80km resolution
• Regional system over current LAPS domain, at
  37.5km.
• Regional system over current TXLAPS domain, at
  37.5km.
• Mesoscale system over current MesoLAPS domain, at
  12.5km.
• Mesoscale system for TC forecasting - run
  on-demand, at 15km.
• City-based systems (Brisbane, Sydney,
  Victoria-Tasmania, Adelaide, Perth), at 5km.

62
Phase I - recommended by ACCESS Operational NWP
Study Group
63
Towards operational implementation NWP Expert
Group recommendations

• Phase two suite
• With phase one complete, we will have sufficient
  experience and confidence with the ACCESS
  framework to be able to consider the issue of
  significantly different model suites for phase
  two.
• One of the most important drivers for this is to
  minimise the level of support required for the
  suite as a whole, by reducing the number of
  component systems.
• Suite rationalisation is not a straightforward
  task however, requiring a careful balance of
  stakeholder needs, computing and scheduling
  limitations, and modelling concerns such as
  resolution change across nested systems, and
  boundary-condition influence and errors.

64
Towards operational implementation NWP Expert
Group recommendations

• Phase two suite
• Global system at 40km resolution
• Mesoscale system over current LAPS domain, at 10
  - 12km
• City-based systems (Brisbane, Sydney,
  Victoria-Tasmania, Adelaide, Perth), at 2 - 4km
• Note the much larger domain for the mesoscale
  system (LAPS, rather than MesoLAPS domain).
• Elimination of the LAPS-level system is possible
  due to the increase in global model resolution -
  a nesting "jump" in resolution from the 40km
  global to the 10 - 12 km mesoscale should present
  no significant difficulties
• Again, no data assimilation will be performed at
  the city-scale, though at this point we should be
  planning as to how we might approach this (radar
  assimilation, etc).

65
Phase II - recommended by ACCESS Operational NWP
Study Group
66
Towards operational implementation NWP Expert
Group recommendations

• Phase two suite
• Phase Two is expected to see the introduction of
  our first ACCESS based ensemble systems, running
  at half the resolution of their global
  counterparts
• Phase Two Suite, Australian Region, Ensemble
  (MOGREPS)
• Global Ensemble, 80km resolution, 24 members
• Mesoscale System Ensemble over current LAPS
  domain, 24km resolution, 24 members.

67
Towards operational implementation NWP Expert
Group recommendations

• Phase two suite
• The tropical system configuration in phase two is
  unclear at this point
• An equivalent system to TCLAPS will obviously
  still be required
• It may be possible to eliminate TXLAPS, as the
  LAPS-equivalent was able to be eliminated once
  the global resolution reached 40km.
• Currently, TXLAPS uses different assimilation and
  modelling strategies to LAPS - for example the
  former includes a diabatic nudging scheme. At
  this point, its not obvious we'll have the
  effort, even in phase two, to port all of these
  features into the ACCESS framework.
• A plausible tropical system configuration in
  phase two would be to run the TC system inside
  the ACCESS global system.
• There are risks associated with this approach,
  hence a detailed assesement of the tropical
  performance of the global system will be needed
  before making any final decision

68
Towards operational implementation
69
Operational NWP Issues SMS

• SMS ECMWF Supervisor Monitor Scheduler
• Powerful GUI environment for operational (
  research) job scheduling and monitoring
• Core component of BoM NMOC Operations for last
  decade 24x7 operations, n000 jobs per day
• Bureau SMS version has not been updated, but
  still working fine at this point
• ACCESS has always considered recommended SMS as
  scheduler of choice in planning for ACCESS-UM
  operational NWP systems

70
(No Transcript)
71
http//www.accessimulator.org.au/index.html
72
http//gale.ho.bom.gov.au/bm/internal/daas/access_
nwp/index.html
73
ACCESS Test climate simulations with upgraded
UM

• Multi-decade test simulation performed with an
  upgraded version of the UM (UM6.3 PC2)
• Version features new PC2 cloud scheme instead
  of the standard
• PC2 may be used in HadGEM3
• Follows AMIP protocol - use observed SST and
  sea-ice from 1979 - 2000
• Mainly as a learning experience - for model
  itself and for diagnostics

74
Summer rainfall
75
Winter rainfall
76
Southern Oscillation Index
77
Leading SH storm track modes crossing Australia
in JulyObservations and Models
Observation-based (1975-94 reanalysis)
CSIRO MK3.0
ACCESS
300hPa Streamfunction Amplitude
78
AusCOM Arctic sea ice thickness (m)
February
August
79
AusCOM Antarctic sea ice thickness (m)
February
August
80
Collaboration with Met Office

• The Bureau and CSIRO have signed a research
  license agreement with the MetOffice
• A Science Plan for collaboration with the Met
  Office has been agreed to and some projects have
  commenced
• There have been ongoing and fruitful
  communications between the ACCESS Group Leader
  and the Met Office external collaboration manager
• There have been regular communications between
  ACCESS scientists and their Met Office
  counterparts prompt responses from Met Office
  have led to rapid solutions to problems
  encountered by ACCESS
• A number of ACCESS scientists have made extended
  visits to the MetOffice these have been very
  useful

81
Relationship with universities

• University representatives on Science Advisory
  Group C. Jakob, M. England
• University representatives on all ACCESS Teams
• June Science Advisory Group meeting will be
  solely devoted to University issues C. Jakob,
  M. England will prepare an agenda after
  consultation with University colleagues
• A small group of scientists from Melbourne and
  Monash Universities will shortly go through a
  familiarisation tutorial on how to run the ACCESS
  modelling system

82
New Collaborations

• The Korea Meteorological Agency (KMA) have
  decided to develop an NWP system that very
  closely follows the ACCESS system
• K. Puri visited the KMA from 20-24 January 2008
  at the invitation of the KMA
• Main objectives of visit
• to inform KMA scientists on the ACCESS
  implementation
• discussions with senior scientists
• prepare a draft document on technical cooperation
  between the two agencies discussions with senior
  scientists
• The cooperation includes visits to Melbourne by
  KMA scientists (started in April) and ACCESS
  scientists to Seoul

83
Concluding comments

• Good progress has been made on all fronts
• Local infrastructure has been developed
• All NWP components are working detailed testing
  prior to operational implementation has commenced
• A number of AMIP-type climate runs have been
  completed
• A number of studies with the UM are ongoing
• Good progress has been made in coupling
  AusCOM/CICE and CABLE to the UM
• We expect to have a technically fully coupled ESM
  by the end of 2008 - However a lot of hard work
  lies ahead