Download NCAR Capabilites

New Directions in Climate Research
and Simulation:
IPCC AR5 and US/NCAR Climate
Modeling Activities
Dr. Lawrence Buja
National Center for Atmospheric Research
Boulder, Colorado, USA
Recently, the direction of our climate change
research program dramatically changed.
WAS: Is anthropogenic climate change occurring?
NOW: What will be the of impact of anthropogenic climate
change on coupled human and natural systems?
• Magnitude and speed?
• Direct and indirect impacts?
• Adaptation vs mitigation
• What are our options & limits?
Addressing these new, much more complex, questions requires
• new approaches & priorities,
• new science capabilities,
• new collaborators/partners
Image courtesy of Canada DND
CCSP 2.1a Mitigation Simulations
MSE3 Climate Topics Summary
DOE’s ten-year vision to use exascale computing to
revolutionize DOE’s approaches to energy, environmental
sustainability and security global challenges.
Exascale systems provide and unprecedented opportunity for
science to use computation not only as an critical tool along with
theory and experiment in understanding the behavior of the
fundamental components of nature but also for fundamental
discovery and exploration of the behavior of complex systems with
billions of components including those involving humans.
Download complete MSE3 Report at
http://www.er.doe.gov/ASCR/ProgramDocuments/TownHall.pdf
HPC dimensions of Climate Prediction
New Science
(new processes/interactions
not previously included)
Better Science
(parameterization → explicit model)
Spatial
Resolution
Timescale
(Length of simulations
* time step)
(simulate finer details,
regions & transients)
Ensemble size
(quantify statistical properties of simulation)
Data Assimilation
(decadal prediction/ initial value forecasts)
Lawrence Buja (NCAR) / Tim Palmer (ECMWF)
HPC dimensions of Climate Prediction
New Science
Better Science
ESM+multiscale GCRM
Code Rewrite
10
Spatial
Resolution
(x*y*z)
10
Regular
AMR
1000
70
1.4°
160km
0.2°
22km
100yr*
20min
10
50
10
Ensemble size
?
1000yr*
3min
Timescale
(Years*timestep)
1000yr * ?
Today
Terascale
5
500
10
10
Climate Model
400
10000
1Km
Earth System Model
2010
10
Petascale
2015
Exascale
Cost Multiplier
10
Data Assimilation
Lawrence Buja (NCAR)
T42 2.8°
310km
FV 2.0°
220km
300
IPCC AR3
1998
Global
General
Atm/Ocn
Circulation
Horizontal Grid Size (Km)
250
Continental Scale
Flow
200
150
IPCC AR4
2004 4TF
T85 1.4°
160km
FV 1.0°
110km
Regional
T170 0.7°
78km
MJO/MLC
Convergence
FV 0.5°
T340 .36°
FV 0.25°
FV 0.1°
55km 2010 500TF
39km
28km CCSM Grand
Challenge
11km
100
50
0
Carbon Cycle
+ BGC Spinups
Lawrence Buja (NCAR)
IPCC AR5
2010 1PF
Sub-Regional
Hurricanes
CCSM at ¼ ° ATM 1/10°OCN
Courtesy Dr. David Bader, PCMDI/LLNL/DOE
IPCC AR4 Modeling Centers & AR5 Timeline
New CCSM Components for IPCC AR5
• Aerosols
– Direct and indirect effects
• Chemistry
– Radiative and air quality issues
• Dynamic Vegetation
– Regrowth following disturbance
• Carbon & Nitrogen Cycle
– Ocean & land biogeochemistry
– Anthropogenic (transient) land use/cover
• Land Ice Sheets
– Sea level Rise & Abrupt Climate change
IPCC AR5 (2013) Scenarios
The AR5 process has much greater coordination between IPCC
WG-I (Physical Science Basis)> WG-II (Impacts, Adaptation,
Vulnerability and WG-III (Mitigation).
1. IPCC “Classic + ” Long-term (Mitigation) Scenarios:
• 100 & 300-year climate change simulations
• Medium resolution
• Core “required” + optional Tier 1 and Tier 2 simulations
• Carbon, Nitrogen & Biogeochemical cycles
• 4 Representative Concentration Pathways (RCPs) from IAM community
• Quantify investment return of mitigation strategies
2. New: Short-term Climate Change “Adaptation” Simulations:
• Short-term (30-year) climate predictions
• Single scenario
• High-resolution (0.5° or 0.25° resolution)
• Designed for impacts, policy and decision making communities.
RCPs in perspective – CO2 emissions
120
100
MES-A2R 8.5
( 900ppm, +4.5°, IIASA )
AIM 6.0
( 671ppm, +3.7°, NIES )
MiniCAM 4.5
( 550ppm, +---°, PNNL )
Emissions (GtCO2)
80
60
40
20
30
IMAGE 2.9
0
IMAGE
252.6
( 424ppm, <2°, MNP )
( 370ppm, <2°, MNP )
2100
2090
2080
2070
2060
2050
2040
2030
2020
2010
2000
From Moss et al., 2008
sions (MtN2O)
Stabilization range (10-90th percentile)
-20
20
15
Baseline range (10-90th percentile)
Post-SRES (min/max)
Selected scenarios (min/max)
NCAR
CMIP5 Long-term Experiments
ensembles
: AMIP &
20 C
Control, RCP4.5
AMIP, & 20C RCP8.5
E-driven 20 C
E-driven
RCP8.5
1%/yr CO2 (140 yrs)
abrupt 4XCO2 (150 yrs)
fixed SST with 1x & 4xCO2
E-driven control
with C-cycle
Coupled carbon-cycle
climate models only
Carbon cycle
sees 1XCO2
(1% or RCP4.5)
All simulations except those
“E-driven” are forced by
prescribed concentrations
NCAR
CMIP5 Decadal Predictability/Prediction Exps
Additional predictions
Initialized in
‘01, ’02, ’03 … ‘09
prediction
with 2010
Pinatubo-like
eruption
prescribed
SST
timeslices
10 year
initialized hindcasts
& predictions, O(3)
30 year
initialized hindcasts
& predictions
O(3)
Alternative
initialization
strategies
extended
ensembles
to O(10)
regional
air quality
NCAR
Temperature at 2030
Averages and Extremes
(IPCC AR4: no C-cycle/dynVeg feedback)
Precipitation at 2030
Averages and Extremes
(IPCC AR4: no C-cycle/dynVeg feedback)
DATA: Earth System Grid Center for
Enabling Technologies (ESG-CET)
ESG Goals
Current ESG Sites
• Petabyte-scale data volumes
• Globally federated sites
• “Virtual Datasets” created through
subsetting and aggregation
• Metadata-based search and discovery
• Bulk data access
• Web-based and analysis tool access
• Increased flexibility and robustness
http://www.earthsystemgrid.org
http://www-pcmdi.llnl.gov
For AR5, ESG will be
expanded to form a
global virtual data center!
From: Earth System Grid Center for Enabling Technologies: (ESG-CET)
Towards a Next Generation
Climate-Weather-ESM
• Existing and future applications require (at least locally)
meso-scale and cloud-scale resolution in a global model
• Current climate models are poor weather models, and
current weather models are poor climate models.
• Opportunity to leverage the diverse interests and
experience of the climate and weather communities to
create and share a next-generation atmospheric
simulation system.
• New dynamic grids and solution methods capable of
efficient operation on petascale computers
Nested Regional Climate Model
NCAR
North Atlantic and North American
Regional Climate Changes
The goal is to simulate
the effects of climate
change on precipitation
across the intermountain
West States and tropical
cyclones, with a focus on
the Gulf of Mexico.
• 36, 12 and 4 km domains nested into CCSM
• 1996-2005, then time slices out to 2050
• Multi-member ensembles for each period
• Dedicated time on NCAR IBM Power 6 (Bluefire) since July:
 24 nodes (~20% of total number of processors)
 36 (12) km simulations use 128 (256) processors per job
 Will use 3.9M processor hours through 11/08
 ~300 Tb of data (to date); 450 Tb total (including earlier runs)
NCAR
Improving Predictions of Regional
Changes in Weather and Climate
The Nested Regional Climate Model
High Resolution Climate Modeling
IPCC (2007)
IPCC (2013)
NRCM
NCAR
Importance of Resolution
18 storms
25 storms
NCAR
Thanks!
Any Questions?
NCAR
National Center for Atmospheric Research
More than Meteorologists….
Weather
Climate
Observations
Simulations
Oceans
Sea-Ice
Hydrology
Mathematicians
Software Engineers
Network Engineers
GIS/Data Management
Solar
Physicists
Energy
Biology
Land Use
Statisticians Risk analysis
Gov’t Affairs
Policy
Law
Economists
Field Programs
Engineers:
Chemical,
Electrical,
Mechanical…
Social Scientists
-Adaptation
-Mitigation
-Impacts
-Health
Chemists
Writers
Communicators
Educators