state of the cesm · state of the cesm thanks to nsf and doe for their long-term support of the...

CESM Advisory Board Meeting, February, 2013

Marika Holland [email protected]

CESM Advisory Board Meeting February 2013

Marika Holland

CESM Chief Scientist

State of the CESM

Thanks to NSF and DOE for their long-term support of the CESM Project



Outline

• CAB Recommendations from 2012

• CESM Planning

• Community Activities

• Science Highlights

• Future Directions



1) Implementation Plan Recommendation for a concise living document

CAB Letter Recommendations: March 2012

2) Transparency of the climate modeling enterprise For the decision process in development and application activities, etc.

3) Documenting decisions on model optimization/tuning

CESM Planning Activities during the last year are addressing these recommendations



CESM Planning Activities: Implementation Plan Draft provided: Concise/living document to be updated as needed

Outline: 1. Introduction – Focused on CESM2; Articulation of best practices/procedures 2. CESM Major Model Releases

• Definition of what these entail, what we expect for CESM2 (compsets, etc.) 3. Imperatives for future releases

• Community research tool for studies of climate processes/interactions 4. Development priorities for CESM2

• Small scale phenomena in all components and their interactions • Improved projections of sea level rise • More complete biogeochemical cycles • Climate/human system interactions • Near-term climate prediction

Note that WGs maintain component priorities; More details in revised science plan 5. Timeline for CESM2 6. Model Assessment process (and associated metrics document) 7. Model development process (and associated model release strategy document) 8. Resources and risk management



CESM Planning: New Model Release Strategy

• Pre-set annual release dates with a standard, firm timeline – Should ease planning for developments and release

resource needs; aid incremental development; provide more consistency

– Targeting May releases

• Configurations (“compsets”) in multiple categories – Scientifically vetted (with control runs/ climate

“assessment”) – Functionally vetted (routine testing) – Development only (no testing; use at own risk)

• Between releases, compsets can shift categories



CESM Planning: Model Validation/Diagnostics • Compiling lists of metrics currently in use for

component and coupled model assessment • Considering improvements to standard diagnostic

packages • Work underway via CVCWG on coupled model

(variability/change) diagnostics package

Courtesy of Adam Phillips and Clara Deser



CESM Planning: New Model Support Policy • Released in November

• Support Includes – bug fixes, – script changes, and/or code for time evolution of

supported machines

• Major releases of CESM supported for 5 years after release date or for 2 years after the next major release, whichever period is longer.

• CCSM3 support expired;

• CCSM4/CESM1 to 2015 at least http://www.cesm.ucar.edu/models/cesmsupportpolicy-2012.html



CESM Planning Activities: Working Group Process/Procedures

By SSC Request, Working Groups are documenting:

• Priorities for short-term (~yearly) component model developments, including scientific motivation

• Updated annually after winter and/or summer meetings.

• The component model development process/protocols,

• The responsibilities of developers,

• How decisions are made, etc.

• Documents to be provided under WG webpages

• Metrics/diagnostics used for component model assessment

Should contribute to a more open and transparent model development and decision making process



CESM Planning Activities: Update to 2009-2015 Science Plan

• Mid-way through the existing Science Plan

• Assessing progress and updating priorities for near-future (next several years) developments (CESM2)

• Draft should be available within months (by Breckenridge)

a) Interaction of the Carbon Cycle, Ecosystems and Climate b) Decadal Climate Projections and Forecasts c) Interaction of Aerosols and Climate d) Interaction of Chemistry and Climate e) Role of the Middle Atmosphere in Climate f) Role of Ice Sheets, Sea Ice and Land in Abrupt Climate Change g) Bounding Future Climate Scenarios h) Simulating Paleoclimates i) Role of Ocean Mesoscale Eddies in Climate j) Interaction with Integrated Assessment Modeling



Redesign of Model Output/Processing Workflow

• Need to streamline future processing/publishing for large intercomparison projects

• Need to reduce CESM data storage

• Maintaining the status quo not possible Effort led by John Dennis (CISL), Jim Edwards (CSEG),

Mariana Vertenstein (CSEG)

CESM Project Updates/Announcements

Category CMIP3 CMIP5

Models 2 (CCSM3, PCM) 5 (CCSM4, CESM1-BGC, CESM1-CAM5, CESM1-WACCM, CESM1-CHEM)

Volume submitted ~9 TB (over 10 months) ~136 TB (over 1 year)



Community Activities



Winter Working Group Meetings

• Ocean Model: 22-23 January; 48 registered participants • Joint/follow on meeting of the CPT on Internal Wave Mixing

• Atmosphere Model, Polar Climate, Whole Atmosphere, Chemistry Climate: • 11-13 February; 126 Registered participants • Follow on meeting on DOE Lab-NCAR CAM development (13-14 Feb)

• Land Ice: 14-15 February; 35 Registered participants

• Societal Dimensions, Land, BGC: • 19-22 February; 83 Registered participants • Joint session with Uncertainty Quantification meeting (20-21 Feb)

• Climate Variability and Change: • 4 March, 43 Registered participants

Coupling Technologies for Earth System Models (CW2013) held 20-22 Feb



CESM Tutorial

• Third Annual CESM Tutorial on 30 July-3 August, 2012

• Thanks to NSF and DOE for co-sponsoring student participation

• 80 Participants out of 180 Applicants

• Involved lectures and practical sessions, requiring a large staff commitment and support from CISL

• Online tutorial materials available and being widely used

• Thanks to Dave Bailey for chairing the organizing committee

• Fourth Annual CESM Tutorial planned for 12-16 August, 2013 • Announcement is out and applications are being accepted • We are again targeting about 80 participants • Thanks to Susan Bates for chairing the organizing committee



CESM Model Releases in Early Feb

• cesm1.1.1 (support for new model features) – CAM-SE/CAM5 development code base – CAM5/CLM4.0 targeted defaults

• Targeting “scientific support” for FV 1°, FV 2° 1850, 20th Century and RCP CMIP5 simulations

• Out of box support for yellowstone, titan

• cesm1.0.5 (support for CMIP5 science) – Out of box support for yellowstone, titan – Scientific support for all CAM4 FV CMIP5 simulations

• ccsm3 (university community request)

To enable computing on Yellowstone and Titan and to make new features available to development community



Community Integrations - Large Ensemble Science Motivation

From Deser et al., 2012, Nature Climate Change

Purpose:

• To robustly determine simulated natural variability

• To assess climate extremes and their changing likelihood

• To investigate detection/attribution of climate changes in the 20th-21st centuries



Planned Community Integrations Large Ensemble

• 30 members (perhaps more)

• 1900-2080 with RCP8.5 forcing scenario

• CESM-CAM5-BGC (1-degree) with prescribed CO2.

• Spin-up runs underway; long control will be available

• Initial state will vary by a round-off level change

• High-frequency output to be saved – Timeslice output for decades in 1850s, 1910s, 1990s, 2070s

for downscaling runs

– Smaller set of continuous daily fields for analysis Thanks to Clara Deser and Jen Kay for leading the planning effort

https://wiki.ucar.edu/display/ccsm/CESM+Large+Ensemble+Planning+Page



Community Integrations: Last Millennium Ensemble

• Motivation: only one full forcing simulation with CCSM4 completed for CMIP5

• New runs: From 850-2005; ensembles of fully-forced + single forcing CESM-CAM5 experiments (23) and fully-forced WACCM runs (2)

• Serving user-community needs for assessing and attributing regional climate changes

CESM1-CAM5 at 2o



Community Integrations – High Resolution Control Science Motivation

Regular lat-lon

Cubed Sphere

Tropical Cyclones 12-km

CAM-SE Run

Slide courtesy of Rich Neale

Mid-west Spring time propagating

systems



• 25km CAM5 coupled to 1-degree ocean

• Multi-century integration (~200 yrs)

• Requires CAM-SE configuration

– Developments still underway

– Hopeful that CAM-SE with new vertical advection will be released in May, 2013

• Integrations unlikely to start until later this year

• Will solicit community input on experiment design, desired output, etc.

• Planning will begin after the May release

Planned Community Integrations High Resolution Control



CCSM4/CESM J. Climate Special Collections

• 43 Papers available via AMS early online release

• Some additional papers still in review

• Document major model components and numerous aspects of simulated variability and change

http://journals.ametsoc.org/page/CCSM4/CESM1



CESM as a Community Resource ~450 people with developer level access to code > 600 (unique) Downloads of CESM Tutorial materials >1500 Registered Users of CESM1.0 Participation at CESM Workshop and WG meetings continues to grow



Science Highlights



(From Hurrell et al., in press, BAMS)

CESM



CCSM4

CCSM3

CESM1-CAM5

Comparisons of CESM-CAM5 and CCSM4 Example: ENSO Statistics

Other aspects: mean state, climate feedbacks, polar change have also been investigated

(Courtesy of Adam Phillips and Clara Deser)

OBS CESM1-CAM5 CCSM4 CCSM3

Ability to simulate

the extended

duration of La Nina Events



Example: Dry Zone Biases and Climate Sensitivity

(Fasullo and Trenberth, 2012, Science)

CMIP Analysis AIRS 500mb RH MJJA (%)

Intensity of subtropical dry zones in models strongly linked to:

• Projected cloud trends • Reflected solar radiation • Model sensitivity

∆TOA SW 21C

Correlated to ECS

R(MJJA RH, ECS)



Example: Dry Zone Biases and Climate Sensitivity

(From Fasullo and Trenberth, 2012, Science)

CAM3

CAM4

CAM5

CMIP Analysis



CMIP5 Model Intercomparison

(Sanderson and Knutti, “Assessing large scale climate properties from CMIP5 model projections”, in prep)



Component Model Intercomparisons Example: CLIVAR WGOMD COREII Experiments

Courtesy of Gokhan Danabasolgu

AMOC Mean (1988-2007) in Depth Space



CESM1 Carbon Cycle

Documented in: Lindsay et al, submitted, J. Clim Long et al., submitted, J. Clim

• Simulations with prescribed emissions and fully prognostic carbon cycle

• 20C increase in CO2 too large

• Not enough uptake by land or ocean

20th Century CO2 Ocean Flux

Land Flux

OBS OBS



Land Model Developments (CLM4.5)

GCP est. (?)

Courtesy of Dave Lawrence

Accumulated carbon from • losses due to land cover

change, • gains due to CO2

fertilization • regional losses or gains

due to climate-carbon feedbacks.

More realistic land carbon uptake results from reduced

N-limitation on CO2 fertilization.

From “forced” CLM experiments

Change in Terrestrial Carbon



Ozone Recovery and Antarctic Sea Ice

Smith, Polvani, and Marsh., 2012, GRL

2056-2065 Response of Antarctic sea ice to fixing ozone depleting substances at 2000 levels relative to RCP4.5

CESM-WACCM 1979-2005 CESM-WACCM 2065-2065 Obs 1979-2005

• Ice loss is 33% greater for the ensemble with no stratospheric ozone recovery

• Suggests that ozone recovery will mitigate Antarctic sea ice loss in the coming decades

No ozone recovery minus

RCP4.5



CESM-CISM Integrations Analysis of Greenland Surface Mass Budget Change

Present-day budgets compare well to RACMO

In 21st Century:

• Higher precipitation

• Larger melt

• Ablation area increases from 9% to 28% of ice sheet

• Equilibrium line ~500 m higher

• SMB increases over 2000m

Average Net Surface Mass Budget

1980-1999 2080-2099

(Courtesy of Miren Vizcaino)



Where We Are Heading

Investigation of small-scale phenomena Improved/Enhanced biogeochemical cycles

Improved predictions of sea level rise Near-term climate prediction

Climate/Human system interactions



Current high resolution simulation underway

Yellowstone Advanced Scientific Discovery J. Small, F. Bryan, J. Tribbia, R. Tomas, D. Bailey, J. Dennis, A. Baker, J. Edwards, J.Caron,

M. Vertenstein, T. Scheitlin, J. Bacmeister

– Resolutions • 25km CAM5/CLM4.0 (spectral element grid)

• 10km POP2 (62 level), CICE (tripole grid)

• 1/2° RTM (lat/lon grid)

• Resolutions allow for resolving ocean eddies and tropical cyclones

– Performance: ~2 model-years/day on 16,000 processors

– Target: multi-decade simulations (grand challenge goal)

Will help inform other planned simulations • 20-21 C high resolution simulations (Washington et al) • Community Integrations • 10km POP2 integration with BGC



ASD Simulation: 10km ocean and 25 km atmosphere J. Small, F. Bryan, J. Tribbia, R. Tomas, D. Bailey, J. Dennis, A. Baker, J. Edwards, J.Caron,

M. Vertenstein, T. Scheitlin, J. Bacmeister

SST, Sea ice cover and Sea Surface Height



Building a Global, Multi-Scale Ocean Model

MPAS development is a partnership between NCAR and LANL.

MPAS supports both quasi-uniform and variable resolution meshing of the sphere.

The MPAS ocean (MPAS-O) model will be coupled into the CESM over the next year.

high-resolution region

Below: Snapshot of kinetic energy from a global ocean simulation with 7.5 km resolution in the North Atlantic. The rest of the global ocean is

resolved with a 38 km mesh.

Slide courtesy of Todd Ringler



Community Ice Sheet Model (CISM)

• Now testing scalable dynamical cores with higher-order ice flow

–SEACISM dycore with Trilinos solvers

–BISICLES dycore with adaptive mesh refinement

–To be included in CISM 2.0, CESM 1.1

Antarctic ice speed, BISICLES model (red = fast flow)

Slide courtesy of Bill Lipscomb



Model Improvements Targeted at Specific Biases Example: Oxygen Minimum Zone Bias

From Moore et al., submitted, J. Clim

OMZs too extensive in CESM

Improvements to

Particulate Organic Matter (POM)

cycling reduce this bias

These are targeted for the CESM1.2

release



New Capabilities Example: Coupling to Data Assimilation Research Testbed

Courtesy of the DART Group

New multi-instance model capability

Coupling to DART

enables:

• Near-term initialized climate prediction

• Transpose AMIP-type experiments



And More…

All component models incorporating improved parameterizations and processes



In summary:

• Community aspects of the project generally remain strong

• CESM planning activities are underway to enhance transparency in the model development process and clarify procedures

• CESM applications continue to increase

• Model developments and improvements are ongoing

We look forward to your advice!