wiki:access/AocDec2020

ACCESS Oversight Committee - Notes from the 10th December 2020

1. ACCESS-NRI

Andy H has followed up with the ANU DVCR. Waiting on Departmental direction. Nothing further at this stage.

2. Round-table - update on activities and issues relating to ACCESS

BoM

  • High-resolution ensemble city model ("ACCESS-CE3") at 2.2 km resolution now operational for three domains (for Brisbane, Sydney, and Vic-Tas). Four additional domains (Adelaide, Perth, Darwin and N.Qld) are now under testing and are planned to become operational progressively over the next 6 months.
  • Licensing for use of Bureau data, in particular for APS3 NWP and BARRA reanalysis output, is being finalised (see under NCI below)
  • Joerg Henrichs is planning on giving a training session on introduction to PSyclone and LFRic. Significant interest from CLEX and CSIRO in attendance.

Universities

  • Much work underway using ACCESS-CM2 and ESM1.5 models (e.g., Paleo runs) and analysis of output.
  • Analysis example – CMIP6 study looking at simulation of compound extreme events, in this case extreme rain and wind. Several CMIP6 models represent these events quite well, and CM2 and ESM1.5 are among the best performers. However, many CMIP6 models are quite weak over N Australia. Overall, there is no correlation between model resolution and skill in representing these composite events, which was an unexpected result. A GRL paper is coming out soon.

CSIRO

  • ACCESS workshop to be held at AMOS Conference in Feb. RL requests input on what needs to be covered.
  • Some CMIP6 scenario runs being extended out to 2300
  • DAMIP simulations complete and being processed for ESGF
  • New paper out by Bock et al (JGR) compares performance of the numerous CMIP6 models, and finds CM2 and ESM1.5 rank highly in the field.
  • CAFÉ Reanalysis paper has received good reviews and is being split into two papers for publishing. The reanalysis fields will be published on AWS.
    • Committee discussion about data storage / publishing and need to consider data solutions as part of next NCRIS road mapping process (starting soon).

NCI

  • Greg Ayers now chair of NCI Board
  • ERA5 from 1979-present is available at NCI. Has all data links etc https://opus.nci.org.au/display/ERA5/ERA5+Community+Home
  • Long awaited and latest Australian Gridded Climate Data (AGCD) set almost ready with Bureau. Lots of careful work here, but expect to release before the end of the year. This will replace out-of-date predecessor datasets (early AGCD and older AWAP)
  • Licensing policy with Bureau is under advanced discussion, expecting to finalise with the Bureau early in the new year. Move to a “default position” of Creative Commons license CC-BY4.0, though each dataset will be resolved within the Bureau. Will be gauging interest from climate community in the use of BARRA dataset once licensing is clarified.
  • ACCESS models tested on the ARM processors of a machine in Japan. While is good to see that the model runs, and the ARM processor has excellent memory bandwidth, overall the Intel Cascade processor on Gadi remains the most powerful processor for running the ACCESS model.
  • Continue to profile and tune
  • Keeping in sync with Justin Freeman and Joerg on PSyclone activity.
  • Pangeo environment at NCI will take progressive steps to next stage of release (inc. Jupyter, dask, etc). Currently have released Pangeo on gadi as batch (start of 2020). Was released knowing that the interactivity is not suitable or clunky for many users. So in addition to the Gadi installation, a new scalable environment (e.g., multiple node execution) is currently in configuration testing on the NCI cloud. Releasing in stages since there are many components to this. The service is initially targeted at climate/weather, earth observation and geophysics, but also expect some use-cases from genomics. We are also progressively releasing user documentation on management, access and examples of our python-based environments (inc Pangeo). SU charging to be consolidated under NCI shares.
  • There will be a training session on data at NCI at the AMOS workshop in February, and more will follow.
  • Many of our services have been transitioning from an older NCI cloud version to newer. Old cloud due to be switched off at the end of 2020. One remaining service is ACCESS-dev. Due to complexity, the final transition may delay to coincide with NCI downtime on Jan 5. ACCESS-dev needs a technology and architecture refresh, but will address after the immediate transition off the current cloud.

COSIMA

  • The renewal COSIMA linkage proposal is funded in full! This will cover ongoing work over 4 years FY21/22 through 24/25. There are 11 partners. There will be a COSIMA meeting in approx 6 months.
  • The output from OMIP runs with the 1⁰ model is going to the ESGF
  • The OMIP runs with the 0.25⁰ model are proceeding and going well
  • An OMIP-style run with the 0.1⁰ model is well progressed, having done three 60-year cycles. This version includes tracers for study of Antarctic Bottom Water formation. There is interest in including the BGC in this version, in work with RM.

NESP

  • NESP2 has been approved, with the climate hub having a 5% cut (from the proposed $40m to $38m), in line with cuts to other hubs. There will be a planning process over the next 6 months, with significant stakeholder engagement. There will be an emphasis on applied science.

UM Partnership activities

  • UM Users Workshop (9-12 Nov) and GC4 Assessment Workshop (2-3 Dec) noted. These workshops were well attended by staff from Bureau and CSIRO. The Southern Ocean warm bias is further improved over previous GC versions, however tropical precipitation biases are not greatly improved, including precipitation biases over Australia.

3. Any other business

  • WGCM
    • Meeting later this month to discuss CMIP7 occurrence and format.
    • CMIP6 currently has commitment from 137 models. 103 models have already submitted the DECK plus Historical simulations.
    • Big interest by WCRP in the next step in resolution for global models to "synoptic weather resolving" scale – at order 10km resolution.
    • Paul Durack is the incoming co-chair of the WCRP WIP. More stringent submission is being considered, with errata data being ~15% of the 9.4 PB currently submitted for CMIP6.
  • ESGF: The US DOE have confirmed their commitment to the ESGF (US operation and development) as a long-term infrastructure (particularly CMIP), but is seeking a wider rebid for its future from other US gov agencies. Livermore (the incumbent) is considering rebid with other agencies in this process, expect the evaluation and decision to be complete in 2021 Q3. Livermore continues to run operations and complete current development – and have funding until 2022 Q3. DOE want a seamless transition to next phase of ESGF in the US. Complete support for international activities that co-develop and maintain. The European infrastructure project (which funds ESGF EU to significant level) also completes in 2022 Q4, and they are working on their process for sustaining into the future.
  • Initial funding for planning a Climate and Resilience service proposal has been announced by the Prime Minister. The funding for the next 6 months will cover the development of some initial offering for this service and the planning for a full blown service. The program if successful is likely to have a 4 years duration. The Bureau has been given the overall responsibility for that development and instructed to work closely with other relevant science based governmental organisations (CSIRO, ABS, GA).
  • Next meeting 25th February 2021

---

Back to AOC meeting notes

Back to main ACCESS wiki page

Last modified 10 months ago Last modified on Dec 21, 2020 8:34:51 AM