UK Earth System Modelling News
December 2018
The UK Earth system modelling project is a joint venture between the Met Office and the Natural Environmental Research Council (NERC) to develop and apply a world-leading Earth System Model.
If you like to get in touch, please email us at
uk***@nc**.uk
INSIDE THIS ISSUE:
Results from the UKESM1 CMIP6 DECK and historical simulations
by Colin Jones, Alistair Sellar, Yongming Tang and Steve Rumbold.
The UKESM1 CMIP6 DECK and historical simulations are close to completion. We present a number of highlights from these simulations, including estimates of the model’s Effective Climate Sensitivity (ECS) and Transient Climate Response (TCR). Read more… |
The release of UKESM1: update
by Jeremy Walton, Alistair Sellar, Yongming Tang, Marc Stringer, Richard Hill, Julien Palmieri, Rich Ellis, Ros Hatcher, Grenville Lister, Joao Teixeira and Colin Jones.
UKESM1 is now frozen and we have been working hard on porting the model to a number of HPC systems, including the NERC ARCHER facility. We outline recent progress in this effort, including methods used to scientifically validate these ports. We are confident a full model release will occur in January 2019. Read more… |
UKESM-hybrid: Focussing resolution where it’s most needed
by Marc Stringer, Richard Hill, Mohit Dalvi, Colin Johnson and Colin Jones.
Following finalization of the scientific evaluation of HadGEM3 GC3.1, the first set of “production” runs for the CMIP6 model intercomparison (Eyring et al., 2016) have now begun. These are being performed at the two spatial resolutions of HadGEM3 GC3.1: N96ORCA1 (CMIP6 tag: HadGEM3-GC31-LL) and N216ORCA025 (HadGEM3-GC31-MM). The former of these 2 constitutes the coupled physical model core of UKESM1. Read more… |
Progress coupling Greenland and Antarctic ice sheets into UKESM1
by Robin Smith, Vicky Lee and Antony Siahaan.
A scientifically functioning version of UKESM1 with an interactive treatment of the Greenland ice sheet is now running. In addition, a technically functioning version of the model with both interactive Greenland and Antarctic ice sheets also exists. The latter configuration, we believe, is presently the only such coupled Earth system – ice sheet model anywhere in the world. Read more… |
RECENT PAST EVENTS
9 -10 July 2018 – UKESM LTSM annual meeting – Met Office, Exeter:This year, the LTSM project celebrated its annual meeting at the Met Office Headquarters in Exeter. The meeting ran for two full days. Day 1 included a special session dedicated to the UKESM1 model release, plus a number of presentations on broader research occurring within the UKESM-LTSM, with invited speakers and posters from the contributing NERC centres. Day 2 was centred around project business, specific break groups and future science plans (download the meeting agenda with links to presentations here). |
29 Oct – 1 Nov 2018 – 14th IEEE international conference on eScience, Amsterdam:By Jeremy Walton. This ongoing series of meetings that examines how the adoption of digital technologies is changing the way in which research is performed – initially, in science, but more recently in the humanities as well. The cross-disciplinary nature of eScience was reflected in the range of presentations at the conference, which discussed work in areas as diverse as using machine learning to automatically decipher handwritten documents, identifying animal calls in the wild, curating and accessing medical images, the coupled modelling of magnetic plasmas and analysing the propagation of news information across media sources. In addition to illustrating the application of techniques from computer simulation, data modelling and machine learning across the research spectrum, the meeting included discussions about the importance of open science – i.e. unfettered access to data, code, models, methodologies, publications and evaluations. The tension between the requirements of open science and current reward and incentive policies in science and research was explored in an interactive session by all conference attendees. In addition to plenary eScience sessions, the conference included four parallel sessions which were more focussed on specific fields: Data Handling and Analytics for Health, Advances in eScience for the Humanities and Social Sciences, Exascale computing for High-Energy Physics and Weather & Climate Science in the Digital Era, which I attended. This included interesting talks on multiscale modelling (resolving clouds in a global atmosphere model), increasing parallelism in climate models via additional component concurrency, using machine learning for weather forecasting (directed at prediction of energy usage) and the enhancement of flood maps using social media reports. I presented our work on the hybrid-resolution version of UKESM, which appeared to generate a positive response, and was complimentary to the discussions of multiscale modelling in other systems. The keynote talk in this session was from Peter Neilly of The Weather Company (now part of IBM) who described their efforts to make weather forecasts part of probabilistic-based decision-making systems. Other contributions from IBM included the downscaling of results from physical weather models using deep neural networks, and improving the pre-season agricultural yield forecasts using machine learning. Finally, one of the posters in that session came from (I think) ECMWF, and which discussed the problem of transforming climate model output into CMIP6 datasets for submission to the Earth System Grid Federation – an issue which rang a few bells with the present author. |
5-6 November 2018 – UKESM core group annual meeting and retreat, Bristol:The UKESM core group recently met for a day at the annual core team retreat in Bristol. The group analysed, reviewed and discussed the following topics: UKESM1 coupled, atmospheric, terrestrial and marine performance, progress and plans for UKESM-IS, UKESM-hybrid and ESMValTool, and Model release and support, as well as short presentations for core group members plus four sessions with break-out groups to discuss a range of topics: (i) ‘What type of observations are we lacking or require for evaluating/analysing UKESM1?’; (ii) What further analysis (leading to) papers etc should we prioritise?; (iii) What developments (scientific, computational etc) should we prioritise for a subsequent UKESM1.x?; (iv) How should we work together in the future?; and (v) What is the potential for using AI/emulators/Big Data? |
22-23 November 2018 – OMIP meeting, LOCEAN-IPSL, Paris:Core group members Julien Palmieri and Andrew Yool (NOC; UKESM1 marine biogeochemistry), together with George Nurser and Andrew Coward (NOC; ocean physics), attended an OMIP workshop on the 22-23 November 2018 at LOCEAN-IPSL in Paris. The workshop was organised by Julie Deshayes (IPSL) and involved European groups using NEMO as the ocean component of their Earth system models, with a view to analyse and discuss preliminary findings of NEMO-OMIP simulations. Modelling groups represented at the meeting included IPSL (France), CNRM (France), CMCC (Italy), EC-Earth (Spain) and UKESM1 (NOC; UK). Day 1 focused on 1-degree NEMO configurations, and results presented by IPSL, CNRM and NOC found diversity both in NEMO configurations and simulation results. A common issue identified across models was the occurrence of extreme polynyas in the sea-ice zone of the Southern Ocean, which had consequences for the physical and biogeochemical realism of simulations. On day 2, IPSL presented a detailed evaluation of a suite of 1/4-degree NEMO configurations, and discussion focused on coordination of OMIP activity across groups using NEMO. This could include common approaches to freshwater balancing, and the use of a consistent set of JRA-55 atmospheric reanalysis forcing. |
TEAM NEWS
Recent additions to the UKESM Core Group: