This event is the first in a trial seminar series happening on Tuesday 19th January at 11am.
The aim of these seminars is to encourage knowledge sharing and best practice between JASMIN users - whilst showcasing the types of science enabled by JASMIN. These hour-long events will include talks from two users with plenty of time for questions, and will happen approximately monthly.
Details about the first planned talks are below. Joining instructions will be sent to the JASMIN user mailing list ahead of the event. If you would like to join the seminar but are not on the mailing list, please contact the JASMIN helpdesk.
If you would like to volunteer to give a talk in a future seminar, please contact the helpdesk with the subject line 'User Seminar Series'.
Speaker 1: Jon Seddon, Met Office
The PRIMAVERA project aimed to understand the effect of resolution in climate models. It developed a new generation of advanced and well-evaluated high-resolution global climate models, to provide simulations and predictions of regional climate with unprecedented fidelity for the benefit of governments, business and society in general.
Over 1.6PB of climate data, from seven different models, was analysed on JASMIN by over 100 scientists from across Europe. In this talk, Jon will share experiences for moving, analysing and sharing large amounts of project data. It will also cover best practice for dealing with this volume of data (e.g. automating processes, agreeing on standards and data request, estimating data volumes, ensuring data requests are efficiently designed)
The project made use of the following JASMIN/CEDA services:
Interactive sci servers - for developing analysis tools
Various storage - including making use of tape
LOTUS batch cluster - to run analyses on the full dataset
CEDA Archive - to publish the full dataset for long term reuse
CEDA's Earth System Grid Federation node - to share the dataset with the global community
Over 70 peer-reviewed articles have been published from this project to date. Join this talk to find out the lessons learnt from undertaking a project of this scale and how best to use JASMIN to accomplish it.
Speaker 2: Andrew Meijers, British Antarctic Survey
This talk will cover how Andrew has used the Cluster as a Service (CaaS) for a Pangeo deployment to produce results for CMIP6 model analysis. CaaS makes use of the parallel compute facilities (LOTUS), but on a JASMIN Cloud tenancy - thus allowing greater flexibility. Pangeo is an open-source community platform for data analysis for the Earth sciences which makes use of Jupyter Notebooks, Dask and other Python tools and packages.
These JASMIN services are being used to undertake interactive analysis of the Southern Ocean in CMIP6 - a very large climate model dataset - specifically looking at how water masses vary between models and how they change their heat/carbon properties under different climate forcing scenarios.
JASMIN is presently the only location in the UK that allows researchers to use HPC resource in an interactive way via Pangeo tools. The overall architecture of using Pangeo in general, Andrew’s specific JASMIN Pangeo deployment, as well as specific tools for CMIP6 analysis will be discussed in this talk. A live demonstration will (hopefully!) also be given.
Join this talk to find out how this new, and potentially very powerful, technology could help you get more out of using JASMIN.