This keynote talk to the RSE Geoscience SIG described some of the activities of what is one of the oldest university based RSE groups in the UK - the NCAS Computational Modelling Services (CMS) group.

Presentation: pdf (3.3 MB)

I began by trying to motivate why climate modellers need the biggest possible computers, and always want a bigger one, and why that leads to data problems.

I then introduced what is needed for academia to run (big) climate models in the UK environment, and the role of CMS in supporting that - from managing the PUMA system on ARCHER2 that orchestrates model workflows, to managing data on JASMIN.

I introduced the roles of the current CMS team and then showcased a selection of projects, from the massive logistical effort of the CANARI large ensemble, through our incipient efforts to get a climate model port to Grace Hopper, to model/data toolkits, and our recent upgrades to and acquisition of responsibility for pyfive (a pure python HDF5 reader).

Some of the take home points were about the scale of the data challenge, and the size of the teams necessary to do leading edge climate modelling.