by Claire Carouge
At the start of 2021, Daniel Eisenberg decided to pursue a career change and left CMS. He was replaced in February by Dr. Annette Hirsch. Annette was previously a postdoctoral fellow at the Centre. She has a lot of experience working with the Community Atmosphere Biosphere Land Exchange model (CABLE) and the Weather Research and Forecasting model (WRF). As such, within CMS, she is now in charge of the coupling effort of CABLE to WRF and LIS (the Land Information System).
Since coming to the team, Annette has improved a plotting tool to quickly compare the outputs from LIS and WRF between simulations and with observations and reanalyses. CABLE coupled to WRF is also now working at low resolution (50km) with an updated version of CABLE. This updated version is based on the version of CABLE used in ACCESS-CM2 but contains additional improvements and bug fixes. Work is underway for solving issues with higher resolutions. For licensing reasons, please contact the CMS team at firstname.lastname@example.org for access to the model.
ECMWF has published a retraction notice for a large portion of the ERA5 data. Considering the advice from ECMWF that the retracted data is not fit for research, the data has been immediately removed from the official ERA5 collection at NCI. NCI will endeavour to download the corrected data in a timely fashion, please be patient if your work is affected.
Please note that the erroneous data was not removed from the temporary collection for ERA5 maintained by the Centre as this collection will be decommissioned on 28 April and should not be used anymore. The decommission follows the plan to hand over the maintenance of the ERA5 collection to NCI. All the data CLEX had in the temporary collection is now available in the official ERA5 collection.
This quarter we continued some work with the ACCESS-ESM1.5 model. Scott Wales had to review the land use change part of the setup for the Last Millennium simulation. Some issues with the original setup prevented the model from advancing through the years. This has been corrected and the Last Millennium simulation is now being generated. Holger Wolff is also starting work on porting the SSP585 configuration for ACCESS-ESM1.5 to the payu setup. It might be worth remembering the submissions to CMIP6 with ACCESS-ESM1.5 were done using an old scripts-based workflow. Configurations of interest for further work will first need to be ported to the payu workflow system which requires a bit of time. CMS can help with porting those configurations upon request.
There has been and will be a few changes impacting CLEX personnel at NCI which CMS has and will continue to participate in. First, the virtual machine Accessdev which allows us as well as CSIRO and the Bureau of Meteorology to run the ACCESS models has been updated. This update has solved a few stability issues that had developed during 2020. Unfortunately, a few other issues are still being worked out. Second, to lower the amount of knowledge necessary to run a simulation at NCI, Aidan has developed the qcost tool. Given the characteristics of a simulation for the PBS system (the walltime, the number of cpus, the queue etc.), qcost will return the cost of that simulation in SU. This way users can compare the cost of different choices (eg. the queue used) without knowing all the details themselves. Lastly, we have started a process to reorganise the CLEX projects at NCI to realign those projects with the new Research Programs of the Centre. This means some CLEX users will be moved to a different project in the coming months.
Finally, some work has started to strengthen our data management practices. Our data manager, Paola Petrelli, has been reviewing our data training. This revised training will be offered to all of our Centre’s staff during the year. In parallel, work has started to create the CLEX code collection and the CLEX data collection. These will allow us to present the data and codes produced by CLEX in a unified location independent of where the code resides or where the data has been published.