Posted by d.barclay on 22 March 2023 - 11:00am Photo by Michael Webb, University of Cambridge

By SSI Fellow Luke Abraham, National Centre for Atmospheric Science and University of Cambridge.

In my last blog, I discussed using the AWS cloud for hosting virtual machines for use in online training courses that were necessary during the pandemic. With the support of the NERC Advanced Training Short Courses scheme and the Software Sustainability Institute, I was able to be back in the Yusuf Hamied Department of Chemistry’s Computer Lab for a course on the United Kingdom Chemistry and Aerosols model in December 2022.

Back in the lab

Of the many things I’d forgotten about during the pandemic, one was just how much work it is to organise an in-person course, especially one lasting 5 days. I was able to make use of the developments using AWS and be able to connect to these from the managed Linux desktops. However, I also needed to develop scripts to copy the relevant SSH keys to the correct login accounts provided for the students. This kept all the benefits of using a cloud-based system and made it as seamless as possible for the students to use.

I also updated all the training materials and used the Jupyter notebooks for all the python exercises. Previously I had only used the Met Office Iris python library to interface with the model's proprietary file formats. On this occasion, I decided to use the NCAS cf-python library as well. Both libraries produce output files that conform to the Climate and Forecast (CF) Metadata Conventions used in climate modelling. As an Iris user, I had wanted to use cf-python for a while, and this was the perfect opportunity to compare the two libraries. These notebooks also provide a range of examples for the students to use in their work.

Due to the NERC funding, the students did not need to arrange their own accommodation or meals, which meant I needed to find dates that would fit within the Department’s teaching schedule and the availability of suitable accommodation. The first week of December worked for both the Department and Homerton College, which I had used for accommodation before the pandemic.

Catering was more difficult to arrange than I expected due to a reduction in suppliers since January 2020, when I last ran the course in Cambridge. Only one was able to cater for the group's dietary requirements, but one was all we needed! They also had a varied enough menu to ensure that lunch was different from day to day.

With the lecturers scheduled and the catering booked, the students arrived on the 5th of December. The benefits of in-person training were immediately clear, especially for a course of this length. The tasks can be quite involved, and the students were able to work together more easily in the lab environment. During online training, I found that in the Zoom sessions and Slack channel, there was little interaction between students. In this environment, students who had never met before were soon working together to solve problems. 

Being in-person also allowed me to run poster sessions during the week, and showcase posters from students and demonstrators. This encouraged us all to discuss our work beyond the training and enabled demonstrators to help students with their PhD or postdoctoral work. The workshop dinner also allowed the students, lecturers, and demonstrators to interact and get to know each other outside of the main course schedule.

Future courses

While it was tiring and occasionally stressful for me as the organiser, I certainly believe that in-person training is far better for the students. Feedback showed that they enjoyed the course, with an average score of 9.4 out of 10. I also asked whether the students would prefer the course to be online or in-person in future, with in-person courses being the clear favourite. While the technology used to deliver online courses has improved greatly over the past few years, in-person courses can provide a better overall experience.