Details of Award
NERC Reference : NE/M006255/1
NCAS Training for the Unified Model
Training Grant Award
- Lead Supervisor:
- Dr G Lister, University of Reading, Meteorology
- Grant held at:
- University of Reading, Meteorology
- Science Area:
- Atmospheric
- Overall Classification:
- Atmospheric
- ENRIs:
- Environmental Risks and Hazards
- Global Change
- Natural Resource Management
- Pollution and Waste
- Science Topics:
- Atmospheric Kinetics
- Boundary Layer Meteorology
- Large Scale Dynamics/Transport
- Radiative Processes & Effects
- Water In The Atmosphere
- Abstract:
- The Unified Model (UM) is a software tool developed jointly by the Met Office and the academic community for numerical weather prediction, climate studies, Earth-System modelling, atmospheric chemistry interactions, and more. The model comprises many components including; the computer code; the infrastructure surrounding the central computer code enabling the user to interact with the model to configure it to perform specific experiments; the system to configure the model to run on one of many supercomputer platforms; the supporting software to undertake post-processing and analysis of the model output; and the tools to manage, maintain, and develop the whole system as an efficient software engineering structure. Post-graduate students and in general those new to numerical modelling generally have little or no experience of working with complex software systems. The UM is a very powerful tool, but with that power comes a high degree of complexity which often proves daunting to new users and in some cases has proved too difficult to overcome. The UM training proposed here has been developed over several years to address issues which cause beginners particular difficulty and to help them recognize classes of problems and how to either solve the problems or how to communicate the appropriate issues to the relevant support teams for resolution. We will provide the trainees with accounts on ARCHER (the national supercomputer service) where the majority will be running their models, and accounts on PUMA, the Reading-hosted server acting as a central resource for UM jobs. NCASCMS maintain a UM help-desk with a ticketing system which records all help queries and their resolutions. We have used our knowledge of common issues raised as help-desk queries in developing the training program in order to target common stumbling blocks. Trainees will gain first-hand experience using PUMA and ARCHER and a thorough understanding of the relationship between them and how the UM system interacts with them; no other UM training offering provides this highly targeted delivery nor this level of access to compute resources. On completion of the course, the trainee will be able to build and run any FCM-based version of the UM and will have a working knowledge of UM files and utilities sufficient to allow them to do science with their model output; users intending to develop the model will have a working knowledge of the software engineering tools required to do so. In addition, they will have the confidence to approach an appropriate support team, NCAS-CMS, ARCHER support, or the Met Office for example, to effectively explain issues they can not resolve themselves. The training comprises lecture-style delivery of material covering the following: Introduction to the UM system UM File Formats UM pre- and post-processing tools Flexible Configuration Management System interleaved with a significant practical component summarized as follows: Set up Setting up the PUMA and ARCHER environments Verification of PUMA - ARCHER communications UMUI (Unified Model User Interface) Exploring the UMUI Creating and copying experiments and jobs Comparing jobs, viewing job histories Importing and Exporting jobs PUMA-ARCHER ARCHER architecture Job submission and output Errors resolved on PUMA (a series of exercises designed to expose common pitfalls) Errors resolved on ARCHER (more advanced issues including compilation and run-time errors) Model logging behavior Supercomputer processor decomposition Diagnostic output Check-pointing Model submission infrastructure and build configurations Pre- and post-processing tools Rose infrastructure Flexible Configuration Management Tutorial Create branches and working copies Commit changes to the repository Resolve merge conflicts
- NERC Reference:
- NE/M006255/1
- Grant Stage:
- Completed
- Scheme:
- Doctoral Training
- Grant Status:
- Closed
- Programme:
- Advanced Training
This training grant award has a total value of £31,520
FDAB - Financial Details (Award breakdown by headings)
| Total - Other Costs |
|---|
| £31,520 |
If you need further help, please read the user guide.