Consultancy
NeSI delivers specialist computational science expertise to the research sector, embedding NeSI’s team members within research teams during the course of a project. NeSI experts come from a variety of science domain backgrounds, have extensive programming experience, and can assist with code optimisation, parallelisation, custom code development, and many other tasks. Through these collaborations, NeSI lifts researchers’ productivity, efficiency and skills using research computing tools and resources.
Strengthening our specialist science capabilities
During 2018, NeSI focused on broadening uptake of its Consultancy service by a wider range of organisations and improving visibility of the work undertaken by the NeSI Consultancy team. The efforts paid off, with 2018 being the first year that NeSI succeeded in identifying and working alongside research groups from all Collaborators (University of Auckland, NIWA, University of Otago, and Manaaki Whenua – Landcare Research) and the first time NeSI worked with AgResearch on a Consultancy project.
The Consultancy team also looked closely at how the service is promoted, with the goal of ensuring all NeSI researchers are aware of the Consultancy service and how to access it. Existing processes were optimised and new tactics were put in place, including presenting talks at various conferences, advertising the service within the compute allocations process, and adding information about Consultancy to the platform’s login interface.
With the growing opportunities for researchers in data science, NeSI wanted to build its internal expertise in this area. Unfortunately, we were unable to identify a suitable candidate for a data science role in our team, but recruitment is likely to continue in 2019.
Consultancy projects in 2018
| Project name | Contributions and benefits realised | Science outcomes | Principal investigator |
|---|---|---|---|
| Global Climate Change | This project enabled the use of the second global climate model that will have been run and tested in New Zealand. The results from this project can be compared with those obtained from the HadGEM model currently used by NIWA. Without this consultancy project the researcher would not have been able to carry out this research on NeSI. | Understanding the relationships between atmospheric CO2, water vapour, clouds, and global circulation for better prediction of climate change. Investigating how changes in CO2 level is linked to changes in atmospheric humidity and clouds, which feedback on the radiative energy budget of the Earth’s climate system, and in turn affecting climate sensitivity. | Tra Dinh University of Auckland, Department of Physics |
| Optimisation of tracking algorithm for precipitation systems | Code is now able to run on the new platforms. Without these changes the researcher will not be able to run the full analysis and therefore will not be able to publish results from this work. The researcher was also upskilled in version control and software engineering practices. Performance gains of 3.5 - 5.5x were realised against the initial working version. | Tracking coastal precipitation systems in the tropics. Coastal precipitation plays an important role in the economy of island nations. Impacts from too much or too little precipitation can range from losses in agricultural productivity, to unexpected infrastructure costs, to spikes in sales of particular products or services. | David Coppin University of Auckland, Department of Physics |
| NZESM Model Development | NIWA NeSI team member helped migrate the climate model to the new platforms. Accelerated the set-up of the New Zealand Earth System Model (NZESM) on the new platforms; improved efficiency; reduced core-hour usage; ensured downstream users get data as quickly as possible. 10.3 million core hours usage on the new platforms in 2018. | Producing useful climate simulations using a hierarchy of models. The NZESM is at the core of the Deep South Challenge, a large-scale science challenge project that investigates climate change and its consequences for New Zealand. | Olaf Morgenstern NIWA, Deep South Challenge |
| Code Optimisation for Genotyping-By-Sequencing | Researchers learnt how to optimise and parallelise R codes and have already applied this to other codes they are working on. Performance improvements of up to 20x speed up. | Incorporating new computational methods and tools into analysis of genetic relatedness and construction of genetic linkage maps. | Jeanne Jacobs AgResearch |
| Porting Community Earth System Model (CESM) to NeSI | Assisted in code porting, so that the researcher could use NeSI. This involved comparing available compilers and choosing the best performing one. CESM is a community code, so having it available on the NeSI platforms should be useful for other researchers going forward. | Exploring how changes in the sun, particularly an effect known as energetic particle precipitation or “solar wind”, influences things like ozone balance in the polar atmosphere and how that impacts other climate elements. | Annika Seppälä University of Otago, Department of Physics |
| Modelling discretized point processes with extra zeros and long-range dependence | Consultancy enabled better throughput and efficiency on the NeSI platforms and support for more complex models and larger datasets. Performance improvements of ~3x speed up overall. This project enabled the development of more advanced models for earthquake modelling. Code has also been shared with the broader scientific community. | Modelling the first of its kind in the type of seismic events analysis. Will help forecast large destructive earthquakes. | Ting Wang University of Otago, Department of Maths and Statistics |
| GRATE Sediment Transport Calculator | This project is underway but it is expected that it will increase the usability of the code, and make it suitable for running on NeSI platforms. We are also aiming to improve performance of the code. | GRATE is a research tool that can be applied to numerous practical problems in NZ rivers. It will be an effective educational tool, as well as a valued resource for assessing field data. | Jon Tunnicliffe University of Auckland, School of Environment |
| Generation of sub-Poissonian radiation fields of large photon number | Introduced research software engineering best practices by implementing a new build system, test cases and improving the accuracy of the calculations. Improved the performance of individual simulations, saving 39% of run time. Implemented scripts for automating parameter sweeps involving running large numbers of simulations in parallel, which will save time compared to manually running those simulations. | This code will be applied to simulations of open quantum systems, in particular simulating the fundamental interaction of light and matter. | Howard Carmichael University of Auckland Department of Physics |
| Monitoring land cover changes with TMASK_SENTINEL | The code was parallelised, resulting in a 53x speedup over the original, serial version. The master - worker approach implemented here can be used as a framework to parallelise other similar codes (upskilling the researcher). | The code is intended to play an important role in monitoring land use in New Zealand and overseas. | James Shepherd Manaaki Whenua - Landcare Research |
| Conservative Interpolation for Forecasting Weather Hazards | A method for conservative interpolation was developed during this consultancy project, enabling the remapping (regridding) of vector fields from and to arbitrary unstructured grids made of quadrilateral cells. | Achieve conservation of water and energy when regridding vector fields between different grids, e.g. from the cubed-sphere grid used by the weather/climate LFRic code to a longitude-latitude grid. Can be used to compute the amount of water flux and energy entering a region over time to much higher accuracy than was previously the case. | Michael Uddstrom NIWA |
