- Dr Sam Nichol, CSIRO
Note: There is a top-up scholarship funded by CSIRO available for this project for one PhD student who holds a QUT scholarship, an Australian Government Research Training Program (RTP) scholarship, or some similar scholarship. This top-up provides additional stipend ($7,000 per annum) and generous support costs (including travel).
Animals do not follow simple paths when they move through heterogeneous landscapes, so modelling and managing their populations is a complex mathematical problem. In order to optimise planning of population management, we must account for spatial and temporal variability in the location and densities of populations. Fortunately, technological progress has handed us large, real-time datasets of species locations, which provide up-to-date, high resolution information.
Tools from operations research and artificial intelligence can theoretically incorporate this data to produce optimal plans dictating where and when to act. The resulting plans should maximise the probability of achieving a specific management goal in complex environmental systems. However, these tools are designed for small problems, and cannot computationally cope with the enormous size of the new datasets of species movement. Realistic ecological datasets are very large, and very few of these mathematical techniques are computationally efficient enough to deal with this data.
In collaboration with CSIRO, we are studying how to optimally manage species that are threatened (e.g. bilbies) and others that are invasive (e.g. rats), when different levels of data are available. This project is perfectly timed to take advantage of massive increases in available data. New data calls for new tools. Given the increasing ease of data collection, robust and transparent planning should be developed for both data collection, and subsequent application of data for decisions. First, there is a need to improve tactical decision making about where and when to collect data about populations, which can be expensive and should be carefully targeted. Then, once we have enough data to model populations, we need to determine where to implement management actions to control populations to achieve a better ‘bang for your buck’.
These are complex problems, and they will need complex solutions. The project will require a mixture of mathematical modelling, computational simulation, and operations research analysis. Students will need to develop skills to collaborate with scientists from broad disciplines, including mathematics, computer science, biology, and ecology. The project will be supervised by a QUT lecturer in Operations Research, a QUT Assistant Professor in Applied and Computational Mathematics, and a Research Scientist at CSIRO Land and Water.
This project will investigate the use of approximate and online optimisation methods to provide optimal recommendations for real-time control of large datasets at a scale large enough for realistic application (e.g. with structured data such as GIS layers). If successful, the project will unlock powerful optimisation tools that will aid conservation managers to massively improve their management efficiency, making bigger improvements in the environment and sustainability using a finite pool of resources.
Skills and experience
The candidate should have skills (and ideally undergraduate or graduate studies) in mathematical modelling, operations research, computer science, or statistics. Programming skills, or the strong desire to develop programming skills, are essential. Strong written and interpersonal communication skills are essential.
You may be able to apply for a research scholarship in our annual scholarship round.
- Operations research
- Decision science
- Optimal control
- Dynamic programming
- Approximate computational solutions
- Population modelling
- Spatial modelling
Contact Dr Kate Helmstedt for more information.