Monday, 6 December 2021, 11:00-12:00pm
How statistical modelling and machine learning can offer insights about the sustainability of our precious water resources.
Water is a precious resource, essential for agriculture and for drinking. However, water is under increasing global pressure and vulnerability due to rising populations and changing climate patterns. Because Australia is such a dry continent, protecting our water resources is of particularly critical importance and government agencies are working hard to establish and maintain reliable monitoring networks. Modelling the resulting rich data can offer critical insights into the nature and sustainability of this resource so that effective and equitable decisions can be made regarding its usage. In this presentation, we will talk about various approaches to modelling underground aquifers, which are a significant water source for many parts of Australia. After a very brief review of classical process-based approaches based on hydro-geological theory, our main focus will be more data-driven approaches. In particular, we will discuss several strategies including dynamic regression modelling that draw on the theory of time series, as well as machine learning approaches such as neural network models and their extensions to the time series domain. We consider both “local” models focussed on the time-series analysis of a single bore as well as “global” models that incorporate data from many different bores from the same catchment. We will see that there are advantages and disadvantages to the various approaches and that this is an area with rich opportunity for further methodological exploration. We will finish up with a discussion about the potential and limitations of using these modelling strategies to project.
Zoom link: https://macquarie.zoom.us/j/85383656950
Enquiries: Hayley Prescott
Email: hayley.prescott@mq.edu.au
We look forward to having you join us for this special edition of The Moyal Medal.
Our warmest regards,
The Moyal Medal Committee Macquarie University
|