Modelling interventions influencing nutrient loads into Port Phillip Bay using Bayesian Networks
Dr. Steven Mascaro, Dr. Mitzi Bolton, Mr. Ross Pearson.
Every year nutrients and other pollutants enter Port Phillip Bay through many different routes. Some of these nutrients benefit the marine environment, for example, becoming part of the food chain. However, too much of these can also harm it, for example, encouraging marine pests to grow at the expense of weaker marine plants and animals. Working with government water experts, we are using Bayesian Networks (a form of causal probabilistic modelling) to understand what can be done to reduce the amount of nutrients entering the Bay, and work out which of those things should be done first.
Related SDGs and targets: 6.3, 6.5, 11.6, 12.4, 14.1, 14.2
PhD Research: Measuring Environmental Indicators using Satellite Data and Machine Learning
Ms. Lynn Miller; Supervisors: Prof. Geoff Webb, Assoc. Prof. Chris Rüdiger
Building on deep learning techniques, this project seeks to improve the utility of live fuel moisture content (LFMC) estimates from Earth observation data by addressing two problems. The first is how to forecast future LFMC levels, so these estimates can be used in advance planning for a bushfire season. The second problem is how to adapt models learnt in one region of the world to another region, to enable these models to be used in regions without enough field data to train models.