Event Summary: Ways to mix knowledge and data to solve land and water problems

On the afternoon of Tuesday 22 October we welcomed Edoardo Bertone, Senior Lecturer at Griffith University and Jack Coates-Marnane, Principal Scientist – Catchments and Waterways, Healthy Land & Water, to share their knowledge and experience of Bayesian Network models and tools.

Edoardo led us through an overview of:

• Thomas Bayes’ theorem,
• the key concepts of probabilities,
• key elements of Bayesian network structures,
• the interplay of cause/ effect nodes and
• their applications.

We were given the opportunity to play with our ability to guess the probabilities – the mathematicians in the room loved it – and start to map out the key elements to prepare a Bayesian Network for a management response.

With a mix of education and application, there was something for everyone along the QWMN pipeline within Edoardo’s presentation!

After a delicious afternoon tea, Jack Coates-Marnane presented an overview of his knowledge and experience with Bayesian Network applications in water and natural resource management. The most recent being the QWMN RDI project that Healthy Land & Water, Mirror Analytics and QUT undertook to revolutionise riparian management!

It was a special treat to be guided through the resulting visualisation tool, which is still in development, to get a sense of what is possible for guiding riparian land management.

Key takeaways from the session from both Edoardo and Jack’s presentations were:

• these frameworks and tools can be applied for a very field-specific problem, but one of their advantages is to enable modelling of interdisciplinary systems (where other models might fail), by integrating different types of quantitative/expert data
• they allow us to model a system where limited (numerical) data is available
• by allowing us to better quantify the uncertainty behind predictions, their output can be used to prioritise where to focus management efforts, assist in understanding the probability of a scenario and assess the risk associated with an action (not to make final decisions)
• there are a few limitations such as they can’t account for feedback loops.

Thank you again for your time and efforts in presenting such a comprehensive and dynamic session Edoardo and Jack!