Webinar Series - Decision-Support Modelling: Principles and Practice
A series of four webinars extending over four weeks will be presented by John Doherty (author of PEST) and Catherine Moore (CSIRO), with help from and Jeremy White (USGS, and coauthor of PEST++ and pyEMU). Admission is free.
Description
The webinar series is supported by the Groundwater Modelling Decision Support Initiative (GMDSI).
The target audience is broad. It includes those who build models, those who commission the building of models, those who review models, and those who are affected by decisions made on the basis of models. It includes those who believe in models and those who are sceptical of models. We attempt to explain important, and sometimes forgotten, principles in ways that are easy to understand. We believe that failure to understand these principles can lead to a pandemic of modelling failures.
We begin with the premise that predicting the future of an environmental system is fraught with uncertainty, especially if management of that system is about to change. Nevertheless, decisions must be made; risks must be assessed. Uncertainties must therefore be quantified. At the same time, uncertainties must be reduced if possible, for this constitutes a return on investments in expensive data. But the blind pursuit of an impressive fit between model outputs and field data is not enough to ensure simulation integrity. Sometimes it is not even wise. Things are a little more nuanced than this.
We discuss how numerical simulation can work best in this conceptual framework. We show how it has the potential to strengthen the decision-making process, but also to undermine it if some fundamental truths are ignored (as they often are). We demonstrate the centrality of data assimilation in achieving its potential. We challenge some of the existing modes of model deployment, and present some viable alternatives - grounded in logic and supported by theory. We provide an overview of current data assimilation and uncertainty analysis technology. We show that it is not “black magic”, or even very hard to understand. Its principles are simple; they are based on common sense.
Our aim is to assist those who build decision-support models to think of ways to build them better, and to empower those who commission or rely on models to ask the right questions of those who build them.
Sessions
The webinars will be presented on Wednesday over four consecutive weeks from 3:00-4:00pm Brisbane Time. The presentation duration is approximately 50 minutes, with 10-15 disucssion time to conclude each session.
Session 1 (20th May 2020) - Groundwater Modelling for Decision Support: Concepts and Fundamentals. Webinar Footage - Click here. Power Point Presentation - Click here
Session 2 (27th May 2020) - Repercussions for Model Construction, Deployment and Reporting
Session 3 (3rd June 2020) - Overview of Data Assimilation Technology – Part 1: Calibration and Linear Analysis
Session 4 (10th June 2020) - Overview of Data Assimilation Technology – Part 2: Uncertainty Analysis and Optimization
Presenters
Who should attend
This is not a modelling course. Nor is it a course on data assimilation. It is a course in logic and common sense as it applies to using numerical modelling for environmental management, and for maximizing returns in data investments.
The principles which we discuss should be understood by modellers. They should also be understood by those who pay for models, and those whose activities are regulated by modelling. They enable non-modellers to ask modellers the right questions; they enable both modellers and non-modellers to understand both the strengths and limitations of numerical modelling as a decision support technology.
Information
Upon registering, you will be sent instructions on how to join the webinar.