Feb Webinar - Demystifying uncertainty and its policy repercussions webinar and May Webinar - So we've ticked the uncertainty box. What happens next?
2023 Uncertainty Webinars
Two upcoming GMDSI webinars explore the topics of groundwater model uncertainty analysis, and how decisions can be made with uncertainty taken into account.
Demystifying uncertainty and its policy repercussions
Date: 28th February 2023 | Time: 1:30 PM - 3:00 PM Adelaide
This webinar will have two presenters.
The first is Chris Li (from CDM Smith). Chris delivered a talk entitled 'The “Cinderella Syndrome” of groundwater modelling, and overcoming it through risk-orientated uncertainty analysis' at the recent Australian Groundwater Conference.
It was highly acclaimed. So we asked him to make his talk a little longer, and allow the rest of us to benefit from hearing it. Here is what he will talk about in his own worlds.
Have you ever wondered about the following questions?
- “This groundwater model can fit field observations really well, but how much faith can we put in its predictions?”
- “Can we perform stochastic modelling without blowing project budgets?”
- “How can we make our uncertainty analysis easier to understand for non-technical stakeholders?”
Chris will explore these topics using real-world project examples.
The second speaker will be Joel Hall from Western Australia Department of Water and Environmental Regulation. Joel summarises his talk as follows.
The challenges that a regulator faces when attempting to use uncertainty analysis in decision-making are of a different nature to that of quantifying uncertainty using a numerical model. In a general sense, the major challenges are non-technical (although there are many technical obstacles that need to be overcome) but are more likely to be institutional and/or arise from gaps in policy. In this session we will discuss some of the major issues and propose a roadmap to make the use of uncertainty analysis in regulatory decision-making the status quo. This will include subjects such as the general lack of guidance in risk-based decision-making where there is a small or medium risk of a “bad thing happening” as well as some of the more realpolitik challenges of changing policy.
So we've ticked the uncertainty box. What happens next?
Date: 9th May 2023 | Time: 1:30 PM - 3:00 PM Adelaide
John Doherty, Jeremy White and Catherine Moore will each talk on issues that occupy the boundaries between uncertainty analysis and decision-making/policy-formulation. They will discuss some of the problems that beset the making of decisions in an uncertain world. The talks are non-technical; the issues are important. They include:
- Uncertainties in uncertainty;
- Do decision-makers really want to put a number on the risks that they are taking?
- Are decisions really based on risk anyway?
- Can there ever be a guidebook for uncertainty analysis, or is it too subjective?
- Necessary and sufficient conditions for evaluation of uncertainty;
- Linkages between uncertainty analysis and monitoring;
- Modelling as a collective, social process.
New Release: Tutorial - Data Space Inversion
GMDSI has just released a new tutorial. This is somewhat similar to the one that preceded it. However this new tutorial includes an extended section on data space inversion (DSI).
Data space inversion can be used to evaluate the uncertainties of predictions made by complex models with complicated parameter fields. It requires an astonishingly small number of model runs.
You will also learn how you can use PEST utilities for tasks such as:
- generating random parameter fields;
- creating covariance matrices to use in regularisation and uncertainty analysis;
- creating pictures of grids, zonation, random parameter fields and particle tracks.
The tutorial includes all files and executables, as well as complete documentation.
Watch this space!
Two new worked examples to be released soon.
Two new worked examples will be published early next year. They are complete. We are just awaiting final approval for their release. They will cover:
- Use of the MODFLOW 6 API in optimizing managed aquifer recharge under uncertainty.
- Use of structural overlay parameters in stochastic simulation of inflow to a tunnel.
PEST Conference
Don’t forget about the upcoming second PEST conference. This will be held in San Diego, California. Subtitled “The Path from Data to Decisions”, it will run from March 6th to March 9th 2023.
The PEST conference will be preceded by a two-day course on theory and practice of PEST/PEST++; it will be followed by a one-day course on automation of PEST++ setup using PyEMU and other Python tools. Attendance at these courses is cheap; prices only need to cover costs.
For more information on the PEST conference, visit the following site:
