I validate a geochemical model. Well, sort of...
In a previous post, I provided a snapshot of the GWD2017 conference I recently attended in Stellenbosch (www.gwd2017.com). In this post, I give an overview of my presentation at the conference.
A lot of my work takes place on mining projects, and most of these projects involve estimating the impact of the project on groundwater quantity and quality. Aquifers are complex and the impacts are usually estimated using a numerical model. The value of a model lies in how well it can reproduce reality. This is evaluated in a groundwater model by checking how well it reproduces groundwater levels measured in the field, or even reproduces borehole test pumping results. However, that addresses the quantity aspect.
The quality aspect is often my responsibility on a project. How will a tailings dam, waste rock dump, or pit backfill affect groundwater quality into the future? Unfortunately, my geochemical models often cannot be checked because the system does not exist yet. However, sometimes it is possible, and that was the topic of my conference paper: "Geochemical simulation of drain waer quality in gold tailings".
I developed an ensemble model (an array of models in which output from one (or more) models serves as input to another) of gold tailings that was able to predict actual measurements of drain water quality on a decommissioned tailings dam. That gave some confidence in using the model to predict post-closure drain water quality from a proposed tailings dam - the actual objective of the project.
It was very satisfying to confirm that my model demonstrated some correlation with a real system.