The world is changing fast.
Building a groundwater model has traditionally required that you follow tracks laid out for you by graphical user interfaces or Python libraries. The same applies for important tasks such as simulation-based harvesting of information from field measurements, wherein PEST/PEST++ and a model work together to reduce uncertainties of decision-critical model predictions.
So, traditionally, your modelling destinations were set for you by other people.
With AI you can set your own modelling destination. Once you have done this, you can use AI to build a path to this destination by writing any bits of software that may be required for tasks that are required along the way. These tasks include (but are not limited to):
- generation of stationary/nonstationary stochastic fields;
- giving a model access to these stochastic fields;
- processing field data to model-ready formats;
- creation of information-rich objective functions;
- design of an innovative parameterization scheme;
- supervision of multiple model runs and collection/analysis of model results;
- assembly of models for complex, multi-stage simulation;
- setting up files for PEST and PEST++;
- implementation of data space inversion;
- interfacing a model with GIS and 3D visualization packages;
- calling the MODFLOW 6 API;
- and much more.
But there is one catch. You need to know what you want. This may sound trivial, but it is huge. Learning how to model is no longer about learning how to use appropriate software packages and following workflows that they lay out for you.
It is about knowing what is possible.
It is about knowing where information resides, and how to use a model to harvest that information.
It is about helping people make to decisions by quantifying and reducing uncertainties of key predictions by creating smooth trajectories for flow of information.
Now, you can formulate your own decision-support modelling destination, and pave the path to this destination using software written specifically for that path by AI.
These are exciting times. And we are all learning.
Download a slideshow prepared by John Doherty (author of PEST) in which John documents some of his AI journey. Perhaps this will help you start your own journey.
