Modern numerical models allow for simulation of complex, coupled processes and geometrically detailed representations of subsurface geology.

However, all predictions made by computer simulations of subsurface processes are subject to uncertainty. We provide an approach to numerical modelling that:

  1. Focuses on the prediction or decision of interest;
  2. Considers uncertainty in the predictions from the start; and
  3. Tailors model construction, data collection and calibration methods to reduce the uncertainty in key model-based decisions.

For simulation of groundwater flow, mass and heat transport, the finite element FEFLOW program and the finite volume MODFLOW family of codes are the principle modelling software packages used. Other specialty software packages are used to simulate more complex or coupled processes such as surface water/groundwater interactions, and density dependent flow and transport simulations.

We use PEST, PEST++, and PyEMU software, along with extensive custom programming in Python to facilitate model calibration, predictive uncertainty analysis and model-based decision support. For over five years, we have employed novel cloud-computing based solutions to access the high performance computing resources necessary for quantitative risk based uncertainty analysis within a reasonable cost for our clients.