Software

Model Partner Software

Numerical simulation on its own provides only limited support for the making of decisions. However, when simulation is undertaken in conjunction with modern-day simulator partner software, a simulator’s parameters can host information that is harvested from field data. At the same time, a model ‘s parameters can express the stochastic consequences of information insufficiency. Management-critical simulator predictions can then express uncertainty while benefiting from uncertainty reduction accrued by simulator-based information harvesting.

So strategic use of simulator partner software in conjunction with simulation itself is an essential component of modern day, scientifically based environmental decision-making.

Simulator-partner software must possess two fundamental specifications for it to assist numerical simulation in realizing its full decision-support potential. These are:

  • a non-intrusive interface with a model
  • an ability to parallelize model runs.

Programs belonging to the PEST and PEST++ suites possess these specifications. To find out more about these suites, as well as software packages that support their use, follow the links below.


How GMDSI is Trying to Help

There are a number of major obstacles to widespread adoption of simulator partner software by the groundwater industry at large. These include the following:

  • A lack of understanding of the tasks performed by simulator partner software, and how performance of these tasks can raise the scientific integrity of environmental management;
  • A lack of knowledge of the theory on which simulator partner algorithms are based;
  • Difficulties in construction of input dataset for simulators and simulator-partner software, especially in contexts where parameters and observation are many and diverse;
  • The large numerical burden that is sometimes incurred by use of simulator partner software.

GMDSI has tried to address the first of these issues by producing a series of monographs and videos that examine decision-support modelling from a scientific/philosophical perspective. The second and third issues are being addressed through education, community-building and continual enhancement of the PEST, PEST++ and PyEMU suites.

In order to address the fourth of these issues, considerable research and development is being devoted to improving the algorithmic bases of the PEST and PEST++ suites. Ongoing developments include:

  • Superior expression of hydrogeological uncertainty, particular as this pertains to connected subsurface permeability;
  • Reduction of model run requirements for implementation of data assimilation and uncertainty quantification;
  • Improved optimization of management outcomes, with full account taken of the uncertainties that are associated with some of these outcomes, and with constraints that these outcomes must respect.