The use of stochastic models offers many advantages when trying to understand and manage uncertainty in hydrogeological studies. However, many challenges arise when trying to implement these types of models, which might lead to question their benefits given the expected costs.
GSMOD was developed with the objective of making the stochastic modelling process as simple as possible, including the following features:
- Capacity to add calibration-constrained Monte Carlo (or quasi-Monte Carlo) simulation capabilities to existing MODFLOW / MT3D models, independently of the original modelling software.
- Different post-processing options such as stochastic hydrographs with confidence bands (Figure 1) and probability of exceedance maps (Figure 2).
- Capacity to filter out the stochastic realizations that do not represent the observed historic behaviour (poor calibration).
- Control over the model executions, including an heuristic solver approach that detects when a model fails to converge and tries to re-run it with more robust solver parameters.
- All pre- and post-processing tools wrapped within an easy to use GUI (Figure 3).
Figure 1: Concentration hydrograph with confidence band (1000 realizations; 423 of 1000 with NRMS transport < 10%)
Figure 2: Example of probability map of exceeding concentration limit (1000 realizations; 423 of 1000 with NRMS transport < 10%)
Figure 3: GSMOD v1.3 Graphical User Interface
