OB31 & South Flank: uncertainty analysis

We were commissioned to calibrate an existing groundwater model to predict groundwater impacts associated with the OB31 project. We calibrated the model using singular value decomposition and shuffled-complex evolution techniques to derive the best possible calibration statistics.  A non-linear uncertainty analysis was undertaken using GLUE, DREAM, and null-space Monte Carlo methods.

In addition, we were commissioned to develop a simplified ‘child’ model from a complex regional ‘parent’ model to assess potential groundwater impacts associated with the South Flank Project. We calibrated the model using PEST and performed a linear predictive analysis to determine which parameters and observations were reducing the uncertainty of groundwater drawdown and pit dewatering rates.

A non-linear GLUE uncertainty analysis was performed to assess the complex parameter interactions expected at upper and lower values for the aquifer and recharge parameters.