Resource Modelling and Model Predictive Uncertainty Assessment

Computer models of fluid movement in underground reservoirs comprise a conceptual-mathematical structure and a set of parameter values which are used to represent a physical reality.  However this physical reality is not fully and directly accessible to us; it is only revealed to us via observations in boreholes and field tests.  

 

On the basis of these observations we form conceptual models of reality but almost always find they lend themselves to multiple conceptual models.   This is particularly so when considering the heterogeneity of subsurface strata, where it is impossible to fully describe the system.   This results in predictions made on the basis of a single model almost always containing some error. 

Describing this uncertainty associated with a model prediction should therefore be a matter of professional integrity for the  modeller, however it presents significant computational and mathematical challenges. 

 

Our research focus is on development of methodologies that allow a robust and rigorous description of model prediction uncertainty for a range of differing modeling problems, that are efficient enough to be used in everyday best modelling.

 

List of projects

 

- Integrated Research for Aquifer Protection (IRAP)

 

Publications

 

Tonkin, M., Doherty, J., and Moore, C. (2007) Efficient nonlinear predictive error variance for highly parametised models.  Water Resources Research (in press April 2007).


Moore C., Doherty J. (2006) The cost of uniqueness in groundwater model calibration. Advances in Water Resources. 29: 605-623.


Moore and Doherty. (2005) Role of the calibration process in reducing model predictive error.  Water ResourcesResearch, Vol. 41(W05020, doi:10.1029/2004WR003501).


Moore, C. (2006) The role of model prediction uncertainty analysis in groundwater allocation.  New Zealand Hydrological Society Conference, Christchurch, November 2006.


Moore, C. (2006) Model Prediction Uncertainty In Groundwater Models.  Proceedings of “NZWWA Water Conference 2006 - Enough for Everyone?”  Oct 11-13, 2006, Christchurch, New Zealand.        


Christensen S., Moore C. and Doherty J. (2006) Comparison of stochastic and regression based methods for quantification of predictive uncertainty of model-simulated wellhead protection zones in heterogeneous aquifers. ModelCARE'2005, Redbook. IAHS Publication 304 (published May 2006) ISBN 1-901-502-58-9


Moore, C. and Doherty J. (2003) Quantifying Model Predictive Uncertainty in Heterogeneous Aquifers (Hypothesis Testing Method).    MODFLOW and MORE 2003: Understanding through Modeling – Conference Proceedings, September 16-19, 2003, International Ground Water Modeling Center (IGWMC) Colorado School of Mines (Poeter, Zheng, Hill and Doherty).