Calibration and Reliability in Groundwater
Modelling: Credibility of Modelling
(Proceedings of ModelCARE 2007 Conference, held in
Denmark, September 2007). IAHS Publ. 320, 2008, 305-309.
Practical tools for
calibration of transient groundwater flow models
PETER F. ANDERSEN & GREGORY W. COUNCIL
GeoTrans, Inc., 1080 Holcomb Bridge Road, Suite 190, Roswell, Georgia 30076, USA
Abstract Several texts provide guidance for calibrating a numerical groundwater flow model using statistical measures such as mean error, mean absolute error, and root mean squared error. For transient models, these statistics may be computed over time for specific locations, over space for specific time periods, or as a composite that covers both space and time. However, these metrics are not always adequate for transient models. In some cases transient trends or patterns in observed hydrographs are poorly matched, and yet the metrics are acceptable. In other cases, transient patterns are matched well, but a persistent under- or overprediction results in unacceptable metrics. Additional calibration methods and metrics are presented that can improve the calibration process and provide better evidence that the model accurately represents transient behaviour. These methods include: (1) focusing on targets that coincide with periods of extremes in stresses; (2) use of duration and double-mass curves; and (3) using short-term and long-term transient variation as a target (in addition to absolute head or flow). Reasonable calibration criteria can be established for these methods based on the observed hydrographs, and depending on project objectives. Practical examples are provided for real and hypothetical data from groundwater models. These metrics, though posed for groundwater flow models, may also be applied for solute transport models.
Keywords groundwater; models; transient; calibration