HR: 1340h
AN: H43A-0494    [Abstracts]
TI: Pilot Points Method for the Characterization of Heterogeneous Fields: Hero or Villain?
AU: * Alcolea, A
EM: andres.alcolea@upc.edu
AF: School of Civil Engineering. Technical University of Catalonia, 1-3, Jordi Girona Street, Barcelona, 08034 Spain
AU: Carrera, J
EM: jesus.carrera@upc.edu
AF: School of Civil Engineering. Technical University of Catalonia, 1-3, Jordi Girona Street, Barcelona, 08034 Spain
AU: Medina, A
EM: agustin.medina@upc.edu
AF: School of Civil Engineering. Technical University of Catalonia, 1-3, Jordi Girona Street, Barcelona, 08034 Spain
AB: Instability/ill-posedness of the inverse problem causes a large number of difficulties in the calibration of model parameters. Instabilities are attributed to overparameterization. Therefore, efforts have concentrated on reducing the number of unknowns by means of parameterization schemes, such as the pilot points method (PPM). The PPM consists of (1) generating an initial drift by conditional estimation/ simulation to a given geostatistical model (basically, variogram and measurements of the hydraulic property, despite more sophisticated models can be used), (2) parameterization of the values of the respective hydraulic property over the model domain on the basis of their measurements and their values at the pilot point locations (model parameters) and (3) optimizing the values of the model parameters in such a way that the interpolated field (step 2) minimizes an objective function measuring the misfit between calculated and measured data (often, only heads are considered). A very small number of pilot points have been traditionally used to reduce instability at the cost of spatial resolution. We contend that spatial resolution may be important and argue for using a large number of pilot points. Stability is then controlled by a plausibility term that penalizes departures of the pilot point values from their prior information. It turns out that appropriate weighting of the plausibility term is critical. Low weights disregard prior information, leading to instability. Large weights disregard information contained in heads, biasing the solution to the initial drift. A geostatistically-based formulation of the inverse problem helps finding an optimal weight of the regularization term.
DE: 1829 Groundwater hydrology
DE: 1846 Model calibration (3333)
DE: 1869 Stochastic hydrology
DE: 9820 Techniques applicable in three or more fields
SC: Hydrology [H]
MN: Fall Meeting 2005