Courses in model calibration, model uncertainty analysis, and the application of the program PEST are presented by the author of PEST, Dr. John Doherty, throughout the world.  Three-day courses have a strong practical aspect, where participants are guided through exercises designed to illustrate important facets of model calibration, experimental design, and predictive analysis.  At these courses, participants are also encouraged to bring their own models to investigate the use of PEST.  One-day workshops are occasionally presented that serve to introduce attendees to the theory and application of inverse modeling.

Calibration and Predictive Uncertainty Analysis for Complex Models
September 15-19, 2008 - Neuchâtel, Switzerland

General    

A course on parameter estimation and uncertainty analysis will be held in Neuchatel, Switzerland over the week of September 15 – 19.

The course will cover many aspects of modern parameter estimation and uncertainty analysis as it is applied to complex environmental models, including state-of-the-art methodologies introduced to recent versions of PEST.

Topics will include:

• traditional and highly parameterized parameter estimation;

• mathematical regularization using Tikhonov and subspace methods;

• the calibration null and solution spaces;

• use of “SVD-assist” for highly efficient calibration of complex models;

• linear uncertainty analysis for parsimonious and highly parameterized models;

• nonlinear uncertainty analysis;

• calibration-constrained Monte-Carlo analysis (including PEST’s unique and powerful “null space Monte Carlo” method);

• sources of model predictive uncertainty;

• optimal model complexity;

• optimization of data acquisition;

• parameter estimation using derivatives-free methods.

See:-

http://www.unine.ch/chyn/pest2008/

for more details.