Workshop Summary

The Groundwater Resources Association (GRA) conducted a three-day workshop "Model Calibration and Predictive Uncertainty Analysis Using PEST" on April 19-21, 2004. GRA is pleased to announce that the principal instructor for the workshop was Dr. John Doherty, author of PEST and Principal of Watermark Numerical Computing; he was assisted by Matt Tonkin of SS Papadopulos & Associates, Inc. The course provided attendees with a foundation for parameter estimation theory, an understanding of how the PEST code was developed to apply parameter estimation theory, and incorporated extensive demonstrations and "hands-on" computer lab exercises from a variety of environmental disciplines, including applications of PEST to groundwater/surface water modeling.

  About PEST

Parameter Estimation (PEST) is a general-purpose parameter estimation utility developed by John Doherty of Watermark Computing. The course will focus upon the use of PEST, the most advanced available technology for groundwater, surface water, and all other environmental model calibrations. Using PEST you can apply state-of-the-art calibration and predictive uncertainty analysis methods on your everyday modeling applications. Using PEST you can:

• Apply advanced regularization techniques for improved numerical stability;
  
• Undertake nonlinear predictive uncertainty analysis of key model outputs;
  
• Simultaneously parameterize one or a number of models on the basis of multiple datasets including heads, flows and contaminant concentrations;
  
• Accommodate geological heterogeneity using advanced spatial parameterization methods;
  
• Combine PEST with the use of stochastic field generation to explore model parameter uncertainty in heterogeneous systems;
  
• Use time-series analytical routines to process calibration data in the surface water modeling context;
  
• Conduct parallel model optimization runs across PC or UNIX networks;
  
• Convert a MODFLOW-2000 parameter estimation dataset to a PEST dataset by typing a simple command.

Perhaps the most exciting advance in parameter estimation technology over the last few years is the combination of PEST's advanced regularization functionality with the use of pilot points as a spatial parameterization device. When employed in regularized parameter estimation, you can use more pilot points than ever before while maintaining unconditional numerical stability. This allows vastly improved model parameterization in complex geological environments where representation of true or potential heterogeneity may be important, either for accurate model predictions, or for allowing the uncertainty of predictions to be fully explored. When used in conjunction with stochastic field generation to produce a set of calibration-constrained parameter fields, you will have the ability to use models in more powerful and informative ways than previously possible.

  Course Attendee Information

Practitioners from a wide range of experience will benefit from this course, whether new to PEST, or with previous PEST experience. To get the most out of the course, attendees should have some modeling experience, preferably in the groundwater or surface water disciplines. However, hands-on labs are GUI-independent, and cover a variety of modeling disciplines, so that anyone interested in model calibration, parameter optimization, or the analysis of numerical model uncertainty could attend.

  Program Agenda - Complete Course Description & Agenda >>