2016 PEST Training Workshop

California Water and Environmental Modeling Forum

Promoting Excellence and Consensus in Water and Environmental Modeling

Technical Workshop
PEST Training Workshop
(In cooperation with Dr. John Doherty, developer of PEST)

Tuesday through Friday, June 21-24, 2016 9:00am to 4:30pm
Wildland Fire Training & Conference Center
3237 Peacekeeper Way, Room W120
McClellan, CA 95652
Directions

Workshop Fee: $650 for CWEMF members, $850 for non-members, and $600 for students.
Registration through PayPal™ (button below) is requested.
Please email elaine.archibald@comcast.net to reserve your spot.

PEST, and its suite of support software, undertake model-independent calibration and uncertainty analysis for complex environmental models. As such, PEST adds value to the use of models for decision-support, for it can bring to the decision -making process the assessment of risk. To find out more about the PEST suite, visit www.pesthomepage.org. The four day course is led by Dr. John Doherty, the developer of PEST.

The first two days cover basic PEST use. This includes traditional parameter estimation and basic linear and nonlinear predictive uncertainty analysis. More advanced topics are covered over the next two days. These include highly parameterized inversion, the role played by manual or mathematical regularization in the inversion process, spatial parameterization devices such as pilot points, advanced linear and nonlinear uncertainty analysis and model-based hypothesis-testing.

The course includes lectures and workshops. However it is intended that it be as informal as possible so that there will be plenty of time for discussion – especially on topics and models that are important to attendees. All participants will be given a flash drive which contains the latest version of PEST and all of its support software and documentation, together with reading material, and a wide range of practical exercises that they can complete time permitting.

The course is recommended not just for those who wish to use PEST with their models, but also for those who use models as a basis for making decisions and wish to understand more about what models can and cannot achieve in this important context.

The agenda is briefly outlined at the end of this notice. Lectures are interspersed with workshops and discussion sessions.

Course Instructor: Dr. John Doherty, Watermark Numerical Computing

Computer Requirements: Attendees will use their own laptops. Flash drives which contains the latest version of PEST, all of its support software and documentation, and other reading material, will be handed out on the first day of the workshop. Additional instructions may be mailed to attendees at least one week before the workshop.

PEST Workshop Agenda

Time	Lecture	Contents
Day 1 morning	Overview of PEST	Provides an overview of the environmental decision-making process and the role that models, history-matching, and history-matching-constrained uncertainty analysis plays in that process.
Day 1 afternoon	Traditional model calibration: theory	This lecture covers formulation of a well-posed inverse problem and the means through which this problem can be solved using traditional methods.
Day 2 morning	Introduction to highly parameterized inversion	The drawbacks of traditional model calibration and the need for a more highly parameterized approach to history matching are discussed.
Day 2 afternoon	Theory and practice of highly parameterized inversion	This lecture introduces mathematical regularisation. It shows how all environmental inverse problems are ill-posed and explores the consequences for parameterization uniqueness and uncertainty
Day 3 morning	Use of PEST in groundwater model calibration	Spatial parameterization devices such as pilot points, and their use in everyday modelling practice are explored. The importance of innovative objective function formulation is also discussed, as well as ways in which this is best done in the groundwater modelling context.
Day 3 afternoon	Use of PEST in surface water model calibration	The unique problems presented in calibration of surface water models are explored, as well as the opportunities that highly parameterized inversion present for reduction and quantification of predictive uncertainty.
Day 4 morning	Model defects	All models are simplifications of reality. The repercussions of this for calibration-induced parameter and predictive bias are explored, as are ways to mitigate these effects as efficiently as possible.
Day 4 afternoon	Models and decisions	Having covered parameter estimation, predictive uncertainty analysis, and model defects, the repercussions for the use of models in the decision-making process are explored.

How to Register

Send an email to elaine.archibald@comcast.net to reserve your spot.
Select one of the registration options in the box, provide the participant’s name
and contact information in the fields below and click the “Buy Now”
button to be redirected to the PayPal™ web site for credit card payment.

Electronic Registration and Payment (select one)

Participant’s name and organization or affiliation:

Participant’s email address and telephone number:

If PayPal ™ doesn’t work for you, you may send in your check to:
CWEMF, P.O. Box 22529 Sacramento, CA 95822.
Or you can bring a check with you on June 21 made payable to “CWEMF”.
For more information, please email elaine.archibald@comcast.net