Introduction to model calibration, uncertainty analysis and predictive analysis
In this three-day hands-on course DHI WASY will provide you with an introductory training in FePEST. FePEST is the graphical user interface of PESY for FEFLOW model.We will explain and discuss the basic PEST definitions required for understanding the FePEST interface and achieving FePEST best practice.
The training covers topics of automatic model calibration, sensitivity analysis and uncertainty analysis (Monte Carlo methods). Practical sessions will be provided so that attendees can gain experience in setting up a FePEST project and the required preliminary preparation in FEFLOW.
With FePEST, the FEFLOW package now features a graphical user interface for easy integration of your FEFLOW models into the PEST framework. FePEST covers the essential PEST methods for automatic calibration and uncertainty/predictive analysis of FEFLOW models such as efficient definition of parameters (including the pilot-point method) and parallelization across local and wide area networks, interaction with the FEFLOW programming interface and immediate feedback of the PEST optimization progress.
- Best practice how to set up a FEFLOW model for PEST/FePEST
- FePEST graphical user interface (definition of observations/parameters)
- Parallelisation settings, FePEST Server and cloud computing (start/stop slave manipulation)
- Pilot points, regularization techniques and prior information (preferred values, preferred parameter distributions and spatial parameter correlation)
- Analysis and interpretation of results in FePEST and FEFLOW (sensitivity maps, parameter identifiability maps, analysis of data worth, etc.)
- General linear predictive uncertainty and error analysis
- Sensitivity analysis (SENSAN utility in PEST)
- Predictive analysis (definition of worst and best case scenarios)
- Introduction to predictive analysis (definition of worst and best case scenarios)
- Monte Carlo and Null Space Monte Carlo methods
- Advanced FePEST operations (IFM-implemented parameters/observations, customised PEST setups, command-line operation, Python manipulations, etc.)
- Hands-on exercises and open discussion
TARGET GROUP AND PREREQUISITES
Groundwater professionals working in consulting companies, public authorities, university and research institutions. Participants are expected to have both a basic knowledge of groundwater modelling as well as computer application.