Performance Assessment of Global Optimization Algorithms in Automatic Calibration of Groundwater Models

Master Thesis (2018)
Author(s)

Sruthi Sathyadevan (TU Delft - Civil Engineering & Geosciences)

Contributor(s)

Ronald van Nooijen – Mentor

Koen van der Hauw – Graduation committee member

Nick van de Giesen – Coach

Willem Jan Zaadnoordijk – Coach

Bernard Meulenbroek – Coach

Faculty
Civil Engineering & Geosciences
More Info
expand_more
Publication Year
2018
Language
English
Graduation Date
12-10-2018
Awarding Institution
Delft University of Technology
Faculty
Civil Engineering & Geosciences
Downloads counter
202
Collections
thesis
Reuse Rights

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.

Abstract

Calibration of groundwater models can be done manually by comparing the measured and computed groundwater heads or automatically using algorithms which do the work for you. Modules for this purpose are included in or linked externally to traditional groundwater modelling software and used. One such module PEST (Parameter ESTimation) uses the Levenburg Marquardt algorithm which is a combination of Gauss Newton and Gradient Descent methods to find the minima of a function. However, PEST is capable of finding only a local minimum for the objective function. A possibility exists that there is a global minimum that better fits our function and could give even better results for the optimization problem in, in our case, parameter estimation in groundwater models. A performance assessment has been done with a Genetic algorithm on a groundwater modelling problem in this study.

Files

License info not available