Quantifying uncertainty in Pareto fronts arising from spatial data

Moritz Hildemann, Judith A. Verstegen*

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

Multi-objective spatial optimization problems require spatial data input that can contain uncertainties. Via the validation of constraints and the computation of objective values this uncertainty propagates to the Pareto fronts. Here, we develop a method to quantify the uncertainty in Pareto fronts by finding the extreme lower and upper bound of the range of optimal values in the objective space, i.e. the Pareto interval. The method is demonstrated on a land use allocation problem with initial land use (for objectives and constraints) and soil fertility (for one objective) as uncertain input data. Pareto intervals resulting from uncertain land use data were wide and irregularly shaped, whereas the ones from uncertain soil data were narrow and regularly shaped. Furthermore, in some objective-space regions, optimal land use patterns remained relatively stable under uncertainty, while elsewhere they were clouded. This information can be used to select solutions robust to spatial input data uncertainty.
Original languageEnglish
Article number105069
JournalEnvironmental Modelling & Software
Volume141
Early online date26 Apr 2021
DOIs
Publication statusE-pub ahead of print - 26 Apr 2021

Fingerprint Dive into the research topics of 'Quantifying uncertainty in Pareto fronts arising from spatial data'. Together they form a unique fingerprint.

Cite this