Incorporating linguistic, probabilistic, and possibilistic information in a risk-based approach for ranking contaminated sites

Authors

  • Kejiang Zhang,

    Corresponding author
    1. Department of Civil Engineering, University of Calgary, 2500 University Drive NW, Calgary, Alberta T2N 1N4, Canada
    • Department of Civil Engineering, University of Calgary, 2500 University Drive NW, Calgary, Alberta T2N 1N4, Canada.
    Search for more papers by this author
  • Gopal Achari,

    1. Department of Civil Engineering, University of Calgary, 2500 University Drive NW, Calgary, Alberta T2N 1N4, Canada
    Search for more papers by this author
  • Yuansheng Pei

    1. State Key Laboratory of Water Environment Simulation, School of Environment, Beijing Normal University, 19 Xinjiekouwai Street, Beijing, People's Republic of China
    Search for more papers by this author

Abstract

Different types of uncertain information—linguistic, probabilistic, and possibilistic—exist in site characterization. Their representation and propagation significantly influence the management of contaminated sites. In the absence of a framework with which to properly represent and integrate these quantitative and qualitative inputs together, decision makers cannot fully take advantage of the available and necessary information to identify all the plausible alternatives. A systematic methodology was developed in the present work to incorporate linguistic, probabilistic, and possibilistic information into the Preference Ranking Organization METHod for Enrichment Evaluation (PROMETHEE), a subgroup of Multi-Criteria Decision Analysis (MCDA) methods for ranking contaminated sites. The identification of criteria based on the paradigm of comparative risk assessment provides a rationale for risk-based prioritization. Uncertain linguistic, probabilistic, and possibilistic information identified in characterizing contaminated sites can be properly represented as numerical values, intervals, probability distributions, and fuzzy sets or possibility distributions, and linguistic variables according to their nature. These different kinds of representation are first transformed into a 2-tuple linguistic representation domain. The propagation of hybrid uncertainties is then carried out in the same domain. This methodology can use the original site information directly as much as possible. The case study shows that this systematic methodology provides more reasonable results. Integr Environ Assess Manag 2010;6:711–724. © 2010 SETAC

Ancillary