By the same authors

How much confidence should we assign to results obtained through Monte Carlo modelling?

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Author(s)

Department/unit(s)

Publication details

Title of host publicationPESTICIDE IN AIR, PLANT, SOIL & WATER SYSTEM
DatePublished - 2003
Pages409-415
Number of pages7
PublisherLA GOLIARDICA PAVESE
Place of Publication27100 PAVIA
EditorsAAM DelRe, E Capri, L Padovani, M Trevisan
Original languageEnglish
ISBN (Print)88-7830-359-3

Abstract

Research investigated the robustness of Monte Carlo techniques when applied to pesticide fate models. Work focused on predictions for leaching of a hypothetical pesticide to groundwater using the PELMO model. The robustness of the Monte Carlo approach was estimated by i) repeating the probabilistic modelling 10 times using different seed numbers (the seed number is used to initiate the generation of random sequences) for various number of model runs; and ii) assessing the dependence of Monte Carlo results on small variations in the procedures for parameterising statistical distributions for Koc and DT50. Monte Carlo results were found to be inherently variable (CV 5 to 211% for 10 replicates) even when a relatively large number of model runs (5000) were undertaken. Results were also found to be affected by slight changes in the parameterisation of probability density functions and in the assignment of correlation between parameters. The levels of variability reported have the potential to affect regulatory decisions. The findings suggest that extreme care should be exercised when undertaking Monte Carlo modelling or assessing the results of probabilistic modelling.

    Research areas

  • uncertainty, risk assessment, probabilistic, Monte Carlo, reliability, DATA SET, UNCERTAINTY, SENSITIVITY

Discover related content

Find related publications, people, projects, datasets and more using interactive charts.

View graph of relations