Methods used for handling and quantifying model uncertainty of artificial neural network models for air pollution forecasting. (December 2022)

• Record Type: Journal Article
• Authors: Cabaneros, Sheen Mclean; Hughes, Ben
• Is Part Of: Environmental modelling & software. Volume 158(2022)
• Publication Date: 2022-12
• Subjects: Air pollution forecasting -- Artificial neural networks -- Uncertainty quantification -- Bayesian -- Monte Carlo simulation -- Fuzzy; Environmental monitoring -- Computer programs -- Periodicals; Ecology -- Computer simulation -- Periodicals; Digital computer simulation -- Periodicals; Computer software -- Periodicals; Environmental Monitoring -- Periodicals; Computer Simulation -- Periodicals; Environnement -- Surveillance -- Logiciels -- Périodiques; Écologie -- Simulation, Méthodes de -- Périodiques; Simulation par ordinateur -- Périodiques; Logiciels -- Périodiques; Computer software; Digital computer simulation; Ecology -- Computer simulation; Environmental monitoring -- Computer programs; Periodicals; Electronic journals; 363.70015118
• DOI: 10.1016/j.envsoft.2022.105529
• Languages: English
• ISSNs: 1364-8152

Abstract

Abstract: The use of data-driven techniques such as artificial neural network (ANN) models for outdoor air pollution forecasting has been popular in the past two decades. However, research activity on the uncertainty surrounding the development of ANN models has been limited. Therefore, this review outlines the approaches for addressing model uncertainty according to the steps for building ANN models. Based on 128 articles published from 2000 to 2022, the review reveals that input uncertainty was predominantly addressed while less focus was given to structure, parameter and output uncertainties. Ensemble approaches have been mostly employed, followed by neuro-fuzzy networks. However, the direct measurement of uncertainty received less attention. The use of bootstrapping, Bayesian, and Monte Carlo simulation techniques which can quantify uncertainty was also limited. In conclusion, this review recommends the development and application of approaches that can both handle and quantify uncertainty surrounding the development of ANN models. Highlights: Methods for handling uncertainty of ANN-based air pollution models were reviewed. Uncertainty surrounding model inputs was predominantly addressed. Ensemble and neuro-fuzzy approaches were popularly adopted. The use of methods that can both handle and quantify model uncertainty was limited.