Validation study of WindTrax reverse dispersion model coupled with a sensitivity analysis of model-specific settings

Francesca Tagliaferri; Marzio Invernizzi; Federica Capra; Selena Sironi

doi:10.1016/j.envres.2023.115401

Validation study of WindTrax reverse dispersion model coupled with a sensitivity analysis of model-specific settings

Environ Res. 2023 Apr 1:222:115401. doi: 10.1016/j.envres.2023.115401. Epub 2023 Jan 31.

Authors

Francesca Tagliaferri¹, Marzio Invernizzi², Federica Capra¹, Selena Sironi¹

Affiliations

¹ Politecnico di Milano, Department of Chemistry, Materials and Chemical Engineering, "Giulio Natta" - P.za Leonardo da Vinci 32, 20133, Milano, Italy.
² Politecnico di Milano, Department of Chemistry, Materials and Chemical Engineering, "Giulio Natta" - P.za Leonardo da Vinci 32, 20133, Milano, Italy. Electronic address: marzio.invernizzi@polimi.it.

PMID: 36731598
DOI: 10.1016/j.envres.2023.115401

Abstract

In last years, atmospheric dispersion models have reached considerable popularity in environmental research field. In this regard, given the difficulties associated to the estimation of emission rate for some kind of sources, and due to the importance of this parameter for the reliability of the results, Backward dispersion models may represent promising tools. In particular, by knowing a measured downwind concentration in ambient air, they provide a numerical value for the emission rate. This paper discusses a critical validation of the WindTrax Backward model: the investigation does not only deal with the strict reliability of the model but also assesses under which conditions (i.e. stability class, number, and location of the sensors) the model shows the greatest accuracy. For this purpose, WindTrax results have been compared to observed values obtained from available experimental datasets. In addition, a sensitivity study regarding model-specific parameters required by WindTrax to replicate the physics and the random nature of atmospheric dispersion processes is discussed. This is a crucial point, since, for these settings, indications on the numerical values to be adopted are not available. From this study, it turns out that the investigated model specific settings do not lead to a significant output variation. Concerning the validation study, a general tendency of the model to predict the observed values with a good level of accuracy has been observed, especially under neutral atmospheric conditions. In addition, it seems that WindTrax underestimates the emission rate during unstable stratification and overestimates during stable conditions. Finally, by the definition of alternative scenarios, in which only a portion of the concentration sensors was considered, WindTrax performance appears better than acceptable even with a small number of concentration sensors, as long as the positioning is in the middle of the plume and not in the strict vicinity of the source.

Keywords: Backward Lagrangian stochastic model; Complex sources; Dispersion modelling; Inverse modelling; Sensitivity analysis; Validation.

MeSH terms

Air Pollutants* / analysis
Climate
Models, Theoretical
Reproducibility of Results

Substances

Air Pollutants