Air pollution modeling using regression models
Închide
Articolul precedent
Articolul urmator
273 4
Ultima descărcare din IBN:
2022-05-13 15:10
Căutarea după subiecte
similare conform CZU
504.3.054:51 (1)
Știința mediului înconjurător (910)
Matematică (1636)
SM ISO690:2012
SVOZILIK, Vladislav, BITTA, Jan. Air pollution modeling using regression models. In: Ecological and environmental chemistry : - 2022, Ed. 7, 3-4 martie 2022, Chișinău. Chisinau: Centrul Editorial-Poligrafic al USM, 2022, Ediția 7, Vol.1, p. 136. ISBN 978-9975-159-07-4.. 10.19261/eec.2022.v1
EXPORT metadate:
Google Scholar
Crossref
CERIF

DataCite
Dublin Core
Ecological and environmental chemistry
Ediția 7, Vol.1, 2022
Conferința "Ecological and environmental chemistry 2022"
7, Chișinău, Moldova, 3-4 martie 2022

Air pollution modeling using regression models

CZU: 504.3.054:51

Pag. 136-136

Svozilik Vladislav1, Bitta Jan2
 
1 Joint Institute of Nuclear Research,
2 VSB - Technical University of Ostrava
 
Disponibil în IBN: 15 martie 2022


Rezumat

Air pollution dispersion modelling using standard Gaussian methodologies is a data-intensive process that requires extensive computing power, high-quality data, and human effort. Land Use Regression (LUR) is an appropriate alternative to standard modelling. LUR work with the presumes that air pollution concentration concludes from environmental factors. These factors are evaluated using spatial analysis and selected based on the ability to express air pollution variability. The main benefits of the LUR modelling are lower compute and data-intensive and substantial elimination of human factor. The significant drawback of LUR modeling is lower accuracy in comparison with Gaussian methodologies. The standard LUR models use linear regression equations for the estimation of air pollution concentrations. We presume that linear regression is not appropriate for describing nonlinear phenomena such as air pollution. Therefore, alternative LUR model was constructed, and linear regression equations were substituted by Artificial Neural Network (ANN)-based regression, which is appropriate to capture non-linear behavior of the phenomena. The presented study assesses two approaches: 1. Construction of the LUR model based on the linear regression using the Gaussian model results dataset or emission data dataset. 2. Construction of the LUR model based on the non-linear regression using the Gaussian model results dataset or emission data dataset. The accuracy of the estimation was evaluated using the coefficient of determination (R2). LUR models constructed based on the linear regression reached 0.639 for emission data and 0.652 for Gaussian model results data. LUR models constructed based on the non-linear reached 0.937 for the LUR model based on emission data and 0.938 for the model based on Gaussian model results. The LUR model based on non-linear regression provides more accurate results.