A Land Use Regression Model to Estimate Ambient Concentrations of PM10 and SO2 in İzmit, Turkey

Yücer E., ERENER A., SARP G.

Journal of the Indian Society of Remote Sensing, vol.51, no.6, pp.1329-1341, 2023 (SCI-Expanded) identifier

  • Publication Type: Article / Article
  • Volume: 51 Issue: 6
  • Publication Date: 2023
  • Doi Number: 10.1007/s12524-023-01704-1
  • Journal Name: Journal of the Indian Society of Remote Sensing
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Aquatic Science & Fisheries Abstracts (ASFA), CAB Abstracts, Geobase, INSPEC
  • Page Numbers: pp.1329-1341
  • Keywords: Air pollution, Land use regression, Particulate matter, Spatial parameters, Sulfur dioxide cross-validation
  • Süleyman Demirel University Affiliated: Yes


The goal of this study is to develop a land use regression (LUR) model, for estimating the intraurban variation of PM10 and SO2 in a highly dense industrialized city of İzmit, Kocaeli, Turkey. The method allows for the simultaneous consideration of transportation, demography, topography, traffic patterns, road patterns, and land use characteristics as estimators of pollution variability. In the study, PM10 and SO2 Concentrations were obtained hourly from National Air Quality Monitoring Network. The mean annual pollution parameters of 2019 were used to evaluate the temporal differences of estimator variables. 102 sample points were used in the study. 72 of the sampling points were used to establish the LUR model and 30 of them were used to test the accuracy of the model. In the model results, the R square value between the pollutant concentrations of the independent variables was 0.876 for SO2 and 0.919 for PM10. It has been determined that the distance to the roads, the density of the industrial areas, and the population density are the main variables that affect the PM10 and SO2 concentrations. In addition, it has been revealed that meteorological variables are effective in the concentration of pollutants. R square values between the observed and predicted values in the validation analysis of the model were determined as 0.90 for SO2 and 0.94 for PM10.This study showed that it can make accurate estimations about air pollution in areas with complex topographic factors, variable meteorological conditions, and industrial activities.