Spatial models to estimate long-term exposure to NO, NO2, and NOx in the mega-city of Tehran, Iran
Amini, H; Shahri, SMT; Henderson, SB; Hosseini, V; Tsai, MY; Unzli, NK; Yunesian, M
Introduction: Land use regression (LUR) models based on measurement campaigns are frequently used to assess long-term exposure to ambient air pollution. Our aim was to develop LUR models based on fixed-site monitoring station data for estimating annual and seasonal spatial variation in nitrogen oxide (NO), nitrogen dioxide (NO2) and nitrogen oxides (NOx) concentrations in the heavily polluted city of Tehran, Iran.
Methods and Materials: We used hourly 2010 data for NO, NO2 and NOx from 23 fixed sites in the urban monitoring network. A total of 210 geographic variables were generated to potentially explain spatial patterns of air pollution concentrations. A standard approach was developed for LUR modeling given the specific situation of Tehran (Sci. Total Env., 2014, 488: 343–353), and LUR models were developed for annual, cool-season, and warm-season NO, NO2, and NOx.
Results: The annual NO, NO2, and NOx concentrations were 88.1, 53.1, and 141.8 ppb, respectively. The correlation for the annual, cool-season, and warm-season averages, ranged from 0.96 to 0.99 for NO, 0.74 to 0.95 for NO2, and 0.94 to 0.99 for NOx. The correlation of NO and NO2 ranged from 0.24 to 0.44. However, the correlation of NO and NOx ranged from 0.93 to 0.98. The correlation of NO2 and NOx ranged from 0.35 to 0.58. The adjusted R2 values ranged from 0.60 to 0.71 for NO models, 0.58 to 0.68 for NO2 models, and 0.50 to 0.73 for NOx models. The most predictive variables were distance to traffic sources and high traffic areas, distance to primary schools, density of official areas, slope, and elevation.
Conclusions: Our models indicate that spatial patterns did not much vary across the seasons. However, different pollutants, in particular NO and NO2, had independent spatial patterns. Overall, our models performed moderately well. In a next step, we will evaluate whether larger number of sites may lead to better models.
Disclosure: Nothing to disclose.