Modelling air pollution for epidemiologic research--part II: predicting temporal variation through land use...

Area of Science:

• Environmental Science
• Epidemiology
• Spatial Analysis

Background:

• Land use regression (LUR) is a common method for modeling intra-urban air pollution, but rarely models temporal variations.
• Few studies have estimated long-term air pollution concentrations for epidemiological research.

Purpose of the Study:

• To estimate annual mean nitrogen dioxide (NO2) and particulate matter (PM10) concentrations in Greater Manchester from 1996 to 2008.
• To utilize these estimates in the Manchester Asthma and Allergy Study (MAAS) birth cohort for health effect assessments.
• To develop and validate LUR models capable of capturing temporal air pollution variability.

Main Methods:

• Recalibrated a 2005 Greater Manchester LUR model using historical NO2 and PM10 data (1996-2008).
• Incorporated temporally resolved variables, including traffic intensity and PM10 emissions.
• Validated models by comparing estimated concentrations with measured data at automatic monitoring stations.

Main Results:

• Successfully recalibrated LUR models for each year between 1996 and 2008.
• Achieved low mean prediction errors (-0.8 µg/m³ for NO2, 0.8 µg/m³ for PM10) and root mean squared errors (6.7 µg/m³ for NO2, 3.4 µg/m³ for PM10).
• Demonstrated the feasibility of modeling temporal air pollution variation using LUR with minimal error.

Conclusions:

• Land use regression models can effectively capture temporal variations in air pollution concentrations.
• The use of air dispersion model data enabled extrapolation over an extended period, overcoming limitations of short-term monitoring.
• This methodology provides valuable long-term air quality data for epidemiological studies, such as the MAAS birth cohort.