This paper presents an attempt to model the trend of emissions through analysis of a time series of PPM10 concentrations in Christchurch, New Zealand. Emissions are not constant over time, but show high seasonality. Fluctuations are removed by creating a time series in which concentrations do not show dependency on ambient air temperature. Remaining meteorological influences are removed through multiple linear regression. Finally, a moving average filter is applied to reveal the low-frequency trend in the residuals of the meteorologically adjusted time series. The modelled trend shows a peak in emissions in 2001-2002 with a steady decrease thereafter.
Keywords: particulate matter, PM10, air pollution meteorology, multiple linear regression, New Zealand, air quality, time series analysis, pollutant emissions, home heating