Open Collections

Ozone and aerosol prediction in the Lower Fraser Valley of British Columbia

University of British Columbia

During summer months, the lower Fraser Valley (LFV) of British Columbia experiences elevated concentrations of ground level ozone and aerosol pollution. Environmental agencies often need to make daily air pollution forecasts for public advisories. Previous studies have demonstrated a relationship between pollution episodes and the various weather regimes influencing the LFV. Statistically based, multivariate linear regression models have been applied to capture relationships between meteorological conditions and ambient pollution concentrations. However, the relationship between atmospheric circulation and pollutant concentrations is quite complex and non-linear. Consequently, this study proposes the use of neural network models to capture the inherently complex pollutant-weather relationship, and to investigate their ability to forecast daily pollutant concentrations when compared to traditional regression models. Several neural network and multivariate regression models have been developed for use in Abbotsford, British Columbia, allowing a comparative study of the two approaches. Results illustrate that neural network techniques do outperform those of traditional regression models. However, neural network models do not dramatically increase forecast ability, therefore practical use in air quality management is still in question.

Thesis/Dissertation

application/pdf

2009-06-29

For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use.

http://hdl.handle.net/2429/9793

Atmospheric Science

University of British Columbia

UBCV

DSpace