Prediction of ozone concentration in ambient air using multivariate methods

Area of Science:

• Environmental Science
• Atmospheric Chemistry
• Statistical Modeling

Background:

• Assessing urban air quality is critical for public health.
• Miskolc, Hungary's second-largest city, faces air pollution challenges due to heavy traffic.
• Understanding the factors influencing ozone concentration is key to developing mitigation strategies.

Purpose of the Study:

• To evaluate ambient air quality in Miskolc using advanced statistical methods.
• To identify key variables and patterns influencing ozone concentrations.
• To model ozone levels and differentiate between photochemical and chemical processes.

Main Methods:

• Multivariate statistical methods were employed, including Principal Component Analysis (PCA) for pattern recognition.
• Modeling techniques such as Multiple Linear Regression (MLR), Partial Least Squares (PLS), and Principal Component Regression (PCR) were utilized.
• Air samples were collected near ground level in a high-traffic area.

Main Results:

• PCA indicated that separating day and night data is essential due to differing factors influencing ozone.
• MLR and PCR demonstrated similar efficiency in modeling ozone when meteorological conditions were stable.
• Excluding nighttime data, PCR and PLS suggested that chemical processes, rather than photochemical ones, were dominant.

Conclusions:

• Day and night data analysis is crucial for accurate air quality assessment.
• Statistical models effectively capture ozone concentration dynamics under specific conditions.
• The interplay between chemical and photochemical processes significantly impacts urban ozone levels.