Air Quality Forecast by Statistical Methods: Application to Portugal and Macao

Area of Science:

• Environmental Science
• Atmospheric Chemistry
• Data Science

Background:

• Air pollution, including nitrogen dioxide (NO2), particulate matter (PM), and ozone (O3), frequently exceeds safe levels in Portugal and Macao.
• Existing air quality regulations include EU directives and WHO guidelines.
• Air quality forecasting is a vital tool for mitigating pollution impacts and protecting public health.

Purpose of the Study:

• To develop and implement a statistical air quality forecasting model for key pollutants.
• To provide reliable next-day predictions for air pollutant concentrations in specific regions.

Main Methods:

• Utilized a combined approach of Classification and Regression Trees (CART) and multiple regression (MR) analysis.
• Developed optimized regression models for forecasting.
• Applied the methodology to the Greater Lisbon Area, Madeira, and Macao.

Main Results:

• The statistical models successfully forecast PM10, PM2.5, NO2, and O3 concentrations with good performance (R² from 0.50 to 0.89).
• Forecasts demonstrated good agreement with observed concentrations and captured trend evolution.
• The models showed effectiveness in predicting pollution episodes.

Conclusions:

• Statistical air quality forecasting is a successful and effective method for anticipating pollution events.
• This approach aids in implementing preventive measures and safeguarding public health.
• The methodology is robust, having been operational for over a decade with continuous updates.