Weather Normalized Modelling for Air Pollution Forecasting

Sammanfattning

Air pollution has emerged as a pressing global challenge, posing significant risks to public health, the environment, and economic stability. Accurate forecasting of air quality is crucial for enabling effective mitigation strategies and protecting vulnerable populations. This study aims to develop a robust Weather Normalized Model (WNM) that integrates meteorological data with air quality information to enhance the accuracy of air pollution predictions in Quito, Ecuador. The research employs a comprehensive methodology that combines data collection, preprocessing, and the development of a Gradient Boosting Machine model. Local air quality data from the Atmospheric Monitoring Network of Quito (REMMAQ) and global weather data from OpenWeatherMap are utilized to train and validate WNM. The study also compares the performance of the WNM against traditional forecasting models, both with and without weather normalization, to assess the added value of incorporating meteorological factors. The results indicate that the Weather Normalized Model (WNM) significantly outperforms the traditional model that does not integrate meteorological factors. This improvement underscores the importance of incorporating weather data into air quality forecasting models. The study also conducted experiments to assess the effectiveness of combining local air quality data with global weather data sourced from OpenWeatherMap. This approach aimed to determine whether utilizing global weather forecasts could enhance the accuracy of air pollution predictions. Furthermore, the research explored the potential of combining local data with a simple and traditional weather forecasting model, such as ARIMA, to address the challenges posed by limited access to real-time weather data for local sources. Overall, these experiments collectively demonstrate the effectiveness of weather normalization and the strategic integration of diverse data sources in improving air quality forecasting models. The successful implementation of the WNM in Quito can serve as a blueprint for other urban centres facing similar air quality challenges, particularly in developing countries where resources and data availability may be limited. This study represents a significant step towards improving public health outcomes, supporting sustainable urban planning, and mitigating the environmental and economic impacts of air pollution.