New publication on air quality in Barcelona!
A team of researchers from the Barcelona Institute for Global Health has developed and compared three types of models—Land Use Regression (LUR), Dispersion (DM), and a Hybrid model (HM)—to estimate outdoor pollutant concentrations in the city of Barcelona. The study focused on measuring the levels of nitrogen dioxide (NO2), fine particulate matter (PM2.5), black carbon (BC) and some PM2.5 constituents (Fe, Cu, Zn) in search of more accurate methods to assess the health effects of pollution.
During the study, two monitoring campaigns were carried out at different points in the city, measuring the concentration of NO2 at 984 residential addresses, and then the levels of NO2, PM2.5, and BC at 34 key locations. The three models were evaluated for their accuracy and performance through cross-validation and the use of monitoring stations.
The results showed that the models varied significantly in their ability to predict pollutant concentrations. The hybrid model (HM), which used machine learning techniques with Random Forest, showed better performance overall, especially for NO2 and BC measurements, with higher coefficients of determination (R²) than the other models. The LUR model also revealed good results, although less consistent than the HM. Meanwhile, the dispersion model (DM) was the one that showed the worst performance, particularly for the prediction of PM2.5.
The researchers highlighted that these results are crucial for future studies on the impact of air pollution on health, underlining the importance of using the most accurate methods to avoid bias in exposure assessment. This study highlights the need to continue improving methods for measuring air quality in urban areas, particularly in cities with high levels of pollution, such as Barcelona, where the effects of pollution on public health are a growing concern.
You can find the article at the following link: https://www.sciencedirect.com/science/article/pii/S0048969724067883?via%3Dihub
This article was written by Alan Domínguez, a pre-doctoral student in the BiSC project.