Analyzing the association between meteorological and air pollution conditions over time and space in Baden-Württemberg (Germany)

Abstract

Background: Understanding the relationships between environmental stressors and relevant variables in health data analysis is necessary and challenging. Due to our earth observation and modelling capabilities, we can access daily meteorological data for the entire state of Baden-Württemberg in Germany. These data include information on meteorological variables (precipitation, temperature, ultraviolet radiation (UV), vapor pressure (VP)) and air pollution variables (PM2.5, PM10, nitrogen dioxide (NO2), ozone (O3)). Baden-Württemberg is a state with both rural and urban areas. This study aims to delve deeper into the connections and interactions among meteorological and air pollution factors. Methods: We conduct a temporal analysis using cross-correlations and a spatial analysis using Local Indication of Spatial Association (LISA). Results: There is a strong connection between meteorological and air pollution factors that follow distinct spatial and seasonal patterns. An example of a strong interdependent relationship is between NO2 and O3, with the Pearson correlation coefficient varying over time. The variables show a negative correlation in January (-0.84), April (-0.47), and October (-0.54), with varying coefficients. July (0.45) shows a positive correlation. The correlation direction of O3 and NO2 changes depending on temperature and UV radiation, confirmed through cross-correlation analysis. In terms of location, urban areas have significantly higher levels of NO2, PM2.5, and PM10 compared to rural areas. O3 experiences the reverse effect. LISA analysis confirmed the presence of distinct hot and cold spots for various environmental stressors. Furthermore, according to linear regression analysis, PM10 is influenced by PM2.5, and vapor pressure is affected by temperature. Conclusion: To better understand the complex interrelationships, it is important to consider multiple stressors. However, some variables can be excluded to reduce the dimensionality of the analysis. For example, analyzing PM2.5 proved to be sufficient for this particular study. While PM10 has larger particles, its characteristics are almost identical to PM2.5.