Dalarna University's logo and link to the university's website

Understanding the bidirectional relationship between air pollution and meteorological variables is crucial for advancing urban climate science and informing environmental policy. This thesis investigates the causal interactions between fine particulate matter (PM₂.₅) and relative humidity (RH) in Quito, Ecuador, a high-altitude city facing unique air quality challenges due to its topography and rapid urbanisation. Using 20 years of hourly data (2004–2024) from four monitoring districts (Belisario, Cotocollao, Carapungo, and El Camal), the study applies Convergent Cross Mapping (CCM), a nonlinear Empirical Dynamic Modeling method, to detect and characterise causal relationships.

The analysis focuses on two representative 100-day windows per district, comparing early and recent years to explore temporal evolution. CCM reveals statistically significant, bidirectional causal links between PM₂. ₅and RH across all sites, with notable shifts over time. In the early 2000s, RH more strongly influenced PM₂.₅, while in recent years, the reverse direction grew stronger, indicating that pollution may indeed now be altering meteorological conditions. Surrogate testing confirms that observed causal signals are not mere products of shared seasonality or random fluctuations. Partial derivative analysis further uncovers state-dependent and evolving interaction strengths.

These findings highlight the growing feedback between pollution and meteorology in Quito and underscore the limitations of traditional correlation-based analyses in complex urban systems. By demonstrating the utility of CCM for environmental causality detection, this study advances understanding of pollution and climate interactions and provides actionable insights for urban resilience planning and air quality management across the world.