Spatiotemporal modelling of air pollution in the Valencian Community using spectral analysis and daily geostatistical mapping

Air pollution constitutes one of the leading modifiable environmental risks to public health, and its evolution can be influenced by meteorological variability and climate change. In this context, air quality in Mediterranean coastal regions exhibits strong seasonal variability and pronounced spatial gradients, driven by urban emissions, meteorology, and complex topography. In this work, the spatiotemporal evolution of 12 atmospheric pollutants in the Valencian Community (Spain) is modelled over 2009–2023 at daily resolution, and a reproducible and transferable methodology for the spatiotemporal analysis of environmental variables is presented, applicable not only to air pollution but also to other datasets with spatial and temporal components.

Data from the Valencian Air Quality Monitoring Network (RVVCCA), collected at 87 monitoring stations, are used. Prior to processing, an exhaustive quality assessment is conducted to validate record consistency, identify and handle missing or anomalous data, and robustly establish the effective study period. Temporal variability is analysed using time-series techniques, including descriptive exploration (basic statistics, variability, and seasonal patterns) and spectral analysis, in order to identify periodicities and dominant signals across different temporal scales.

Spatial analysis follows the methodology proposed by Yao and Journel (1998), which enables the automatic computation of covariance surfaces via the Fast Fourier Transform (FFT), facilitating operational implementation for a large number of days and pollutants. Based on these covariance surfaces, a geostatistical estimation scheme based on Ordinary Kriging (OK) is implemented to generate daily gridded concentration fields over the Valencian Community, enabling the assessment of persistent spatial patterns, regional gradients, and their interannual evolution. The final outcome is an integrated workflow that combines quality control, temporal analysis, and daily geostatistical mapping, providing a solid methodological basis to study the spatiotemporal dynamics of air pollution from monitoring networks.