Abstract
Variations in emission patterns can alter the spatial atmospheric gradients, compromising the effectiveness of air quality networks (AQNs). This study aims to develop a methodological approach to evaluate whether the existing AQNs design remains effective over time, applying combined methodologies (correlation analysis, principal component analysis, k-means clustering, and geospatial interpolation). 2006-2021 annual O 3 and PM 10 concentrations from two Community of Madrid AQNs were examined. Gaps in the original time series were estimated, providing adequate performance outcomes (RMSE, MAE, and MAPE values of 1.12 μg/m 3 , 3.94 μg/m 3 , and 6.83% for O 3 , and 0.73 μg/m 3 , 4.34 μg/m 3 , and 13.77% for PM 10 ) and complying with modeling pollutants criteria set in legislation. Identifying the more representative stations within networks reduced the number of stations by 83% (O 3 ) and 77% (PM 10 ). The proposed methodology was tested comparing the annual distribution gradients between the original and proposed AQNs: spatial validating covering from 2006 to 2021, revealing outcomes agreement ranging from 78.41-99.01% (O 3 ) and 73.78-97.01% (PM 10 ), and temporal validating involving 2005 and 2022 (periods did not included in the previous methodological approach), evidencing a minimum spatial similarity of ⁓90%, providing reliability and temporal independence to the approach. Due to differences in the atmospheric chemistry of O 3 and PM 10 , the recommended methodology cannot be applied to both pollutants simultaneously. 2006–2021 meteorological study evidenced interannual stability. This approach may serve as a harmonized methodology to complement the guidelines set by European legislation on air pollutant monitoring and to foster joint efforts among public entities that manage independent networks.