Accuratepredictionofsurface-levelozone(O3.)concentrationsisessen- tial for ensuring efficient air quality control and safeguarding public health. This study proposes a flexible spatiotemporal modeling framework to estimate daily mean O3. levels across Italy by integrating satellite-derived ozone data with high-resolution environmental predictors and in situ observations. The model relies on a linear regression with dynamic intercepts and slopes to relate in situ O3. measurements to satellite estimates, accounting for additive and multiplicative biases. The time- and space-varying parameters are modeled within a generalized additive model (GAM) framework to capture complex relationships between ozone and environmental drivers. Model evaluation and cross-validation demonstrate superior explanatory and predictive performance compared to other alternatives. The framework identifies key bias patterns linked to elevation, nitrogen dioxide concentrations, and seasonal variation. Overall, the proposed model provides an accurate, scalable, and interpretable framework for obtaining harmonized databases, facilitating environmental monitoring and informing policy development.
Ozone Predictions Through a Generalized Additive Model
Cappello C.
;Palma M.;De Iaco S.
2025-01-01
Abstract
Accuratepredictionofsurface-levelozone(O3.)concentrationsisessen- tial for ensuring efficient air quality control and safeguarding public health. This study proposes a flexible spatiotemporal modeling framework to estimate daily mean O3. levels across Italy by integrating satellite-derived ozone data with high-resolution environmental predictors and in situ observations. The model relies on a linear regression with dynamic intercepts and slopes to relate in situ O3. measurements to satellite estimates, accounting for additive and multiplicative biases. The time- and space-varying parameters are modeled within a generalized additive model (GAM) framework to capture complex relationships between ozone and environmental drivers. Model evaluation and cross-validation demonstrate superior explanatory and predictive performance compared to other alternatives. The framework identifies key bias patterns linked to elevation, nitrogen dioxide concentrations, and seasonal variation. Overall, the proposed model provides an accurate, scalable, and interpretable framework for obtaining harmonized databases, facilitating environmental monitoring and informing policy development.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
