Near-Surface CAMS Ozone Variability in Northwestern Libya: A Physical–Statistical Analysis and Implications for Environmental Sustainability

Tropospheric ozone (O3) is a secondary pollutant governed by nonlinear interactions among nitrogen oxides (NOx), volatile organic compounds (VOCS), and meteorological processes. This study calculated the diurnal and nocturnal variability of near-surface ozone using Copernicus Atmosphere Monitoring Service (CAMS) data, at the 1000 hPa level over a coastal-marine grid cell of northwestern Libya, during 2021. A total of 2920 three-hourly observations; were analyzed using Ordinary Least Squares (OLS) and Generalized Least Squares (GLS) models incorporating AR(1) errors, seasonal effects, and cyclic diurnal components. The annual mean ozone concentration was 92.381 µg m⁻³, with higher daytime concentrations (105.571 µg m⁻³) than nighttime concentrations (79.191 µg m⁻³). Observed concentrations ranged from 27.562 to 202.125 µg m⁻³, while the mean diurnal maximum occurred around 14:00 Local Time (118 µg m⁻³). Concentrations were generally higher during spring and summer, coinciding with elevated temperature and oxidant levels. The final GLS models yielded Pseudo-R² values of 0.694 for the full study period, 0.827 for daytime, and 0.682 for evening/nighttime. For the full-period model, adding the cyclic diurnal component reduced Pseudo-R² slightly from 0.702 to 0.694, while lowering AIC from 20424.903 to 20222.483. However, RMSE increased slightly from 16.368 to 16.592 µg m⁻³. These results indicate that the diurnal component improved the relative fit–complexity trade-off according to AIC, but did not improve all performance metrics. The findings therefore support evaluating temporal model components using multiple criteria rather than relying on a single goodness-of-fit measure.