Improving ozone estimation during rainy-warm seasons from the perspective of weather systems based on machine...

Surface ozone pollution in eastern China is increasingly serious during summer, coinciding with distinct stages of the rainy seasons in this region. This study focuses on the spatiotemporal distribution of ozone concentrations, their synoptic driving factors and estimation during the Meiyu periods from 2015 to 2022. Results show that high ozone levels mainly occur during the interval of Meiyu season (HOP), accounting for 15.1 % of the total Meiyu period, with MDA8 O₃ concentrations in HOP exceeding those in non-HOP (NHOP) by over 20 μg m^-3 at >60 % of sites. Compared to NHOP, HOP is characterized by lower precipitation (-0.17 mm), higher temperatures (0.25 K), increased solar radiation (3.16 kJ m^-2), and reduced relative humidity (-8.56 %). Using the Lindeman, Merenda, and Gold method, it is found that the meteorological factors with the greatest impact on MDA8 O₃ during the 2015-2022 Meiyu season are relative humidity (30.4 %), surface 2 m temperature (28.6 %), and total precipitation (23.2 %). These favorable conditions for ozone production are along with specific state synoptic-scale and large-scale atmospheric circulation systems. On a daily scale, the further south the position of the western Pacific subtropical high pressure ridge (72.2 %) and the stronger the intensities of El Niño-Southern Oscillation and East Asian summer monsoon (19.0 %) will lead to higher O₃ concentration in the Meiyu-affected area. The results of machine learning prove that instead of conventional consideration of meteorological conditions, the introduction of daily atmospheric circulation indices can effectively improve the accuracy of ozone concentration estimation (R²: 0.72, NRMSE: 0.15). These findings provide valuable insights for predicting ozone pollution during the rainy interval period.