Multi-Temporal Scale Response Analysis of Ozone Pollution over Eastern and Central China

Ozone (O3) pollution exhibits complex processes and characteristics across short-term, seasonal, and long-term temporal scales. Existing studies have predominantly focused on individual scales, highlighting the need for further exploration of response relationships across different scales. This study focuses on East Asia (105° to 125°E, 20° to 42.5°N), integrating daily three-dimensional field data of O3, NO2, HCHO, and isoprene from the ECMWF EAC4 reanalysis dataset, along with ERA5 meteorological data, covering the period from 2017 to 2025. Treating O3 and its major precursors as an integrated analytical object, a dual-layer Gaussian low-pass filter was applied to decompose standardized pollutant fields into short-term (ST), seasonal (SN), and long-term (LT) components. Joint Empirical Orthogonal Function (EOF) decomposition was then employed to extract spatial modes and principal component time series reflecting the co-evolution of ozone and precursors at each scale. PINN was incorporated to explore potential physical drivers underlying each mode, while rolling multivariate Ordinary Least Squares (OLS) regression and CNN-BiLSTM neural network were jointly applied to investigate cross-scale response intensity β and its time-varying characteristics. Results indicate that: (1) modal coupling relationships across temporal scales are continuously evolving, with traditional single-factor dominated patterns giving way to multi-factor synergistic modes; (2) cross-scale interactions are themselves changing, and neglecting their evolution may systematically misestimate risks; (3) only the first long-term mode exhibits a marginally significant upward trend (p = 0.048), while change point detection at 2019 and 2022 closely coincides with major emission reduction policy milestones. Continued cross-scale research is essential for adaptive ozone control strategies.