Changes in wintertime visibility across China over 2013–2019 and the drivers: A comprehensive assessment using machine learning method

Abstract

Effective emission reduction measures have largely lowered the particulate concentration in China, but low-visibility events still occur frequently, greatly affecting people's daily life, travel, and health. In the context of carbon neutrality strategy and climate change, the mechanisms governing visibility changes may be undergoing a transformation. To address this critical issue, we have undertaken a comprehensive assessment by employing a novel approach that combines site observations, model-derived datasets, and machine learning techniques. Our analysis of the dataset shows varying degrees of improvement in wintertime visibility in regions such as North China, South China, and the Fenwei Plain over 2013–2019, but an unexpected deterioration (approximately 1 km yr−1) in central and southern China (CSC). We further elucidate key roles of PM2.5 reduction in these regions with visibility improvement; whereas the unsatisfactory visibility trend in CSC was caused by combined effect of relative humidity (RH) increase (47 %), aerosol hygroscopicity (κ) enhancement (9 %), and boundary layer (BLH) reduction (8 %), which greatly overwhelms the effect of PM2.5 reduction recently. Moreover, the study reveals a growing influence of RH on the wintertime visibility, reaching 40 % ± 24 % to the total contribution in 2019, while that of PM2.5 declined to 18 % ± 19 % and is expected to further diminish with emission reduction. Note those often-neglected factors-temperature, wind speed, BLH, and κ, account for over 40 % of the total contribution. Though the importance of aerosol hygroscopic growth to visibility was found decreasing recently, it retains non-negligible impacts on driving inter-annual visibility trends. This study yields innovative insights for air pollution control, underscoring the imperative of region-specific strategies to mitigate low-visibility events.