长三角地区2014~2023年近地面大气臭氧变化特征

Variation Characteristics of Near-Surface Atmospheric Ozone in Yangtze River Delta Region from 2014 to 2023

摘要: 本研究基于2014~2023年长三角地区近地面臭氧浓度及气象数据，结合随机森林模型探讨了大气臭氧的时空变化特征及其驱动因素。通过整合国家青藏高原科学数据中心的高分辨率臭氧数据(空间分辨率1 km，时间分辨率日/月/年)及NCDC气象数据，采用Mann-Kendall趋势检验、Kruskal-Wallis检验等方法分析了臭氧浓度的年际、季节与月变化规律，并结合空间分布差异揭示了人类活动排放与气象条件的交互作用。结果表明；(1) 年际上，臭氧浓度以年均2.7%的速率显著上升(Sen斜率2.69 μg/m³/年，p < 0.01)，2023年达106.1 μg/m³，其中工业排放与交通源增长是主要驱动因素；季节上，夏季浓度最高(111.7 μg/m³)，冬季最低(69.5 μg/m³)，光化学反应强度的季节性变化及气象扩散条件的差异是主要原因；(2) 空间上呈现“北高南低”的分布，南京、合肥等工业密集城市浓度显著高于沿海地区(如杭州)，北部年均浓度较南部高15%~20%；(3) 气象因子中，温度(r = 0.551)与臭氧浓度呈现强正相关，云覆盖率(r = −0.227)通过抑制光化学反应降低浓度。随机森林模型(R² = 0.72)验证了温度的重要性得分最高(1.630)，模型可为臭氧污染预警提供技术支撑。

Abstract: This study investigates the spatiotemporal variations of surface ozone and their driving factors in the Yangtze River Delta region from 2014 to 2023, based on ozone concentration and meteorological data, using a random forest model. By integrating high-resolution ozone data (with a spatial resolution of 1 km and temporal resolutions of daily/monthly/annual) from the National Qinghai-Xizang Plateau Data Center and meteorological data from the NCDC, methods such as the Mann-Kendall trend test and Kruskal-Wallis test were employed to analyze the interannual, seasonal, and monthly variations of ozone concentrations. Additionally, spatial distribution differences were examined to reveal the interactive effects of anthropogenic emissions and meteorological conditions. The results indicate that: (1) Interannually, ozone concentrations increased significantly at an average annual rate of 2.7% (Sen’s slope = 2.69 μg/m³/year, p < 0.01), reaching 106.1 μg/m³ in 2023, with industrial emissions and traffic sources being the primary drivers. Seasonally, concentrations were highest in summer (111.7 μg/m³) and lowest in winter (69.5 μg/m³), primarily due to seasonal variations in photochemical reaction intensity and differences in meteorological diffusion conditions. (2) Spatially, a “higher in the north, lower in the south” distribution pattern was observed, with industrially dense cities such as Nanjing and Hefei exhibiting significantly higher concentrations than coastal areas like Hangzhou. The annual average concentration in the northern region was 15%~20% higher than in the southern region. (3) Among meteorological factors, temperature (r = 0.551) showed a strong positive correlation with ozone concentrations, while cloud coverage (r = −0.227) reduced concentrations by inhibiting photochemical reactions. The random forest model (R² = 0.72) confirmed the highest importance score for temperature (1.630), demonstrating the model’s potential to provide technical support for ozone pollution warnings.