| 城市臭氧污染特征与高影响气象因子:以苏州为例 |
| 摘要点击 2397 全文点击 1778 投稿时间：2022-02-25 修订日期：2022-04-21 |
| 查看HTML全文
查看全文 查看/发表评论 下载PDF阅读器 |
| 中文关键词 臭氧(O3)|污染特征|太阳辐射|相关性分析|机器学习|预测模型 |
| 英文关键词 O3 concentration|pollution characteristics|solar radiation|correlation analysis|machine learning|prediction model |
|
| 中文摘要 |
| 近年来城市臭氧(O3)污染问题日益突出,影响O3污染的关键气象因子尚不明确,因此分析典型城市——苏州的O3污染特征,探究O3污染的高影响气象因子,对该区域大气污染防治具有重要意义.基于苏州环境监测中心2015~2020年4~9月逐小时O3浓度数据及同期气象观测资料,应用相关分析和机器学习方法对其开展相关分析研究.结果表明:① 6年间O3污染高发季,O3污染超标率均达20%以上,O3污染日数和以O3为首要污染物的污染日数占比均逐年上升,O3污染问题日益凸显;② O3浓度存在单峰日变化特点,谷值出现在07:00前后,峰值出现在15:00~16:00;其与气温和太阳辐射能的日内变化趋势较一致,但其浓度峰值出现时刻又滞后于二者.2017年和2019年O3有典型的"周末效应",周末较高的太阳辐照度对O3浓度升高有明显的正相关作用.O3浓度和污染超标率的月变化呈双峰型特征;③ O3污染的发生受到多种气象条件的影响,日照时数在7 h以上,气温在30℃附近,太阳辐照度在350~440 kW·m-2之间,相对湿度在50%~75%之间,O3污染日数达最多,且强度达最强.当偏东风风速小于1.5 m·s-1时,或西南风风速小于3.5 m·s-1时,易发生O3中度污染;④运用机器学习方法构建的最优O3浓度预测模型对4、5、7和9月O3浓度有较好的预测能力,当O3浓度超过200μg·m-3时,预报值存在一定程度的低估和平滑,太阳辐射能对于O3浓度的影响最为明显,相对湿度影响次之,而气温和风重要性低于前二者. |
| 英文摘要 |
| The problem of urban ozone (O3) pollution has become prominent in recent years. However, the meteorological factors associated with O3 pollution remain unclear. Analyzing the characteristics of O3 pollution in Suzhou, as a typical urban city, and exploring the high-impact meteorological factors with O3 pollution are crucial to the prevention and control of air pollution in this region. This study used correlation analysis and machine learning methods to analyze the variation in O3 concentration and the relationship between meteorological driving factors in Suzhou based on the O3 concentration data provided by Suzhou Environmental Monitoring Center and the contemporaneous meteorological observation data in Suzhou from April to September in 2015 to 2020. The results showed that: ① O3 pollution exceeding the standard rate was more than 20% in ozone seasons during the past six years; further, pollution days of O3 and the number of pollution days of O3 as the primary pollutant increased yearly. Evidently, the problem of O3 pollution has become increasingly prominent. ② The diurnal variations in O3 were unimodal with the valley point at 07:00 and the highest peak between 15:00 and 16:00. Similar trends were found in diurnal variations of both air temperature and solar radiation, but the daily highest peak came earlier than that of O3. The results also showed an apparent weekend effect of O3 concentration in 2017 and 2019 and a significant correlation between O3 concentration and solar irradiance during the week. In addition, the monthly variation in O3 concentration and pollution exceeding the standard rate was bimodal. ③The occurrence of ozone pollution was affected by various meteorological conditions. The maximum number of days appeared when daily sunshine hours lasted longer than 7 hours, with a daily maximum air temperature around 30℃, solar irradiance ranging from 350 to 440 kW·m-2, and relative humidity ranging from 50% to 75%, at which time the intensity of pollution was the strongest. When the wind speed of easterly wind was less than 1.5 m·s-1, or the wind speed of southwest wind was less than 3.5 m·s-1, moderate ozone pollution occurred. ④ An optimal prediction model of O3 concentration was established based on machine learning, which had good predictive ability for O3 concentration in April, May, July, and September but did not perform well when O3 concentration exceeded 200 μg·m-3. Meanwhile, it was found that solar radiation had the most obvious effect on O3 concentration, followed by relative humidity, whereas the temperature and wind were less important than the former two factors. |
|
|
|
