算术平均法、百分位数法和滑动平均法空气质量评价对比分析及其揭示的信息

综合采用算术平均法、百分位数法和滑动平均法对空气质量进行评价对比分析,可以更加全面、客观、精细、准确地描述空气质量状况,发现其内在规律,揭示不同城市的问题。对比分析发现,全国169个城市空气质量改善程度用算术均值与百分位数测算存在一定差异,并进一步反映出2017年空气质量改善主要归功于污染较为严重天气的减少,2018年主要归功于空气质量良好及轻度污染天气的增加。169个城市PM2. 5浓度的中位数与算术均值之间具有较好的线性关系且越来越强,反映出重污染天气控制取得成效。各区域和城市百分位数同比变化情况反映了大气污染防治不同阶段的工作重点及成效不同:京津冀及周边"2+26"个城市2017年空气质量改善以降低污染天气为主,2018年以增加优良天气为主;珠三角地区和长三角地区都是以增加空气质量为良的天气为主。3年滑动均值比年均值曲线下降更为平缓,更能准确反映污染治理成效。建议深入研究空气质量信息,采用多种评价方法科学、客观、全面地评价污染治理情况。

Comparative analysis of air quality evaluation by arithmetic method,percentile method and moving average method and the information it reveals

Analyzing and assessing air quality by utilizing arithmetic method, percentile method and moving average method can describe air quality condition in a more comprehensive, objective, and precise way, and reveal air pollution situation for different cities. Arithmetic method and percentile method will lead to different results in evaluating air quality improvement of 169 cities, which reflects that air quality improvement in 2017 is mainly attributed to the decrease in days of severe pollution and air quality improvement in 2018 is mainly attributed to the increase in days of fine condition and slight pollution. There is a good linear relationship between median and arithmetic mean of PM2. 5 concentration for 169 cities, and it is getting stronger,which indicates that heavy pollution control has made progress. The percentile changes compared to that of the past years reflect different work priorities and achievement for different stages of air pollution prevention and control. The "2 + 26" cities in Beijing-Tianjin-Hebei and its surrounding areas mainly reduced days of severe pollution in 2017, but switched to increase days of good and fine air quality in 2018. The Pearl River Delta region and the Yangtze River Delta region are both dominated by increasing days of fine air quality. The 3-year moving mean curve is flatter than the annual mean curve, and it can reflect the pollution control achievement more precisely. It is suggested to deep dive into air quality data and information, and utilize multiple evaluation methods to characterize pollution control progress.