Research on circulation classification and atmospheric pollution mechanisms in Gansu based on machine learning

doi:10.11755/j.issn.1006-7639-2025-04-0563

Abstract

Abstract:

To reveal the influence mechanisms of weather patterns and meteorological factors on air pollution in arid and semi-arid regions, this study utilized the fifth-generation atmospheric reanalysis dataset (ERA5) provided by the European Centre for Medium-Range Weather Forecasts (ECMWF) from 2016 to 2022. The Self-Organizing Map (SOM) neural network was employed to classify weather patterns based on the 700 hPa geopotential height field and wind field, combined with quadratic curve fitting to analyze the nonlinear relationships between meteorological factors and pollutants in typical cities across different climatic zones of Gansu Province. The results indicate that: 1) PM₁₀ and PM_2.5 mass concentrations generally exhibit a negative correlation with temperature, while for O₃ its mass concentration increases nonlinearly with rising temperature. Under low wind speeds (<1 m·s^-1) and high wind speeds (>4 m·s^-1), PM₁₀ and PM_2.5 mass concentrations are elevated, which attributed to local accumulation under calm conditions and dust transport under strong winds, respectively. In contrast, wind speeds between 1 and 4 m·s^-1 favor precursor accumulation, leading to higher mass concentration of O₃. Pollutant concentrations are generally higher under relative humidity between 25% and 75%, though regional differences exist. For instance, in Jiuquan, the PM₁₀ concentration is higher at humidity below 25% due to frequent dust events, the PM_2.5 concentration increases with the rise of relative humidity due to the hygroscopic growth effect, while O₃ concentration decreases with declining humidity owing to reduced consumption under dry conditions. 2) In winter and spring, the dominant weather patterns are the southwestern high-pressure type and the eastern trough type. Under the southwestern high-pressure pattern, strong northwesterly winds in western regions create pollutant transport pathways, whereas the eastern trough pattern results in poor diffusion conditions and stable atmospheric conditions in Gansu, facilitating pollutant accumulation and significantly increasing of PM₁₀ and PM_2.5concentrations. 3) In summer and autumn, high-pressure systems dominate, with abundant solar radiation and high temperatures elevating the boundary layer height. Coupled with warm, moist air transport, these conditions provide a favorable environment for photochemical reactions, leading to higher O₃ concentrations.

Key words: air pollution, synoptic weather patterns, meteorological elements, machine learning

摘要：

为揭示干旱半干旱地区天气形势与气象要素对大气污染的影响机制，基于2016—2022年欧洲中期天气预报中心（European Centre for Medium-Range Weather Forecasts，ECMWF）提供的第五代大气再分析数据集（ERA5），采用自组织映射神经网络（Self-Organizing Map，SOM）对700 hPa位势高度场和风场进行天气分型，并结合二次曲线拟合解析甘肃不同气候区典型城市气象要素与污染物的非线性关系。结果表明：1）PM₁₀和PM_2.5质量浓度与气温整体呈负相关，而O₃质量浓度随气温升高非线性增加；低风速（<1 m·s^-1）和高风速（>4 m·s^-1）下，PM₁₀和PM_2.5质量浓度较高，在静风及强风时分别因局地累积和沙尘输送导致颗粒物质量浓度升高，而1~4 m·s^-1的风速有利于前体物积累，导致O₃质量浓度升高；25%~75%的相对湿度条件下污染物质量浓度较高，但其影响存在区域差异，如干旱区的酒泉，在湿度小于25%条件下由于易发生沙尘天气PM₁₀质量浓度较高，PM_2.5由于吸湿增长作用，质量浓度随相对湿度升高而增加，O₃在低湿条件下的消耗降低，其质量浓度随相对湿度升高递减。2）冬春季，以西南高压型和东部低槽型为主导，西南高压型下西部强西北风形成污染物输送通道，东部低压槽型下甘肃地区扩散条件差，其大气形势较为稳定，污染物易积累，导致PM₁₀和PM_2.5质量浓度显著升高。3）夏秋季，以高压型为主导，充沛的太阳辐射与高温条件促使边界层高度抬升，配合暖湿气流输送，为光化学反应提供有利环境，导致O₃质量浓度较高。

关键词: 大气污染, 天气分型, 气象要素, 机器学习

CLC Number:

P448

LIU Zongrui, WAN Ziyue, ZHAO Yuhan, LIU Weiping, WANG Ruoan, MA Yuxia. Research on circulation classification and atmospheric pollution mechanisms in Gansu based on machine learning[J]. Journal of Arid Meteorology, 2025, 43(4): 563-575.

刘宗瑞, 万紫悦, 赵宇瀚, 刘卫平, 王若安, 马玉霞. 基于机器学习的环流分型与甘肃大气污染机制研究[J]. 干旱气象, 2025, 43(4): 563-575.

Figures/Tables 11

Fig.1 Proportion of primary pollutants in 14 cities of Gansu Province from 2016 to 2022

Fig.2 The fitting relationship between meteorological elements in Lanzhou and the mass concentrations of PM10， PM2.5 and O3 （The circle colors represent the mass concentration ranges of PM10， PM2.5 and O3 at different IAQI levels，the same as below）

Fig.3 The fitting relationship between meteorological elements in Jiuquan and the mass concentrations of PM10， PM2.5 and O3

Fig.4 The fitting relationship between meteorological elements in Qingyang and the mass concentrations of PM10， PM2.5 and O3

Fig.5 The fitting relationship between meteorological elements in Longnan and the mass concentrations of PM10， PM2.5 and O3

Fig.6 The 700 hPa geopotential height field （the color shaded， Unit： dagpm） and wind field （arrow vectors， Unit： m·s-1） based on SOM classification

Fig. 7 Seasonal distribution of weather patterns

Fig.8 Box plots of the mass concentrations of three pollutants under nine circulation patterns in typical cities （The dots represent the mean values， the red lines denote the median， the upper line of the box indicates the upper quartile， the lower line of the box signifies the lower quartile， the upper edge represents the maximum value， the lower edge corresponds to the minimum value， and the triangles denote outliers）

Fig.9 The vertical cross-sections of potential temperature （the color shaded， Unit： K） and wind （arrow vectors， Unit： m·s-1） under different circulation patterns over Lanzhou and Qingyang （The purple lines represent the boundary layer height， the same as below； the brown and green triangles indicate the longitudes of Lanzhou and Qingyang， respectively）

Fig.10 The vertical cross-sections of potential temperature （the color shaded， Unit： K） and wind （arrow vectors， Unit： m·s-1） under various circulation patterns over Jiuquan （The brown triangle indicates the longitude of Jiuquan）

Fig.11 The vertical cross-sections of potential temperature （the color shaded， Unit： K） and wind （arrow vectors， Unit： m·s-1） under various circulation patterns over Longnan （The brown triangle indicates the longitude of Longnan）

References 45

[1]	陈锦超, 2017. 合肥近地面臭氧浓度分布特征及影响因素[D]. 合肥: 中国科学技术大学.
[2]	胡春梅, 陈道劲, 周国兵, 等, 2020. 基于自组织神经网络算法的重庆秋冬季空气污染与天气分型的关系[J]. 气象, 46(9):1222-1 234.
[3]	国家环境保护部, 2012a. 环境空气质量标准(GB3095—2012)[S]. 北京: 中国环境科学出版社.
[4]	国家环境保护部, 2012b. 环境空气质量指数(AQI)技术规定(试行)(HJ633—2012)[S]. 北京: 中国环境科学出版社.
[5]	黄瑶, 陶丽, 刘新超, 等, 2021. 大渡河上游强降水的环流分型及时空分布特征[J]. 沙漠与绿洲气象, 15(4):58-67.
[6]	吉莉, 刘晓冉, 2023. 重庆中心城区气象因子对大气污染影响的研究[J]. 高原山地气象研究, 43(2):113-119.
[7]	李顺姬, 李红, 陈妙, 等, 2018. 气象因素对西安市西南城区大气中臭氧及其前体物的影响[J]. 气象与环境学报, 34(4):59-67.
[8]	马敏劲, 陈然, 曹译丹, 等, 2024. 卷积神经网络研究进展及其在大气科学中的应用[J]. 干旱气象, 42(5):719-733. DOI
[9]	王翠连, 张军, 郑瑶, 等, 2019. 郑州城区PM₁₀、PM_2.5质量浓度变化特征及其对气象因子的响应[J]. 环境保护科学, 45(6):76-83.
[10]	吴进, 李琛, 马志强, 等, 2020. 基于天气分型的上甸子大气本底站臭氧污染气象条件[J]. 环境科学, 41(11):4864-4 873.
[11]	杨雅涵, 翟盘茂, 周佰铨, 2024. 基于SOM的长江流域持续性强降水过程典型环流的客观分型[J]. 气象学报, 82(5):632-644.
[12]	张宸赫, 赵天良, 陆忠艳, 等, 2020. 2014—2019年沈阳市大气污染物变化及气象因素影响分析[C]// 中国环境科学学会科学技术年会论文集(第一卷). 北京: 中国环境科学学会.
[13]	张淑平, 韩立建, 周伟奇, 等, 2016. 冬季PM_2.5的气象影响因素解析[J]. 生态学报, 36(24): 7 897-7 907.
[14]	张莹, 2016. 我国典型城市空气污染特征及其健康影响和预报研究[D]. 兰州: 兰州大学.
[15]	赵熠琳, 李子, 沈劲, 等, 2024. 气象要素与前体物排放对中国区域性臭氧污染的影响[J]. 环境污染与防治, 46(6):876-881.
[16]	赵天良, 张郁青, 宁贵财, 等, 2024. 四川盆地臭氧污染时空变化及平流层入侵影响研究进展[J]. 高原山地气象研究, 44(4):11-20.
[17]	AGATHOKLEOUS E, FENG Z Z, OKSANEN E, et al, 2020. Ozone affects plant, insect, and soil microbial communities: A threat to terrestrial ecosystems and biodiversity[J]. Science Advances, 6(33): eabc1176. DOI: 10.1126/sciadv.abc1176.
[18]	APTE J S, BRAUER M, COHEN A J, et al, 2018. Ambient PM_2.5 reduces global and regional life expectancy[J]. Environmental Science & Technology Letters, 5(9): 546-551. DOI: 10.1021/acs.estlett.8b00360.
[19]	BEI N F, LI X P, TIE X X, et al, 2020. Impact of synoptic patterns and meteorological elements on the wintertime haze in the Beijing-Tianjin-Hebei region, China from 2013 to 2017[J]. Science of the Total Environment, 704: 135210. DOI: 10.1016/j.scitotenv.2019.135210.
[20]	CHAMBERS S D, PODSTAWCZYŃSKA A, 2019. Improved method for characterising temporal variability in urban air quality part II: Particulate matter and precursors in central Poland[J]. Atmospheric Environment, 219:117040. DOI: 10.1016/j.atmosenv.2019.117040.
[21]	CHEN Z Y, XIE X M, CAI J, et al, 2018. Understanding meteorological influences on PM_2.5 concentrations across China: A temporal and spatial perspective[J]. Atmospheric Chemistry and Physics, 18(8): 5 343-5 358. DOI: 10.5194/acp-18-5343-2018.
[22]	CHENG B W, MA Y X, ZHAO Y H, et al, 2024. Influence of topography and synoptic weather patterns on air quality in a valley basin city of Northwest China[J]. Science of the Total Environment, 934: 173362. DOI: 10.1016/j.scitotenv.2024.173362.
[23]	GAO Z Q, DO K, LI Z R, et al, 2024. Predicting PM_2.5 levels and exceedance days using machine learning methods[J]. Atmospheric Environment, 323: 120396. DOI:10.1016/j.atmosenv.2024.120396.
[24]	GONG S L, LIU Y L, HE J J, et al, 2022. Multi-scale analysis of the impacts of meteorology and emissions on PM_2.5 and O₃ trends at various regions in China from 2013 to 2020 1: Synoptic circulation patterns and pollution[J]. The Science of the Total Environment, 815:152770.DOI:10.1016/j.scitotenv.2021.152770.
[25]	GUAN Q Y, LI F C, YANG L Q, et al, 2018. Spatial-temporal variations and mineral dust fractions in particulate matter mass concentrations in an urban area of northwestern China[J]. Journal of Environmental Management, 222: 95-103. DOI: 10.1016/j.jenvman.2018.05.064. PMID
[26]	HU F, XIE P H, XU J, et al, 2025. Impacts of synoptic weather patterns on Hefei’s ozone in warm season and analysis of transport pathways during extreme pollution events[J]. Journal of Environmental Sciences, 156: 371-384. DOI: 10.1016/j.jes.2024.06.032.
[27]	HU X M, HU J, GAO L, et al, 2021. Multisensor and multimodel monitoring and investigation of a wintertime air pollution event ahead of a cold front over Eastern China[J]. Journal of Geophysical Research: Atmospheres, 126(10): e2020JD033538. DOI: 10.1029/2020JD033538.
[28]	HUANG J, PAN X, GUO X, et al, 2018. Health impact of China’s Air Pollution Prevention and Control Action Plan: An analysis of national air quality monitoring and mortality data[J]. The Lancet Planetary Health, 2(7): e313-e323. DOI: 10.1016/S2542-5196(18)30141-4.
[29]	JIN X P, CAI X H, HUANG Q Q, et al, 2021. Atmospheric boundary layer-free troposphere air exchange in the North China plain and its impact on PM_2.5 pollution[J]. Journal of Geophysical Research: Atmospheres, 126(9): e2021JD0 34641. DOI: 10.1029/2021JD034641.
[30]	JUN M J, GU Y, 2023. Effects of transboundary PM_2.5 transported from China on the regional PM_2.5 concentrations in South Korea: A spatial panel-data analysis[J]. PLoS One, 18(4): e0281988. DOI: 10.1371/journal.pone.0281988.
[31]	KANG H Q, ZHU B, LIU X H, et al, 2021. Three-dimensional distribution of PM_2.5 over the Yangtze River Delta as cold fronts moving through[J]. Journal of Geophysical Research: Atmospheres, 126(8): e2020JD034035. DOI: 10.1029/2020JD034035.
[32]	LI J D, LIAO H, HU J L, et al, 2019. Severe particulate pollution days in China during 2013-2018 and the associated typical weather patterns in Beijing-Tianjin-Hebei and the Yangtze River Delta regions[J]. Environmental Pollution, 248: 74-81. DOI: 10.1016/j.envpol.2019.01.124. PMID
[33]	LI M G, WANG L L, LIU J D, et al, 2020. Exploring the regional pollution characteristics and meteorological formation mechanism of PM_2.5 in North China during 2013-2017[J]. Environment International, 134: 105283. DOI: 10.1016/j.envint.2019.105283.
[34]	LIU Y L, SHI G M, ZHAN Y, et al, 2021. Characteristics of PM_2.5 spatial distribution and influencing meteorological conditions in Sichuan Basin, Southwestern China[J]. Atmospheric Environment, 253: 118364. DOI: 10.1016/j.atmosenv.2021.118364.
[35]	MA S M, XIAO Z M, ZHANG Y F, et al, 2020. Assessment of meteorological impact and emergency plan for a heavy haze pollution episode in a core city of the North China plain[J]. Aerosol and Air Quality Research, 20(1): 26-42. DOI: 10.4209/aaqr.2019.08.0392.
[36]	MO J Y, GONG S L, ZHANG L, et al, 2021. Impacts of long-range transports from Central and South Asia on winter surface PM_2.5 concentrations in China[J]. Science of the Total Environment, 777: 146243. DOI: 10.1016/j.scitotenv.2021.146243.
[37]	RUPAKHETI D, YIN X F, RUPAKHETI M, et al, 2021. Spatio-temporal characteristics of air pollutants over Xinjiang, Northwestern China[J]. Environmental Pollution, 268: 115907. DOI: 10.1016/j.envpol.2020.115907.
[38]	SUN Z B, ZHAO X J, LI Z M, et al, 2021. Boundary layer structure characteristics under objective classification of persistent pollution weather types in the Beijing area[J]. Atmospheric Chemistry and Physics, 21(11): 8 863-8 882.
[39]	WANG T H, DU H D, ZHAO Z Z, et al, 2022. Impact of meteorological conditions and human activities on air quality during the COVID-19 lockdown in Northeast China[J]. Frontiers in Environmental Science, 10: 877268. DOI: 10.3389/fenvs.2022.877268.
[40]	WU K, ZHU S P, MAC KINNON M, et al, 2023. Unexpected deterioration of O₃ pollution in the south coast air basin of California: The role of meteorology and emissions[J]. Environmental Pollution, 330: 121728. DOI:10.1016/j.envpol.2023.121728.
[41]	YOO J W, PARK S Y, JO H Y, et al, 2024. Assessing the role of cold front passage and synoptic patterns on air pollution in the Korean Peninsula[J]. Environmental Pollution, 348: 123803. DOI: 10.1016/j.envpol.2024.123803.
[42]	YOO J W, PARK S Y, LEE K, et al, 2022. Impacts of plateau-induced lee troughs on regional PM_2.5 over the Korean Peninsula[J]. Atmospheric Pollution Research, 13(7): 101459. DOI: 10.1016/j.apr.2022.101459.
[43]	YOU T, WU R G, HUANG G, 2018. Differences in meteorological conditions between days with persistent and non-persistent pollution in Beijing, China[J]. Journal of Meteorological Research, 32(1): 81-98.
[44]	ZHANG H, YUAN H O, LIU X H, et al, 2018. Impact of synoptic weather patterns on 24 h-average PM_2.5 concentrations in the North China plain during 2013-2017[J]. Science of the Total Environment, 627: 200-210.
[45]	ZHOU C, WEI G, ZHENG H, et al, 2019. Effects of potential recirculation on air quality in coastal cities in the Yangtze River Delta[J]. Science of the Total Environment, 651: 12-23.