基于机器学习方法的气溶胶对西藏高原地区雨季降水的影响

doi:10.3969/j.issn.1673-503X.2024.03.017

摘要/Abstract

摘要：

利用2001—2021年年降水量资料, 2021年雨季6—9月和非雨季露点温度、降水量等资料, PM_2.5、硫酸盐、硝酸盐等大气污染物资料, 基于机器学习方法对西藏高海拔地区拉孜降水量进行建模, 并量化了影响拉孜地区降水量的气象因素和气溶胶粒子。结果表明: 露点温度是影响拉孜地区降水量的关键变量, 其对雨季和非雨季降水量的贡献分别为74%和66%。硝酸盐是气溶胶组分中对降水量影响最大的变量, 其在雨季和非雨季占气溶胶总贡献的61%和71%, 表明硝酸盐气溶胶对降水有重要作用。

关键词: 高海拔地区, 降水成因, 机器学习

Abstract:

Using the precipitation data from 2001 to 2021 and the data of atmospheric pollutants such as PM_2.5, sulfate and nitrate, we chose Lazi as the study area and use the machine learning (ML) method to decipher the complex relationship between precipitation and its influencing factors. After that, the contribution of each input variable to the precipitation in was is quantified by this ML method. The results indicate that the dew point temperature is the most crucial elements affecting the precipitation in Lazi, contributing 74% to precipitation in rainy season and 66% to precipitation in non-rainy season. Among aerosol components, nitrate shows the greatest influence on precipitation, accounting for 61% and 71% to the precipitation in rainy and non-rainy seasons, respectively. This result means that nitrate aerosols play an important role in precipitation.

Key words: High altitude region, Precipitation mechanism, Machine learning algorithm

中图分类号:

达瓦次仁,落追,洪一航,次仁卓嘎. 基于机器学习方法的气溶胶对西藏高原地区雨季降水的影响[J]. 气象与环境学报, 2024, 40(3): 138-144.

Ciren DAWA,Zhui LUO,Yihang HONG,Zhuoga CIREN. Impact of aerosol on rainy season precipitation in Xizang Plateau based on machine learning method[J]. Journal of Meteorology and Environment, 2024, 40(3): 138-144.

图/表 4

图1

表1

图2

图3

参考文献 24

1	常姝婷. 全球变暖背景下青藏高原夏季大气水汽特征及对区域气候的影响[D]. 兰州: 兰州大学, 2018.
2	路红亚, 杜军, 袁雷, 等. 1971—2012年珠穆朗玛峰地区极端降水事件变化研究[J]. 冰川冻土, 2014, 36 (3): 563- 572.
3	刘蕾, 陈茂钦, 蓝柳茹, 等. 桂北山区两次突发性大暴雨触发及维持机制分析[J]. 气象与环境学报, 2022, 38 (5): 15- 24.
4	刘达之, 姚聃, 梁旭东. 基于四维变分同化的"6·23"阜宁龙卷大涡模拟研究[J]. 气象与环境学报, 2022, 38 (6): 1- 9.
5	邢莉, 傅宗玫. 有机气溶胶对中国境内云凝结核数量的贡献研究[J]. 北京大学学报: 自然科学版, 2015, 51 (1): 13- 23.
6	张瑶, 吴昊, 张东梅, 等. 超大城市新粒子生成事件对云凝结核贡献分析[J]. 环境科学学报, 2022, 42 (9): 372- 383.
7	Huang R J , Zhang Y L , Bozzetti C , et al. High secondary aerosol contribution to particulate pollution during haze events in China[J]. Nature, 2014, 514 (7521): 218- 222. doi: 10.1038/nature13774
8	Dao X , Lin Y C , Cao F , et al. Introduction to the national aerosol chemical composition monitoring network of China: objectives, current status, and outlook[J]. Bulletin of the American Meteorological Society, 2019, 100 (12): ES337- ES351. doi: 10.1175/BAMS-D-18-0325.1
9	Geng G N , Zhang Q , Tong D , et al. Chemical composition of ambient PM_2.5 over China and relationship to precursor emissions during 2005-2012[J]. Atmospheric Chemistry and Physics, 2017, 17 (14): 9187- 9203. doi: 10.5194/acp-17-9187-2017
10	Liu S G , Geng G N , Xiao Q Y , et al. Tracking daily concentrations of PM_2.5 chemical composition in China since 2000[J]. Environmental Science & Technology, 2022, 56 (22): 16517- 16527.
11	Bao M Y , Zhang Y L , Cao F , et al. Highly time-resolved characterization of carbonaceous aerosols using a two-wavelength Sunset thermal-optical carbon analyzer[J]. Atmospheric Measurement Techniques, 2021, 14 (6): 4053- 4068. doi: 10.5194/amt-14-4053-2021
12	Yu M Y , Zhang Y L , Xie T , et al. Quantification of fossil and non-fossil sources to the reduction of carbonaceous aerosols in the Yangtze River Delta, China: insights from radiocarbon analysis during 2014-2019[J]. Atmospheric Environment, 2023, 292, 119421. doi: 10.1016/j.atmosenv.2022.119421
13	Miyakawa T , Kanaya Y , Komazaki Y , et al. Intercomparison between a single particle soot photometer and evolved gas analysis in an industrial area in Japan: implications for the consistency of soot aerosol mass concentration measurements[J]. Atmospheric Environment, 2016, 127, 14- 21. doi: 10.1016/j.atmosenv.2015.12.018
14	Hong Y H , Cao F , Fan M Y , et al. Impacts of chemical degradation of levoglucosan on quantifying biomass burning contribution to carbonaceous aerosols: a case study in Northeast China[J]. Science of the Total Environment, 2022, 819, 152007. doi: 10.1016/j.scitotenv.2021.152007
15	姜建芳, 侯丽丽, 齐梦溪, 等. 天津市采暖季PM_2.5中碳组分污染特征及来源分析[J]. 生态环境学报, 2020, 29 (6): 1181- 1188.
16	黄炯丽, 陈志明, 莫招育, 等. 广西玉林市大气PM₁₀和PM_2.5中有机碳和元素碳污染特征分析[J]. 环境科学, 2018, 39 (1): 27- 37.
17	Shin J Y , Ro Y , Cha J W , et al. Assessing the applicability of random forest, stochastic gradient boosted model, and extreme learning machine methods to the quantitative precipitation estimation of the radar data: a case study to gwangdeoksan radar, South Korea, in 2018[J]. Advances in Meteorology, 2019, 2019, 6542410.
18	Li Q L , Zhu Q Y , Xu M W , et al. Estimating the impact of COVID-19 on the PM_2.5 levels in China with a satellite-driven machine learning model[J]. Remote Sensing, 2021, 13 (7): 1351. doi: 10.3390/rs13071351
19	Grange S K , Carslaw D C . Using meteorological normalisation to detect interventions in air quality time series[J]. Science of the Total Environment, 2019, 653, 578- 588. doi: 10.1016/j.scitotenv.2018.10.344
20	Grange S K , Carslaw D C , Lewis A C , et al. Random forest meteorological normalisation models for Swiss PM₁₀ trend analysis[J]. Atmospheric Chemistry and Physics, 2018, 18 (9): 6223- 6239. doi: 10.5194/acp-18-6223-2018
21	Grange S K , Uzu G , Weber S , et al. Linking Switzerland's PM₁₀ and PM_2.5 oxidative potential (OP) with emission sources[J]. Atmospheric Chemistry and Physics, 2022, 22 (10): 7029- 7050. doi: 10.5194/acp-22-7029-2022
22	Lovric M , Pavlovic K , Vukovic M , et al. Understanding the true effects of the COVID-19 lockdown on air pollution by means of machine learning[J]. Environmental Pollution, 2021, 274, 115900. doi: 10.1016/j.envpol.2020.115900
23	Wright M N , Ziegler A . Ranger: a fast implementation of random forests for high dimensional data in C++ and R[J]. Journal of Statistical Software, 2017, 77 (1): 1- 17.
24	Liu F T, Ting K M, Zhou Z H. Isolation forest[C]//Proceedings of 2008 Eighth IEEE International Conference on Data Mining. Pisa: IEEE, 2008: 413-422.

变量	全年	雨季	非雨季	显著性分析
降水量/mm	37.3±64.9	98.0±82.9	8.6±20.0	P < 0.01
U10/(m·s^-1)	0.8±0.8	0.1±0.5	1.2±0.7	P < 0.01
V10/(m·s^-1)	0.9±0.7	0.7±0.6	1.0±0.7	P>0.05
D2M/℃	-6.8±11.1	5.6±2.6	-12.7±8.4	P < 0.01
T2M/℃	4.7±5.3	10.7±1.3	1.9±4.0	P < 0.01
LAI_HV/(m²·m^-2)	0.0±0.0	0.0±0.0	0.0±0.0	P>0.05
LAI_LV/(m²·m^-2)	0.7±0.1	0.8±0.0	0.7±0.0	P>0.05
SP/(hPa)	590.0±2.9	591.8±1.5	589.2±3.0	P < 0.01
PM_2.5/(μg·m^-3)	11.7±4.5	9.1±1.4	13.0±4.9	P < 0.01
SO₄^2-/(μg·m^-3)	2.1±0.7	1.8±0.5	2.3±0.7	P < 0.01
NO₃^-/(μg·m^-3)	2.4±1.1	1.4±0.3	2.8±1.0	P < 0.01
NH₄⁺/(μg·m^-3)	1.9±0.7	1.3±0.3	2.2±0.7	P < 0.01
OM/(μg·m^-3)	3.8±1.3	3.1±0.7	4.2±1.4	P < 0.01
OC/(μg·m^-3)	2.4±0.8	1.9±0.4	2.6±0.9	P < 0.01
BC/(μg·m^-3)	0.7±0.3	0.6±0.1	0.8±0.3	P < 0.01
SOC/(μg·m^-3)	1.0±0.4	0.7±0.2	1.1±0.4	P < 0.01
Others/(μg·m^-3)	0.8±2.7	0.8±1.2	0.8±3.1	P < 0.01
SOC/OC	0.4±0.1	0.4±0.1	0.4±0.1	P < 0.01

[1]	吴宇恒,白景昌. 基于遥感和机器学习的中国典型城市碳排放驱动因子及预测分析[J]. 气象与环境学报, 2024, 40(1): 105-112.
[2]	张煦庭, 刘慧, 刘瑞芳, 巨菲, 刘嘉慧敏, 高星星, 黄少妮, 王楠. 基于极端梯度提升算法的西安市逐小时PM_2.5浓度预报研究[J]. 气象与环境学报, 2023, 39(1): 44-54.
[3]	王涛, 王乙舒, 赵春雨, 王小桃, 秦美欧, 沈玉敏, 侯依玲, 赵建云. 基于机器学习方法的辽宁省初霜冻日期预测模型研究[J]. 气象与环境学报, 2022, 38(4): 47-56.
[4]	王远谋,李家启,陈施吉,唐家萍,夏佰成,韩世刚. 基于机器学习的重庆长江航道雾图像特征识别研究[J]. 气象与环境学报, 2021, 37(1): 106-112.