Severe convective weather systems are often accompanied by abundant lightning, and the key discharge characteristics of lightning are closely related to charge regions. However, existing detection methods cannot achieve comprehensive observation of these charge regions, which limits the in-depth understanding of lightning discharge processes.To address this issue, an identification algorithm for charge regions in convective systems was developed based on a multi-channel threshold method combined with a clustering algorithm. The probability of detection (POD), false alarm rate (FAR), and critical success index (CSI) were used as quantitative metrics to evaluate the algorithm’s performance. The results show that the algorithm achieves high identification accuracy, with a POD of 71.67%, a FAR of 3.83%, and a CSI of 69.67%.A comparative analysis was carried out between the proposed algorithm and an existing single-channel threshold algorithm for convective cloud cluster identification, based on the same dataset. Distinct differences were found in their results: charge regions tend to be sporadically distributed, while convective cloud clusters appear as large contiguous patches; the spatial scope of charge regions does not fully overlap with that of convective cloud clusters, and charge regions are likely to be located at the peripheries of cloud clusters; additionally, not all convective cloud clusters contain charge regions. It is concluded that there is a certain spatial displacement between charge centers and convective centers. The research findings provide valuable references for optimizing charge region detection techniques and deepening the study of lightning discharge mechanisms.
CHEN Xi-fan, TAN Yong-bo, WANG Hai-chao, CHEN Hao-chen, DENG Jie.
Identification of Charge Regions in Convective Systems Based on Multi-threshold Clustering[J].
Science Technology and Engineering, 2026, 26(23): 9786-9795 DOI:10.12404/j.issn.1671-1815.2507223
需要说明的是,当前,雷暴云中电荷的直接观测主要通过原位电场探空和闪电辐射源定位反演[6],基于目前的观测现状,难以全面地获取对流系统中电荷结构信息[33-35],而中尺度数值模式WRF-ELEC(weather research and forecasting with the electrification extra package)对对流系统动力和微物理具有较强的描述能力,可以详细描述起电过程、空间电荷结构及放电过程,能够提供算法所需的宏微观参量和电荷结构信息。该模式在雷电物理研究中有成熟应用, 有很多学者在此基础上开展了大量相关研究[36-39],较好地验证其模拟数据对真实情况的反映能力。因此,本文算法所使用的数据均由中尺度WRF-ELEC模式模拟得到。
采用整合起放电参数化方案的WRF-ELEC(V3.3.1)[40-42]模式模拟对流天气过程,以获取算法识别、验证所需的三维数据。海上的强对流天气对海洋活动及周边地区影响显著,是相关研究领域的一个热点。基于此,选取中国毗邻海域(105°E~135°E,1°N~20°N)作为研究的目标区域。开展模拟前,利用全球闪电定位系统(world wide lightning location network,WWLLN)提供的闪电定位资料筛选出目标区域内闪电活动频繁的区域,依据闪电活动的时空分布特征,确定强对流天气过程的大致时空范围。下载对应时空范围内的美国国家环境预报中心(National Centers for Environmental Prediction,NCEP)1°×1°再分析资料,作为模式模拟的初始场数据输入WRF-ELEC模式,对强对流天气过程进行模拟,最终输出模拟得到的垂直风风场、温度、组合反射率、冰晶混合比和数浓度、雪混合比和数浓度、霰混合比和数浓度、云滴混合比和数浓度、位势高度、总空间电荷密度的三维数据。强对流天气过程一般不超过十几个小时,每经过数小时,对流系统的生成消亡、分布范围及内部电荷分布等变化较大,因此从强对流天气过程的模拟结果中,每两小时提取一次上述参量,以形成多个个例。为清晰呈现数据层次,对相关概念作出如下界定:将一次完整的强对流天气过程(从发生到消亡的全生命史)定义为一个“事件”;将一个“事件”中提取的一次数据定义为一个“个例”;每个“个例”中包含多个独立的电荷区,一个电荷区称为一个“样本”。通过中尺度WRF-ELEC模式对目标区域内不同类型的对流系统进行模拟,累计模拟21个事件,获得146个个例,共10 413个样本,构建形成数据集。
1.2 研究方法
1.2.1 聚类算法
本文算法用到的聚类算法为DBSCAN[43-44](density-based spatial clustering of applications with noise),是一种通用的机器学习聚类算法,其主要特点是无需指定聚类中心或聚类数量就能够处理任意形状的聚类,而且具备处理噪声的能力以及对数据顺序不敏感。DBSCAN根据邻域半径ε和minPts对样本点进行聚类(minPts表示成为核心样本点的点在邻域半径内的最少样本点数)。聚类中的样本点被分为核心样本点、边界样本点和噪声样本点。如果一个样本点周围ε范围内有至少minPts个其他样本点,则该点被定义为核心样本点。边界样本点在某个核心样本点的ε范围内,但周围少于minPts个样本点。核心样本点允许扩散,继续搜索它的邻近点,而边界样本点可以被搜索到,但不会进一步搜索它的邻近点。噪声样本点则是不会被核心样本点搜索到的,相对孤立的点。在minPts为3的情况下聚类原理如图1所示,其中点A、B为核心点,半径内分别有4个点和3个点;点C半径内只有2个点,且落在核心点B范围内,属于边界点;点D、E均为噪声点。
YangJun, LiuYan, LiJie, et al. Research on lightning type recognition based on multi-scale residual networks[J]. Science Technology and Engineering, 2025, 25(27): 11483-11489.
[3]
MazurV, RuhnkeL H. Model of electric charges in thunderstorms and associated lightning[J]. Journal of Geophysical Research: Atmospheres, 1998, 103(D18): 23299-23308.
ZhangYijun, XuLiangtao, ZhengDong, et al. Review on inverted charge structure of severe storms[J]. Journal of Applied Meteorological Science, 2014, 25(5): 513-526.
[6]
TanY, TaoS, ZhuB. Fine-resolution simulation of the channel structures and propagation features of intracloud lightning[J]. Geophysical Research Letters, 2006, 33(9): 2005GL025523.
[7]
TanY, TaoS, ZhuB, et al. A simulation of the effects of intra-cloud lightning discharges on the charges and electrostatic potential distributions in a thundercloud[J]. Chinese Journal of Geophysics, 2007, 50(4): 916-930.
QieXiushu, LiuDongxia, JiangRubin, et al. Detection, mechanism, and forecasting of lightning and thunderstorms[J]. Journal of Meteorological Research, 2025, 83(3): 833-854.
ZhuJie, ChenPan, TangShunxian, et al. Research on quality control technology of Chinese satellite-based lightning data based on multi-source data[J]. Science Technology and Engineering, 2022, 22(27): 11783-11791.
[12]
WilsonC T R. Investigations on lightning discharges and on the electric field of thunderstorms[J]. Philosophical Transactions of the Royal Society of London. Series A, Containing Papers of a Mathematical or Physical Character, 1921, 221: 582-593.
[13]
SimpsonG C, ScraseF J. The distribution of electricity in thunderclouds[J]. Proceedings of the Royal Society of London. Series A-Mathematical and Physical Sciences, 1937, 161: 309-352.
[14]
SimpsonG C, RobinsonG D. The distribution of electricity in thunderclouds, II[J]. Proceedings of the Royal Society of London. Series A. Mathematical and Physical Sciences, 1941, 177: 281-329.
[15]
MarshallT C, RustW D. Electric field soundings through thunderstorms[J]. Journal of Geophysical Research: Atmospheres, 1991, 96(D12): 22297-22306.
[16]
StolzenburgM, MarshallT C, RustW D, et al. Horizontal distribution of electrical and meteorological conditions across the stratiform region of a mesoscale convective system[J]. Monthly Weather Review, 1994, 122(8): 1777-1797.
XinXueqi, ZhaoYang, TianJianbing, et al. Analysis of thunderstorm and its CG lightning characteristics in the eastern Qinghai-Tibet Plateau[J]. Science Technology and Engineering, 2019, 19(18): 47-57.
ZhangQin, ZhangXiao, GouAning, et al. Analysis of lightning characteristics of mesoscale convective system in two strong convective weather in Shandong province[J]. Science Technology and Engineering, 2018, 18(28): 6-14.
CuiXuedong, GuYuan, XuZhenyu, et al. Characteristic analyses on multi-return stroke of lightning parameters over Zhejiang based on 3-D lightning location system data[J]. Science Technology and Engineering, 2021, 21(23): 9706-9712.
ZhangYijun, MengQing, LuWeitao, et al. The charge structure and ground lightning characteristics of two supercell thunderstorms[J]. Chinese Science Bulletin, 2005, 50(23): 2663-2675.
QieXiushu, LiuDongxia, LiFengquan, et al. Charge structure of an isolated thunderstorm on the Tibetan Plateau and the formation of bolt-from-the-blue lightning[J]. Chinese Science Bulletin, 2025, 70(7): 944-954.
[27]
ColemanL M, MarshallT C, StolzenburgM, et al. Effects of charge and electrostatic potential on lightning propagation[J]. Journal of Geophysical Research: Atmospheres, 2003, 108(D9): 4298.
[28]
PilkeyJ T, UmanM A, HillJ D, et al. Rocket-triggered lightning propagation paths relative to preceding natural lightning activity and inferred cloud charge[J]. Journal of Geophysical Research: Atmospheres, 2014, 119(23): 427-456.
[29]
GoyensC, LauwaetD, SchröderM, et al. Tracking mesoscale convective systems in the sahel: relation between cloud parameters and precipitation[J]. International Journal of Climatology, 2012, 32(12): 1921-1934.
[30]
ChenD, GuoJ, YaoD, et al. Mesoscale convective systems in the Asian monsoon region from advanced himawari imager: algorithms and preliminary results[J]. Journal of Geophysical Research: Atmospheres, 2019, 124(4): 2210-2234.
SongDonghao, WangWenming, WangMinzhong, et al. Classification of millimeter-wave radar cloud echo data based on lightweight gradient boosting machine[J]. Science Technology and Engineering, 2025, 25(17): 7072-7079.
[35]
MaddoxR A. Mesoscale convective complexes[J]. Bulletin of the American Meteorological Society, 1980, 61: 1374-1387.
BaiJie, WangHongqing, TaoZuyu. The recognition and tracking of severe convective cloud from ir images of GMS[J]. Journal of Tropical Meteorology, 1997, 3(2): 192-201.
[38]
SteinerM, Houze JrR A, YuterS E. Climatological characterization of three-dimensional storm structure from operational radar and rain gauge data[J]. Journal of Applied Meteorology and Climatology, 1995, 34(9): 1978-2007.
[39]
BiggerstaffM I, ListemaaS A. An improved scheme for convective/stratiform echo classification using radar reflectivity[J]. Journal of Applied Meteorology, 2000, 39(12): 2129-2150.
[40]
HintonG E, OsinderoS, TehY W. A fast learning algorithm for deep belief nets[J]. Neural Computation, 2006, 18(7): 1527-1554.
ZhengYiqin, YangXiaofeng, LiZiwei. Detection of severe convective cloud over sea surface from geostationary meteorological satellite images based on deep learning[J]. Journal of Remote Sensing, 2020, 24(1): 97-106.
ZhouKanghui, ZhengYongguang, HanLei, et al. Advances in application of machine learning to severe convective weather monitoring and forecasting[J]. Meteorological Monthly, 2021, 47(3): 274-289.
LiuDongxia, QieXiushu, WangZhichao, et al. Characteristics of lightning radiation source distribution and charge structure of squall line[J]. Acta Physica Sinica, 2013, 62(21): 526-534.
LinXiaohong, ZhangWenjuan, FanNengzhu, et al. Lightning activity in the pre-TC squall line of typhoon Lekima (2019) observed by FY-4A LMI and its relationship with convective evolution[J]. Remote Sensing Technology and Application, 2021, 36(4): 873-886.
LiuHengyi, DongWansheng, XuLiangtao, et al. 3D spatial-temporal characteristics of initial breakdown process in lightning observed by broadband interferometer[J]. Journal of Applied Meteorological Science, 2016, 27(1): 16-24.
ZhengDong, ZhangYijun, MengQing, et al. Relationship between lightning activities and surface precipitation in thunderstorm weather in Beijing[J]. Journal of Applied Meteorological Science, 2010, 21(3): 287-297.
[53]
StolzenburgM, RustW D, SmullB F, et al. Electrical structure in thunderstorm convective regions: 1. mesoscale convective systems[J]. Journal of Geophysical Research: Atmospheres, 1998, 103(D12): 14059-14078.
[54]
FierroA O, MansellE R, MacGormanD R, et al. The implementation of an explicit charging and discharge lightning scheme within the WRF-ARW model: benchmark simulations of a continental squall line, a tropical cyclone, and a winter storm[J]. Monthly Weather Review, 2013, 141(7): 2390-2415.
SunLing, QieXiushu, EdwardR. Mansell, et al. Feedback effect of electric field force on electrification and charge structure in thunderstorm[J]. Acta Physica Sinica, 2018, 67(16): 378-389.
[57]
SrivastavaA, KunduS S, PawarS D, et al. Evaluation of WRF-ELEC model to forecast lightning over the North Eastern region of India[J]. Meteorology and Atmospheric Physics, 2023, 135(4): 39.
GeSuan, LiYang, YangShixuan, et al. Effects of WRF parameterization schemes on wind field simulation in different terrain areas in Northern Xinjiang[J]. Science Technology and Engineering, 2024, 24(31): 13277-13287.
[60]
MansellE R, ZieglerC L, BruningE C. Simulated electrification of a small thunderstorm with two-moment bulk microphysics[J]. Journal of the Atmospheric Sciences, 2010, 67(1): 171-194.
[61]
SaundersC P R, KeithW D, MitzevaR P. The effect of liquid water on thunderstorm charging[J]. Journal of Geophysical Research: Atmospheres, 1991, 96(D6): 11007-11017.
[62]
ZieglerC L, MacGormanD R, DyeJ E, et al. A model evaluation of noninductive graupel-ice charging in the early electrification of a mountain thunderstorm[J]. Journal of Geophysical Research: Atmospheres, 1991, 96(D7): 12833-12855.
[63]
EsterM, KriegelH P, SanderJ, et al. A density-based algorithm for discovering clusters in large spatial databases with noise[C]// Proceedings of the 2nd International Conference on Knowledge Discovery and Data Mining (KDD-96). Washington: AAAI, 1996: 226-231.
[64]
HutchinsM L, HolzworthR H, BrundellJ B. Diurnal variation of the global electric circuit from clustered thunderstorms[J]. Journal of Geophysical Research: Space Physics, 2014, 119(1): 620-629.
[65]
MarshallT C, RustW D, StolzenburgM. Electrical structure and updraft speeds in thunderstorms over the southern Great Plains[J]. Journal of Geophysical Research: Atmospheres, 1995, 100(D1): 1001-1015.
[66]
BruningE C, MacGormanD R. Theory and observations of controls on lightning flash size spectra[J]. Journal of the Atmospheric Sciences, 2013, 70(12): 4012-4029.
[67]
TakahashiT. Riming electrification as a charge generation mechanism in thunderstorms[J]. Journal of Atmospheric Sciences, 1978, 35(8): 1536-1548.
[68]
JayaratneE R, SaundersC P R. Charge on ice crystals in laboratory clouds[J]. Journal of Geophysical Research: Oceans, 1983, 88(C9): 5494-5496.
[69]
TownsendJ T. Theoretical analysis of an alphabetic confusion matrix[J]. Perception & Psychophysics, 1971, 9(1): 40-50.
YuXiaoding, ZhengYongguang. Advances in severe convective weather research and operational service in China[J]. Acta Meteorologica Sinica, 2020, 78(3): 391-418.
[74]
李扬. 基于深度学习的强对流临近预报方法研究[D]. 南京: 南京信息工程大学, 2023.
[75]
LiYang. Using deep learning for severe convective nowcasting[D]. Nanjing: Nanjing University of Information Science and Technology, 2023.
ZhengYongguang, ChenJong, ZhuPeijun. Climatological distribution and diurnal variation of mesoscale convective systems over China and its vicinity during summer[J]. Chinese Science Bulletin, 2008, 53(10): 1574-1586.
[78]
VelascoI, FritschJ M. Mesoscale convective complexes in the Americas[J]. Journal of Geophysical Research: Atmospheres, 1987, 92(D8): 9591-9613.
[79]
FuR, Del GenioA D, RossowW B. Behavior of deep convective clouds in the tropical Pacific deduced from ISCCP radiances[J]. Journal of Climate, 1990, 3(10): 1129-1152.
[80]
ZhangW, ZhangY, ZhengD, et al. Lightning distribution and eyewall outbreaks in tropical cyclones during landfall[J]. Monthly Weather Review, 2012, 140(11): 3573-3586.
ZhangTengfei, YinLiyun, ZhangJie. Evolutions and cloud-toground lightning features of two mesoscale convective thunder storm systems in Yunnan[J]. Quarterly Journal of Applied Meteorological Science, 2013, 24(2): 207-218.