基于DTW与K-means算法的河北场雨及雨型分区特征研究

doi:10.12082/dqxxkx.2021.200343

地球信息科学学报 ›› 2021, Vol. 23 ›› Issue (5): 860-868.doi: 10.12082/dqxxkx.2021.200343

• 地理空间分析综合应用 • 上一篇下一篇

基于DTW与K-means算法的河北场雨及雨型分区特征研究

李雨欣¹^,²(), 王瑛¹^,²^,^*(), 马庆媛¹^,², 刘天雪¹^,², 司丽丽³, 俞海洋³

1.北京师范大学环境演变与自然灾害教育部重点实验室,北京 100875
2.应急管理部-教育部减灾与应急管理研究院,北京 100875
3.河北省气象灾害防御中心,石家庄 050021

收稿日期:2020-07-03 修回日期:2020-09-18 出版日期:2021-05-25 发布日期:2021-07-25
通讯作者: *王瑛（1974— ）,女,云南曲靖人,博士,教授级高级工程师,主要从事灾害风险评估研究。E-mail:wy@bnu.edu.cn
作者简介:李雨欣（1996— ）,女,浙江宁波人,博士,主要从事灾害风险评估研究。E-mail:lyx2019@mail.bnu.edu.cn
基金资助:
国家重点研发计划项目(2017YFC1502505)

Research on the Characteristics of Rainfall Events and Rain Pattern Zoning in Hebei based on Data Mining Technology

LI Yuxin¹^,²(), WANG Ying¹^,²^,^*(), MA Qingyuan¹^,², LIU Tianxue¹^,², SI Lili³, YU Haiyang³

1. Key Laboratory of Environmental Change and Natural Disaster of Ministry of Education, Beijing Normal University, Beijing 100875, China
2. Academy of Disaster Reduction and Emergency Management, Ministry of Emergency Management and Ministry of Education, Beijing 100875, China
3. Hebei Meteorological Disaster Prevention Center, Shijiazhuang 050021, China

Received:2020-07-03 Revised:2020-09-18 Online:2021-05-25 Published:2021-07-25
Contact: WANG Ying
Supported by:
National Key R&D Program of China(2017YFC1502505)

摘要/Abstract

摘要：

深入挖掘气象站点的观测降雨数据,研究区域降雨的雨型规律,对于洪涝灾害预警和减灾措施制订有重要意义。本文基于河北省2005—2017年3189个站点逐小时降雨观测数据,进行“场雨”的划定,进而提取历史上各场雨的累积雨量、时长指标。采用数据挖掘技术中的DTW相似性算法进行场雨雨型的自动归类,将场雨分成Ⅰ—Ⅶ共7种雨型,包括峰值在前、中、后期的3种单峰型降雨,以及3种双峰型降雨和均匀型降雨,结果显示：河北降雨过程以Ⅰ型前期单峰值降雨、Ⅲ型中期单峰值降雨居多,二者占总降雨场次的70%以上,但空间分布上存在差异;通过K-means聚类,将河北地区分成3个雨型区：① 区, Ⅰ型和Ⅲ型降雨为主,分布在燕山丘陵气候区、冀东平原气候区和山前平原气候区。② 区,Ⅲ型、Ⅰ型、Ⅵ型、Ⅶ型降雨并重,在冀北高原气候区,承德市南部等分散分布。③ 区,Ⅲ型降雨为主,主要分布在石家庄市南部、邯郸市、邢台市大部分地区。本文将DTW相似性算法和K-means聚类方法相结合的数据挖掘技术,可以在未来的气象大数据分析中得到更多的应用。

关键词: 场雨, 雨型, 数据挖掘, DTW相似性算法, K-means聚类, 暴雨灾情, 区划, 河北

Abstract:

In-depth data mining of rainfall data and characterizing the rainfall events from meteorological records play an important role in formulation of flood disaster warning and mitigation measures. This paper analyzed a large number of hourly rainfall observations to quantify the rainfall pattern, cumulative rainfall, duration, and other indicators of rainfall events in Hebei province. First, we generated independent rainfall events based on the hourly precipitation data of 3189 stations in Hebei from 2005 to 2017. We counted the rainfall events for each station by calculating the time interval between rainfall events. Rainfall events with time interval less than 6h were counted as a single rainfall event. Otherwise, they were regarded as independent rainfall events. Then, we calculated the occurrence time, end time, duration, hourly accumulation, and total accumulation of each rainfall event. Finally, rainfall events were divided into seven types (I-Ⅶ), including three types of single-peak rainfall (i.e. single peak in the front, middle, and end), three types of double-peak rainfall, and uniform rainfall. The Dynamic Time Warping (DTW) algorithm was used for rainfall events classification. The results show that the rainfall in Hebei province was dominated by typeⅠ (single-peak rainfall in the front) and type Ⅲ (single-peak rainfall in the middle), which accounted for more than 70% of the total rainfall events with a significant spatial variation. The type Ⅳ (uniform rainfall) was the least with a proportion of less than 5%. The type Ⅱ (single-peak rainfall in the end) and three types of double-peak rainfall accounted for less than 25%. Through K-means clustering, the Hebei province was divided into 3 rain-type regions: district A, with type I and type Ⅲ rainfall mainly distributed in the Yanshan hilly climate region, the eastern Hebei plain climate region, and the piedmont plain climate region; district B, with type Ⅲ, type Ⅰ, type Ⅵ, and type Ⅶ rainfall scattered in the northern Hebei plateau climate zone and southern Chengde city; and district C, with type Ⅲ rainfall dominated in the southern part of Shijiazhuang City, Handan City, and most of Xingtai City. In this paper, the data mining method that combines DTW similarity algorithm and K-means clustering can be applied in future meteorological big data analysis.

Key words: field rain, rainfall type, data mining, DTW, K-means clustering, rainstorm disaster, zoning, Hebei

李雨欣, 王瑛, 马庆媛, 刘天雪, 司丽丽, 俞海洋. 基于DTW与K-means算法的河北场雨及雨型分区特征研究[J]. 地球信息科学学报, 2021, 23(5): 860-868.DOI:10.12082/dqxxkx.2021.200343

LI Yuxin, WANG Ying, MA Qingyuan, LIU Tianxue, SI Lili, YU Haiyang. Research on the Characteristics of Rainfall Events and Rain Pattern Zoning in Hebei based on Data Mining Technology[J]. Journal of Geo-information Science, 2021, 23(5): 860-868.DOI:10.12082/dqxxkx.2021.200343

图/表 10

图1

图2

图3

图4

图5

图6

表1

表2

图7

表3

参考文献 20

[1]	国家气候中心. GB/T 33680-2017,暴雨灾害等级[S]. 北京: 中国标准出版社, 2017.
	[ National Climate Center. GB/T 33680-2017, Rainstorm disaster level[S]. Beijing: Standard Press of China, 2017. ]
[2]	Melillo M, Brunetti M T, Peruccacci S, et al. An algorithm for the objective reconstruction of rainfall events responsible for landslides[J]. Landslides, 2015,12(2):311-320. doi: 10.1007/s10346-014-0471-3
[3]	莫洛科夫. 雨水道与合流水道[M]. 北京: 建筑工程出版社, 1959.
	[ Mo B. The rain water and confluent channel[M]. Beijing: Architectural Engineering Press, 1959. ]
[4]	岑国平, 沈晋. 城市设计暴雨雨型研究[J]. 水科学进展, 1998,9(1):41-46.
	[ Cen G P, Shen J. Research on rainfall pattern of urban design storm[J]. Advances in Water Science, 1998,9(1):41-46. ]
[5]	Keifer G J, Chu H H. Synthetic storm pattern for drainage design[J]. Journal of the Hydraulics Division, 1957,83(4):1-25.
[6]	Huff F A. Time distribution of rainfall in heavy storms[J]. Water Resources Research, 1967,3(4):1007-1019. doi: 10.1029/WR003i004p01007
[7]	Pilgrim D H, Cordery I. Rainfall temporal patterns for design floods[J]. Journal of the Hydraulics Division, 1975,101(1):81-95. doi: 10.1061/JYCEAJ.0004197
[8]	张兴奇, 徐鹏程, 顾璟冉. SCS模型在贵州省毕节市石桥小流域坡面产流模拟中的应用[J]. 水土保持通报, 2017,37(3):321-328,333.
	[ Zhang X Q, Xu P C, Gu J R. Application of SCS model to simulate runoff in slope field at Shiqiao small watershed in Bijie City of Guizhou Province[J]. Bulletin of Soil and Water Conservation, 2017,37(3):321-328,333. ]
[9]	殷水清, 王杨, 谢云, 等. 中国降雨过程时程分型特征[J]. 水科学进展, 2014,25(5):617-624.
	[ Yin S Q, Wang Y, Xie Y, et al. Characteristics of intra-storm temporal pattern over China[J]. Advances in Water Science, 2014,25(5):617-624. ]
[10]	银磊, 陈晓宏, 陈志和, 等. 广州市典型雨量站暴雨雨型研究[J]. 水资源研究, 2013,2(6):409-414.
	[ Yin L, Chen X H, Chen Z H, et al. Study on storm pattern of typical rainfall station in Guangzhou[J]. Journal of Water Resources Research, 2013,2(6):409-414. ]
[11]	Mayer-Schonberger V, Cukier K. Big data: A revolution that will transform how we live, work, and think[M]. London: John Murray, 2013.
[12]	何清, 李宁, 罗文娟, 等. 大数据下的机器学习算法综述[J]. 模式识别与人工智能, 2014,27(4):327-336.
	[ He Q, Li N, Luo W J, et al. A survey of machine learning algorithms for big data[J]. Pattern Recognition and Artificial Intelligence, 2014,27(4):327-336. ]
[13]	Trafalis T B, Santosa B, Richman M B. Learning networks in rainfall estimation[J]. Computational Management Science, 2005,2(3):229-251. doi: 10.1007/s10287-005-0026-0
[14]	李进讷. 基于DSCAN优化算法与决策树优化算法的气象时空数据挖掘技术研究[D]. 昆明:云南大学, 2018.
	[ Li J N. Research on spatio-temporal meteorological data mining technology based on dscan optimization algorithm and decision tree optimization algrithm[D]. Kunming: Yunnan University, 2018. ]
[15]	李正欣, 郭建胜, 王瑛, 等. DTW距离的过滤搜索方法[J]. 控制与决策, 2018,33(7):1277-1281.
	[ Li Z X, Guo J S, Wang Y, et al. Filtering search method for DTW distance[J]. Control and Decision, 2018,33(7):1277-1281. ]
[16]	Serrà J, Arcos J L. An empirical evaluation of similarity measures for time series classification[J]. Knowledge-Based Systems, 2014,67:305-314. doi: 10.1016/j.knosys.2014.04.035
[17]	王彬雁, 赵琳娜, 巩远发, 等. 北京降雨过程分型特征及短历时降雨重现期研究[J]. 暴雨灾害, 2015,34(4):302-308.
	[ Wang B Y, Zhao L N, Gong Y F, et al. Characteristics of temporal pattern and return period of short-duration rainfall at Beijing Observatory[J]. Torrential Rain and Disasters, 2015,34(4):302-308. ]
[18]	吴焕丽, 崔可旺, 张馨, 等. 基于改进K-means图像分割算法的细叶作物覆盖度提取[J]. 农业机械学报, 2019,50(1):42-50.
	[ Wu H L, Cui K W, Zhang X, et al. Improving accuracy of fine leaf crop coverage by improved K-means algorithm[J]. Transactions of the Chinese Society for Agricultural Machinery, 2019,50(1):42-50. ]
[19]	Jain A K, Murty M N, Flynn P J. Data clustering[J]. ACM Computing Surveys, 1999,31(3):264-323. doi: 10.1145/331499.331504
[20]	《河北省气象灾害风险地图集》编辑委员会. 河北省气象灾害风险地图集[M]. 北京: 科学出版社, 2018.
	[ Editorial committee of the Atlas of meteorological disaster risk in Hebei province. Atlas of meteorological disaster risk in Hebei province[M]. Beijing: Science Press, 2018. ]

	雨型
	Ⅰ	Ⅱ	Ⅲ	Ⅳ	Ⅴ	Ⅵ	Ⅶ
数量/个	58 300	7893	128 932	522	2485	20 070	23 660
比例/%	24.10	3.26	53.31	0.22	1.03	8.30	9.78

分区名称	站点数/个		雨型
分区名称	站点数/个		Ⅰ型	Ⅱ型	Ⅲ型	Ⅳ型	Ⅴ型	Ⅵ型	Ⅶ型
① 区	1620	数量/个	33 899	4062	66 669	271	1340	10 569	11 747
① 区	1620	比例/%	26.37	3.16	51.86	0.21	1.04	8.22	9.14
② 区	531	数量/个	9115	1607	19 086	102	535	3747	4521
② 区	531	比例/%	23.55	4.15	49.30	0.26	1.38	9.68	11.68
③ 区	1038	数量/个	15 195	2168	43 306	162	584	5728	7449
③ 区	1038	比例/%	20.37	2.91	58.06	0.22	0.78	7.68	9.99

雨型分区	场雨信息						灾情信息
雨型分区	地市	区县	开始时间	历时/h	雨型	累积量/mm	死亡人数/人	受灾面积/hm²	直接经济损失/万元
①	保定市	定州市	2013060711	10	Ⅰ	97.2	1	3950.3	2735.1
	沧州市	黄骅市	2010071811	20	Ⅲ	133.4	0	2700	3700
	衡水市	冀州市	2013081403	6	Ⅰ	30.8	0	6730	21 145
	秦皇岛市	抚宁县	2012072601	12	Ⅲ	52.8	2	2295	2750.5
	唐山市	玉田县	2006081003	8	Ⅲ	133.1	0	22 400	2341
	唐山市	玉田县	2009072218	2	Ⅰ	38.3	1	7377	2266
	张家口市	赤城县	2007070314	4	Ⅰ	16.4	6	2000	568
②	承德市	丰宁县	2006062819	11	Ⅶ	32.0	0	711	746
	承德市	丰宁县	2009082623	1	Ⅲ	16.2	0	1267	1160
	邯郸市	武安市	2010080416	5	Ⅰ	48.2	0	343.3	210
	秦皇岛市	昌黎县	2012072603	11	Ⅲ	38.2	0	9409	6082.4
	石家庄市	平山县	2009080101	3	Ⅰ	24.4	0	3202	1538.4
	邢台市	临西县	2013071522	4	Ⅵ	28.9	0	1413	973
③	保定市	唐县	2011082504	10	Ⅲ	27.8	0	1128	2980
	沧州市	盐山县	2010071910	20	Ⅲ	155.9	0	1100	417
	衡水市	武强县	2013070113	12	Ⅲ	143.9	0	919	1000
	秦皇岛市	青龙县	2012072518	15	Ⅲ	49.9	0	4000	3500
	石家庄市	赞皇县	2009082600	11	Ⅲ	28.2	0	590	720
	邢台市	新河县	2013070116	11	Ⅲ	136.1	0	540.6	647.1

基于DTW与K-means算法的河北场雨及雨型分区特征研究

Research on the Characteristics of Rainfall Events and Rain Pattern Zoning in Hebei based on Data Mining Technology

RichHTML

PDF (PC)

赞

可视化

摘要/Abstract

引用本文

使用本文

图/表 10

参考文献 20

相关文章 15

Metrics

本文评价

推荐阅读 0

[1]	陈传明, 龚杉, 杨峰, 肖振兴, 俞庆英. 基于停留区域识别的子轨迹异常检测方法[J]. 地球信息科学学报, 2023, 25(4): 684-697.
[2]	吴子豪, 刘耀林, 冯向阳, 陈奕云, 闫庆武. 基于多尺度地理加权回归的土壤镉污染局部影响因子分析[J]. 地球信息科学学报, 2023, 25(3): 573-587.
[3]	孔云峰. 基于迭代局部搜索的区划问题算法研究[J]. 地球信息科学学报, 2022, 24(9): 1730-1741.
[4]	赵志远, 黄永刚, 吴升, 邬群勇, 汪艳霞. 基于时空热点分析的城市交通违法行为特征识别方法[J]. 地球信息科学学报, 2022, 24(7): 1312-1325.
[5]	姜晓, 白璐斌, 楼夏寅, 李梅, 刘晖. 基于多尺度时空聚类的共享单车潮汐特征挖掘与需求预测研究[J]. 地球信息科学学报, 2022, 24(6): 1047-1060.
[6]	张晗, 邬群勇. 基于LDA和优化蚁群的OD流向时空语义聚类算法[J]. 地球信息科学学报, 2022, 24(5): 837-850.
[7]	李军, 刘举庆, 游林, 董恒, 俞艳, 张晓盼, 钟文军, 杨典华. 多源大数据支持的土地储备智能决策模型集研究[J]. 地球信息科学学报, 2022, 24(2): 299-309.
[8]	陈伟杰, 赵楠, 张婕姝, 宋炳良. AIS数据在集装箱港口服务效率的应用研究[J]. 地球信息科学学报, 2022, 24(1): 153-164.
[9]	胡添, 刘涛, 杜萍, 余贝贝, 张萌生. 空间同位模式支持下城市服务业关联发现及特征分析[J]. 地球信息科学学报, 2021, 23(6): 969-978.
[10]	姚博睿, 秦昆, 罗萍, 朱炤瑗, 漆林. 特殊事件中国际关系网络时序演化分析[J]. 地球信息科学学报, 2021, 23(4): 632-645.
[11]	刘亚溪, 宋辞, 刘起勇, 张知新, 王席, 马佳, 陈晓, 裴韬. 重庆市新型冠状病毒肺炎流行时空特征及其与人群活动性的关系[J]. 地球信息科学学报, 2021, 23(2): 222-235.
[12]	谢聪慧, 吴世新, 张晨, 孙文涛, 何海芳, 裴韬, 罗格平. 基于谱系聚类的全球各国新冠疫情时间序列特征分析[J]. 地球信息科学学报, 2021, 23(2): 236-245.
[13]	尹凌, 刘康, 张浩, 奚桂锴, 李璇, 李子垠, 薛建章. 耦合人群移动的COVID-19传染病模型研究进展[J]. 地球信息科学学报, 2021, 23(11): 1894-1909.
[14]	陈芳淼, 黄慧萍, 贾坤. 时空大数据在城市群建设与管理中的应用研究进展[J]. 地球信息科学学报, 2020, 22(6): 1307-1319.
[15]	胡最. 传统聚落景观基因的地理信息特征及其理解[J]. 地球信息科学学报, 2020, 22(5): 1083-1094.