2020 , Vol. 22 >Issue 2: 187 - 197

Scenarios Construction and Spatial-temporal Deduction of Typhoon Storm Surge

• RAO Wenli 1 ,
• LUO Nianxue , 2, *
Expand
• 1. Beijing Global Safety Technology Company Limited, Wuhan 430000, China
• 2. School of Geodesy and Geomatics, Wuhan University, Wuhan 430079, China
LUO Nianxue, E-mail:

Request revised date: 2019-11-20

Online published: 2020-04-13

Supported by

National Key Research and Development Project of China(2017YFC1405300)

Abstract

Due to abruptness of typhoon storm surge, continuity of the evolution time and uncertainty of the path, it is hard for emergency decision-makers to make correct decisions in emergency rescue. To solve this problem, this article applies "scenario-response" to the typhoon storm surge. Firstly, based on the analysis of the typhoon storm surge scenarios and the conceptual model of the scenario elements, we extract the key scenario elements by means of data collection and attribute recognition. Then, we construct the dynamic scenario network of the typhoon storm surge by the method of frame representation. Secondly, we analysis the evolution and path of typhoon storm surge. Thirdly, we construct dynamic scenario network of typhoon storm surge with the dynamic Bayesian network method. Finally, we calculate the state probability of scenarios with the prior state probability and conditional probability and realize the key scenario deduction of the typhoon storm surge. In the end of the essay, we simulated an experiment for the influence of typhoon on the coastal cities of Guangdong Province from 11 to 17 on September 16 in 2018. The experiment results show that the probability of dykes, seawater inversion, floods and landslides respectively are 85%, 81%, 74%, 54%. The conclusion is drawn as follows: (1) The structure and content of each scenario element in the scenario construction process are different and interactional. Frame representation can reasonably characterize complex heterogeneous scenario elements data. (2) The evolution path of the situation is determined by many factors such as the situation itself, the disaster-bearing body, and emergency management. Decision makers need to comprehensively consider the emergency team and the rational use of resources when making decisions. (3) From the construction of the storm surge scenario to the deduction, the whole process has clear ideas and intuitive results, which is conducive to the promotion and application in marine disasters. The tentative application of "scenario-response" in storm surge events provides new emergency ideas and solutions for storm surge control.

RAO Wenli , LUO Nianxue . Scenarios Construction and Spatial-temporal Deduction of Typhoon Storm Surge[J]. Journal of Geo-information Science, 2020 , 22(2) : 187 -197 .

1 引言

“情景—应对”在突发事件中成功应用的同时,相继也提出了系统动力学[9]、演化博弈理论[10]、贝叶斯网络推理[11]、案例推理[12]等方法,为情景推演方法多样化提供了基础。如王循庆等[10]基于演化博弈理论构建了危化品安全监管演化博弈模型,并进行了情景推演模拟仿真;王宁等[12]以共性知识元模型为基础,基于案例匹配进行了突发事件情景推演。

2 研究方法

2.1 风暴潮情景结构与知识表示

2.1.1 情景结构知识单元

$S = S t 1 , S t 2 , … , S tn$

$S t = ( S , A , D )$

$K = ( N , C , G t )$

图1 风暴潮情景结构

Fig. 1 Structure of scenario in storm surge

2.1.2 情景知识框架表示

图2 风暴潮情景统一建模语言图

Fig. 2 UML(Unified Modeling Language) diagram of storm surge

2.2 风暴潮情景时空推演方法

2.2.1 情景演化规律

图3 风暴潮情景演化规律

Fig. 3 Evolution of storm surge scenario

2.2.2 情景状态概率计算

$P ( X 1 , … , X n ) = ∏ 1 n P X | P ∏ X n$

2.2.3 情景链构建

（1）确定网络节点变量。通过专家经验打分或层次分析法计算风暴潮情景中属性权重值来确定关键属性,关键属性值即为节点变量。具体可分步完成：① 根据风暴潮情景要素的分类,通过知识框架表示法结构化采集情景要素;② 风暴潮研究领域专家根据经验以打分形式评价情景要素从而得到关键要素;③ 根据关键要素知识框架表示,确定关键要素的类型、取值范围。
（2）确定网络节点变量之间的关系。根据第一步的结果得到网络节点变量,接着分析确定网络节点变量之间的关系,用有向边连接一个节点到另一个节点以表示节点变量之间的因果关系,从而形成完整的情景动态贝叶斯网络。
（3）确定网络节点变量的概率。需先确定没有父节点的网络节点变量的先验概率,即 $P ∏ X i$ ;然后采用专家打分方式,组织风暴潮领域专家对有父节点的网络节点变量分配概率,对同一节点变量的不同专家给出的条件概率计算算术平均值后,确定节点变量的条件概率,即 $P X i | P ∏ X i$ ;最后通过已确定的节点变量概率计算情景状态概率[26],从而得出下一时刻情景的状态概率,实现风暴潮的情景推演,为合理应急决策提供科学依据。

3 实证分析

3.1 实例情景及数据来源

2018年第22号台风“山竹”（简称山竹）9月15日19时其中心位于广东江门台山市东偏南方向690 km的南海东北部海面上,16日下午移动到广东珠海到湛江一带沿海登陆。据广东省民政厅报告,16日11—16时,强降雨主要分布在江门、东莞、深圳、中山、惠州等地,时段雨量大于100 mm的站有101个,16时沿海出现最高319 cm的风暴增水。17时山竹登陆,风力14级,中心最低气压955 hPa。主要灾情统计显示,大鹏和盐田发生海水倒灌事件,10宗水库溢洪,地质灾害1起,次生灾害13起等。

表1 实例分析使用的数据与来源

Tab. 1 Datas and sources using in instance analysis

3.2 山竹情景与情景链构建

图4 山竹情景要素信息

Fig. 4 Scenario elements of Mangkhut

表2 山竹事件情景链（部分推演）

Tab. 2 Scenario chain of Mangkhut (partial deduction)

图5 山竹2018年9月16日11时预测路径

Fig. 5 Forecast path of at 11o'clock of Mangkhut on September 16, 2018

图6 山竹2018年9月16日11时风圈影响范围内受损承灾体

Fig. 6 Damaged body within the influence range of Mangkhut at 11 o'clock on on September 16, 2018

图7 不同机构预测山竹2018年9月16日17时增减水情况

Fig. 7 Results of predicting the increase and decrease of water of Mangkhut by different agencies at 17 o'clock on September 16, 2018

表3 网络节点变量类型与取值集合

Tab. 3 Variable type and value set of network node

3.3 山竹情景概率计算

表4 网络节点的先验概率与条件概率（部分数据）

Tab. 4 Prior probability and conditional probability of network nodes (partial data)

$P(D0=F)$ 0.05 $P(S0=T|D0=T, A0=N)$ 0.80
$P(A0=P)$ 0.70 $P(S0=T|D0=F, A0=P)$ 0.70
$P(A0=N)$ 0.30 $P(S0=T|D0=F, A0=N)$ 0.40

$P(D1=F)$ 0.04 $P(S1=T|D1=T,A1=P,S0=F)$ 0.70
$P(A1=P)$ 0.75 $P(S1=T|D1=T, A1=N, S0=T)$ 0.85
$P(A1=N)$ 0.25 $P(S1=T|D1=T, A1=N, S0=F)$ 0.60
$P(S1=T|D1=F, A1=P, S0=T)$ 0.55
$P(S1=T|D1=F, A1=P, S0=F)$ 0.40
$P(S1=T|D1=F, A1=N, S0=T)$ 0.30
$P(S1=T|D1=F, A1=N, S0=F)$ 0.25

图8 基于动态贝叶斯网络的山竹情景推演

Fig. 8 Scenario deduction of Mangkhut based on dynamic Bayesian network

4 结论与讨论

（1）“情景—应对”能辅助应急决策。风暴潮灾害来势猛、速度快、破坏力强,具有潜在次生衍生危害和很强的“情景依赖”,传统的“预测-应对”模式已难以应对处置。同时,目前风暴潮研究主要集中在影响因素分析,危险性评估、数值模拟等方面,普遍缺乏对整个灾情态势发展与控制的研究。“情景—应对”是新型决策方法,通过情景推演,能认知当前情景随着时间变化时可能出现的情景及造成危害的程度,进而生成应对方案辅助应急决策[2]
（2）框架表示法能合理刻画情景结构与情景间关系。风暴潮形成的灾害具有共生性、依赖性、复杂性及综合性,多种影响因子的相互作用,导致风暴潮情景的复杂性。情景数据来源不一、结构多样,而情景是整个风暴潮事件中的关键元素。框架表示法能层次化梳理多源异构数据,合理、正确地从大量灾情信息中提取出情景数据,分类、组织并刻画成情景。
（3）合理的应急决策可以使风暴潮情景朝期望方向演化。情景链的关键情景及下一情景演化方向由情景本身、承灾体自身特点、应急管理等共同决定,具有不确定性。动态贝叶斯网络方法能保障不确定性情景演化的可靠性。由于不同应急管理会使情景朝不同方向发展,应急决策者需综合考虑情景现状、应急队伍与资源能力而做出合理决策。

 [1] 孙湘平 . 中国近海区域海洋[M]. 北京: 海洋出版社, 2006. [ Sun X P. Offshore seas of China[M]. Beijing: Ocean Press, 2006. ]
 [2] 舒其林 . 非常规突发事件的情景演变及"情景–应对"决策方案生成[J]. 中国科学技术大学学报, 2012,42(11):936-941. [ Shu Q L . Study on scenario evolvement and alternative generation of "scenario-response" decision-making in unconventional emergencies[J]. Journal of University of Science and Technology of China, 2012,42(11):936-941. ]
 [3] Chang M S, Tseng Y L, Chen J W . A scenario planning approach for the flood emergency logistics preparation problem under uncertainty[J]. Transportion Research Part E: Logistics and Transportion Review, 2007,43(6):737-754.
 [4] Mahmoud M, Liu Y Q, Hartmann H . A formal framework for scenario development in support of environmental decision-making[J]. Environmental Modelling & Software, 2009,24(7):798-808.
 [5] 吴倩, 谈伟, 盖文妹 . 基于动态贝叶斯网络的民航突发事件情景分析研究[J]. 中国安全生产科学技术, 2016,12(3):169-174. [ Wu Q, Tan W, Gai W M . Study on scenario analysis of civil aviation emergency based on dynamic Bayesian network[J]. Journal of Safety Science and Technology, 2016,12(3):169-174. ]
 [6] 孙超, 钟少波, 邓羽 . 基于暴雨内涝灾害情景推演的北京市应急救援方案评估与决策优化[J]. 地理学报, 2017,72(5):804-816. [ Sun C, Zhong S B, Deng Y . Scenario deduction based emergency rescue plan assessment and decision optimization of urban rainstorm water-logging: A case study of Beijing[J]. Acta Geographica Sinica, 2017,72(5):804-816. ]
 [7] 张恭孝, 杨荣华, 崔萌 , 等. 基于"情景–应对"模式的危险化学品泄漏应急决策系统的构建[J]. 化工环保, 2018,38(5):609-614. [ Zhang G X, Yang R H, Cui M , et al. Construction of the emergency decision system for hazardous chemicals leakage based on "scenario-response" model[J]. Environmental Protection of Chemical Industry, 2018,38(5):609-614. ]
 [8] 徐元元 . 基于动态贝叶斯网络的高层建筑火灾应急决策研究[D]. 郑州:郑州大学, 2019. [ Xu Y Y . Study on the decision making of high-rise buildings fire emergency based on for on dynamic Bayesian network[D]. Zhengzhou: Zhengzhou University, 2019. ]
 [9] 仲秋雁, 路光, 王宁 . 基于知识元模型和系统动力学模型的突发事件仿真方法[J]. 情报科学, 2014,32(10):15-19. [ Zhong Q Y, Lu G, Wang N . Simulation method research on emergencies based on system dynamics model and knowledge element model[J]. Information Science, 2014,32(10):15-19. ]
 [10] 王循庆, 李勇建, 孙晓羽 . 基于演化博弈的危化品安全监管情景推演研究[J]. 中国安全生产科学技术, 2017,13(1):115-121. [ Wang X Q, Li Y J, Sun X Y . Study on scenario inference of hazardous chemicals safety supervision based on evolutionary game[J]. Journal of Safety Science and Technology, 2017,13(1):115-121. ]
 [11] 宋英华, 刘含笑, 蒋新宇 , 等. 基于知识元与贝叶斯网络的食品安全事故情景推演研究[J]. 情报学报, 2018,37(7):712-720. [ Song Y H, Liu H X, Jiang X Y , et al. Research on scenario evolution of food safety incidents based on knowledge element and Bayesian network[J]. Journal of the China Society for Scientific and Technical Information, 2018,37(7):712-720. ]
 [12] 王宁, 谢晓珊, 刘海园 . 基于案例的突发事件推演规则验证方法[J]. 系统工程学报, 2019,34(2):145-157,237. [ Wang N, Xie X S, Liu H Y . Validation method of emergency deduction rules based on cases[J]. Journal of Systems Engineering, 2019,34(2):145-157,237. ]
 [13] 曾银东 . 福建宁德海洋工程风暴潮灾害风险特征参数分析[J]. 应用海洋学学报, 2017,36(4):500-511. [ Zeng Y D . Risk characteristics of storm surge hazards on marine project in Ningde[J]. Journal of Applied Oceanography, 2017,36(4):500-511. ]
 [14] 张敏, 罗军, 胡金磊 , 等. 雷州市沿海风暴潮淹没危险性评估[J]. 热带海洋学报, 2019,38(2):1-12. [ Zhang M, Luo J, Hu J L , et al. Inundation risk assessment of storm surge along Lei Zhou coastal areas[J]. Journal of Tropical Oceanography, 2019,38(2):1-12. ]
 [15] 原楠, 陈新平, 陈学恩 , 等. 罗源湾海域台风风暴潮数值模拟研究[J]. 海洋通报, 2019,38(1):20-30. [ Yuan N, Chen X P, Chen X E , et al. Numerical simulation of typhoon storm surge at the Luoyuan Bay[J]. Marine Science Bulletin, 2019,38(1):20-30. ]
 [16] 汤富平, 郭见兵, 余华芬 , 等. 基于GIS的台风风暴潮淹没情景模拟方法与平台开发[J]. 地理与地理信息科学, 2019,35(1):6-11,2,19. [ Tang F P, Guo J B, Yu H F , et al. Scenario simulation method and platform of typhoon storm surge inundation based on geographic information system[J]. Geography and Geo-information Science, 2019,35(1):6-11,2,19. ]
 [17] Zhu X H, Li X Y, Wang S Y , et al. Scenarios conversion deduction method of natural disaster based on dynamic Bayesian networks[C]. Proceedings of 2017 2 nd International Conference on Computational Modeling: Simulation and Applied Mathematics , 2017: 267-271. ]
 [18] 孙佳, 左军成, 黄琳 , 等. 东海沿岸台风及风暴潮灾害特征及成因[J]. 河海大学学报(自然科学版), 2013,41(5):461-465. [ Sun J, Zuo C J, Huang L , et al. Characteristics and causes of typhoon and storm surge along coast of East China Sea[J]. Journal of Hohai University (Natural Sciences), 2013,41(5):461-465. ]
 [19] 范维澄, 闪淳昌 . 公共安全与应急管理[M]. 北京: 科学出版社, 2018. [ Fan W C, Shan C C. Public safety and emergency management[M]. Beijing: Science Press, 2018. ]
 [20] 李春娟 . 突发事件应急管理知识系统演化研究[D]. 秦皇岛:燕山大学, 2015. [ Li C J . Research on the evolution of emergency management knowledge system[D]. Qinhuangdao: Yanshan University, 2015. ]
 [21] 冯文娟, 杜云艳, 苏奋振 . 台风时空过程的网络动态分析技术与示例[J]. 地球信息科学学报, 2007,9(5):57-63. [ Feng W J, Du Y Y, Su F Z . Dynamic network analyzing technology of typhoon spatio- temporal process and illustration researching[J]. Journal of Geo-information Science, 2007,9(5):57-63. ]
 [22] 陈泽强, 陈能成, 杜文英 , 等. 一种洪涝灾害时间信息建模方法[J]. 地球信息科学学报, 2015,17(6):644-652. [ Chen Z Q, Chen N C, Du W Y , et al. A method of modelling flood event[J]. Journal of Geo-information Science, 2015,17(6):644-652. ]
 [23] 朱晓寒, 李向阳, 王诗莹 . 自然灾害链情景态势组合推演方法[J]. 管理评论, 2016,28(8):143-151. [ Zhu X H, Li X Y, Wang S Y . Scenarios combination deduction method of natural disaster[J]. Management Review, 2016,28(8):143-151. ]
 [24] 张明红 . 基于案例的非常规突发事件情景推理方法研究[D]. 武汉:华中科技大学, 2012. [ Zhang M H . Study on unconventional emergency scenario reasoning method the cases-based[D]. Wuhan: Huazhong University of Science and Technology, 2012. ]
 [25] 徐磊 . 基于贝叶斯网络的突发事件应急决策信息分析方法研究[D]. 哈尔滨:哈尔滨工业大学, 2013. [ Xu L . Research on emergency events decision making information analysis method based on Bayesian network[D]. Harbin: Harbin Institute of Technology, 2013. ]
 [26] 巩前胜 . 基于动态贝叶斯网络的突发事件情景推演模型研究[J]. 西安石油大学学报(自然科学版), 2018,33(2):119-126. [ Gong Q S . Research on scenario deduction model of emergency based on dynamic Bayesian network[J]. Journal of Xi'an Shiyou University(Natural Science Edition), 2018,33(2):119-126. ]
 [27] 张凯华 . 化工园区火灾事故情景推演及应急知识匹配研究[D]. 青岛:中国石油大学(华东), 2017. [ Zhang G H . Research on scenario deduction and emergency knowledge matching of fire accident in chemical industry zone[D]. Qingdao: School of Economics and Management China University of Petroleum (East China), 2017. ]
 [28] 杨伟, 李彤 . 非常规灾害事件情景演化的概率性生长模式——基于台风莫拉克的探索性案例研究[J]. 电子科技大学学报(社科版), 2013,15(5):14-19. [ Yang W, Li T . Probabilistic growing mode of scenario evolution in unconventionality disaster: Exploratory case study of typhoon Morakot[J]. Journal of University of Electronic Science and Technology of China (Social Sciences Edition), 2013,15(5):14-19. ]
Options
Outlines

/

 〈 〉