基于深度学习的海洋承灾体评价方法研究  

Research on the Evaluation Method of Marine Disaster-Bearing Body Based on Deep Learning

作　　者：邬满 刘金 文莉莉 严小敏 许贵林 WU Man;LIU Jin;WEN Lili;YAN Xiaomin;XU Guilin(Technology Innovation Center of Marine Information,Ministry of Natural Resources,Tianjin 300171,China;Guangxi Academy of Sciences,Nanning 530007,China;Nanning Normal University,Nanning 530001,China)

机构地区：[1]自然资源部海洋信息技术创新中心,天津300171 [2]广西科学院,广西南宁530007 [3]南宁师范大学,广西南宁530001

出　　处：《测绘地理信息》2022年第6期91-95,共5页Journal of Geomatics

基　　金：国家自然科学基金(U20A20105);广西科技重大专项(桂科AA22067072);自然资源部海洋信息技术创新中心2019年度开放基金。

摘　　要：针对北部湾区域海洋灾害频发、承灾体灾变响应机理不明、缺乏智能预警监测技术手段等问题,提出一种基于深度学习与地理网格评价的海洋承灾体脆弱性评价方法。该方法基于海洋大数据分析与全球网格剖分技术,构建海洋数据集装箱,在影像特征基础上,将海洋环境数据,海洋承载体脆弱性数据,强台风作用的预报数据(如风力强度、风向、风量、降水量),海岛树木、房屋、海堤、人口等数据的变化以及历年灾害、经济损失等数据,采用深度学习方法进行训练,得到最终的海洋承灾体智能评价模型,可以实现对海洋承灾体的动态评估。该方法用于承灾体响应机理研究、承灾体评价以及灾变预警,为强台风来临时防灾减灾精细化管理提供依据。In view of the frequent occurrence of marine disasters in the Beibu Gulf region,the unknown response mechanism of disaster bearing bodies,and the lack of intelligent early warning and monitoring technology,this paper proposes a vulnerability assessment method of marine disaster-bearing body based on deep learning and geographical grid assessment.Based on the analysis of ocean big data and the technology of geographic grid division,the ocean data container is constructed via this method.On the basis of image features,the data of ocean environment,the vulnerability data of ocean bearing body,the forecast data of strong typhoon(such as wind force intensity,wind direction,air volume,precipitation),the changes of island trees,houses,seawall,population and other data,as well as disasters and economy loss over the years are trained by deep learning method,and the final intelligent evaluation model of marine disaster-bearing body is obtained,which can realize the dynamic evaluation of marine disaster-bearingbody.This method can be used to study the response mechanism of the disaster-bearing body,evaluate the disaster-bearing body and give early warning of the disaster.

关 键 词：深度学习 海洋承灾体 智能预警与监测 脆弱性评价 地球剖分网格编码 防灾与减灾 

分 类 号：P237[天文地球—摄影测量与遥感] P208[天文地球—测绘科学与技术]