基于水平集的洪涝淹没范围时空模拟方法  

Approach to simulating the spatial-temporal process of flood inundation area

作　　者：张连翀 李国庆[1,3] 于文洋[1,3] 冉全[1,2] 

机构地区：[1]中国科学院遥感与数字地球研究所数字地球实验室,北京100094 [2]中国科学院大学,北京100049 [3]海南省地球观测重点实验室,海南572023

出　　处：《国土资源遥感》2017年第1期92-96,共5页Remote Sensing for Land & Resources

基　　金：国家重点研发计划项目"多源遥感监测数据在线融合及协同分析云平台"(编号:2016YFB0501504);中国科学院数字地球重点实验室开放基金项目"面向按需处理的遥感信息模型自动化计算方法"(编号:2015LDE005)共同资助

摘　　要：遥感技术能及时获取洪水空间分布特征信息,已成为洪涝灾害监测与损失评估的重要依据。然而受天气和环境等因素影响,不能全天时接收遥感影像导致部分数据缺失,无法提供动态连续的洪涝淹没过程资料。以2013年汛期黑龙江流域八岔段溃口淹没区为例,基于多时相GF-1卫星晴空遥感影像提取的洪涝淹没范围信息,将洪涝淹没过程转化为水平集函数的偏微分方程数值求解问题,利用空间迎风差分格式和时间欧拉差分格式模拟了从8月24日到10月8日洪水涨退过程的逐日淹没范围。精度评价结果表明,洪涝淹没范围的模拟结果与同时期遥感影像提取结果的Kappa系数分别为0.921 2和0.893 2;与同时期洪涝淹没范围的统计数据相比略偏低,但相对误差都小于10%。该方法的模拟结果与影像提取结果和实际统计数据都具有较好的时空一致性,为不依赖先验资料的洪涝灾害应急响应决策提供了科学依据。Remote sensing data, as important information for flood disaster monitoring and loss assessment, can timely obtain the spatial - temporal distribution characteristics of flood. However, as it is restricted by weather conditions, it cannot form a dynamic and continuous process data. In this study, multi - temporal GF - 1 satellite remote sensing clear images were used to extract the flood extent area based on bacha breach on the Heilong River in 2013. The flood inundation process was transformed into a numerical problem of partially differential equations by level set function. Finite difference method both in space and time was used to simulate tile results of daily flood inundation area from August 24 to October 8. The results show that, compared with remote sensing data, the spatial - temporal consistency and the Kappa coefficients are 0. 921 2 and 0. 893 2; Compared with statistic data, the relatively error is less than 10%. This method has provided a scientific basis for the decision of flood disaster emergency response without prior information.

关 键 词：洪涝淹没范围 时空模拟 水平集 遥感 黑龙江洪水 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]