航天遥感监测油库火灾污染环境行为的理论与方法研究

Since China has an increasing number of oil depots, the probability of fire explosion accidents is high and it is known that these accidents are harmful. Ground field monitoring, as the main means of oil depot fire pollution monitoring at present, is of low technical level and already difficult to meet the needs concerned. Space remote sensing is the best way to monitor a large scope of oil depot fire pollution in a rapid, real-time, dynamic, time-saving and effortless way. However, till now few researches have been carried out on space remote sensing monitoring of emergent oil depot fire pollution, needless to say relevant theories and methods. The project takes the oil depot fire accident as research object, utilizes multiple remote sensing data sources (visible light, infrared, hyperspectral and radar), GIS data and ground site data to conduct research on oil depot fire pollution sources as well as the principles and methods of monitoring their environmental behaviors in a rapid and dynamic way. Also, the project establishes a rapid acquisition model of oil depot target remote sensing monitoring data, explores the characteristic substances of fire pollution and the remote sensing monitoring mechanism of fire-pollution-source rapid identification, and puts forward a set of space remote sensing monitoring-based theories and methods of analyzing oil depot fire pollution environmental behaviors. Through the above researches, the project will improve the level of emergent oil depot fire pollution rapid identification and a wide range of dynamic monitoring in China, so as to provide theoretical and technical support for improving emergency monitoring and rapid response capability in oil depot fire pollution emergencies.

我国拥有油库数量众多且规模日益扩大，油库发生火灾爆炸事故的概率高、危害极大。现有油库火灾污染监测手段主要以地面现场监测为主，总体技术水平滞后，难以满足需要。航天遥感是快速、实时、动态、省时省力监测大范围油库火灾污染的最佳途径。然而，国内外鲜有学者开展突发油库火灾污染航天遥感监测研究，缺乏航天遥感监测油库火灾污染的理论与方法。本项目以油库火灾事故为研究对象，在可见光、红外、高光谱和雷达等多源遥感数据、GIS数据和地面现场数据的支持下，系统研究油库火灾污染源及其环境行为的快速动态监测原理与方法，建立油库目标遥感监测快速提取模型，探索油库火灾污染特征物质以及火灾污染源快速发现的遥感监测机制，提出一套基于航天遥感监测的油库火灾污染环境行为分析的理论与方法，提高我国突发油库火灾污染的快速发现与大范围动态监测水平，为提高我国应对突发油库火灾污染的应急监测与快速响应能力提供理论与技术支撑。

油库火灾爆炸事故危害巨大，而现有的油库火灾污染监测理论与方法严重滞后，不能满足突发油库火灾污染应急监测与快速响应的需要。.本项目综合利用可见光、红外、高光谱等多元遥感数据和地面现场数据，探索出油库火灾特征污染遥感监测机制：（1）测试了油料及其他可燃物火焰光谱，获取了油料燃烧火焰特征光谱数据源，创建了基于油料池火焰识别指数(OPFDI)的油料火焰光谱识别模型。（2）基于天—空—地一体化监测油料火灾污染外场实验，研究提出了烟气斑块的提取模型，其提取精度约为93%；探索了油料燃烧过程中污染物产生扩散的特点与规律，并为航天遥感信息提取模型的构建与校正提供了依据。（3）研究利用实际火灾事故的Landsat 8卫星影像，创建了基于监督分类的PCA—SVM火点像元提取模型及基于非监督分类的火点像元的分类树模型，结果表明两类模型均具有较好的鲁棒性，TPR可达0.8以上；提出了油料火灾烟气识别指数(NSDI)，该指数增强了烟气污染区域的遥感识别信息；构建了基于Landsat 8 卫星第7波段的短波红外数据的油料火焰温度反演模型，研究表明该模型的温度反演值优于传统热红外反演结果。.创建了油库目标遥感监测快速提取模型—研究提出一种基于小波与分水岭变换的遥感图像分割方法，可准确快速的提取油库目标。结合天—地一体化油料火灾污染外场实验数据，数值模拟研究了基于空间遥感信息的油库火灾大气污染扩散过程，并以重庆某油库为例进行了真实条件下的油库火灾污染数值模拟分析；构建了基于遥感技术的油库火灾大气污染损害评估指标体系，并研发出油库火灾大气污染模拟预测与评估的专用软件模块。.该项目成果可为我国突发油库火灾污染的快速发现识别、大尺度动态监测提供基础理论与技术支持，在提高我国突发油库火灾污染应急监测与快速响应能力、促进新型卫星传感器的研发，维护人民生命财产安全、大幅减少国家经济损失，维护国防安全及保护生态环境等方面均具有极其重要的意义和应用前景。