【作者】 覃先林；

【导师】 李增元； 易浩若； Weimin Hao；

【作者基本信息】 中国林业科学研究院 ， 森林经理学， 2005， 博士

【摘要】 森林火灾是一种世界性的严重自然灾害。它分布广、发生频度高,破坏森林资源,干扰人民正常生活秩序,造成全球性环境污染,越来越受到各国政府的重视。如何采用遥感、地理信息系统技术等现代高新技术对森林火险等级进行预报和对森林火灾进行监测,成为当前国内外研究的热点。本文在查阅大量文献及系统分析的基础上,以遥感和地理信息系统等现代高新信息技术为主要手段,研究了以中等空间分辨率的遥感数据(MODIS)和地理信息系统技术相结合的林火预警应用方法。 本文在综述国内外相关研究的基础上,分别对MODIS数据的处理方法及其在林火预警中的应用方法进行了研究。主要开展了MODIS数据几何校正方法的对比研究,MODIS数据在全国森林可燃物分布与分类、植被状态的反演、植被燃烧信息的提取以及森林火险等级预报等方法的研究。通过开展本研究工作,获得了基于GIS平台的、以MODIS为主要信息源的卫星遥感数据在林火预警中的应用方法。即: (1) 获得了MODIS L1B数据的几何校正和融合等处理方法 本文对比研究了利用GLT方法和GCP方法对MODIS L1B数据的几何校正的优缺点,并提出了对MODIS L1B数掘几何校正处理的建议。同时,获得了不同空间分辨率的MODIS L1B数据集间的融合方法,并利用程序实现了融合算法。 (2) 获得了基于MODIS数据的大尺度森林可燃物分类方法 本文在国内外前人已做的植被、土地覆盖/土地利用等分类研究的工作基础上,借鉴已有的分类成果,首次在国内提出了基于MODIS数据的大尺度的森林可燃物分类体系。在制定的分类体系指导下,采用非监督分类和监督分类相结合的综合分类方法,对我国的森林可燃物分类方法进行了研究与实践,获得了全国可燃物分布图;并探讨了中等空间分辨率的卫星数据在大尺度分类研究中的结果检验方法,即利用GIS技术,将分类结果与专题数据叠加进行精度验证。 (3) 获得了森林火险等级预报因子的处理方法 本文将输入森林火险预警模型中的数据是否经常随着时间的变化而将其分为动态数据和静态数据。通过实验,分别获得了对这两大类数据的处理方法。即: ① 对于静态数据,采用建立背景数据库后再数量化的方法参与模型中计算。 ② 对于动态数据中的每日气象数据,在对比三种空间插值方法在气象数据栅格化结果的精度与效率的基础上,获得了利用反间距空间插值方法对气象数据各因子进行空间插值处理,从而实现了利用已有气象站数据预估其他没有气象观测站设置地区的气象数据的方法;对于植被长势的估测,利用MODIS数据获得的相对绿度指数作为其植被更多还原

【Abstract】 Forest fire is a kind of worldwide natural calamity. It is extensively distributed with high occurrence frequency and destroys forest resources thus disturbing normal living order of people and leading to environmental deterioration. It has been paid more attention to by many governments. It’s a study hotspot in worldwide how to using Modern high-tech information technologies, such as Remote Sensing (RS) and Geographic Information Systems (GIS), to forest fire danger rating predicting and forest fire monitoring. Based on the review of internationally and domestically published papers, Moderate spatial resolution RS data (MODERATE RESOLUTION IMAGING SPECTRO RADIOMETE, MODIS) and GIS Techniques have been used as main techniques in this application study on forest fire early warning and monitoring.Based on the review of internationally and domestically published papers on MODIS, research on how to process MODIS data and their application in forest fire danger prediction have been carried out. Main work includes comparison study of geometric correction method on MODIS data, Chinese national forest fuel distribution and classification using MODIS data, Chinese national vegetation status, Chinese vegetation burning information retrieval and forest fire danger rating. Application methods based on remote sensing data that use GIS as the platform and MODIS as the main information source have been developed in forest fire prediction. Summarization of this dissertation is as follows:(1) Approaches on geometric correction and fuse of MODIS LIB data have been obtainedThe advantages and disadvantages of using geometric correction to process MODIS LIB data by Georeference from Input Geometry (GLT) method and Geometry Control Point (GCP) method were discussed through the comparison study. A followed suggestion about Geometric Correction of MODIS LIB data has been given. Fuse method of different spatial resolution of MODIS LIB data has also been obtained and program of fuse implement has been developed.(2) Classification of Large Scale Forest Fuels Map Based on MODIS Data has been generated.Based on previous researches and achievements on the classification of vegetation, landcover, and land utilization, large scale forest fuels classification system based on MODIS data has been bring forward in China for the first time. Under this classification system, China national forest fuels class has been studied by using an integrative unsupervised and supervised classification method. A national forest fuels distribution map has been produced. At the same times, the classification’s precision and its verification by middle spatial resolution satellite data has been discussed. The precision of this classification was verified by overlaying the classifying results with background thematic data in GIS.(3) The processing method of different forest fire danger rating factors has been obtainedData used by forest fire danger rating model was classified into static data and dynamic data based on whether these data were stable against time. After lab test, the processing method of these two types of data was obtained as the followings:a) For the static data, after it was corrected by the background database, it was digitalized before being inputted into the rating model program.b) For the dynamic data, three kinds of dynamic data resource have been used in the forest fire danger rating system. The first one is the daily weather observation data. Based on the precision and efficiency of three kinds of spatial interpretation method, the best way is the use of Invert Distance Weight (IDW) spatial interpretation method. IDW can predict the weather condition of the region where there is no weather observation station by using weather observation station data from other regions. The second one is the growth of vegetation. Relative greenness parameter has been used to estimate the forest fuels’ growth. The third one is the moisture content of forest fuel, which can be estimated by an integrative method using near infrared reflectance (NIR) channels and short wave infrared reflectance (SWIR) channels. The result shows that the best way to estimate moisture content of forest fuels is using the rate of channel 2 to channel 7 of MODIS according to comparing the results of channel 2 to channel 5, channel 6 and channel 7.(4) National Forest Fire Danger Rating Prediction method has been obtainedIn this paper, the dynamic and static data related with forest fire danger rating prediction have been quantified and used to calculate fire danger index. The fire danger index is the quantified index and grading criterion for Chinese national forest fire danger rating prediction. Consequently, it helps the realization of describing forest fire danger rating from quality to quantification. The method has been verified by the tests in North East of China. The results demonstrated the RS and GIS based method is useful in forest fire prediction.(5) Method to retrieve the vegetation burning information using MODIS data has been obtainedBase on previous research achievement of fire monitoring using MODIS data, the information of related MODIS bands’ data in daytime was analyzed and the accuracy of the result by using Surface Bright Temperature method and Bright Temperature-NDVI method was compared. A model was developed to identify forest fire by using Bright Temperature-NDVI method and integrating GIS technology. In experiment, the totally precision of identification was above 80%, which can satisfy the need of forest fire check-up. At the same time, it can also identify the vegetation type of fire.更多还原