煤田火灾是世界性的自然灾害，不仅造成巨大的经济损失，而且造成严重的环境污染。然而，现阶段煤火的发现主要通过口耳相传、逐级反映的方式，被动性强，滞后性明显；煤田火区的勘探与监测工作更是任务重、风险高、费时费力。遥感技术具有大面积同步观测、成像速度快、信息量丰富、周期短、时效性强等优点，已成为煤田火区地表温度异常信息提取、时空动态监测和地裂缝信息提取的重要手段。 本文以新疆大泉湖火区和乌尊布拉克火区为研究区，以火区地表的温度特征和结构特征为研究目标，首先利用ETM+和TIRS遥感影像，通过地表温度反演算法、密度分割法和人工阈值法开展煤田火区地表温度异常信息提取与时空动态监测研究。同时，利用ZY-3卫星影像，通过建立知识模型和三维可视化技术，提取煤田火区地裂缝信息。主要的研究内容如下： (1)煤田火区地表温度反演。利用ETM+和TIRS遥感影像，依次通过普适性单通道法、Weng算法、覃志豪单窗法和劈窗算法反演煤田火区的地表温度。最后，将反演结果和卫星过境时的地表实测数据进行对比与分析。结果表明：利用ETM+数据进行反演的普适性单通道法精度最高，RSME为0.87℃。 (2)煤田火区动态监测。在获得地表温度分布信息的基础上，利用人工阈值法和密度分割法提取温度异常区。选定重点验证区，将温度异常分布图与磁法勘探图进行叠加分析，计算二者的重叠率及反演的准确率。结果表明：重叠率为82.71%，准确率为80.17%。选定2004年、2009年和2014年的三期ETM+遥感影像，依次提取其温度异常区，统计和分析煤田火区地表温度异常的时空动态变化，结果表明：温度异常分布呈整体减少、局部发育的特征；2004-2009年和2009-2014年，年均分别减少9.95%和8.22%。究其原因，系火区于2008年经过了初步治理的缘故。 (3)煤田火区地裂缝自动提取。基于高分辨率的ZY-3卫星影像，增强处理后，充分利用地裂缝的亮度知识、NDVI知识、纹理知识和火区范围等知识建立知识模型，实现煤田火区地裂缝的自动提取。依次通过图层叠加和建立火区三维可视化模型的方法，定性定量的分析地裂缝的提取效果。结果表明：重叠率为70.63%，划定了煤火的燃烧等级；温度特征与结构特征的相互验证可提高探测煤火的准确性。

Coal fire is a global natural disaster, not only caused huge economic losses, but also caused serious environmental pollution. However, the discovery of coal fires, mainly through word of mouth and reflecting gradually, passive resistance and lag significantly. The survey work in coalfield fire area is effortlessly. Remote sensing technology has a wide range of observations, obtaining a large amount of information, access rapidly, real-time and so on. It has become a important mean to extract surface temperature anomaly information, temporal dynamic monitoring and extract information of ground fissures in coalfield fire area. This paper takes the coal fire area of DaQuanhu and WuZunbraque in Xinjiang Uygur Autonomous Region as research area. Firstly, extracting temperature anomaly information and monitoring real-time by retrieve algorithms, manual threshold method and density slicing method based on ETM+ and TIRS remote sensing images. At the same time, extracting ground fissures information in coalfield fire area by the establishment of knowledge model and 3D visualization techniques based on ZY-3 remote sensing image. (1) Surface temperature retrieving in coalfield fire area. With the images of ETM + and TIRS, we retrieved the LST by the single window algorithm, generalized single channel algorithm, weng algorithm and split window algorithm respectively in coalfield fire area. Finally, comparing and analyzing the results of retrieved with measured data. The results show that: the precision of retrieving by generalized single-channel method with ETM+ data is the highest, the RSME is 0.87℃. (2) Dynamic monitoring in coalfield fire area. On the basis of retrieving LST, we used manual threshold method and density slicing method to calculate threshold and extract temperature anomaly area. Then we selected key authentication region to overlap the temperature anomaly map with magnetic prospecting map, calculate the accuracy rate and overlap rate. The results show that: the overlap rate is 82.71%, accuracy rate is 80.17%. Selected in 2004, 2009 and 2014 three ETM+ images, extracted temperature anomaly sequentially, counted and analyzed the dynamic changes of temperature anomaly information, the results showed: anomaly temperature distribution is overall reduction and local development; from 2004 to 2009 and 2009 to 2014, the annual average reduction are 9.95% and 8.22%. The reason was that this fire area had been governed preliminarily in 2008. (3) Ground fissures extraction in coalfield fire area. Based on the high resolution ZY-3 image, after enhencement, taking full use of the brightness feature, NDVI feature, texture feature and the fire area range to build a knowledge model and extract ground fissures automatically. Then analysising the effect of ground fissures extraction qualitatively and quantitatively through the method of layer overlapping and 3D visualization model building. The results show that: the overlap rate was 70.63%, the fire area can be classified in three grades; mutual authentication between temperature information and structural information can improve the accuracy rate of detecting coal fire.