随着新一代静止气象卫星的发射，高频次和高时效的观测特性对于火点探测具有独特优势。本文基于Himawari-8新一代静止气象卫星高频次观测特点，提出有利于火情初期火点判识的时序探测方法。与传统的极轨气象卫星遥感火情监测采用的上下文法不同，时序探测法判识火点的方法依据为探测像元亮温在观测时间上的差异。研究结果显示，在无云及无异常热源条件下，相邻时次中红外亮温差异较小，当前后时次亮温差达到3K时，可判识出火点，而上下文法的阈值均在6 K以上，时序法的火点判识阈值较上下文法明显降低，探测相应的亚像元火点面积减小一倍以上，从而提高了火情判识的灵敏度，实现火点早期发现。本文介绍了时序法火点判识方法，并以黑龙江桦川县的星地同步观测实验进行验证，研究表明，时序法较上下文法在初发火点探测灵敏度方面有明显优势，时序法和上下文法的结合可提高气象卫星对火情发展过程的监测能力。

Wild fires, especially large-scale wild fires, in forests, grasslands and farmlands have a significant influence on crop productivity, atmospheric pollution, biodiversity, climate change and public health. In recent years, the increasing events of forest fires in China, US, Australia, and Amazon Rain Forests and grassland fires in Mongolia have caused a large number of causality. Due to its great influences, growing emphasis has been placed on the monitoring of wild fires based on remote sensing products, such as using MODIS, NOAA and other polar orbit meteorological satellites. With the launch of a new generation of geostationary meteorological satellites, the characteristics of high frequency and real-time observation have obvious advantages for fire spot detection. Based on the high frequency observation characteristics of Himawari-8 that a new generation geostationary meteorological satellite, the objectives of this paper proposes a temporal sequence detection method to extract the initial fire spot of fire behavior. This method of geostationary meteorological satellite will greatly improve the method of fire identification, give full play to the advantage of temporal sequential, and realize the early detection of fire by remote sensing.The method of study for identifying fire points is based on the pixel temperature brightness difference in observation times and its rate, which is different from the conventional contextual method used in remote sensing fire monitoring of polar orbit meteorological satellites. According to the brightness temperature change value of detected pixel at the same position and different time, when the brightness temperature change value of the current and subsequent times exceeds the threshold, the pixel can be identified as a fire point. The change of observation methods has brought about a great improvement in monitoring sensitivity and timeliness in fire monitor.The results showed that under the condition of cloud-free and no abnormal heat source, the mid-infrared bright temperature have little difference between the adjacent times. Generally, the brightness temperature change of minute interval is less than 0.5 K. When the bright temperature rate between the current time and later time reaches 3K, the fire spot can be identified, while the threshold of contextual method is above 6 K. Compared with the contextual method, the temporal sequential method reduces the threshold of recognition by half and increases the sensitivity by more than twice. The fire spot detection threshold of the temporal sequential method is significantly lower than that of the contextual method.This paper introduces the method of temporal sequence for fire spot detection, and verifies it with the satellite and ground synchronous observation experiment in huachuan county, heilongjiang province. The conclusion demonstrated that this method is benefit to find early fire spot. If the threshold of contextual method is used alone, the early fire is difficult to be obtained, but the method of temporal sequence can make up for this defect. Temporal sequence method is used in the early stage of fire identification, context method is used in the middle stage, The combination of the temporal sequence method and the fluctuation of contextual can improve the meteorological satellite monitoring ability of fire development process.