基于MODIS数据火点监测指数研究

Fire-monitoring method of FPI-NDVI based on MODIS data

根据火点辐射与背景辐射的差异及MODIS数据火灾监测原理,利用MODIS数据22通道(火点辐射)和31通道(背景辐射)亮温构建了一个火点指数FPI。根据只有当植被覆盖率达到一定程度时才会发生火灾的情况,利用归一化植被指数(NDVI)对其火点提取精度进行了修正,去除了地表不是植被、或植被稀少的不足以引起火灾的区域,有效减少了因裸土或城市热岛引起的误判。就湖南2008年3月1日和黑龙江2004年10月15日特大森林火灾的情况,提取了几种地物地FPI、NDVI值,分析了FPI、NDVI值与不同地表类型的关系。其中,云是干扰火点FPI值的最大因素,地表植被是干扰燃烧植被的最大因素,并确定了FPI和NDVI的阈值。经过分析,相对于MODIS绝对火点识别算法,其精度及提取速度上都有提高。

The present research intended to introduce the author＇s fire-monitoring method of FPI-NDVI, or a fire point index-FPI index by using the fire point radiation MODIS data and the 22 channels as well as the 31 channels （background radiation）. According to the fire-monitoring principle and the difference between the background radiation and fire point radiation, fire accidents can take place usually only when the vegetation coverage reaches a certain extent. It was just for this reason that the conditions of the fire-occurrence were revised and the possibilities of fire-cansing disasters were excluded by removing the areas with no vegetation and the places with scarce vegetation, it was necessary to remove the liability for the bare soil or the urban thermal islands to get on fire. By taking the fire that took place in Hunan province on Ma. 1,2008 and Heilongjiang province on Oct. 15, 2004, as an exampie, in particular, this paper drew a few kinds of geographical features of the principle and analyzed the values of NDVI as well as the relations between the different surface types. Next, cloud was taken as the biggest factor that helps to define FPI and the NDVI threshold value with the vegetation being the key factor influencing the fire burning. Three strong points can be found in the use of MODIS absolute fire point recognition algorithm, which can be illustrated as follows： first of all, it is the noise ingredient for the effective attenuation-detecting instrument; secondly, it may have some counter-balanced influence on the air movement in a certain degree; third, it helps the fire-detectors＇ to narrow their scanning attention. Thus, it can be concluded that the suggested method in this paper helps to enhance its monitoring precision with greater extraclion speed.