Estimating Biomass Burned Areas from Multispectral Dataset Detected by Multiple-Satellite
YU Chao1, 2, CHEN Liang-fu1*, LI Shen-shen1, TAO Jin-hua1, SU Lin1
1. State Key Laboratory of Remote Sensing Science, Institute of Remote Sensingand Digital Earth, Chinese Academy of Sciences, Beijing 100101, China 2. University of Chinese Academy of Sciences, Beijing 100049, China
摘要: 露天生物质燃烧是重要的大气污染物排放源,导致空气质量恶化并引起气候变化。卫星遥感数据能够提供大尺度、多时相的监测信息,然而燃烧火点监测和火烧迹地监测两种方式都存在着局限性。以美国东南部地区为研究区域,通过结合卫星遥感获取的高分辨率燃烧面积数据及多时相的燃烧火点数据,建立时空匹配模型估算露天生物质燃烧过火面积。通过分析植被燃烧前后的光谱变化特征,基于高分辨率的Landsat-5 TM4波段(0.84 μm)与7波段(2.22 μm)数据, 利用差分归一化燃烧比(dNBR: the differential normalized burn ratio)提取燃烧面积数据;而燃烧火点数据则通过分析燃烧植被的热红外光谱特征利用MODIS 4与11 μm波段数据提取。结果显示,该地区燃烧面积与燃烧火点数量之间相关系数达0.63,并且二者之间的比例关系随植被类型而发生变化,林地、草地、灌木、耕地和沼泽五种植被类型对应的像元燃烧面积分别为0.69,1.27,0.86,0.72和0.94 km2。通过与美国火灾中心(national interagency fire center, NIFC)地面调查数据比对,模型估算的美国东南部过火面积数据较为精确,而同期的MODIS燃烧面积产品(MCD45)及燃烧源清单产品(global fire emissions database, GFED)遗漏了该区域大量的小面积燃烧事件。因此,本研究建立的过火面积估算模型能够提供更为精确的排放源参数信息,有利于区域空气质量模式准确地模拟露天生物质燃烧排放状况。
关键词:露天生物质燃烧;MODIS;Landsat;燃烧火点;过火面积;差分归一化燃烧比
Abstract:Biomass burning makes up an important part of both trace gases and particulate matter emissions, which can efficiently degrade air quality and reduce visibility, destabilize the global climate system at regional to global scales. Burned area is one of the primary parameters necessary to estimate emissions, and considered to be the largest source of error in the emission inventory.Satellite-based fire observations can offer a reliable source of fire occurrence data on regional and global scales, a variety of sensors have been used to detect and mapfires in two general approaches: burn scar mapping and active fire detection. However, both of the two approaches have limitations. In this article, we explore the relationship between hotspot data and burned area for the Southeastern United States, where a significant amount of biomass burnings from both prescribed and wild fire took place.MODIS (Moderate resolution imaging spectrometer) data, which has high temporal-resolution, can be used to monitor ground biomassburning in time and provided hot spot data in this study. However, pixel sizeof MODIS hot spot can’t stand for the real ground burned area. Through analysis of the variation of vegetation band reflectance between pre- and post-burn,we extracted the burned area from Landsat-5 TM (Thematic Mapper) images by using the differential normalized burn ratio (dNBR) which is based on TM band4 (0.84 μm) and TM band 7(2.22 μm) data. We combined MODIS fire hot spot data andLandsat-5 TM burned scars data to build the burned area estimation model, results showed that the linear correlation coefficient is 0.63 and the relationships vary as a function of vegetation cover. Based on the National Land Cover Database (NLCD), we built burned area estimation model over different vegetation cover, and got effective burned area per fire pixel, values for forest, grassland, shrub, cropland and wetland are 0.69, 1.27, 0.86, 0.72 and 0.94 km2 respectively. We validated the burned area estimates by using theground survey data from National Interagency Fire Center (NIFC), our results are more close to the ground survey data than burned area from Global Fire Emissions Database(GFED) and MODIS burned area product (MCD45), which omitted many small prescribed fires. We concluded that our model can provide more accurate burned area parameters for developing fire emission inventory, and be better for estimatingemissions from biomass burning.
Key words:Biomass burning;MODIS;Landsat;Fire Hot Spot;Burned Area;the differential normalized burn ratio (dNBR)
[1] Wiedinmyer C, Akagi S K, Yokelson R J, et al. Geosci. Model. Dev., 2011, 4(3): 625. [2] van der Werf G R, Randerson J T, Giglio L, et al. Atmospheric Chemistry and Physics, 2006, 6: 3423. [3] Naeher L P, Brauer M, Lipsett M, et al. Inhal. Toxicol., 2007, 19(1): 67. [4] Langmann B, Duncan B, Textor C, et al. Atmospheric Environment, 2009, 43(1): 107. [5] Wiedinmyer C, Quayle B, Geron C, et al. Atmospheric Environment, 2006, 40(19): 3419. [6] Soja A J, Al-Saadi J, Giglio L, et al. J. Appl . Remote Sens., 2009,3(1):031504. [7] Giglio L, Descloitres J, Justice C O, et al. Remote Sensing of Environment, 2003, 87(2-3): 273. [8] TIAN Qing-jiu, WANG Ling, BAO Ying, etal(田庆久,王 玲,包 颖,等). Scientia Sinica Informationis(中国科学-信息科学), 2011, 41(Suppl.): 117. [9] Giglio L, Loboda T, Roy D P, et al. Remote Sensing of Environment, 2009, 113(2): 408. [10] Randerson J T, Chen Y, van der Werf G R, et al. J. Geophys Res-Biogeosci., 2012,117(G4):1. [11] Giglio L, van der Werf G R, Randerson J T, et al. Atmospheric Chemistry and Physics, 2006, 6: 957. [12] Smith R, Adams M, Maier S, et al. Remote Sensing of Environment, 2007, 109(1): 95. [13] Tansey K, Beston J, Hoscilo A, et al. J. Geophys Res-Atmos, 2008, 113: D23112. [14] Zhang X, Hecobian A, Zheng M, et al. Atmospheric Chemistry and Physics, 2010, 10(14): 6839. [15] Chander G, Markham B L, Helder D L. Remote Sensing of Environment, 2009, 113(5): 893.