林业科学  2018, Vol. 54 Issue (5): 135-142 PDF
DOI: 10.11707/j.1001-7488.20180515
0

#### 文章信息

Yang Wei, Jiang Xiaoli

Review on Remote Sensing Information Extraction and Application of the Burned Forest Areas

Scientia Silvae Sinicae, 2018, 54(5): 135-142.
DOI: 10.11707/j.1001-7488.20180515

### 作者相关文章

1. 太原师范学院地理科学学院 晋中 030619;
2. 太原师范学院城镇与区域发展研究所 晋中 030619

Review on Remote Sensing Information Extraction and Application of the Burned Forest Areas
Yang Wei1, Jiang Xiaoli2
1. School of Geography Science, Taiyuan Normal University Jinzhong 030619;
2. Institute of Urban and District Development, Taiyuan Normal University Jinzhong 030619
Abstract: Forest fires are one of the most important disturbances of terrestrial ecosystems, atmospheric environment and global change. Traditional fire statistical data were based on field investigation which is difficult to collect and spatialize. The development of remote sensing technology provides a new method for fire study. This paper aims to summarize the relevant studies of burned area based on remote sensing, including data development and data application, and bring up a prospect for future research. This paper firstly reviewed researches on the burned area from two aspects: 1) Existing data product set of burned area and its development. Firstly, the existed burned area data were introduced from several aspects, such as name of dataset, remote sensing data source, coverage, time span, spatial resolution and time resolution. Secondly, algorithms for mapping burned area were compared, including thermal anomaly based method, Spectralanomaly based method, and a combination of them. When fire happens, there are obvious temperature anomaly characteristics which are the base of active fire detection. There are also obvious changes in the vegetation dynamics pre- and post-fire, which are the base of mapping burned area. Most algorithms for mapping burned area only consider one characteristic (temperature anomaly or vegetation dynamics) in mapping burned area, which could cause a relative high error. Thirdly, accuracy of current burned area products was summarized from omission errors, commission errors and overall accuracy. 2) The applications of burned area products in fire hazard assessment, fire emissions and dynamic global vegetation model are reviewed. Finally, the paper points out, the future remote sensing based burned area study may include three aspects: 1) Long-term series burned area data are needed for understanding the long-term fire, climate and vegetation interaction. 2) High accuracy burned area products are needed in regional fire study. 3) New outputs such as fire energy, fire temperature and fire size are expected by different end-users which can expand the scope of the application of burned area data.
Key words: burned area    remote sensing    ecological disturbance    forest fire

1 火烧迹地数据产品及提取方法 1.1 火烧迹地数据产品

1.2 火烧迹地提取方法

1.3 火烧迹地数据产品精度验证

2 火烧迹地数据产品应用研究

2.1 火灾风险评估

2.2 火灾燃烧排放估算

 ${M_i} = A \times {\rm{AFL}} \times \beta \times {\rm{E}}{{\rm{F}}_i}。$

2.3 全球植被动力学模型

2.4 火灾后生态系统监测

3 结论与展望

1) 构建长时间序列火烧迹地数据产品集。火灾的发生为随机事件，长时间序列数据能够更好的刻画火灾机制(Moreira et al., 2011)，分析火灾发生频率与气候变化的相关性(Achard et al., 2008)以及评价火灾对于全球碳循环的影响(Jing et al., 2000)。而现有的火烧迹地数据产品时间序列均较短，如时间序列较长的MODIS火灾数据产品只能提供2000年以来的数据，并未涉及2000年之前，而GBS数据提供了1982至1999年的火灾迹地数据产品。如何对不同的数据产品进行整合，或者开发适用于不同遥感数据源的火烧迹地信息提取算法，建立长时间序列火烧迹地数据产品将成为未来火烧迹地研究的一个重要方向。

2) 开发高精度的火烧迹地数据产品集。如前所述，现有的火烧迹地数据产品虽然数量较多，但整体而言精度较低。充分考虑火灾发生时以及火灾发生前后的各种异常特性，真实反映火灾发生所带来的地表特征变化，才能建立更加精确、稳定的提取算法，提高火烧迹地数据产品精度。此外，从目前使用的数据来看，多为时间分辨率较高而空间分辨率较低的遥感数据。而现有的数据如高分2号，时间分辨率为5天，比MODIS、AVHRR等数据低，但其多光谱数据空间分辨率可达到4 m，开发针对此类卫星遥感数据或者综合利用2类卫星遥感数据的火烧迹地提取方法，将大大提高火烧迹地数据产品精度。

3) 建立多样化的火烧迹地数据产品集。现有的火烧迹地数据产品集均只包含了火烧像元的时间、空间位置以及面积等信息，但对于火灾发生时的其他特征却并未涉及，如火烧能量(fire energy)(Luigi et al., 2009Ellicott et al., 2009)、火焰温度(Fire temperature)以及火灾规模(fire size)(Eckmann et al., 2008)等，而这些特征在相关研究中均有着重要的应用。此外，现有的火烧迹地数据集对于热带地区的火灾(彭光雄等，2008)以及秸秆焚烧(Kant et al., 2008)的探测能力有限。改良现有火烧迹地数据产品集，增加新的监测内容，将进一步扩大火烧迹地数据产品集的应用范围。