基于极轨卫星与静止轨道卫星数据的火点联合反演研究

Active fire detection is one of the hot issues in the management and assessment of wildfires in forest and grassland. From the aspect of applications in remote sensing, spatiotemporal resolution, cloud and smoke coverage, fire intensity and other factors often result in omission and commission errors of fire identification results. In this project, a comprehensive active fire detection framework based on polar-orbiting and geostationary orbit satellite data is proposed. The project plans to use data from Chinese satellites Fengyun-3 and Fengyun-2 to perform the research. Various complex conditions are taken into consideration, including different vegetation, surface space distribution, and seasonal climate. To further explore the internal mechanism of active fire detection, the project focuses on studying the optimization method of algorithms as well as the corresponding spatiotemporal feature parameters. By establishing the relationship of the radiation energy, the number of pixels and other parameters between related data from Fengyun III and Fengyun II, the project can finally achieve a comprehensive active fire detection structure based on polar-orbiting and geostationary orbit satellite data. This study will be helpful in filling the blank where there are few methods for the joint detection with by domestic satellites. Error rates caused by clouds, smoke and water bodies will also be reduced. On the other hand, this research can accumulate the theoretical and practical achievements for the further study in active fire detection with advanced satellite observation data in the future.

火点识别是森林和草原火灾管理和评估研究中的热点问题之一。从遥感火点识别角度来看，数据时空分辨率、云和烟雾覆盖、火势强度等因素影响常造成火点识别结果的漏判和误判。本项目提出基于极轨卫星与静止轨道卫星数据的火点联合反演研究，以风云三号、风云二号系列卫星作为实例，研究不同植被类型、季节气候条件等各类复杂因素对火点识别的影响机制，研究火点识别过程相关的时空特征参量优化选取方法，通过建立风云三号与风云二号数据间对应火点像元的燃烧辐射能量、像元数目等参量之间的关系模型，实现极轨卫星与静止轨道卫星数据相结合的火点识别。本研究成果将针对性地克服由云、烟雾、水体等因素带来的火点识别干扰，有助于填补国产风云系列卫星进行火点联合反演的方法缺失，为未来卫星组网观测数据在火点识别方面的应用提供先期的理论方法积累。

火点识别是森林和草原火灾管理和评估研究的热点问题，本项目将具有同期观测特性的静止轨道卫星与极轨卫星数据相结合，研究了基于国产风云系列卫星的火点联合反演的方法。具体研究要点包括：对火点识别中存在的内部（火势强度等）和外部（云、烟雾等）影响因素开展量化研究；对影像中的邻域窗口统计量、反射特征参量、热特征参量等火点识别特征实现了优化分析；以风云三号、风云二号系列卫星数据为实例，分析了极轨卫星与静止轨道卫星数据在时间尺度的一致性，建立了多源数据间火点像元数等参量的关系模型，实现了基于静止及极轨卫星的火点联合反演。本项目算法设计中新增加了动态阈值和时间序列的研究内容，使得火点识别精度较国际、国内火点算法精度显著提升。研究设计了多源数据组合方式下火点识别结果的新型验证模式，克服了云、烟雾、水体等干扰因素对火点识别的影响，有助于填补风云系列卫星进行火点联合反演方法的缺失，为未来卫星组网观测数据在火点识别、燃烧面积提取等方面的应用提供了方法基础。