中国大兴安岭地区火灾后复杂条件下森林恢复遥感监测建模研究

Fire is one of the most common and widespread forest disturbance factors. It not only affects the forest regeneration rate and forest composition as well as structure, but also produces an effect on the nutrient cycling and energy flow of forest ecosystems. The Greater Hinggan Mountain area of China is extremely rich in forest resources, while also acts as one of the most serious regions of forest fires. Taking the burned area in this region as examples, the project proposes to conduct the post-fire forest recovery monitoring under complicated conditions based on Landsat TM/ETM+/OLI and Quickbird data. First, focusing on some specific forest fires in the current year and historic years, on the basis of building an ecological niche coefficient and extracting the fire severity, the post-fire forest regeneration and vegetation recovery trajectories was detected and monitored, with a comparison of differences in forest composition and structure under different ecological niche conditions, fire severity levels and post-fire recovery stages. Aiming at these differences, the number and type of reflecting endmembers were determined using linear spectral unmixing. Then based on Li-Strahler Geometric-Optical model, the forest scenes in the typical plots of burned areas were simulated and modeled. Based on modeling, the corresponding reflecting components were extracted and their characteristics were analyzed. The validation was performed by further inversion of forest parameters (e.g. LAI). The conclusions can be used to study the influence mechanism of fire on forest ecosystems and detect the characteristics of post-fire forest recovery trajectory under complicated conditions, thus providing advanced techniques as well as rich information for forest health research and management.

火灾是影响最为广泛的森林扰动因素之一，它影响了森林的更新速度和组成结构，且干扰其物质循环和能量流动。中国大兴安岭森林资源极其丰富，但也是森林火灾最严重的地区之一。本项目基于Landsat、Quickbird等遥感数据，面向该地区开展火灾后复杂条件下森林恢复监测研究。首先选取现状年和历史年份的森林火灾，基于地形和气候条件构建生态区位系数，提取火灾严重度，监测灾后的森林恢复进程，针对不同生态区位条件、严重度级别和恢复阶段的火烧迹地森林组成和结构特征差异，采用线性光谱分解的思想，确定一定数量和类型的反射端元，基于不同分量的Li-Strahler几何光学模型对火烧迹地典型样地森林场景进行建模，提取各反射分量并分析其光谱特征，通过进一步对叶面积指数等参数的反演进行验证。研究成果可用于探究火灾对于森林的影响机制和监测灾后复杂条件下森林恢复进程，进而为森林健康研究和管理提供先进的技术和丰富的信息。

森林火灾是影响最为广泛的森林扰动因子之一，它既影响森林的组成结构和生长更新，又干扰其物质循环和能量流动，火灾后森林恢复监测对于森林监测管理、全球碳循环研究具有重要意义。中国大兴安岭森林资源极其丰富，但也是森林火灾最严重的地区之一。.本项目基于Landsat、ALOS/PALSAR、GF等时间序列遥感影像，面向该地区开展火灾后复杂条件下森林恢复遥感监测建模研究。首先收集大兴安岭地区的森林火灾数据和多源遥感影像，构建森林火灾相关因子和遥感影像数据库；在基于多元光谱指数提取森林火烧迹地和烧毁森林面积的基础上，通过分析全国第七次和第八次森林资源清查数据中大兴安岭样地的数据（树龄、树高、胸径、郁闭度）对火烧迹地森林更新进行统计分析，结果表明大兴安岭地区森林火灾后最先恢复的先锋森林类型为山杨、白桦等阔叶树种，后续逐渐过渡为樟子松、落叶松等针叶树种。其次，分析地形和气候因子对于火烧迹地植被恢复的影响，构建物候表征参数，基于2000—2018年的时间序列Landsat与MODIS影像，利用STARFM融合算法构建各年份植被生长期相对一致的反射率影像，消除物候对于植被恢复的影响，分析森林恢复进程。再次，基于时间序列Landsat数据、ALOS/PALSAR数据等分析不同的严重度等级、不同的灾后森林更新方式、不同的恢复阶段的森林恢复进程差异，对比各遥感光谱指标参数的监测精度差异。结果表明火灾最严重地区在灾后30年内稳步恢复，但仍未恢复到灾前水平，且处于天然更新恢复方式下的森林呈现出和人工更新、人工促进方式下的森林完全不同的恢复进程。基于RF-VIMP算法分类烧毁森林时，TCW、TCB和SWIR1、SWIR2是最为关键的光谱参数。此外，新构建的雷达植被指数RRVI、RNDVI对于监测也有效。最后，基于Li-Strahler几何光学模型对火烧迹地典型样地的森林场景进行建模，通过进一步对森林郁闭度参数的反演进行了精度验证。.项目研究成果可用于探究火灾对于森林的影响机制和监测灾后复杂条件下森林恢复进程，为当地森林资源监测管理提供了丰富的基础数据和先进的技术支撑。