红树林冠层高光谱探测及其群落类型识别研究

It is very important to obtain the information about the changes of mangroves and the distribution of mangrove types, which is required for decision-making.The remote sensing discrimination and spatial distribution mapping of mangrove community is the base of study on landscape structure and its dynamic analysis of mangrove wetland, coastline change, change of sea level elevation, global environment and climate change. However, there was very few research on remotely sensed discrimination of mangrove community, especially hyperspectral remote sensing. The project will propose the mangrove discrimination index and mangrove community discrimination index, and develop the hyperspectral remote sensing extraction mode of mangrove community types based on Hyperion imagery, TM imagery, DEM, tide level and the habitat investigation of mangrove community. Thus the accurate mapping of mangrove community will carry out. Firstly, the instantaneous seawater boundary line was extracted and then mangrove suitable growth region was derived. Secondly, in order to accurately distinguish mangrove from the other coastal land vegetation, the reflectance spectra of mangrove, DEM,temperature and moisture information will be used to put forward the mangrove discrimination index. Thirdly, the optimal phase, optimal bands and optimal spectral features of mangrove community discrimination will be obtained by means of the reflectance spectra of different mangrove canopy and its components. Fourthly, the mangrove community discrimination index will be erected by using reflectance spectral characteristics, the levels of mangrove community spatial distribution, tide level and moisture information. Finally, we will develop the hyperspectral remote sensing extraction mode of mangrove communities types and remote sensing discrimination and spatial distribution mapping of mangrove community was accomplished.

及时准确地掌握红树林群落现状信息可为保护和恢复红树林生态系统提供重要的决策依据。对红树林群落进行遥感分类在实际应用中具有较大的意义。高光谱遥感技术因其丰富的光谱信息被运用到红树林遥感信息的提取研究中，前人多运用航空高光谱数据对红树林群落加以识别，但具有费用高、观测范围小、无法长时间进行观测等缺点。星载高光谱遥感则可以克服航空高光谱遥感的局限性，实现大范围的红树林的群落的识别及其动态监测。.本项目拟通过选择红树林典型分布区，基于"星地"同步试验，探讨采用星载高光谱遥感技术实现红树林群落类型精细识别的可行性，并以EO-1/Hyperion星载高光谱影像为主要数据源，结合TM影像、DEM、潮位及红树林群落生境调查信息，提出红树林判定指数和红树林群落识别指数，构建红树林群落类型高光谱遥感提取模式，以实现红树林群落的精确制图。

快速准确地掌握红树林群落现状信息可为保护和恢复红树林生态系统提供重要的决策依据。对红树林群落进行遥感分类在实际应用中具有较大的意义。高光谱遥感技术因其丰富的光谱信息被运用到红树林遥感信息的提取研究中，前人多运用航空高光谱数据对红树林群落加以识别，但具有费用高、观测范围小、无法长时间进行观测等缺点。星载高光谱遥感则可以克服航空高光谱遥感的局限性，实现大范围的红树林的群落的识别及其动态监测。本项目以深圳湾红树林保护区为研究区，基于红树林的植被及湿地特征，首先通过提取瞬时海边线，并结合红树林相内陆分布的范围，采用缓冲区分析获得红树林的生长适宜区；然后结合不同潮位的Landsat TM影像，在充分考虑红树林湿地周期性地被海水浸淹的特点，综合亮温、湿度、潮位变化、DEM (Digital Elevation Model)等信息，构造了温湿度指数 (Temperature Moisture Index, TMI)及红树林识别指数(Mangrove Recognition Index, MRI)，并以此准确地提取出红树林分布区，然后结合EO-1 Hyperion星载高光谱影像，获得红树林分布区内的高光谱影像；最后采用连续小波变换方法，通过比较不同的变换尺度，找出不同红树林品种之间的特征差异，运用多范围光谱特征拟合(Multi Range Spectral Feature Fitting)实现了对红树林品种的高精度识别和制图。