Extraction Of Forest Types And Estimation Of Forest Canopy Closure From Hyperspectral Remote Sensing

Forest planning and management require information about forest resources. Some of the most important information are the spatial distribution of forest types and the forest crown closure. Both of them are an important parameter in ecological, hydrological and climate models. Their measurement in fields are difficult and time-consuming. This is particularly true over large areas. Therefore, the directly measures of forest parameters are only practical on experimental plots of limited site. Consequently, investigating forest parameter estimation over large area is problematic. Remote Sensing techniques, particularly the use of satellite imagery, may provide a practical means to measure forest parameters at the landscape scale or even global scale. With remote sensing techniques, scientists have made use of methods that correlate remotely sensed data with regional estimates of a number of forest ecosystem variables, including forest crown closure, canopy temperature, etc.However, most remote sensing systems in the past decades rely on measured reflectance data in a few broad wavelength bands. Current research is focused on the mapping of forest stands and determining their quantitative parameters using reflectance spectra recorded by airborne and spaceborne imaging spectrometery and its analysis methods such as spectral unmixing. The hyperspectral sensors have been developed to provide more than 220 spectral contiguous and very narrow bands across a full spectral range from 0.4 to 2.5 um, such as Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) and EO-1 Hyperion, the first spaceborne hyperspectral imager in the world. So that much smaller spectral difference of objects can be detected by hyperspectral image. This is especially important for vegetation recognition because the spectral characteristics of different vegetation are similar and difficult to be classified. In contrast to traditional multispectral sensors, the hyperspectral sensors are expected to improve the ability of observing the earth surface, and it becomes one of the most important leading research fields of remote sensing.The recognition and classification of forest parameters is the basis of forestry remote sensing. In this dissertation, taking Wangqing Forestry Bureau, Jilin Province of China as a study area, focuses on the research on extraction of forest types information and estimation of forest crown closure using hyperspectral remote sensing data (EO-1 Hyperion data). The research contents and major conclusions of this paper can be summarized as following:1. Review of hyperspectral remote sensing application in forestry...