chapter  5
42 Pages

Hyperspectral Remote Sensing of Vegetation Bioparameters ����������������������������������

ByRuiliang Pu, Peng Gong

Imaging spectroscopy, as a new remote-sensing technique (i.e., “hyperspectral remote sensing”), is of growing interest to Earth remote sensing. Hyperspectral remote sensing refers to a special type of imaging technology that collects image data in many narrow contiguous spectral bands (<10-nm bandwidth) throughout the visible and solar-reected infrared portions of the spectrum (Goetz et al. 1985). Since many Earth surface materials show diagnostic absorption features that are from 20-to 40-nm spectral resolution (Hunt 1980), spectral imaging systems, which acquire spectral data in contiguous narrow bands at <10-nm resolution, can produce data with sufcient resolution for direct identication of those materials with diagnostic spectral features. However, traditional remote sensing

CONTENTS

5.1 Introduction ........................................................................................................................ 101 5.2 Spectral Characteristics of Typical Bioparameters ........................................................ 103

5.2.1 Leaf Area Index, Specic Leaf Area, and Crown Closure ............................... 105 5.2.2 Species and Composition ...................................................................................... 106 5.2.3 Biomass .................................................................................................................... 107 5.2.4 Pigments: Chlorophylls, Carotenoids, and Anthocyanins .............................. 107 5.2.5 Nutrients: Nitrogen, Phosphorous, and Potassium .......................................... 108 5.2.6 Leaf or Canopy Water Content ............................................................................ 108 5.2.7 Other Biochemicals: Lignin, Cellulose, and Protein ......................................... 109

5.3 Analysis Techniques and Methods ................................................................................. 109 5.3.1 Derivative Analysis ............................................................................................... 109 5.3.2 Spectral Matching .................................................................................................. 110 5.3.3 Spectral Index Analysis ........................................................................................ 111 5.3.4 Analysis of Absorption Features and Spectral Position Variables ................. 118 5.3.5 Hyperspectral Transformation ............................................................................ 120 5.3.6 Spectral Unmixing Analysis ................................................................................ 122 5.3.7 Hyperspectral Image Classication .................................................................... 124 5.3.8 Empirical/Statistical Analysis Methods ............................................................. 126 5.3.9 Physically Based Modeling .................................................................................. 127

5.4 Summary and Future Directions .................................................................................... 129 Acknowledgments ...................................................................................................................... 130 References ..................................................................................................................................... 130

systems, which usually are called “multispectral remote sensing” systems and acquire data in a few discrete wide bands (usually >50-nm bandwidth), cannot resolve these spectral features (Goetz et al. 1985; Vane and Goetz 1988). Therefore, the value of hyperspectral remote sensing lies in its ability to acquire a complete reectance spectrum for each pixel in an image, and it is developed for improving identication of materials and quantitative determination of physical and chemical properties of targets of interest, such as minerals, water, vegetation, soils, and human-made materials.