1. 北京大学空间信息集成及其应用重点实验室，地球与空间科学学院，北京 100871
2. School of Mechanical and Aerospace Engineering, Nanyang Technological University，639798, Singapore
3. 河海大学地球科学与工程学院，江苏 南京 211100
Using a Polarization Method to Reduce the Vegetation Inversion Error Caused by Strong or Weak Reflection Intensity
ZHAO Shou-jiang1, YANG Bin1, JIAO Jian-nan2, YANG Peng1, WU Tai-xia3*, WANG Xue-qi1, YAN Lei1*
1. The Beijing Key Lab of Spatial Information Integration and Its Applications, School of Earth and Space Sciences, Peking University, Beijing 100871, China
2. School of Mechanical and Aerospace Engineering, Nanyang Technological University，639798, Singapore
3. School of Earth Sciences and Engineering, Hohai University, Nanjing 211100, China
Abstract：Whether the development of multi angle remote sensing, or the development of polarization and hyperspectral remote sensing all have the same purpose. They use the characteristics and spatial characteristics of electromagnetic waves to accurately identify all the surface of the earth’s object. Any single method and means cannot fully describe and reflect all the features of the ground. Polarization measurement is one of the indispensable technologies in target recognition and recognition technology, and has become a research hotspot in the field of target recognition in the world in recent years. Since the effects of strong and weak reflection intensity on vegetation remote sensing cannot be ignored in quantitative remote sensing inversion, which renders the reflected radiant signal as either saturated or too weak to be detected. Polarization is an important method for the quantitative remote sensing of vegetation. Consequently, it is necessary to develop a method to overcome the vegetation inversion error caused by strong and weak reflection intensities, which is the goal of our present research. If the reflected radiant signal is either too strong or too weak, it will affect the accuracy of remote sensing. Polarized light from vegetation can provide useful information, especially when the reflected radiant signal is saturated so that the sensor cannot obtain enough useful non-polarization information. This study developed a polarization method to overcome the vegetation inversion error caused by strong or weak reflection intensity using a ground-based polarized field imaging spectrometer system. The FISS-P polarization imaging spectrometer system was used to study the effect of reflection intensity on the utility of remote sensing vegetation NDVI and DoLP. The experiment was conducted at the Olympic Science and Technology Park of Chinese Academy of Sciences in Beijing. When targets are sampled, the vegetation with strong reflectivity, low reflectivity and moderate reflectivity is measured respectively. Meanwhile, the DoLP of target vegetation’s different bands (470, 555, 670, 864 nm) are calculated and analyzed. The degree of vegetation density (NDVI) is very low due to signal saturation and shadow effect, resulting in severe inversion error. However, strong reflection has little effect on DoLP. As the ground-based field imaging spectrometer system (FISS-P) provides high-spatial-resolution images with polarization information, we can determine the spectrum-polarization characteristics of single pixels in shaded and strong reflection areas. On the basis of the imaging spectral information, the physical properties of the ground objects are analyzed by using the polarization of light. In this paper, the Stokes component is used to characterize the polarization components of the reflected light, and the degree of polarization of the reflected light is characterized by linear polarization (DoLP). Signal saturation and shadow effects result in very low values for dense vegetation on the Normalized Difference Vegetation Index (NDVI), causing serious inversion error. However, strong reflection has few effects on the degree of linear polarization (DoLP). This study showed that polarization can improve vegetation inversion accuracy by using the appropriate band when the reflected radiant signal is saturated, and the relative error of the average NDVI is 33.8%, while that of DoLP (670 nm) is only 6.3%,the relative errors of DoLP (555, 864 nm) in other bands are much larger. The results of this study show that strong reflection can be ignored when identifying vegetation, however, the shadow (weak reflection) effects could not be ignored. FISS-P images are an effective tool for calculating polarization and non-polarization parameters for sample types with different reflection intensities. In conclusion, the polarization effect can improve the vegetation inversion accuracy when the reflected radiant signal is saturated compared with non-polarization methods. This study analyzed the reduction in error caused by strong and weak reflection intensities using a polarization method. And there are still some problems need to be solved in order to further reveal the relationship between the shadow (weak reflection) effects and DoLP of vegetation.
Key words：Degree of linear polarization；Field imaging spectrometer system；NDVI；Reflection intensity
作者简介: ZHAO Shou-jiang, (1996—), Ph. D. student, the School of Earth and Space Sciences, Peking University
e-mail:
1601110606@pku.edu.cn
引用本文:
赵守江，杨 彬，焦健楠，杨 鹏，吴太夏，王雪琪，晏 磊. 利用一种偏振方法减小由强反射或弱反射强度引起的植被反演误差[J]. 光谱学与光谱分析, 2018, 38(10): 3315-3320.
ZHAO Shou-jiang, YANG Bin, JIAO Jian-nan, YANG Peng, WU Tai-xia, WANG Xue-qi, YAN Lei. Using a Polarization Method to Reduce the Vegetation Inversion Error Caused by Strong or Weak Reflection Intensity. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2018, 38(10): 3315-3320.
[1] Goetz A F H. Remote Sensing of Environment, 2009, 113：S5.
[2] Tong Q, Xue Y, Zhang L. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2014, 7(1)：70.
[3] Jacquemoud S, Verhoef W, Baret F, et al. Remote Sensing of Environment, 2009, 113：S56.
[4] Lewis P, Gómez-Dans J, Kaminski T, et al. Remote Sensing of Environment, 2012, 120：219.
[5] Litvinov P, Hasekamp O, Cairns B, et al. Journal of Quantitative Spectroscopy and Radiative Transfer, 2010, 111(4)：529.
[6] Suomalainen J, Hakala T, Puttonen E, et al. Journal of Quantitative Spectroscopy and Radiative Transfer, 2009, 110(12)：1044.
[7] Peltoniemi J, Hakala T, Suomalainen J, et al. Journal of Quantitative Spectroscopy and Radiative Transfer, 2009, 110(17)：1940.
[8] Kotchenova S Y, Vermote E F. Applied Optics, 2007, 46(20)：4455.
[9] Sun Z, Zhang J, Zhao Y. Geoscience and Remote Sensing Letters, 2013, IEEE, 10(1)：170.
[10] Wu T, Zhang L, Cen Y, et al. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2013, 6(2)：724.
[11] Zhang L, Huang C, Wu T, et al. Sensors, 2011, 11(3)：2408.
[12] Adeline K R M, Chen M, Briottet X, et al. ISPRS Journal of Photogrammetry and Remote Sensing, 2013, 80：21.
