Retrieval Of Bidirectional Reflectance And It's Application In Changbai Mountain Area

Changbai Mountain is located in the southeastern region of Jilin Province. Its geographical coordinates are 127°40'～128°16'E,41°35'～42°25'N. Changbai Mountain is located on the border between China and North Korea. It is the birthplace of the Tumen River, Yalu River and Songhua River. It is a famous scenic area of national level-5A.The highest peak is Baiyun, 2691 meters above sea level. There are 16 peaks over 2500 meters above sea level, with a total area of 8,000 square kilometers. The vegetation in Changbai Mountain distribute around the Tianchi,extend along the main mountain ranges, showing zonal distribution ring, and a clear vertical mountain landscape. There are alpine tundra zone, betula ermanii forest zone, coniferous forest zone and mixed broad-leaved Korean pine forest zone. It is important to study the structure features of the vegetation on Changbai Mountain because of its typical vegetation features and its vital status in ecological environment.Bidirectional reflectance is a basic phenomenon that all objects in nature reflect the electromagnetic wave. It reflects the spatial distribution of all objects. Bidirectional reflectance is closely related to the composition and surface structure of the objects. As a result it can calculate the three-dimensional structural information of the features on the ground. Currently the research is mainly focused on the bi-directional reflectance distribution function model. It is hot spots of quantitative remote sensing study to create a regional, large-scale BRDF model and to inverse the structural parameters of the ground using BRDF.The MODIS(Moderate Resolution Imaging Spectroradiometer) and MISR (Multi-angle Imaging Spectroradiometer) provide a large number of multi-angle, multi-spectral, multi-resolution image data. And it makes the study of bidirectional reflectance develop rapidly in recent years. In this paper,the MODIS BRDF data is used,of which the resolution is 500m. The detailed analysis and description of algorithm for the products are presented. The BRDF product use the kernel-driven semiempirical Ambrals BRDF model, utilizing five variants of kernel functions characterizing isotropic,volume and surface scattering. We linear regress the 16-day observation of the multi-angle remote sensing data, to get the parameters of the model. Then calculate the BRDF in all solar zenith angle and observation zenith angle.In this paper I selected the MODIS BRDF data from 2000 to 2008. After resampling and projection transformation,the BRDF values of Changbai Mountain are inversed. The values are multi-time, multi-angle and multi-band.Then the spectral characteristics of different features, as well as the BRDF characteristic shape of different vegetation types at different times are analysed. I summarize the changes of BRDF in different types of vegetation,and analysis the effects that the height and slope bring to BRDF. The multi-angle BRDF factor are discussed and applied.Chosing the BRDF values of September 14, 2001 No. 6-band as the standards to interpret the vegetation cover types in Changbai Mountains, and got the vegetation cover classification results. Then evaluate the accuracy based on the vegetation classification results of ETM+ images. The overall classification accuracy of BRDF is 75.19%,and the coefficient kappa is 0.6213. Pine and broad-leaved forest achieve the accuracy of 81.77% and 93.17%,respectively. The results show that the MODIS BRDF products has achieved a high overall classification accuracy. On the one hand, it confirmed the BRDF has the capabilities to identify and classify the vegetation structure. On the other hand,it lays a good foundation for the further application of MODIS BRDF product for large-scale classification.Finally, the thesis discussed the way to improve the classification accuracy using BRDF and multi-angle BRDF factor.