並列タイトル等擬似的多方向観測データを用いた針葉樹林・広葉樹林の特徴抽出
タイトル（掲載誌）CEReS環境リモートセンシングシンポジウム資料集
一般注記出版タイプ： VoR
type:text
[ABSTRACT] Forest classes for global land cover data is important for global environmental studies. However, it is difficult to correctly classify needle-leaves forest and broad-leaves forest using phenology information extracted from observation data as a nadir view. In this study, we attempted to differntiate needle-leaves and broad-leaves forests using MOD09GA and MYD09GA. To produce pseudo multidirectional satellite data, we selected several days of data with scannig widths overtapping along the crosstrack direction. We used the pseudo multidirectional data sets to study reflectance on four bands, as well as the variation with solar zenith and relative azimuth for NDVI (Normalized Difference Vegetation Index) and GVI (the green chlorophyll index). Data selection followed criteria aimed to reduce as much as possible the influence of viewing direction and terrain conditions. We found that NDVI and GVI change depending on the solar zenith and relative azimuth. Using data in limited ranges of solar zenith and relative azimuth, it was not possible to clearly show differences between the characteristics of needle-leaves forest and broad-leaves forest from pseudo multidirectional observation data. However, we found differences for both NDVI and GVI between nadir and stant data depending in the surface state of the forest.
連携機関・データベース国立情報学研究所 : 学術機関リポジトリデータベース（IRDB）（機関リポジトリ）