LBA-ECO CD-34 PROCESSED HYPERION IMAGES FOR THE ENTIRE AMAZON BASIN

This data set contains 20 multispectral surface reflectance images collected by the EO-1 satellite Hyperion sensor at 30-m resolution and covering the entire Amazon Basin for 2002 - 2005. All images were converted to GeoTiff format for distribution. The respective ENVI *.hdr files are included as companion files and contain image projection and band information. The selected multispectral images were processed using ENVI software as described in Chambers et al. (2009). Bands with uncalibrated wavelengths and those with low spectral response were removed leaving a spectral subset of generally 196 bands (some images have fewer). A cloud mask was developed using 2-d scatter plots of variable reflectance bands to highlight clouds as regions of interest (ROIs), allowing clouds and cloud edges to be masked. A de-streaking algorithm was then applied to the image to reduce variance in balance between the vertical columns. Apparent surface reflectance was calculated for this balanced image using the atmospheric correction algorithm ACORN in 1.5pb mode (AIG-LLC, Boulder, CO). The images (18 of the 20) were georeferenced using the corresponding Advanced Land Imager (ALI) satellite images.本数据集包含20幅由EO-1卫星Hyperion传感器（Hyperion sensor）采集的多光谱地表反射率图像，空间分辨率为30米，覆盖2002年至2005年的整个亚马逊流域。所有图像均转换为GeoTiff格式用于分发。配套提供对应的ENVI *.hdr文件，其中包含图像投影与波段信息。 本数据集所选用的多光谱图像均按照Chambers等人（2009年）的研究方案，使用ENVI软件完成预处理。研究人员移除了波长未校准以及光谱响应较低的波段，最终保留约196个波段的光谱子集（部分图像波段数略少）。通过可变反射率波段的二维散点图将云标记为感兴趣区域（Regions of Interest，ROIs），以此生成云掩膜，实现云层与云边缘区域的掩膜处理。随后对图像应用去条带算法，以降低垂直列间的亮度平衡方差。使用大气校正算法ACORN的1.5pb模式（AIG-LLC，美国科罗拉多州博尔德市）对该经亮度平衡后的图像计算表观地表反射率。其中18幅图像通过对应的先进陆地成像仪（Advanced Land Imager，ALI）卫星图像完成地理配准。