Valores_de_11_indices_espectrales_de_vegetacion_en_11_tipos_de_bosques_en_cuba.xlsx

The objective of obtaining the data was to make a spectral description of eleven types of forests in Cuba using spectral indices of vegetation.The 11 indexes were calculated from 11 images produced by the Landsat 8 satellite and downloaded from the Glovis portal belonging to the USGS in 2017 and corresponding to the months of February to March. The images were preprocessed to obtain reflectance values, eliminate atmospheric effects and mask water bodies and clouds. Finally, with a histogram matching, a map was obtained for Cuba of each spectral vegetation index.The management categories and the name of the protected areas were obtained from the map of protected areas given by the National Center for Protected Areas (CNAP). The names and location of forest types were obtained from the unpublished map of Estradas et al. in 2011 and to reduce the errors of this map regarding the area of ​​real forest, a binarized arboreal coverage map (Tree-Cover of Landsat) was used, for which it was assumed that the values ​​of 1, which represented tree cover 50% or more were forests. This map was vectorized and overlaid the map of Estrada et alAll values ​​were extracted from 10,000 points generated within the binarized coverage map vectors and superimposed on the forest type map of Estrada et al., In ArcGis v 10.1. The points were cleaned and filtered to eliminate sampling errors, so in the end there was a total size of 9513 cases. The coordinates of each point are also shown with a UTM WGS 84 N17 projection本数据集的采集目标为利用植被光谱指数对古巴境内11类森林开展光谱特征描述。11种植被光谱指数由2017年下载自美国地质调查局（USGS）旗下Glovis平台的11景Landsat 8卫星影像计算得到，影像采集时段为2月至3月。研究人员首先对影像进行预处理，以获取反射率值、消除大气影响，并掩膜去除水体与云区；随后通过直方图匹配法，生成古巴全境各植被光谱指数的空间分布图。保护区管理类别与保护区名称源自国家保护区中心（CNAP）发布的保护区地图。森林类型的名称与分布范围则取自埃斯特拉达等人2011年编制的未公开地图。为修正该地图在实际森林面积测算上的误差，研究采用了二值化林地覆盖图（Tree-Cover of Landsat），其中将像元值为1、代表林地覆盖率≥50%的区域定义为森林。将该覆盖图进行矢量化后，与埃斯特拉达等人的森林类型地图进行叠加。随后在ArcGIS v10.1软件中，于二值化覆盖图的矢量化区域内生成10000个采样点并提取相关数据。对采样点进行清洗与过滤以消除采样误差，最终得到9513条有效样本。所有采样点的坐标均采用UTM WGS 84 N17投影坐标系。