斯里兰卡30 m土地覆盖分类数据集(2018)

斯里兰卡30m土地覆盖分类数据集(2018)是基于2018年Landsat OLI生成4季度合成图，计算植被指数、水体指数、灰度共生矩阵非相似性、同质性特征，基于像元挑选样本点，使用随机森林分类器进行监督分类，实现了斯里兰卡7个主要土地覆盖类型的识别，分别为农田、森林、草地、湿地、水域、人造地表和裸地，经目视解译验证，总体精度为0.85左右。该数据集空间分辨率为30 m，数据存储格式为.tif，由1个数据文件组成，数据量为479 MB（压缩为1个文件8.41 MB）。

Sri Lanka 30 m Land Cover Classification Dataset (2018) was developed based on 4-season composite images generated from 2018 Landsat OLI data. Vegetation indices, water indices, gray-level co-occurrence matrix (GLCM) dissimilarity and homogeneity features were calculated. Sample points were selected on a per-pixel basis, and a random forest classifier was employed for supervised classification, enabling the identification of 7 major land cover types in Sri Lanka, namely cropland, forest, grassland, wetland, water body, artificial surface and bare land. Verified via visual interpretation, the overall accuracy of this dataset is approximately 0.85. It has a spatial resolution of 30 m, is stored in .tif format, consists of one single data file, with an uncompressed size of 479 MB (compressed to 8.41 MB as a single file).

搜集汇总

背景与挑战

背景概述

该数据集基于2018年Landsat OLI卫星影像，通过四季度合成和随机森林监督分类方法，生成了斯里兰卡30米空间分辨率的土地覆盖分类图，涵盖农田、森林等7个主要类型，总体精度约为0.85，适用于土地覆盖监测和地理分析。数据以TIFF格式存储，总大小479 MB。