果洛藏族自治州高寒草地分布数据集（1990-2020）

高寒草地作为青藏高原地表覆盖重要的组成部分，其时空分布是探究东亚、南亚的生态屏障安全和区域经济社会可持续发展的重要支撑。果洛藏族自治州（以下简称“果洛州”）作为高寒草地广泛分布的典型区域，既是青藏高原主要牧区的组成部分，也是黄河的源区，高寒草地的时空分布数据不仅是探究青藏高原生态屏障安全和畜牧产业可持续发展的基础，也是探索黄河流域自然和经济社会的高质量发展策略制定与落地的重要参考。本数据集首先引入Mann-Kendall-Sneyers检验方法，并参考ESA Wordcover 10m 2020数据集、谷歌地球中高分影像，基于Landsat卫星系列遥感数据计算的长时间序列归一化差值植被指数（NDVI），构建形成了一套适合于长时序遥感分类的稳定样本点数据，主要包括草地、水体、林地和其他共四类，其次，基于稳定样本点和Landsat卫星系列遥感数据，利用随机森林分类算法，提取了1990-2020年每5年一期的果洛州草地分布数据，采用混淆矩阵的精度验证结果显示草地生产者精度和用户精度均保持在0.9左右，总体精度和Kappa系数则分别在0.89、0.77以上；最后参考《生态环境状况评价技术规范》（HJ192-2015），基于草地的提取结果和NDVI，通过计算植被覆盖度，将果洛州草地划分为草地低覆盖度区域、草地中覆盖度区域、草地高覆盖度区域和其他区域，形成果洛州高寒草地分布及对应的覆盖度等级数据。

As a critical component of the land cover on the Qinghai-Tibet Plateau, the spatiotemporal distribution of alpine grasslands serves as an important support for exploring the ecological barrier security in East and South Asia and the sustainable development of regional economy and society. Golog Tibetan Autonomous Prefecture (hereinafter referred to as "Golog Prefecture"), a typical region with widespread alpine grasslands, is not only a part of the main pastoral areas on the Qinghai-Tibet Plateau but also the source region of the Yellow River. The spatiotemporal distribution data of alpine grasslands not only lays the foundation for exploring the ecological barrier security of the Qinghai-Tibet Plateau and the sustainable development of the animal husbandry industry, but also provides an important reference for the formulation and implementation of high-quality development strategies for the natural, economic and social systems of the Yellow River Basin. Firstly, this dataset introduces the Mann-Kendall-Sneyers test method, and refers to the ESA WorldCover 10m 2020 dataset and high-resolution imagery from Google Earth. Based on the long-time-series Normalized Difference Vegetation Index (NDVI) calculated from Landsat satellite series remote sensing data, a set of stable sample point data suitable for long-time-series remote sensing classification is constructed, which mainly includes four categories: grassland, water body, woodland and others. Secondly, based on the stable sample points and Landsat satellite series remote sensing data, the Random Forest classification algorithm is used to extract the grassland distribution data of Golog Prefecture every 5 years from 1990 to 2020. The accuracy verification results using the confusion matrix show that both the producer's accuracy and user's accuracy of grassland remain around 0.9, while the overall accuracy and Kappa coefficient are above 0.89 and 0.77, respectively. Finally, referring to the Technical Specifications for Ecological Environment Status Assessment (HJ192-2015), based on the grassland extraction results and NDVI, the grasslands in Golog Prefecture are divided into low-coverage grassland areas, medium-coverage grassland areas, high-coverage grassland areas and other areas by calculating vegetation coverage, forming the alpine grassland distribution and corresponding coverage grade data of Golog Prefecture.