Ultra-high Resolution Land Use Data Set of Typical Villages in Northeastern Tibetan Plateau

This dataset was collected by a research team during a field investigation in the Hehuang Valley of Qinghai Province from July to August 2022. Using the DJI Mavic2pro equipped with a Hasselblad L1D-20c camera, 55 typical villages were selected in the Hehuang Valley and over 4600 aerial photographs were obtained using drone photogrammetry technology as raw data. Using AgisfphotoScan 1.25 software to synthesize orthophoto images with a spatial resolution of 0.05m. The vector data of human settlement boundaries in villages was extracted through visual interpretation. Based on the object-oriented human-machine interaction interpretation method, 55 typical village land use datasets in 2022 were obtained (including forests, grasslands, forest land, cultivated land, water bodies, roads, unused land, and building land, totaling 8 categories). By establishing 1050 sample points and using confusion matrix analysis, it was found that the overall accuracy of the dataset was 96.86%, with a Kappa coefficient of 0.95. It can accurately reflect the spatial form, land use composition, and surrounding environment of typical villages. Aerial photographs all have longitude, latitude, and altitude information, providing ultra-high resolution data sources for village spatial structure analysis, land use mapping, and analysis work, effectively assisting in the improvement of human housing and rural revitalization strategies.本数据集由某研究团队于2022年7月至8月在青海省河湟谷地开展野外调查期间采集。研究团队选取河湟谷地内55个典型村落，使用搭载哈苏L1D-20c相机（Hasselblad L1D-20c）的大疆御2专业版（DJI Mavic2pro）无人机，通过无人机摄影测量技术获取了超过4600张航拍照片作为原始数据。使用AgisfphotoScan 1.25软件合成空间分辨率为0.05米的正射影像。通过目视解译提取村落人居环境边界矢量数据；基于面向对象的人机交互解译方法，获取2022年55个典型村落的土地利用数据集，涵盖森林、草地、林地、耕地、水体、道路、未利用地及建筑用地，共计8类。通过布设1050个样本点并采用混淆矩阵分析，结果显示该数据集整体精度达96.86%，Kappa系数为0.95，可精准反映典型村落的空间形态、土地利用构成及周边环境。所有航拍照片均携带有经纬度与海拔信息，可为村落空间结构分析、土地利用制图与分析工作提供超高分辨率数据源，有效助力人居环境改善与乡村振兴战略的推进实施。