Assessing the suitability of urban-oriented land cover products for mapping rural settlements - accuracy data and code

The data covers two independent accuracy assessmets, based on 2 sets of randomly selected points, for three urban-oriented land cover products: Global Urban Footprint - GUF, European Settlement Map - ESM and Open Street Map - OSM) to detect rural settlements in the Carpathian ecoregion. The study area covers the Carpathians located in 5 countries: Czechia, Hungary, Poland, Romania and Slovakia.<br> The accuracy assessment was conducted with two independent approaches regarding the selection of verification points. In the first approach, a set of verification points was defined through the stratified random selection of 500 points per country (2500 points in total) within the area of any of the three analysed settlement datasets combined, hereafter called the ‘total settlement mask’. In the second approach, a set of verification points was defined through the stratified random selection of 500 independent points per country (2500 points in total) within the ‘total settlement mask’ buffered by 100 m, hereafter called the ‘buffer mask’. The dataset is accompanied with Python code.<br> <br>

本数据集包含两项独立的精度评估任务，基于两组随机选取的验证点，针对三款面向城市的土地覆盖产品——全球城市足迹（Global Urban Footprint, GUF）、欧洲居民地图（European Settlement Map, ESM）以及开放街道地图（Open Street Map, OSM）——展开，用于识别喀尔巴阡生态区的农村居民点。研究区域涵盖分布于捷克、匈牙利、波兰、罗马尼亚及斯洛伐克5个国家的喀尔巴阡山脉。

本次精度评估采用两种独立的验证点选取方案。方案一：在三款待分析居民点数据集的合并覆盖区域（下文简称“总居民点掩膜”）内，按国别实施分层随机抽样，每个国家选取500个验证点，总计2500个。方案二：在经100米缓冲区扩展后的“总居民点掩膜”范围内，按国别分层随机选取500个独立验证点，总计2500个，该采样区域下文简称“缓冲掩膜”。本数据集附带Python代码。