PVCF v1: A High-Resolution, Spatially Consistent Multi-Source Fusion Dataset for Photovoltaic Infrastructure in China

The global energy transition has driven a rapid expansion of clean energy infrastructure, with photovoltaic (PV) systems playing a central role. As the world’s largest producer and installer of PV systems, China has witnessed sustained growth in installed capacity over recent years. Reliable, high-resolution spatial data on PV infrastructure are essential for informing energy policy, optimizing grid integration, assessing land use conflicts, and evaluating ecological impacts. While several publicly available PV datasets for China exist, they vary substantially in time periods, data sources, identification methodologies, and data precision, creating challenges for users who require consistent, high-quality inputs for their analyses. To systematically evaluate and integrate these heterogeneous datasets, this study develops a validation framework incorporating data fusion, DeepLab V3+ validation, and manual annotation. Applying this framework and based on accessible PV dataset, we construct the China’s High-Resolution Photovoltaic Comparison and Fusion (PVCF) dataset, a high-resolution PV vector dataset for China updated to December 2025. The results reveal substantial inconsistencies among the nine source datasets, in which low-confidence regions, defined as areas identified as PV by only a single dataset, constitute the largest proportion and cover a substantial area. Building upon 3,033.82 km2 of high-confidence PV areas from the nine existing datasets, defined as regions identified by more than five sources, the PVCF further incorporates 5,701.94 km2 of manually annotated, refined, or newly identified PV patches, together with 1,627.50 km2 of PV area validated using DeepLab V3+, yielding a total area of 10,363.26 km2. The northwestern provinces collectively accounting for 5,136.82 km2 (49.70%) of the national total, predominantly in the form of large clustered utility scale solar farms, while eastern China is dominated by fragmented and dispersed distributed and rooftop systems. As the most recent high-resolution PV vector dataset for China, PVCF provides a robust data foundation for applications in PV policy planning, biodiversity assessment, and ecological research.全球能源转型推动清洁能源基础设施快速扩张，光伏（photovoltaic, PV）系统在此进程中扮演核心角色。作为全球最大的光伏系统生产与安装国，中国近年来的光伏装机容量持续增长。可靠的高分辨率光伏基础设施空间数据，对于制定能源政策、优化电网并网、评估土地利用冲突以及测算生态影响至关重要。目前已有多套面向中国的公开光伏数据集，但这些数据集在时间范围、数据来源、识别方法与数据精度上差异显著，给需要获取一致、高质量分析输入的用户带来了挑战。为系统性评估并整合这些异质数据集，本研究构建了一套融合数据融合、DeepLab V3+验证与人工标注的验证框架。基于可获取的光伏数据集并应用该框架，我们构建了中国高分辨率光伏对比与融合（Photovoltaic Comparison and Fusion, PVCF）数据集——这是一套更新至2025年12月的中国高分辨率光伏矢量数据集。研究结果显示，9套源数据集之间存在显著不一致性，其中仅被单套数据集标记为光伏区域的低置信度区域占比最高，覆盖面积广阔。基于9套现有数据集中被超过5套数据源标记的高置信度光伏区域（总面积3033.82 km²），PVCF数据集进一步纳入了5701.94 km²经人工标注、优化或新识别的光伏斑块，以及1627.50 km²经DeepLab V3+验证的光伏区域，总覆盖面积达10363.26 km²。西北各省合计覆盖5136.82 km²，占全国总量的49.70%，以连片集中的公用事业级大型光伏电站为主；而中国东部地区则以碎片化、分散式的分布式光伏与屋顶光伏系统为主。作为中国目前最新的高分辨率光伏矢量数据集，PVCF为光伏政策规划、生物多样性评估以及生态研究等应用场景提供了坚实的数据基础。