A dataset of multi-source typical land cover classification in the Field Park in Wuhan (2020)

<p indent="0mm">Land cover classification of remote sensing imagery is a crucial method for acquiring surface information, and is essential for accurately understanding the characteristics and changes of the Earth’s surface. Given the limitations of existing land cover classification datasets, such as single data sources and insufficient spectral information, this study constructs a dataset of multi-source remote sensing land cover classification with rich spectral information and high spatial resolution. This dataset was collected from Wuhan Field Park and its surrounding areas, with data processing and labeling performed using ArcGIS and Labelme software. To assess the quality of the dataset, various land cover classification models were trained and validated for accuracy. This dataset includes three types of data sources: visible light, thermal infrared, and five single-band multispectral data, with spatial resolutions of <sc>0.0700 m,</sc> <sc>0.2684 m,</sc> and <sc>0.1540 m,</sc> respectively. It covers six typical land cover types: water bodies, vegetation, bare land, buildings, brick roads, and asphalt roads. In total, it consists of 612 image tiles with corresponding labels. This dataset provides data support for research fields such as multi-source data fusion and analysis, as well as remote sensing land cover classification.