Farmland shelterbelts in Northeast China and four other types of land cover sample points

Field investigations were conducted from June to September 2023 to collect spectral data of farmland shelterbelts using Sentinel-2 satellite imagery. A multi-source reference system was established by integrating the latest high-resolution satellite images, ground survey data, and high-definition Google Earth imagery. The sample data collection procedure consisted of the following steps: (1) Field surveys at 1,800 sampling points provided accurate ground truth data for spectral characteristics and spatial distribution of different land cover types. (2) Manual interpretation of high-resolution Google Earth imagery yielded an additional 6,451 sampling points, significantly expanding spatial coverage. This process resulted in a multi-class sample dataset comprising 8,251 points across five major land cover categories: farmland shelterbelts (2,730), forest (1,763), cropland (1,965), built-up areas (954), and water bodies (839).