Long-term crop mapping in the three provinces of Northeast China based on multi-source sample fusion

翻译:This dataset documents the annual spatial distribution of three major grain crops (rice, maize, and soybean) in Northeast China (Heilongjiang, Jilin, and Liaoning provinces) from 1985 to 2024. The dataset consists of high-resolution (30 m) crop classification maps generated by integrating Landsat series satellite imagery and machine learning techniques. These maps were developed using multi-source geospatial data, including historical crop classification maps, land use/cover products, and cultivated land resource distributions. The methodology involved constructing feature sample sets, performing multi-source data fusion, and conducting spatial overlay analysis. To ensure temporal consistency, a model transfer-based sample expansion technique was employed to augment historical data availability. A long-term crop classification model was subsequently trained to produce the final high-quality crop maps. All data are provided in GeoTIFF format to support a wide range of research and application needs. Data Format:GeoTIFFCoordinate Reference System:EPSG:4326 (WGS 1984 geographic coordinate system)Pixel Value Definitions:0:Other land types (including non-farmland and non-target crops)1:Rice2:Maize3:Soybean