This dataset is published as supplementary materials of the article: Cubaud, M., Le Bris, A., Jolivet, L., & Olteanu-Raimond, A. M. (2024). Assessing the transferability of a multi-source land use classification workflow across two heterogeneous urban and rural areas. International Journal of Digital Earth, 17(1). https://doi.org/10.1080/17538947.2024.2376274 It is composed of the Land Use datasets for Gers and Rhône, constructed and annotated by the IGN (French National mapping agency) in the frame of the OCcupation des Sols à Grande Echelle (Large Scale Land Use Land Cover, OCS GE). To these datasets were added attributes constructed using multiple sources of data. The attributes are listed in the file "attributes_table.xlsx", and their construction process is detailed in the article. The attributes are used in the article to characterize the polygons and to predict their Land Use. In the article, attributes constructed from Land Files are also used, but they cannot be included due to privacy and license reasons. The construction of the OCS GE is described in https://geoservices.ign.fr/sites/default/files/2022-11/DC_OCS_GE_1-1.pdf (in French) The date of Gers dataset is 2019 and the date of Rhône dataset is 2020. The file NomenclatureOCSGE.csv presents the Land Use classes. Files OCS_GE_Rhone.gpkg : Rhône Land Use dataset; OCS_GE_Gers.gpkg : Gers Land Use dataset; attributes_table.xlsx : list of the attributes; NomenclatureOCSGE.csv : description of the Land Use classes, with their legend color. Usage The dataset files can be opened in GIS software like QGIS. The following code allows to read them in Python. import geopandas as gpd ocs_ge_gers = gpd.read_file("OCS_GE_Gers.gpkg") ocs_ge_rhone = gpd.read_file("OCS_GE_Rhone.gpkg") The repository https://github.com/mcubaud/multi-source-Land-Use-classification shows the code used in the article to construct the attributes and predict the Land Use from these attributes.