The Soil and Water Assessment Tool (SWAT) is an ecohydrological model widely applied to assess water quality and watershed management. This tool also has the advantage of building watershed models even with limited monitoring data availability. The essential data required by this tool includes digital elevation models (DEM), land use maps, climate data, and soil data. Nonetheless, the availability of spatial data is still often a challenge in developing hydrological models, especially in developing countries such as Indonesia. This research will compare the accuracy of freely available data in Indonesia in facilitating the development of hydrological models from SWAT in the Rawa Pening catchment area. This research is crucial since Rawa Pening Lake is a priority lake for revitalization, so the research results will help provide suggestions regarding presenting data in SWAT modeling. This research compares SWAT models built from different land use and DEM (Digital Elevation Models) data. The land use data being compared is the result of processing from the Google Earth Engine (GEE) platform using machine learning with land use data from government agencies, namely the Ministry of Environment and Forestry, while the DEM data being compared is SRTM and DEMNAS data. The validation results using R, R2, RMSE, and NSE show that, in general, the model built from land use from GEE is the best compared to the other models. In modeling SWAT in Indonesia, we recommend using good-quality land-use data. Utilizing supervised classification through Random Forest (RF) algorithms within GEE can facilitate the acquisition of this data. To reduce computation time, the DEM can be SRTM with a small sacrifice of accuracy.