A self-supervised contrastive learning framework for giant panda habitat suitability modelling with the ecologically constrained background sampling strategy

This dataset contains environmental predictor layers and processed modeling inputs used for giant panda habitat suitability modeling in Giant Panda National Park (GPNP), China. The data support the manuscript entitled “A self-supervised contrastive learning framework for giant panda habitat suitability modeling with the ecologically constrained background sampling strategy.” The dataset includes multi-source environmental, anthropogenic, and landscape predictors used to characterize habitat conditions across GPNP. All raster layers were preprocessed to a common spatial resolution of 250 m and clipped to the GPNP study area for model training, evaluation, and habitat suitability mapping. The associated source code is available at: https://github.com/cathy0227/simclr-gpnp-habitat-suitability.