Mapping resort expansion and ecosystem impacts using high resolution satellite imagery

The dataset supports resort mapping, land use and land cover (LULC) change analysis, and ecosystem impact assessment across the Yucatán Peninsula, Mexico. resort_data_layer.zip includes the following spatial datasets: Building footprints extracted from high-resolution satellite imagery, used for calculating building morphology and landscape metrics and identifying resort-related buildings. Points of Interest (POIs) collected using tourism-related keywords (e.g., hotel, resort, villa, and all-inclusive resort), used as reference data for resort identification and classification. Resort prediction results, generated using an XGBoost classification model. The output layer contains the attribute resort_pred, where 1 indicates predicted resort parcels and 0 indicates non-resort parcels. Parcel boundaries extracted from high-resolution satellite imagery, representing the basic spatial units used for resort prediction. Land cover classification maps (2025) derived from satellite imagery, including major classes such as forest, pavement, and pools. Mapped resort boundaries (1995–2025) documenting the spatial distribution of resort areas at seven time points (1995, 2000, 2005, 2010, 2015, 2020, and 2025). Mapped forest boundaries (1995–2025) documenting forest distribution throughout the study period. Land use and land cover (LULC) datasets for the three states of the Yucatán Peninsula (Yucatán, Quintana Roo, and Campeche), including both the complete study area and the 5 km coastal buffer. LULC datasets for Quintana Roo, including the entire state and the corresponding 5 km coastal buffer. Mangrove loss excluding resort expansion, representing mangrove loss areas after removing regions overlapping with mapped resort expansion, allowing distinction between losses associated with resort development and those caused by other factors. These datasets were used to support resort mapping, landscape pattern analysis, land use transition analysis, and ecosystem impact assessment presented in the associated manuscript. resort_prediction_method.zip integrates high-resolution satellite imagery, landscape metrics, and machine learning techniques to automatically identify and delineate resort areas. It demonstrates the complete processing pipeline from parcel extraction to final resort prediction. The archive includes: ArcPy scripts for automated parcel extraction and preprocessing. ArcGIS Pro project files containing the complete geoprocessing workflow. Python scripts for landscape metric calculation and XGBoost-based resort classification. Sample datasets representing a subset of the study area for workflow demonstration and testing. The workflow consists of the following major steps: Automated parcel extraction from high-resolution satellite imagery. Building footprint processing and regularization. Spatial analysis using ArcGIS tools, including Spatial Join, Near, Buffer, Erase, Merge, Dissolve, and Multipart to Singlepart. Calculation of parcel-level landscape metrics and building characteristics. Training an XGBoost model using manually labeled resort and non-resort samples. Automated prediction of resort parcels and generation of resort boundary maps. The included datasets represent a small experimental subset of the study area and are provided solely to demonstrate and validate the complete workflow. They enable users to reproduce the methodology without requiring the full project dataset. The complete GIS datasets used in the study are provided separately in the accompanying data archive (resort_data_layer.zip).