This thesis presents a data-driven framework for improving precision irrigation management in apple orchards in the province of Cuneo, Italy. In situ soil matric potential sensors, a weather station, and remote sensing using Copernicus Sentinel-2 satellite data were used. The research focuses on transforming disparate observations into a coherent analytical dataset for a predictive machine learning model. Soil water potential was collected every 30 min and averaged into daily median values to coordinate with daily meteorological data, and also for irregular overpasses from Sentinel-2. Satellite Scene Classification Layer (SCL) filtering is adapted for cloud masking and calculates vegetation/water indices (NDWI, NDVI, NDRE, EVI, MSI, CCCI).