Effective identification of crop types plays a vital role in modern agricultural monitoring. Itenables more precise resource management, yield forecasting, and support for food securitystrategies. In this study, we integrate field boundary data provided by the Swedish AgricultureAgency with multi-source remote sensing data and machine learning to classify crop typesacross Sweden for multiple growing seasons.Using the Google Earth Engine (GEE) platform, we processed satellite imagery fromSentinel-1 (Synthetic Aperture Radar, SAR), Sentinel-2, and Landsat-8/9 to derive both optical(e.g., Normalized Difference Vegetation Index, NDVI) and radar-based features over a definedarea of interest. These features were aggregated into monthly and weekly composites to capturekey phenological stages of various crops. Field-level statistics were extracted and used to traina classifier, incorporating limited crop types present in the national dataset.This expanded methodology demonstrates the potential of combining official field parceldata with satellite-based Earth observation (EO) time series for efficient, scalable, and com-prehensive crop mapping across the entire country. This workflow offers a replicable and data-driven approach that supports sustainable agricultural practices and national-scale agriculturalmonitoring.