Accurate and spatially explicit information on the condition of agricultural landscapes is essential for monitoring developments and advancing management practices in agricultural systems. A real-world example is the need for reliable, high-resolution crop maps as needed by the Swiss national greenhouse gas inventory. The inventory currently relies on crop distribution data aggregated at the municipality level, limiting the ability to capture spatial differences. Future inventories aim to transition toward fully spatially explicit representations, requiring robust, high-resolution crop type maps.  In this work, we generate national-scale distribution maps for 36 crop types and 6 grassland classes across Switzerland using satellite image time series. We employ an attention-based deep learning model trained on the “Swiss Crops” dataset, which is annotated from farmer declarations and contains 9.3M polygons (8.7M ha) covering the years 2019-2024. To ensure robustness under real-world conditions, we train models on temperature-informed samples in a cross-year setting and evaluate their ability to generalize to unseen years. This explicitly addresses inter-annual variability in crop development driven by climatic fluctuations and management practices. Preliminary results show F1-scores above 0.85 for most majority crops and above 0.7 for most minority crops. Meadow intensity classes (intensive vs. extensive) can be reliably distinguished (F1 ≈ 0.80 and 0.65), while performance in distinguishing pasture intensity remains limited.  Our results demonstrate that the proposed approach generalizes well throughout Switzerland and remains stable under substantial year-to-year variation, making it suitable for operational applications. All maps and labels will be made freely available, forming one of the largest national-scale, multi-year satellite benchmark datasets for crop classification and segmentation. The produced crop and grassland maps provide a key building block for spatially explicit greenhouse gas accounting and other agro-environmental assessments.