This thesis investigates the effectiveness of using varying data sizes to transfer crop type classification models from one year to the other, with a focus on balancing data sufficiency and model accuracy. The significance of crop detection through satellite imaging lies in its potential to enhance agricultural productivity and resource management. Machine learning techniques, particularly long short-term memory (LSTM) models, have become instrumental in interpreting satellite data due to their predictive accuracy and adaptability. However, applying models trained in one year to subsequent years poses challenges due to variations in environmental conditions and agricultural practices. To address these challenges, in this thesis, we explore the cost-benefit of fine-tuning existing models versus developing new ones based on the quantity of new data. Using smaller datasets for fine-tuning is more computationally efficient and reduces the cost of data collection. Experiments conducted using satellite data from farms in southern Alberta reveal that smaller datasets, with fewer than 25 fields per class, can effectively fine-tune models for accurate interannual classification, while larger datasets are more conducive to training new models. This highlights the key challenge of optimizing data usage for crop classification, balancing data sufficiency and computational efficiency. Additionally, this thesis contributes to the field by selecting the best combination of bands and information from Sentinel-1 and Sentinel-2 satellites. Another significant contribution is the incorporation of crop rotation as a feature for crop classification, which enhances the model's predictive capabilities. The findings of this research offer valuable insights for optimizing data use in crop classification, benefiting both academic research and practical agricultural applications.