Monitoring agricultural lands and estimating crop area are essential components of water resources management and agricultural planning in semi‑arid regions. This study aimed to evaluate the capability of multi‑temporal optical and radar satellite data for classifying summer and autumn crops in the Qazvin irrigation network. For this purpose, time‑series satellite images corresponding to key crop growth stages were processed, and the performance of five machine learning classification algorithms—CatBoost, LightGBM, Random Forest, XGBoost, and Support Vector Machine (SVM)—was compared for mapping major crops including corn, tomato, alfalfa, rapeseed, wheat, and barley. The classification outputs were assessed using overall accuracy and the Kappa coefficient. The results indicated that boosting‑based methods, particularly CatBoost with a Kappa coefficient of 0.83 in summer and LightGBM with 0.85 in autumn, provided superior performance in crop discrimination and reducing pixel‑level noise. In contrast, SVM, despite its acceptable statistical accuracy, showed lower spatial coherence due to its sensitivity to noise and over‑estimation along field boundaries. The resulting crop maps illustrated well‑structured spatial patterns and enabled reliable estimation of crop‑specific cultivated areas. Overall, the findings demonstrate that integrating multi‑temporal radar and optical data with advanced machine learning algorithms offers an effective and scalable approach for regional crop mapping and agricultural monitoring.