Oil palms have large economic value and are grown extensively across Southeast Asia. However, growth of the oil palm industry comes at the expense of the environment as forests are cleared to grow oil palms. Oil palm plantations need to be monitored to balance economic growth and environmental sustainability. Synthetic Aperture Radar (SAR) imagery allows for the cost-effective and frequent mapping of the extent of oil palm plantations over large areas. This paper aims to develop an oil palm plantation mapping model using X, C and L band SAR and compare their relative performance. The models are developed with the Feature Pyramid Network based on annotations acquired over Batu Pahat, Malaysia. X-band has the best Dice Score of 0.9 for oil palm plantations and the highest overall accuracy of 81.78%. Repeat pass satellite images captured 6 months later were then inferred with the 3 models to identify changes to the land cover. X-band also has the best accuracy in change detection as it has the best land cover classification performance overall. The plantation maps add semantic meaning to the land cover changes. This paper successfully developed a model that can generate frequently updated and detailed oil palm plantation maps, which can be used to detect changes in the oil palm plantation extent promptly.