Crop planting structure extraction in irrigated areas includes a range of dynamic parameters which require proper spatial and temporal resolution remotely sensed data. The paper seeks to extract crop planting structure by employing multi-temporal images from multi-sensors. Landsat enhanced thematic mapper plus (ETM+) images and moderate resolution imaging spectroradiometer (MODIS) normalized difference vegetation index (NDVI) monthly data were res-merged to produce a mega data tube, which was then classified using ISO cluster algorithm. Spectral signature of each class was extracted and identified using spectral matching technique taking crop coefficient curve as reference. In the way Zhanghe Irrigation system in southern China was classified into four classes: rice-rapeseed rotation, rice-wheat rotation, single summer crops, and double economic crops. Accuracy assessment suggests good agreement with statistical data and 91% classification accuracy when using IKONOS high resolution images as Ground Truth data. The application demonstrates the method a cost-efficient and robust approach to extract crop planting structure at irrigation system scale.