Remote sensing monitoring of maize planting area at town level

In order to extract maize planting acreage rapidly and accurately, taking Changchun city in Jilin province as study area, a multi-layer decision tree classification model was constructed based on multi-temporal HJ-1A/1B CCD images and digital elevation model (DEM), which introduced with multi-information including planting structure, phenological characteristics, terrain feature of the study area, spectral characteristics and vegetation index. The precision of classification results was evaluated by spatial agricultural census data at town level. The results indicated that the method could improve maize identification accuracy, which reached up to 92.57%. The method can meet the demand for large-scale and multi-temporal agricultural monitoring system and resolve the spatiotemporal contradiction effectively in the large-scale crop acreage monitoring system.