Using Time Series Sentinel-1 Images for Object-Oriented Crop Classification in Google Earth Engine

The purpose of this study was to evaluate the feasibility and applicability object-oriented crop classification using Sentinel-1 images in Google Earth Engine (GEE). In study, two areas (Keshan farm Tongnan town) with different average plot sizes Heilongjiang Province, China, were selected. research time consecutive years (2018 2019), which used verify robustness method. growth period (May September) each area composited three intervals (10 d, 15 d 30 d). Then, composite segmented by simple noniterative clustering (SNIC) according finally, training samples processed input into a random forest classifier for classification. results showed following: (1) overall accuracy method combined image represented great improvement compared pixel-based large plots (increase 10%), applicable scope depends on size area; (2) shorter interval was, higher was; (3) features high importance mainly distributed July, August September, due differences these months; (4) optimal segmentation closely related resolution size. Previous studies usually emphasize advantages Our not only emphasizes but also analyzes constraints classification, is very important follow-up synthetic aperture radar (SAR).