Crop Classification and Representative Crop Rotation Identifying Using Statistical Features of Time-Series Sentinel-1 GRD Data

Compared with a monoculture planting mode, the practice of crop rotations improves fertilizer efficiency and increases yield. Large-scale rotation monitoring relies on results classification using remote sensing technology. However, limited accuracy cannot satisfy accurate identification patterns. In this paper, mapping scheme combining random forest (RF) algorithm new statistical features extracted from time-series ground range direction (GRD) Sentinel-1 images. First, synthetic aperture radar (SAR) stacks are established, including VH, VV, VH/VV channels. Then, named objected generalized gamma distribution (OGΓD) introduced to compare other object-based for each polarization. The showed that OGΓD σVH achieved 96.66% overall (OA) 95.34% Kappa, improving around 4% 6% compared backscatter in VH polarization, respectively. Finally, annual crop-type maps five consecutive years (2017–2021) generated RF. By analyzing five-year sequences, soybean-corn (corn-soybean) is most representative study region, soybean-corn-soybean-corn-soybean (together corn-soybean-corn-soybean-corn) has highest count 100 occurrences (25.20% total area). This offers insights into monitoring, giving basic data government food planning decision-making.