Accurate extraction of crop planting types in Southwest mountainous areas based on hierarchical extraction

Abstract

The spatial distribution of crop planting structure plays a crucial role in reflecting the layout and arrangement of farmland, making it highly significant for the advancement of precision agriculture and agricultural resource inventory. The extraction of crop planting structure through remote sensing satellite earth observation technology serves as a fundamental pillar for the development of precision agriculture informatization. This study focuses on the application of precision agricultural remote sensing in monitoring complex mountainous regions, employing the cognitive theory of remote sensing atlas to support the process. By integrating high-spatial resolution satellite images, medium-resolution time series satellite images, and ground observation data, a hierarchical zoning approach was developed, using farmland parcels as the basic spatial unit for crop type identification. A case study was conducted in Wushan County, Chongqing, serving as a demonstration zone. Vegetation indexes and phenological indexes were extracted by combining the high spatial resolution of GF satellites with the temporal resolution of middle-resolution satellite images. Subsequently, a multi-feature crop identification model at the parcel scale was constructed, facilitating the precise extraction of crop planting types in the southwest mountainous areas.