Mapping annual 10-m soybean cropland with spatiotemporal sample migration.

Area of Science:

• Agricultural Science
• Remote Sensing
• Geospatial Analysis

Background:

• China is the largest soybean importer and fourth-largest producer globally, necessitating precise agricultural data.
• Accurate mapping of soybean cultivation areas is crucial for global food security and agricultural management.
• Challenges exist in collecting and verifying ground survey data for diverse crop types.

Purpose of the Study:

• To develop a reliable method for mapping soybean planting areas in China.
• To create high-resolution (10-meter) soybean maps for key producing provinces.
• To support soybean growth monitoring, yield estimation, and sustainable agricultural practices.

Main Methods:

• A spatiotemporal migration method utilizing vegetation indices' temporal characteristics.
• Feature extraction from optical time-series images, incorporating Synthetic Aperture Radar (SAR) data for data gaps.
• Classification using the random forest algorithm on a feature space including phenological curve integrals and a concavity-convexity index.

Main Results:

• Development of the ChinaSoybean10 maps, detailing soybean cultivation from 2019 to 2022 in ten major provinces.
• Achieved an overall accuracy of approximately 93% for the generated soybean maps.
• Demonstrated significant alignment with statistical yearbook data, confirming map reliability.

Conclusions:

• The proposed method effectively maps soybean planting areas with high accuracy.
• The ChinaSoybean10 maps provide valuable data for agricultural monitoring, policy development, and food security initiatives.
• This research contributes to sustainable agriculture through improved resource management and yield prediction.