Work

Mapping crop residue cover in global black soil regions based on a new multi-band spectral index and an integrated thresholding approach

Year: 2026

Type: preprint

Abstract: Crop residue cover is a defining feature of conservation tillage, supporting soil health, carbon sequestration, and agricultural productivity, particularly across the world’s black soil regions. However, large-scale monitoring remains challenging due to the spectral heterogeneity of post-harvest fields and the scarcity of ground-truth samples. Here, we propose a new multi-band crop residue cover spectral index (MCRCSI), and an accuracy-weighted composite thresholding (AWCT) for robust crop residue mapping on single-date multispectral imagery. MCRCSI integrates the cellulose- and lignin-related SWIR contrast of crop residues with their joint Red and near-infrared reflectance response, enhancing separability from soil and photosynthetic vegetation. AWCT integrates thresholds derived from Otsu’s method, k-means clustering, and Gaussian mixture models using accuracy-based weights, enabling adaptive threshold determination across different cropping systems. Across eight representative sites in the United States, Ukraine, Northeast China, and Argentina, we evaluated the MCRCSI-AWCT framework using 15,591 independent validation samples. It achieved a mean overall accuracy (OA) of 93.6% and a mean F1 score of 94.0%, with site-level OA ranging from 81.0% to 98.0% (SD = 5.6%). MCRCSI was sensitive to changes in crop residue fraction, with distinct responses at low (<20%) and high (>80%) cover levels, indicating potential for fraction retrieval. Its strong separation of unharvested crops from post-harvest crop residue also suggests potential for harvest event detection. Together, these findings provide a scalable framework for mapping crop residue cover and monitoring field management practices over time.

Source: Research Square

DOI: 10.21203/rs.3.rs-11627777/v1