Estimating Crop Residue Cover Considering Soil Heterogeneous Background Using Sentinel-2

Abstract:

Keeping crop residues on the soil surface is regarded as an effective anti-erosion way to promote the organic matter balance and nutrient cycles in the soil. Traditional measurement methods of crop residue cover are time-consuming. Remote sensing is an alternative approach for crop residue cover estimation. In this study, using field hyperspectral reflectance, field UVA images and Sentinel-2 MSI as data sources, spectral index (Normalized difference residue index, NDRI) was constructed to evaluate crop residue cover. But the spectral reflectance is influenced by many factors in space. Soil properties are heterogeneous no matter in large scale or small scale. The slope and intercept of soil line were added to construct soil adjusted residue indexes (SARI) of Sentinel-2 to reflect the soil spatial heterogeneity. The evaluation models were built by linear regression (LR) using NDRI and SARI. The results showed that LR model established by SARI (R² = 0.63) was better than the model used by NDRI (R² = 0.14). This is a novel approach for rapid monitoring and accurate diagnosis of crop residue cover.