Remote sensing estimation of winter wheat residue cover with dry and wet soil background

Abstract

Estimation of crop residue cover is important for energy balance in agroecosystem and sustainable development of agriculture. We evaluated the dimidiate pixel model, widely used for estimating photosynthetic vegetation cover, for non-photosynthetic vegetation (such as winter wheat residue) cover estimation. In this study, based on spectral and cover data of winter wheat residue in dry and wet soil backgrounds, the spectral curves of winter wheat residue and soil were identified, the applicability of non-photosynthetic vegetation indices in dimidiate pixel model was analyzed, and the potential of dimidiate pixel model to estimate winter wheat residue cover was explored. In dry soil background, a lignocellulose absorption trough near 2100 nm in the spectral curve of residue-soil mixed scene was observed, and the absorption trough became deeper with increasing residue cover. The normalized difference tillage index (NDTI) had the best correlation with the measured cover of winter wheat residue, and the dimidiate pixel model constructed on the basis of this index was able to accurately estimate the winter wheat residue cover (R2=0.64, RMSE=0.16, RE=26.32 %). In wet soil background, the ability of non-photosynthetic vegetation index to distinguish between winter wheat residue and soil was reduced by soil moisture. The results of this study provide effective insights into the estimation of winter wheat residue cover under different soil moisture conditions, and provide a useful reference for the study of remote sensing estimation of crop residue cover in a large region. The dimidiate pixel model using NDTI can also be used to estimate non-photosynthetic vegetation cover of natural vegetation.