Comparison of Different Crop Residue Indices for Estimating Crop Residue Cover Using Field Observation Data

Abstract

Crop residues, which remained in the farmland after crop harvesting, have positive impacts on the carbon sequestration and the prevention of soil erosion. Remote sensing methods can efficiently assess crop residue cover and related tillage intensity in a region. Up to now, many tillage indices have been proposed to estimate crop residue cover. Although the tillage indices based on the hyper-spectral image (e.g., Cellulose Absorption Index (CAI) and Lignin-Cellulose Absorption (LCA) index) have a better result in crop residue cover assessment, it's short of free remote sensing data having the corresponding band width to them, after the Earth Observing-1 (EO-1) and the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) were out of effect successively. Landsat OLI and Sentinel-2 can provide enough free data, but the band width is too broad for CAI and LCA. Thus, it's of great importance to seek the best indices based on the broad-spectral image to estimate the crop residue cover. The objective of this research is to compare the effect of common tillage indices based on the broad-spectral, and then chose the most suitable indices to estimate the winter wheat crop residue cover in Yucheng county of China. The indices involved Normalized Difference Index 5 (NDI5), Normalized Difference Index 7 (NDI7), Simple Tillage Index (STI), Normalized Difference Tillage Index (NDTI), Normalized Differential Senescent Vegetation Index (NDSVI), Crop Residue Cover Index 1 (CRC1)and Normalized Difference Residue Index (NDRI). Thirty fields were measured with the area of each was 10000m2. Nine sub-flelds(1m×1m) were placed uniformly in one field and the mean spectra and crop residue cover of them were used to stand for the field's. Reflectance spectral was measured in the field using the Analytical Spectral Devices (ASD), it covered the 350-2400nm wavelength region. Digital photos were taken in the field and analyzed indoors to obtain the area ratio of crop residue. Simple linear regression was applied on the tillage indices counted from the field reflectance data and the crop residue cover. The coefficient of determination (R2) between NDTI and crop residue cover was highest with 0.480 while the STI had an approximate R2 of 0.478. The R2 of NDI5, NDI7, NDSVI, CRC1, NDRI with the crop residue were 0.382,0.460,0.294,0.123 and 0.420, which reveals NDTI was the most suitable indices in this field. In the study area, the winter wheat crop residue cover mainly concentrated on 80%-95%, which proved conservation tillage practices have been generalized in this study and the behavior of burning straw have become less for the ban of the policy and the improvement of farmers' environment awareness.