Evaluation of spectral angle index from Landsat TM image for crop residue cover estimation

Crop residue, as one of important factors of eco-agricultural system, can influence the flow of nutrients, carbon, water, and energy in terrestrial ecosystems. Remote sensing approaches have been investigated to obtain the crop residue cover (CRC) over large areas rapidly and objectively. While Cellulose Absorption Index (CAI) is a reliable indicator for crop residue cover estimation, it can only be applied to hyperspectral sensor, which usually has a narrow swath width. A new approach utilizing the shape of the spectrum of consecutive bands of MODIS data was introduced to distinguishing soil, crop residue and live crops. Since Landsat 5 TM's band configuration is similar to MODIS and provides fine spatial resolution (30 m) for agriculture in China, it offers an opportunity to assess crop residue cover over large areas. The objective of this research is to evaluate the Landsat-based angle indices for estimating CRC. Four spectral angle indices were derived by surface reflectance of Landsat TM. The results showed that shortwave angle slope index (SASI) is an ideal index for crop residue cover estimation. Simple linear regression reveals a coefficient of determination (R2) of 0.7895 between measured CRC and SASI for the calibration dataset and RMSE of 0.0886 between measured and predicted CRC for validation dataset. Crop residue cover for Hongxing farm was estimated using Landsat-based SASI and sorted into three categories, corresponding to intensive (30% cover) tillage. Results in this study demonstrated that Landsat-based SASI was capable of crop residue cover estimation.