The amount of senesced biomass in vegetation plays an important role in estimation of carbon storage and plant stress. In this paper, the spectral predictors for estimating senesced biomass were evaluated based on field spectral and corresponding biophysical parameter measurements during the growing seasons of 2009 and 2010 in the desert steppe of Inner Mongolia. Results showed the cellulose absorption index (CAI) was the best one among senesced vegetation coverage indices and band depth indices. The model involving CAI yielded the highest coefficient of determination (R² = 0.67) and the lowest root mean square error of leave-one-out cross validation (RMSECV = 17.9 g m⁻²) compared with normalized difference index (NDI) (R² = 0.21, RMSECV = 27.6 g m⁻²), soil-adjusted corn residue index (SACRI) (R² = 0.29, RMSECV = 26.2 g m⁻²), modified soil-adjusted crop residue index (MSACRI) (R² = 0.1, RMSECV = 29.5 g m⁻²), dead fuel index (DFI) (R² = 0.28, RMSECV = 26.3 g m⁻²), lignocellulose absorption depth (LCD) (R² = 0.56, RMSECV = 20.5 g m⁻²) and lignocellulose absorption area (LCA) (R² = 0.54, RMSECV = 21.1 g m⁻²). The results of this study suggest that CAI has good potential to estimate senesced biomass in desert steppe areas.