Accurate estimation of non-photosynthetic biomass is critical for modeling carbon dynamics within grassland ecosystems.We evaluated the cellulose absorption index(CAI),widely used for monitoring non-photosynthetic vegetation coverage,for non-photosynthetic biomass estimation.Our analysis was based on in situ hyperspectral measurements,during the growing seasons of 2009 and 2010,in the desert steppe of Inner Mongolia.ASD(Analytical Spectral Device)-derived and Hyperion-derived CAI were found to be effective for non-photosynthetic biomass estimation,yielding relative error(RE) values of 26.4% and 26.6%,respectively.The combination of MODIS(Moderate Resolution Imaging Spectroradiometer)-derived(MODIS2 MODIS5)/(MODIS2 +MODIS5) and(MODIS6 MODIS7)/(MODIS6 +MODIS7) showed a high multiple correlation(multiple correlation coefficient,r=0.884) with ASD-derived CAI.A predictive model involving the two MODIS indices gave greater accuracy(RE=28.9%) than the TM(Landsat Thematic Mapper)-derived indices.The latter were the normalized difference index(NDI),the soil adjusted corn residue index(SACRI),and the modified soil adjusted crop residue index(MSACRI).These indices yielded RE values of more than 42%.Our conclusions have great significance for the estimation of regional non-photosynthetic biomass in grasslands,based on remotely sensed data.
