Remote sensing is a valuable and effective tool for monitoring and estimating aboveground biomass(AGB) in large areas.The current international research on biomass estimation by remote sensing technique mainly focused on forests,grasslands and crops,with relatively few applications for desert ecosystems.In this paper,Landsat Thematic Mapper(TM)/Enhanced Thematic Mapper Plus(ETM+) images from 1988 to 2007 and the data of 283 AGB samples in August 2007 were used to estimate the AGB for Mu Us Sandy Land over the past 30 years.Moreover,temporal and spatial distribution characteristics of AGB and influencing factors of climate and underlying surface were also studied.Results show that:(1) Differences of correlations exist in the fitted equations between AGB and different vegetation indices in desert areas.The modified soil adjusted vegetation index(MSAVI) and soil adjusted vegetation index(SAVI) show relatively higher correlations with AGB,while the correlation between normalized difference vegetation index(NDVI) and AGB is relatively lower.Error testing shows that the AGB-MSAVI model established can be used to accurately estimate AGB of Mu Us Sandy Land in August.(2) AGB in Mu Us Sandy Land shows the fluctuant characteristics over the past 30 years,which decreased from the 1980s to the 1990s,and increased from the 1990s to 2007.AGB in 2007 had the highest value,with a total AGB of 3.352×106 t.Moreover,in the 1990s,AGB had the lowest value with a total AGB of 2.328×106 t.(3) AGB with relatively higher values was mainly located in the middle and southern parts of Mu Us Sandy Land in the 1980s.AGB was low in the whole area in the1990s,and relatively higher AGB values were mainly located in the southern parts of Uxin.In 2007,AGB in the whole area was relatively higher than those of the last twenty years,and higher AGB values were mainly located in the northern,western and middle parts of Mu Us Sandy Land.
