Several indices of vegetation and soil salinity have been developed to quantitatively evaluate soil salinization.This study was conducted to assess the soil salinity levels in the Fubei region (FG),Manas River Basin (MRB),and Werigan-Kuqa River Delta Oasis (WKRDO),which are distributed in the northern and southern Tianshan Mountains in Xinjiang,China.Ground measurements and remote sensing data were used to evaluate the sensitivity of vegetation and soil salinity indices to soil salinity variation in farmland and salt-affected land.A random sampling approach was used to collect soil samples from FG (n=37,only at 0-10-cm depth),MRB (n=58),and WKRDO (n=38).A total of 14 broadband indices encompassing vegetation and soil salinity indices were extracted from Landsat images.The correlation coefficient based on linear and non-linear models (10 models) between these indices,Landsat bands,and soil salinity was examined.The results showed that the extended enhanced vegetation index (EEVI) was the most effective for explaining the soil salinity variation at depths of 0-10 cm in two modes (all samples and partial samples with soil salinity (soil salt content)>0.3%) in FG.With the mode of all samples and partial samples (soil electric conductivity <2 dS/m) in MRB,band 2 yielded the best results for assessing the soil salinity of cultivated lands at the early stage of crop growth in April.The maximum depth of the significance test by using indices for detecting variation of soil salinity in this area was 30 cm.For all samples in WKRDO,the salinity index (SI-T) interpreted more variation of soil salinity than that by other indices at depths of 0-10 and 10-20 cm,and the three-band maximal gradient difference index (TGDVI) exhibited the highest significant correlation with salinity at 20-40 cm.In the mode of partial samples (soil salinity >2 dS/m),the most sensitive index for variation of soil salinity at 0-10,10-20,and 20-40 cm were band 5,TGDVI,and EEVI.In addition,the correlation of other indices (excluding those mentioned above) and soil salinity was highly dependent on land cover heterogeneity and sample period,and showed no significant relationships (p > 0.05 or p > 0.01).These results are preliminary conclusions,but in general,the soil salinity in Xinjiang dominated by different salt types was successfully assessed by broadband vegetation and soil salinity indices extracted from the Landsat images.However,relationships between remote sensing indices and soil salinity within fields are highly complex and require further investigation with additional samples and by using various soil salinity classifications.