Component content reversion based surface sediment classification in muddy intertidal flat by means of hyperspectral remote sensing data

Combining within situ sedimentary sampling and survey,a satellite-based hyperspectral Hyperion image was successfully used for the surface sediment classification in a muddy intert idalregion.Analysis indicated that under the low tidal level there were strong correlations between the surface spectrum reflect ance in the short wave infrared band and the contents of sand,silt and clay,respectively.For the 2102 nm band,the correlation coefficient reached -0.8954,0.9070 and 0.6547 correspondingly.So this band was chosen for the linear regression model const ruction and the sand,silt and clay contents were quantit atively reversed from their corresponding reflectance values.After that a linear equilibrium corrective method was applied for the amendment of these initial inversed content values.On the basis of these corrected sand,silt and clay contents,Shepard triangular graph method was used and the sediment classification for each pixel was automatically obtained with ENVIIDL program.The spatial distribution charact eristics of the surface sediment type for the whole intertidal flat were then acquired with the total accuracy of 87.9%.Results indicated that there were three kinds of sediment distributed in the study area,namely,silty-sand,sandy-silt and silt with the percent age of 18.9%,64.6% and 16.5%,respectively.The distribution trend of the three sediments was adaptive perfectly with the vegetation distribution patterns,hydrodynamic conditions and the sources of the sediments of this area.