S.A.R Nouredini; A.A Bonyad
Volume 9, Issue 1 , October 2017, , Pages 93-110
Abstract
Reflectance of different of land surface phenomena on remote sensing data was influenced by different conditions including atmospheric conditions. Variety methods of atmospheric correction have been developed for remove and reduction of its effects. In this study three atmospheric correction methods: ...
Read More
Reflectance of different of land surface phenomena on remote sensing data was influenced by different conditions including atmospheric conditions. Variety methods of atmospheric correction have been developed for remove and reduction of its effects. In this study three atmospheric correction methods: Dark Object Subtraction (DOS), Fast Line-of-sight Atmospheric Analysis of Spectral Hypercubus (FLLASH) and Second Vector Simulation of Satellite Signal in the Solar Spectrum (6SV) have been applied on OLI sensor of Landsat8 inthe forest regions of Guilan province. Numbers of 10 vegetation indices were extracted from each image. Forest area was extracted on various indices detected by global land cover layer. Forest areas segmented on Landsat8 image by object-based method. In the total 91 segments, randomly were selected. Forest canopy density of any segment plot estimated on Google images using 20×20 m network dotted. Person test was used for correlation between indices and training samples and two linear and nonlinear regression models were used for forest canopy density estimation. The results confirmed that 6SV method dominates than other methods in the forest regions of Guilan province. The lowest root means square error (RMSE) with 17.72 was shown in the green atmospherically resistant vegetation index (GARI) extracted from DOS. The results indicated that the lowest RMSE was in atmospherically resistant vegetation index (ARVI) using 6SV, FLAASH and OLI original image with 18.38, 15.87 and 21.78 respectively. The results of this study were shown that use of atmospheric correction methods in preparing vegetation indices is cause of increasing information accuracy from satellite images. Reduction of atmosphere effects in preprocessing before modeling is necessary and suggestible.
