EVALUATION of ATMOSPHERIC CORRECTION METHODS for SENTINEL-2 IMAGERY in the SPECTRAL IDENTIFICATION of POPLAR (POPULUS DELTOIDES BARTR.) SPECIES

Abstract

Spectral libraries can be used for the analysis of the spectral properties of tree species, as well as for the detection and mapping of tree species through remotely sensed data. This study aims to statistically analyze the performance of atmospheric correction methods in calibrating ground spectral measurements with the Sentinel-2 imagery. In this context, field spectral measurements of Populus deltoides Bartr. species cultivated in Akyazi district of Sakarya province, Turkey obtained with the ASD (Analytical Spectral Devices) FieldSpec-3 high resolution (Hi- Res) spectroradiometer (350nm-2500nm) were used as the main data source. In order to evaluate the effects of atmospheric correction methods in identifying poplar species, ENVI QUAC and Sen2Cor processor were applied to Level-1C product of Sentinel-2 data. Moreover, atmospherically corrected Sentinel-2A imagery was also utilized for the comparison of the reflectance values of poplar trees. Furthermore, five widely used spectral vegetation indices (i.e., NDVI-1, NDVI-2, GRNDVI, CVI, and CLgreen) were estimated for comparison of field measured reflectance spectral data and atmospherically corrected Sentinel-2 data. Linear regression (LR) model and normalized Euclidean distance were utilized to estimate the similarity level between field spectra obtained for poplar trees and the spectral signatures from atmospherically corrected Sentinel-2A products, ENVI QUAC and Sen2Cor processor. The LR results showed that highest R2 values were estimated as 0.996 and 0.997 for Sen2Cor processor and Sentinel-2A image, respectively. Moreover, calculated normalized Euclidean distance results verified the findings that atmospherically corrected imagery better represents the spectral signature of the poplar trees. The results of this study revealed that atmospherically corrected Sentinel-2 image exhibits great potential for integration in spectral resampling of field spectra measurements of poplar species. © 2022 Elsevier B.V., All rights reserved.