Assessment of the effectiveness of atmospheric correction methods using standard calibration targets, ground measurements and aster images

Abstract

This paper presents the results obtained by collecting systematically the following: - spectro-radiometric measurements of dark and white targets that can be served pseudo-invariant targets using the GER 1500 and SVC HR-1024 field spectroradiometers; sun-photometer measurements using MICROTOPS II sun-photometer for measuring the AOT and water vapour. These field data assists the assessment of the effectiveness of the existing atmospheric correction algorithms either simple or sophisticated such as atmospheric modeling. The in-situ measurements have been already collected on June-August 2008 and from March to July 2009. Six targets are selected to be served as calibration targets in order to assess the effectiveness of the existing atmospheric corrections. For this paper only the following atmospheric correction methods have been tested: darkest pixel atmospheric correction, covariance matrix method, multi-temporal normalization method, 6S model, 'new method using pseudo-invariant targets and empirical line empirical line regression method'. The accuracy assessment has been done based on the reflectance and aerosol optical thickness. Results of the accuracy assessment based on the reflectance retrieval values are presented. The duration of the project is three years started from 2008 up to 2011 and further measurements are still undertaken. The most effective atmospheric correction applied for the three ASTER bands was the 'New method using pseudo-invariant targets and empirical line' (Hadjimitsis et al., 2009) and then the Darkest Pixel correction method (which has been provided a reasonable correction only for bands 1, 2 bands). The future task consists of creating 'a complete database of the spectral signatures of several target's that can be served non-variant and calibration targets