Evaluation of Atmospheric Correction Methods in Identifying Urban Tree Species With WorldView-2 Imagery

摘要: The radiance recorded at a sensor is not fully representative of Earth surface section features but altered by atmosphere. In this study, we evaluated three atmospheric correction (AC) methods (a typical empirical modeling method, radiative transfer approach, and combination the both methods) in identifying urban tree species/groups with high-resolution WorldView-2 (WV2) imagery City Tampa, FL, USA. We tested whether AC were necessary species discrimination. In situ spectral measurements taken from tops canopy crowns delineated WV2 imagery. Two-sample $\bm{t}$ -tests, repeated measures ANOVA (RANOVA) tests, linear discriminant analysis (LDA), classification regression trees (CART) classifiers used to test difference between in situ spectra atmospherically corrected image discriminate species/groups. experimental results demonstrate that 1) the line-based relatively more effective than transfer-based model correct data, due lacking accurate reliable parameters run 2) the processing was unnecessary seven particular case, most likely because data acquired on single date covered small area ( ${303}\;\mathbf{km}^{2}$ ). study also indicate compared nonparametric classifier CART, parametric LDA produced higher overall accuracy (55% vs. 48%) for