138 Bulgarian Academy of Sciences. Space Research and Technology Institute. Aerospace Research in Bulgaria. 25, 2013, Sofia DETECTION AND ASSESSMENT OF ABIOTIC STRESS OF CONIFEROUS LANDSCAPES CAUSED BY URANIUM MINING (USING HYPERSPECTRAL EO-1/HYPERION DATA) Lachezar Filchev Space Research and Technology Institute – Bulgarian Academy of Sciences e-mail: lachezarhf@space.bas.bg Abstract The paper presents the results from a study aiming at detecting and assessing abiotic stress in coniferous landscapes of Black pine monocultures (Pinus nigra L.) by using ground-based geochemical data, models of technogenic pollution, and satellite hyperspectral data from EO-1/Hyperion. Four vegetation indices (VIs), such as, MCARI, TCARI, MTVI 2, and PRI, as well as the shape and area attributes of the red-edge spectral features, have been studied in order to detect changes caused by the geochemical pollution. The analysis was performed on 4 test sites sized each 30m × 30 m, to correspond to EO-1/Hyperion pixel subspace, in homogenous coniferous landscapes in the Teyna River basin – a left tributary to Iskar River. In effect, the red-edge position index for unstressed coniferous landscapes (Black pine) was found to be at about λ=683 nm, whereas the stressed ones’ red-edge position is at λ=671 nm. It was found that VIs, such as TCARI/MCARI and Z c has highest correlation (r 2 =0.63; F: 5.20 at F: <0.1), followed by MTVI 2 and Z c (r 2 =0.42; F: 2.48 at F: <0.21), and PRI and Z c (r 2 =0.30; F: 1.34 at F: <0.33). Subsequently, the four test sites chosen were subdivided in two pairs, corresponding to the modeled Z c values after Inverse Multiquadratic Function (IMF) from the set of Radial Basis Function (RBF), into unstressed (test sites No. 2 and No. 5) and stressed coniferous landscapes (test sites No. 10 and No. 11). The hypothesis tested is that the clustering of EO-1/Hyperion VIs corresponds to Z c clustering using hierarchical clustering method (Ward). As a result, the MTVI 2 and TCARI/MCARI values group their first cluster relatively farther than Z c but the PRI does not reflect the Z c clusters. This lead to the conclusion that the satellite hyperspectral narrowband VIs are still not sensitive enough to study the geochemical background and vegetation stress of geochemically stressed coniferous landscapes by uranium mining without a priori ground based data.