Discriminação entre solos contaminados com hidrocarbonetos e vegetação em estado avançado de senescência através de espectroscopia de refletância
Reflectance spectroscopy and hyperspectral remote sensing techniques have been successfully applied to direct and indirect detection of hydrocarbons soils. However, spectral similarities between soils contaminated by hydrocarbons and senescent vegetation may cause significant misclassification results. An experiment was conducted between April – May 2010 in real scale to simulate small volume hydrocarbons (PHC) spills along pipelines. Three trenches (with 300 m3 each) were installed in an experimental farm in order to represent non-contaminated soils (CTR); soils contaminated with gasoline (GSL), and soils contaminated with diesel (DSL). The simulation consisted of daily spills of 0.67 L/m3 of each fuel in an Oxisol. This type of soil is widespread in tropical countries and particularly in Brazil. Vegetation canopy and soil spectral data were acquired in-situ over this controlled contamination field experiment. Main absorption features were identified in both, vegetation canopy and soils, and these absorption bands were used to recognize optimal spectral intervals to discriminate between these targets. Six spectral intervals were selected: (i) 399 – 750 nm; (ii) 756 – 1085 nm; (iii) 1116 – 1284 nm; (iv) 1665 – 1789 nm;(v) 2138- 2233 nm; and (vi) 2236 – 2436 nm. Continuum removed was applied in each interval in order to enhance the spectral features. Discrimination between bare soils and senescent vegetation was possible used features in intervals (ii), (iii) and (vi). These ranges showed nanometric spectral differences between these targets, making complex to separate them, even considering mineralogical characteristics of soil.