Assinaturas espectrais de gossans associados a mineralizações de Pb-Zn-Ba na Bacia de Irece (BA) : um estudo baseado em dados de sensores TM e ASTER

AUTOR(ES)
DATA DE PUBLICAÇÃO

2003

RESUMO

The aim ofthis work was to develop strategies for detection ofgossans associated to Pb-Zn-Ba deposits in the Irece basin (BA), as well as their distinction ftom generic laterites, using remote sensing data and techniques. Samples of gossans collected in the study area were analysed by reflectance spectroscopy methods, using a high resolution FieldSpec FR spectrometer covering nanometric wavebands within the 350nm-250Omn interval. Goethite, limonite and hematite were identified by interactive and software-driven automatic interpretation as the gossan s main mineraIs, based on their diagnostic spectral features. Using data available in the literature combined with lab-measured reflectance data convolved to TM and ASTER sensors, it was possible to realize a spectral signature for both genuine and false gossans in terms of multispectral resolution. Using these data, gossans are clearly distinct ftom laterites by the lack of absorption bands around 2.2~m (linked to the presence of kaolinite) and by the lack of important emissivity features around lO~m (akin to a mixture of kaolinite and silica). Using this notion, the main spectral intervals employed to detect and differentiate these two materiaIs were defined as followed: O,350-1,OO~m, covering visible (VIS) and near-inftared (NIR) wavebands, to detect iron oxides and hydroxides; I,OO-2,50~m, covering shortwave inftared (SWIR) wavebands to map kaolinites; and 8,OO-14,OO~m, covering the thermal inftared (TIR) to map kaolinite- and -silica-rich surfaces (only possible with ASTER thermal multispectral data). The detection of gossans, on the basis ofthe conceptual model, was made by exclusion ofvariables not typically linked to these surfaces, using the following corollary: "iron-rich surfaces that are deficient in kaolinite and silica loads (typical of laterites), are probably genuine gossans, to which metal deposits are also potentially linked". Image processing techniques applied to data gathered in the VIS, NIR and SWIR portion of the spectrum comprised a rigorous procedure for atmospheric correction of the data using the Atmospheric CORrection Now (ACORN) software. A radiative transfer model (MODTRAN) was applied to ASTER and TM calibrated data. Several image processing techniques were functional for the VIS-SWIR data, including color composites, band ratios, principal component analyses and hyperspectral image processing techniques adapted to multispectral data, such as Spectral Angle Mapper (SAM) and Mixture Tuned Matched Filtering (MTMF). ASTER TIR data were converted to emissivity and enhanced via color composites, decorrelation stretch, adapted pseudo-ratios, Minimum Noise Fraction and by hyperspectral image processing (SAM e MTMF) techniques. The integration of the conceptual model and image processing techniques tailored to detect genuine gossans, isolating them ftom other representative surfaces in the Irece basin, proved very successful and the results were substantial using ASTER imagery. These achievements using ASTER multispectral data in the Irece basin demonstrate these data can offer detailed mineralogical information, ftom which abundance maps of key mineral endmembers used in exploration for base-metal deposits can be derived, even under tropical conditions.

ASSUNTO(S)

processamento de imagens analise espectral

ACESSO AO ARTIGO

Documentos Relacionados