Abstract: Oil sands, as an important part of the world's energy of 21st century, have been studied by a lot of geological workers on their geological characteristics of formation and distribution of metallogenic belt and so on, rather than remote sensing exploration. The basis of identification of different features by remote sensing technology is the reflectivity difference at the same wavelength position and the variation of reflectivity with wavelength.In this study, the authors got the spectral reflectance characteristics through analyzing the spectral data of oil sands. In order to recognize oil sands on remote sensing satellite image of Landsat 8 OLI, based on the oil spectrum characteristics and the correspondingcorrelation analysis, two recognition models were proposed and applied to the two research areas of Lenghu and Mahai in Qaidam basin. The results show that it is feasible to use remote sensing image to detect oil sands prospecting target area; the oil sands extraction model based on correlation analysis is better than the spectral feature analysis model; the ratio method is better in improving the oil sands remote sensing information in Lenghu area, while the difference method is better in Mahai area.

Key words: oil sands, spectral reflectance, Landsat 8 OLI, remote sensing, identify