The identification of hydrocarbon alteration minerals with hyperspectral remote sensing can be used to detect hydrocarbon microseepages and locate underground oil-gas reservoirs. Taking the Three-lake region in eastern Qaidam basin where natural gas is accumulated as an example, the authors resampled hyperion hyperspectral data to overcome the impact of unobvious target minerals and low SNR of the sensor. Through defining the indicative signature for detecting hydrocarbon alteration minerals with hyperspectral remote sensing, mineral components corresponding to image end-members were determined effectively by using the spectral analysis method of spectral matching combined with spectral linear fitting. Comparative identification results with algorithms of spectral angle matching (SAM) and linear spectral unmixing (LSU) show that the method for reducing image band numbers and determining end-members can effectively enhance the precision of identification for hydrocarbon alteration minerals with hyperspectral remote sensing.