Legend 
Hydrocarbon 
Cloud Shadow 
Camelback Mountain 
Sebei Gas 
«> BE 
Figure 2. Extraction of Hydrocarbon Seeps 
2.5 The Extraction of Alteration Minerals Using Hyperion 
data 
The identification of alteration minerals using Hyperspectral 
Remote Sensing can detect hydrocarbon microseepage and 
locate oil/gas deposits indirectly. This article combines the 
method Linear Spectral Unmixing (LSU) and algorithms of 
Spectral Angle Matching (SAM) for determining the mineral 
composition counterparts of hyperspectral Remote Sensing 
endmember. Identification precision is enhanced by using 
methods of subtracting the hyperspectral image bands rationally 
and algorithms of determining endmembers. What is more, 
integrating materials derived from field surveys of geology, a 
complex progress ensures the accuracy of image endmembers' 
corresponding minerals concentration and composition. 
According to the geochemical data of mineral composition, our 
studying area is characterized by three main alteration minerals: 
clay minerals (of which illite is the represent type), carbonate 
minerals (of which calcite is the represent type) and other 
minerals (of which rock salt is representing). 
Spectrums of the three minerals noted above are obtained from 
reference JPL spectral library. Application of linear mixing for 
different proportions of the three minerals, control increment of 
5% each time, a group of simulation spectrums are derived, 
with intervals containing both clay and carbonate's common 
absorption features at wavelengths from 2.0 to 2.5um. Efforts to 
find the best matching simulation spectrum for endmember 
have been made by using SAM and SFF spectral analysis 
methods at the same time. Supposing that simulation spectrum's 
counterpart minerals composition can be termed as the 
corresponding composition of mixed minerals to endmembers, 
the mixed minerals concentration and composition as 
counterpart of endmembers are determined by combining the 
field survey geology materials simultaneously. 
Figure 3 shows the identification results of alteration minerals 
using SAM algorithms. Rock salt and carbonate minerals of 
high concentration are concentrated at the upper-middle part of 
the image, where the true colour composite image appears a 
greyish high reflection area. The area is characterized of 
development of yardang landform and the surface of the land is 
covered by saline sandy soil (of which silicon and rock salt are 
the main form), and the abundance of fragments of carbonate, 
sand, and rock salt might be result in the weathering and 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B7, 2012 
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia 
   
sedimentary movement. Meanwhile, a minor concentrate also 
exist in the top of Sebei Gas Field’s 2 well, for the reason that 
seeping hydrocarbons reaching the near-surface as well as the 
moisture environment near marsh generate a circumstance of 
deoxidization leading to the carbonate's alteration. What is 
more, the moisture surface environment can also facilitate the 
rising of concentration of rock salt. 
The main distribution of high concentration of illite and 
carbonate near Sebei Gas Field’s 2 well and marsh area 
coincidents with the petroleum indication of our studying area. 
As is known to us that Gas Reservoir is a concentrating place of 
hydrocarbon, the exploit of gas and the moisture condition near 
marsh area contribute greatly to the region’s alteration of clay 
and carbonate minerals. Banded spreading areas similar as 
known petroleum indications along Camelback Mountain 
anticlinal structure can be illustrated by that seeping 
hydrocarbons are present in the form of transversely cutting off 
the structure, and in fact cause the abundance of clay minerals 
and carbonate minerals alteration. 
Figure 4 shows the identification result of alteration minerals by 
using LSU algorithms. Identification result exposed more 
serious problems such as mixture type of minerals, overlapping, 
and the underrate of distribution area (Figure 4) than SAM, 
which turns out that the areas of alteration minerals derived 
from LSU method are smaller than that of SAM, as well as 
some difference in locations of minerals. 
Several minerals do not emerge in the anticipated area 
according to the identify result. For example, illite and 
carbonate of high concentration seldom appear in the SeBei Gas 
Field and Camelback Mountain anticlinal structure. The 
difference might be attributed to the algorithms. SAM measures 
similarity by calculating the angle between the N-Dimension 
space reference spectral (endmembers spectral) and the 
unknown spectral (Hyperion image spec tral) whose result is 
subject to spectral shape, but of little relevant to spectral 
reflectance. However, method LSU suffers from effects both 
spectral shape and reflectance. The SeBei Gas Field 2 well 
Camelback Mountain anticlinal structure leading to a 
development of yardang landform, as well as its near to marsh, 
a moisture soil altogether contribute to the depressing of 
reflectance, posing a situation that while identification worked 
for SAM algorithm, it does not feasible for LSU method. 
Conclusion: based on alteration minerals’ diagnostic absorption 
spectral feature and the low signal-to-noise ratio of Hyperion, 
assuming the mixture of the 3 minerals’ spectrum noted above 
consists the endmember spectrum of image in this study, a 
combination of method Linear Spectral Unmixing (LSU) and 
algorithms of Spectral Angle Matching (SAM) can effecively 
determine the mineral composition counterparts of 
hyperspectral image endmember. A resampling result of 
Hyperion’s 175 bands has been used to identify the alteration 
minerals using both SAM and LSU methods. Accessing the 
identification precisions of different methods, the comparison 
result indicates that a resampling-based SAM method fits the 
known gas reservoir distribution best.