reflectance
000 1 : ; : A ; s ; ;
400 600 800 1000 1200 1400 1600 1800 2000 2200 2400
Wavelength (nm)
CREAR 200m: PRA Ü 5m 1.8m PEE SAN 1.50% so Z Om
N 235m —--40mi ---- 45m
seen D pb eee BEOmb 0 85m - 9.0mi
Figure l. Comparison of different amount of crude oil in soil
spectrums
Deducing from the above analysis that when the content of
crude oil in soil is little, the diagnose characteristic of double
absorption at 1748nm and 2330nm for petroleum hydrocarbons
in soil is very weak or not, and vulnerable to be interfered by
other soil composition, so that reduces the accuracy of detection
result using the feature for petroleum hydrocarbons in soil;
when the crude oil is up to certain amount, it can make a
hyperspectural remote sensing detection for soil with petroleum
hydrocarbons using the diagnosis characteristic of spectrum.
2. Hyperion Images Processing and Analysis
As the analysis method for hyperspectral data developed by
AlG(Analytical Imaging and Geophysics LLC), we select 196
independent valid bands from the original 242 bands; get rid of
22 bands affected strongly by vapour within the spectrum range
around 1356-1417nm, 1820-1932nm and over 2395nm.
Checking out the remain 175 bands one by one, and repairing
the bad line with average on its adjacent columns or rows;
removing the serious strip abnormal in some bands(especially
SWIR band) using global balance method; correcting the smile
effect in images by minimal noise transformation; convert the
radiation image calibrated into apparent reflectance image with
ENVI FLAASH atmosphere correction module; at last, based
on the Landsat7 ETM image in the same coverage area,
choosing the ground control points(97 points)by interactive
method and correcting Hyperion image with a correct
polynomial by the most neighboring sampling and control the
precision within half of a pixel.
Extracting some spectrum of typical pixels in the SeBei gas
field from the Hyperion reflectance image. The following
features can be known: (1) water absorption at 1.35um and
0.93um is very obvious, the water absorption at 1.4um and
1.9um is strong and wide, and these water bands have large
influence on neighbouring data; (2) the absorption feature of
ferric ion is visible at 0.5um and0.7um, and the absorption
bands of water and oxygen result in less obvious performance
for the absorption ferrous ion; (3) the absorption of hydroxyl
ion at 2.2 is obvious; (4) carbonate ion absorption at 2.31-
2.35um is evident, and some curves show the double absorption
feature at 2.31um and 2.35um; and this show the absorption
feature of Hydrocarbon key(C-H)within the bands; (5) the
absorption of hydro carbons occur within bands range from
1.72 to 1.75um; but is not obvious, as the content hydro carbons
in soil is small, and the presence of mixed pixel due to the low
spatial resolution of 30m.
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
2.4 The Extraction of Hydrocarbon Information Using
Hyperion data
The hydrocarbon microseepage information of oil/gas deposits
either onshore or offshore can be directly detected using
hyperspectral remote sensing data, of which identifying the
distinct absorption features related to micoseepage is the key
element. Taking full advantage of remote sensing technology,
microseepage information can be determined through the
spectral absorption signature in the above surface. The
established ^ Three-Band-Ratio algorithm amplifies the
absorptive signature, which utilizes the ratio of the spectral
absorption feature in 1748nm or 2330nm (point b and
reflectance: Rb) of hydrocarbon in soil and the reflectance (Ra,
Rc) of two points (a, c) in the shoulder of the above absorption
feature. The ratio can be calculated as follows:
HI = (4, roi UR zR, 1)
Ac — À
Where Ra; Aa, Rc;Ac= the reflectance/wavelength pairs for the
two shoulder points of the absorption feature.
Values of HI can be a good indicator of hydrocarbon
microseepage information: if HI>0, the value means the
existence of hydrocarbon microseepage; additionally, the lager
the value, the lager the hydrocarbon concentration it represent.
Based on the Three-Band-Ratio algorithm and Absorption-
Depth method noted above, Remote Sensing images are used to
explore the presence of hydrocarbon. For Hyperion data, the
reflectance values for Aa, Ab, Ac, wavelengths of 1699.4nm,
1729.7nm, and 1749.79nm are chose respectively. The values
of HI result from the Hyperion hyperspectral data
(EO1H1370342005223110KV.L1R) required in august 11th,
2005 are shown in Figure2 (a), and Figure2 (b). Results of the
two methods are almost consistent comparing the images: high
values locating in the Camelback Mountain gas-bearing
structure-north part of our study area as well as a nearly linear
distribution along with the Camelback Mountain structure.
Additionally, there also exist some high value points in area of
SeBei Gas Field. Since human eyes are more sensitive to color
hues than to gray tones, a colorful hydrocarbon microseepage
image is composited using bandl55 (1699.4nm), band165
(1800.29nm) and the above calculated HI gray image
(Figure2(c)). In the Figure 2, Blue hues which express the
information of hydrocarbon, indicate that the deeper of the its
hue, the more hydrocarbon microseepage concentration.
Image Figure 2(c) shows that hydrocarbon microseepage areas,
for which the blue hue region of the image stands, located
obviously in two areas: the south SeBei Gas Field and the North
Camelback Mountain gas-bearing structure of the study area.
Given that hydrocarbon, concentration might be high in soil
near Gas Field, and it is reasonable that there might be
obviously hydrocarbon microseepage along the Camelback
Mountain gas-bearing structure fraction, information concluded
from the calculation coincides well with known the natural gas
and anomaly gas distribution.
