5/25/2018 RPD15

1/24

1

Seepage hydrocarbon detection using

Hyperspectral remote sensing analysisMuhammad Ikhwan Jamaludin [G02450]

Assoc. Prof. Dr.Abd. Nasir Matori

1.0 IntroductionIn total, 85% of oil and gas reservoirs have the phenomenon of macro/ -micro leakage (Freeman,

2003). The phenomenon referring to the abnormal signs of hydrocarbon component and escape to the

content of the soil and sea surface which above the hydrocarbon reservoirs (Tian, 2012). In theory,

there are two types of hydrocarbon seepage phenomenon which are macro seepage and micro

seepage. Macro seeps are visible oil and gas seeps, such as found at Kampung Minyak, Kudat

Peninsula, Northern Sabah. This oil seepage located in a tidal mangrove swamps and surrounded by

an adjacent green area which are primary and secondary vegetation (Muda, 2010). While microseeps

are the invisible oil and gas seeps, escape to the soil surface and into the air.

Early this year, the national oil corporation, PETRONAS announced the discovery of onshore

petroleum reservoir via the Adong Kecil West-1 well in Block SK333 (Department, 2013). Located

about 20 km northeast from Miri town, the Adong Kecil West-1 well, was drilled by JX Nippon Oil

and Gas Exploration (Onshore Sarawak) Limited as the operator for Block SK333, along together

with PETRONAS Carigali Sdn. Bhd. Up till now, there are only three discoveries have been found at

onshore Sarawak, namely in Miri, Asam Paya and Adong Kecil West fields.

In the present-day, geological surveys have been used to detect possibilities of oil and gas

reservoir. The surveys were done by geologists examining for hydrocarbon deposits beneath the

Earths surface. Through this survey, petroleum or gas reservoir will be determined where the areas

are geologically alike. After the possible earth surface containing hydrocarbon deposits has been

identified, seismic exploration takes place. Through seismology, geophysicists are able to artificially

create vibrations on the surface and record how the signals are reflected back, revealing the properties

of the geologic beneath.

As the above methods mentioned, it is clearly being time consuming, expensive and requires

many manpower. Based on the micro seepage phenomenon and the revolution of satellite remotesensing, method of hydrocarbon exploration could be rejuvenating especially on the onshore

5/25/2018 RPD15

2/24

2

underground deposits. According to Oyundari, (2008), remote sensing through hyperspectral analysis

is a tool that offers a non-destructive investigation method and has a significant added value to

traditional methods such as drilling of the subsurface and monitoring of the possibilities of a large

area which involve much technical expertise.

Referring back to Malaysia cases, almost 84.51% of the earth's surface are covered with primary

and secondary forest as reported in Forestry Statistics for the year 2011. It is not recommended to

find the potential macro/microseepage through satellite imagery of the earths surface directly. A

possible way to map the area is by using the spectra of stress vegetation through hyperspectral remote

sensing analysis (Freeman, 2003). In addition, hyperspectral remote sensing can be used to

differentiate the types of vegetation in leaf water, chlorophyll, cellulose, and leaf structure (Freeman,

2003).

Method in detecting hydrocarbon seepage generally uses remote sensing application which uses

the hyperspectral analysis imaging to detect the stress of vegetation that is affected by the

hydrocarbon. A few studies focused on the hydrocarbon detection were conducted by foreign scholars

(e.g. Noomen, 2003, and Shu-Fang, 2008). For example a study by Noomen et al., (2003),

investigates the existence of natural gas seepage influence on plants using a Probe-1 image, which is

one of the hyperspectral remote sensing sensors (Figure 1). While a study by Shu-Fang, (2008)

highlighted the extraction procedure in Dongsheng of Inner Mongolia, using hyperspectral remotesensing technology, through EO-1 Hyperion data based micro-seepage phenomena.

It is undeniable that the ability of remote sensing application in oil spill detection, involved large

area surveillance, tactical assistances in crisis and monitoring purposes. On the other hand, for

hydrocarbon seepage discovery, hyperspectral remote sensing has the ability to provide data on the

possible ways of exploration. In a nutshell, high spectral resolution spectroscopic or hyperspectral

measurement either from airborne or satellite devices has a vast potential to be used in the detection of

hydrocarbon seepages.

The aim of this research is to detect seepage hydrocarbon through the stress of vegetation. In this

proposal paper, the outline of the study and the early result of pre-processing using FLAASH are

presented. It is expected a relatively new method for oil-gas reservoir exploration by using

hyperspectral remote sensing, based on the theory of hydrocarbon seepage information will be

implemented in the case study of Adong Kecil, Miri. Image processing of hyperspectral remote

sensing is combined with laboratory analysis in developing the spectral library will be used to the

further methodology in detecting hydrocarbon seeps through the stress of vegetation as it is expected.

5/25/2018 RPD15

3/24

3

Figure 1: Example of seeps influence the vegetation (Source: Noomen et al., 2003)

2.0 Problem Formulation2.1 Literature Review

2.1.1 Introduction

This study generally takes into account the detection of hydrocarbon seeps through the stress of

vegetation using hyperspectral remote sensing analysis which concerns on the spectral reflectance of

the plants itself, where and how they happened as there is a growing interest in finding possibilities of

new oil and gas deposits focussing on onshore discovery. Thus it is very important to solve these

questions in order to get a comprehensive understanding of the phenomena happened. This topic

becomes an interest for this study as the environmental issue becomes as global awareness nowadays

plus the fact that seeps is an important as indicating petroleum reservoir or source of pollution which

plays a crucial role in humans life. Nowadays, most of reservoir possibilities have been done through

geology and seismic exploration as it has the ability to detect the type of soil deposits in detail.

2.1.2 Seepage hydrocarbon phenomenaAs scientifically explained, seepage phenomena are due to pressure differences in the Earths

subsurface, hydrocarbons can migrate from reservoirs to shallower levels and eventually to the

surface. Later on the surface, the hydrocarbons escape and consequently become natural source of

pollution to soil or water, if it is in the form of tar and oil (heavy hydrocarbons), or in the form of

gases (light hydrocarbons), which contribute to the global warming. Apart from the above mentions,

natural hydrocarbon seeps, both macro and micro has been a major indicator of the possibilities of

petroleum deposits location (Etiope, 2009, Shu-Fang 2008, Noomen 2003, and Shu-Fang 2008).

Schumacher (1999) explained that macroseepage refers to the visible oil gas seeps, while

5/25/2018 RPD15

4/24

4

microseepage is an elevated concentration of hydrocarbon elements in soils or its above the petroleum

reservoirs.

According from Etiope (2009), there are at least 4 important key points of gas and oil seepages,

which are (1) as an indicator of petroleum or natural gas reservoirs, (2) indication of the occurrence ofa fault, (3) represent a geo-hazard for societal community and industry and (4) as a natural sources of

greenhouse gas. There are many ways in the detection of hydrocarbon existence; one of the ways is

through stress of vegetation.

2.1.3 Vegetation stress

It is agreed by many scholars that natural gas including hydrocarbon may influence plant in many

ways, as for example through the root system, air, and aerial contact. Smith (2002) stated that the root

system was the main medium how the gas may be taken into the plant. In addition, the chemical

contains in the soil properties also affect the plant stress of the leaking gas. While there is no harm

when involve an aerial expose to the plant.

Smith (2002) added that the age of the trees might have a significant effect to the hydrocarbon.

He stated that the young tree was able to adapt to the seeps elements than trees that more mature. It is

due to the length of the root that younger trees are more shallower rooting system than older trees that

had deep roots. And because of the high concentration of gas in the soil, the roots were unable to grow

towards the surface to get oxygen (Smith 2002).

The other effects of natural gas leakage include growth and reproduction, and decreased numbers

of individuals, or a change in the green color of the leaves. If any of the said scenarios happen, it will

automatically change in the reflectance curves of the plant.

2.1.4 Hyperspectral remote sensing

It is recorded that the very first airborne hyperspectral remote sensing data in Malaysia were

acquired in 2004 over some area of Selangor by using UPM-APSBs Airborne imaging Spectrometer

for different Applications (AISA) (Jusoff, 2007). The missions were aimed to map the individual

timber species identification and forest inventory. The situation marked as a starting point for

hyperspectral remote sensing emerging in Malaysia.

In general each material in the earth has its own unique spectral signature which has been used

for the purpose of identification. It marked the arrival of typical multispectral remote sensing that

produce a few broad bands that is limiting the differential in order to produce detailed identification

(Singh, 2010). Nowadays, with the merging of technology, hyperspectral remote sensing has been

introduced which allowed a more detail identification consisting of a large number of narrow,

contiguously spaced spectral bands (Figure 2).

5/25/2018 RPD15

5/24

5

Hyperspectral remote sensing also known as spectrometers is a sensor that can be used in

laboratories, or mounted on aircraft and satellites as for global exploration that include in mapping an

earth element. Remote sensing in oil and gas exploration has listed at less destruction technology; also

give lots of positive impact to the economy, safety and efficiency (Shu-Fang 2008).

Figure 2: The concept of hyperspectral imagery (source: Shippert, 2004).

It is understandable that each photon of light emitted some sort of energy level in certain

wavelength. It includes light and other forms of electromagnetic radiation which usually refers in

form of wavelengths. Visible light for an example has a wavelength between 0.4 and 0.7 microns,

while X-rays wavelengths smaller than about 1 micron (figure 3).

Figure 3: The electromagnetic spectrum (source: Shippert, 2004).

Reflectance occurs when the light hitting a material, it will be reflected back as being absorbed or

transmitted. The reflectance of an element can be viewed through a reflectance spectrum measured

towards a range of wavelengths. The reflection and absorptions will create patterns across wavelength

that was used to identify certain elements. Figure 4 below shows reflectance spectra measured by

laboratory spectrometers of three elements leaf (green), talc (mineral), and silty loam (soil). By using

spectrometers, the reflectance was measured at many narrow, closely space wavelength bands, so the

spectra appear to be continuous curves. If using an imaging sensor (airborne or satellite) the resulting

images record a reflectance spectrum for each pixel in the image (figure 2). Later, spectral library is

5/25/2018 RPD15

6/24

6

needed to further analyze the reflectance spectra. This spectrum pixel data can give information about

the surface element.

Figure 4: Reflectance spectra measured by laboratory spectrometers (source: Shippert, 2004).

The multispectral image limits the classification techniques into broad categories. While

hyperspectral imagery extends the options to detailed image analysis, such as similar materials can be

differentiated in depth. Example of Hyperspectral satellite sensors and airborne are NASAs HyperionEO-1 sensor and FTHSI sensor (satellite), AVIRIS, HyMap and PROBE-1 (Airborne). Refer to Table

1 and 2 for the current and recent hyperspectral sensors and providers.

5/25/2018 RPD15

7/24

7

Table 1: List of Hyperspectral sensors and data providers (source: Shippert, 2004).

Satellite sensors Manufacturer Numbers of Bands Spectral Range

Hyperion on EO-1 NASA Go-ddard

Space Flight Center

220 0.4 to 2.5 m

FTHSIon Mighty Sat II Air Force ResearchLab 256 0.35 to 1.05 m

Airborne Sensors Manufacturer Numbers of Bands Spectral Range

AVIRIS(Airborne Visible

Infrared Imaging

Spectrometer

NASA Jet

Propulsion Lab

224 0.4 to 2.5 m

HYDICE

(Hyperspectral Digital

Imagery CollectionExperiment)

Naval Research Lab 210 0.4 to 2.5 m

PROBE-1 Earth Search

Sciences

Inc.

128 0.4 to 2.5 m

CASI(Compact Airborne

Spectrographic Imager)

ITRES ResearchLimited

up to 228 0.4 to 1.0 m

HyMap Integrated

Spectronics

100 to 200 Visible to thermal

Infrared

EPS-H

(Environmental Protection

System)

GER Corporation VIS/NIR (76), SWIR1

(32),

SWIR2 (32), TIR (12)

VIS/NIR

(.43 to 1.05 m),

SWIR1

(1.5 to 1.8 m),

SWIR2

(2.0 to 2.5 m),and TIR

(8 to 12.5 m)

DAIS 7915(Digital Airborne Imaging

Spectrometer)

GER Corporation VIS/NIR (32), SWIR1(8),

SWIR2 (32), MIR (1),

TIR (6)

VIS/NIR(0.43 to 1.05 m),

SWIR1

(1.5 to 1.8 m),

SWIR2(2.0 to 2.5 m),

MIR

(3.0 to 5.0 m),

and TIR

(8.7 to 12.3 m)

DAIS 21115(Digital Airborne Imaging

Spectrometer)

GER Corporation VIS/NIR (76), SWIR1(64),

SWIR2 (64), MIR (1),

TIR (6)

VIS/NIR(0.40 to 1.0 m),

SWIR1

(1.0 to 1.8 m),SWIR2

(2.0 to 2.5 m),

MIR(3.0 to 5.0 m),

and TIR

(8.0 to 12.0 m)

AISA

(Airborne Imaging

Spectrometer)

Spectral Imaging up to 288 0.43 to 1.0 m

5/25/2018 RPD15

8/24

8

Table 2: Upcoming Space-Based Hyperspectral sensors (source: Shippert, 2004).

Satellite Sensor Sponsoring Agencies

ARIES-1 ARIES-1 Auspace Ltd

ACRES

Earth Resource Mapping Pty. Ltd.

Geoimage Pty. Ltd.CSIRO

PROBA CHRIS European Space Agency

NEMO COIS Space Technology Development Corporation

Naval Research Laboratory

PRISM European Space Agency

2.1.5 Hyperion EO-1 sensor

The Hyperion EO-1 as a part of National Aeronautics and Space Administrations (NASA) NewMillennium Program, was planned to improved earth observations by demonstrating new technologies

and strategies. Launched on November 21, 2000, the Earth Observation 1 or EO-1 satellite contains 3

main instruments which are Advanced Land Imager (ALI), Hyperion Imaging Spectrometer and the

Linear Etalon Imaging Spectral Array (LEISA) Atmospheric Corrector (LAC) (Ungar 2003).

ALI is a prototype for a new generation of Landsat-7 Thematic Mapper while LEISA is a high

spectral resolution wedge imaging spectrometer designed to measure atmospheric water vapor

content. Last but not least, Hyperion Imaging Spectrometer is the first high resolution hyperspectral

imaging to orbit the earth.

Both Hyperion and ALI spectral bands, cover the visible (VIS), near-infrared (NIR) and

shortwave infrared (SWIR) regions, is basically a push-broom experimental hyperspectral missions

that captures data in 220 spectral channels. As designed for one-year life spends, the EO-1 has now

successfully completed its twelfth year in orbit at the end of 2012. Now, the EO-1 data became

available at no cost at USGS archive, available for researches all around the world (Spencer, 2009,

Riebeek, 2010).

5/25/2018 RPD15

9/24

9

Figure 5: A schematic is shown the scene collections from ALI, Hyperion and Landsat-7(Source: Shippert, 2004).

ALIHyperion

Band Designations Band Names (wavelength, um)

Pan Pan (0.4800.690)

Continuous Spectra 0.42.4um

242 Bands

Bandwidth:10nm

BlueMS-1p (0.4330.453)

MS-1 (0.4500.515)

Green MS-2 (0.5250.605)

Red MS-3 (0.6330.690)

NIR

MS-4 (0.7750.805)

MS-4p (0.8450.890)

SWIR

MS-5p (1.201.30)

MS-5 (1.551.75)

MS-7 (2.082.35)

Spatial Resolution Pan: 10m, MS:30m 30km

Swath width 37km 7.7 km

Table 3: Characteristics comparison between ALI and Hyperion instruments.

2.1.6 Application of remote sensing in detecting hydrocarbon

The detection of hydrocarbon seeps through remote sensing application has been used over the

past 20 years since the introduction of the technology. It is proven in minimizing the destruction that

offers from the traditional methods of investigating seepage and also crude oil pollution. Stated by

Meijde (2008), optical remote sensing has been tested by many scholars for onshore exploration and

detection of seeps at the Earths surface since 1984. As stated in the above section, there are various

scholars successfully detect hydrocarbon elements through hyperspectral remote sensing application,

either from pipe leakage, seeps, stress of vegetation and also from oil spills (Smith, 2002, Kuhn, 2002,

Noomen, 2003, and Shu-Fang, 2008).

The normal vegetation in the electromagnetic radiation (figure 3) is controlled by absorptions andreflection features caused by vegetation pigments such as chlorophyll and carotenoids (organic

5/25/2018 RPD15

10/24

10

pigments that are found in plants), while the near infrared (NIR) and shortwave infrared (SWIR) is

because of four main water absorption features (figure 6). In addition, unhealthy plant could recognize

by an increase in reflectance in the absorption features and decrease in NIR reflectance. Noomen

(2003) stated that it will shift the slope between red and NIR called the red edge position(REP).

Based on research conducted by Kuhn (2003) and Tian (2012) claim that the presence of the

hydrocarbon in the stressed plant can be detected between the range (absorption features) 1.73 um and

2.31 um (figure 6).

Figure 6: (left) the typical vegetation reflectance curve, with indication ofmajor absorption features (right).

2.2 Objective

The objectives of this research include:

i. To explore the correct methodology in detecting hydrocarbon seepage usingHyperspectral remote sensing analysis

ii. To develop a spectral library of vegetation stress of the study area in form of laboratorysimulation

iii. To detect hydrocarbon seepage from Hyperspectral Hyperion EO-1 image based on thespectral library result using the Hydrocarbon Index (HI) and Spectrum Angle Mapper

(SAM)

5/25/2018 RPD15

11/24

11

2.3 Problem Statement

Because of the difference in pressure on the underground of the Earths, hydrocarbons seep to

higher levels and most probably rise to the surface. The escaped hydrocarbons oxidize and influence

the area habitat (vegetation and soils) that alter the formation of mineralogy and botanical freak.

Noomen (2003) indicate that natural seepages are a phenomenon of hydrocarbon deposits below its,

apart from being sources of pollution on local and global scales.

As mention above, the typical way in detecting hydrocarbon is through geological surveys and

seismology experiment which clearly are is a long way process and even costly. Hyperspectral remote

sensing (imaging spectrometry) is the acquisition of many narrow, contiguous spectral bands

(Freeman, 2003). Due to oil and gas seeps have been proven to modify the minerals at the earth's

surface, there is a high possibility to detect macro/micro seepages of oil and gas through the stress of

vegetation on the earth's surface using hyperspectral remote sensing analysis.

2.4 Significant of research

To oil and gas company

i. Relatively new method of detection of seepage hydrocarbon through the stress of vegetation

especially using the Hyperion EO-1 sensor

ii. Less money for exploration, because of the direct detection through the air (satellite or

airborne), no drilling (man operation being done on the ground) and cover very mass area.

iv. Less time consuming, different from the typical oil and gas exploration which required othersfield of expertise to detect the location of the possible reservoir.

To environment

i. Protecting the environment, through hyperspectral remote sensing analysis, there will be less

harm to the environment and also the adjacent area due to the analysis required satellite

hyperspectral image and also laboratory experiment.

3.0 Materials and MethodThe data were collected from primary and secondary data. The primary data include

hyperspectral remote sensing image by EO-1 satellite platform and the required plants which are palm

oil in developing the spectral library of stress vegetation in the form of lab scale simulation. The

secondary data include land use and topographic maps, and spectral library database (USGS or JPL

spectral library) as to compare the spectral signature of the designated plants.

5/25/2018 RPD15

12/24

12

Based on the latest information, the area, Adong Kecil, Miri, Sarawak is already being cleared

out for oil and gas exploration. Referring back to the main focus of the study is to identify the stress of

vegetation inhabited the area, back dated image is the most suitable to carry out the experiment. Most

probably from the past 3 or 5 years image/scene is the best to identify the vegetation stress influence

by hydrocarbon seeps. An EO-1 Hyperion image was acquired from the USGS archive. Table 4 below

shows the detail of the hyperspectral EO-1 satellite image.

Table 4: Hyperspectral EO-1 detail

SPACECRAFT_ID EO-1

SENSOR_ID Hyperion

Spatial Resolution 30m

Swath Width 7.5 kmSpectral Resolution 10 m

Total No. Of Band 224

Date of Acquisition 22 August 2009

Figure 7: Map of Adong Kecil, Miri, Sarawak and image coverage area

Methods of detection are based on the stress of vegetation that lives in the study area. To

understand with the complexity of a dynamic phenomenon of hydrocarbon seepage which is a

requirement are the development of spectral library, site identification and delineation, and spectral

band analysis using Spectral Angle Mapper (SAM) and Hydrocarbon Index (HI)

According to Ellis (2002), there are two ways in obtaining highly accurate sample of the spectral

signatures of the materials, which is through indoor reflected light measurement in the lab or in the

outdoor environment using a portable spectrometer. All this highly accurate sample measurements

than will be combined in various spectral libraries that have been established worldwide (Ellis, 2002

& Baolin, 2010). Therefore, the development of the spectral library database is needed as to acquire

MIRI

Image coverage

area

5/25/2018 RPD15

13/24

13

the unique spectral signature of the plant. The statement is supported by (Werff, 2006) which stated

that the spectral library of different plant species is unique, means its bound that study area only.

A lab scale experiment will be conducted as to develop the spectral signature of the required

plant dominating the study area has been mentioned earlier. Through the survey, the main plantation

habitats the area is palm oil or kelapa sawit (Elaeis Guineesis). Therefore the experiment will be

conducted to recognize the unique spectral signature between stressed palm oil with hydrocarbon and

the normal plant. A handheld hyperspectral spectrometer instrument will be used to record the unique

spectral signature between the normal and stressed palm oil plant. At least 3 samples of plant with

different crude oil composition will be used for the experiment (Ren, 2008).

Pre-processing of the satellite image includes geometric and radiometric correction using

FLAASH technique is carried out on the image before could be used. All of the processing work and

analysis will be done using ENVI imaging software. Later, image processing such as image

classification and image post classification will take place. According to Kuhn (2002), a study in

detecting hydrocarbons using hydrocarbon index (HI) (1) is successful. It showed that hydrocarbon

index is reliable to be applied to airborne hyperspectral remote sensing data. While Spectral Angle

Mapper Classification (2) or SAM is an automated method of comparing of image spectra through the

available spectral library. In this research, both methods will be used during spectral analysis stage. In

addition, supported by Hou (2011), it is proved then both methods are feasible in detectinghydrocarbon element.

Hydrocarbon Index and Spectrum Angle Mapper (SAM)

According to Kuhn (2002), a study in detecting hydrocarbons using hydrocarbon index (1) is

successful. It showed that hydrocarbon index is reliable to be applied to airborne

hyperspectral remote sensing data. While Spectral Angle Mapper Classification (2) or SAM is

an automated method of comparing of image spectra through the available spectral library. In

this research, both methods will be used based in detecting hydrocarbon. In addition,

supported by Hou (2011), it is proved then both methods are feasible in detecting

hydrocarbon element.

5/25/2018 RPD15

14/24

14

Hydrocarbon Index (HI)

Various scholars agreed that the absorption features at 1.73 um and 2.31 um are spectral

signatures of hydrocarbons existence (Kuhn 2002, Tian 2012) (refer to figure 8). The 2.31 um

has been used for feature to detect hydrocarbons, while 1.73 um features it close representmain water absorption maximum. Kuhn (2002) stated that by using HyMap (refer table 1), the

1.73 um features could be identified clearly either in radiance or reflectance (refer subtitle

2.1.3).

,RRRR

BA

AC

AC

ABHI

Hydrocarbon Index algorithm (Kuhn, 2002) where Ra; a, Rc; c = the

reflectance/wavelength pairs for the two shoulder point of the absorption feature. Tian (2012)

agreed that if HI > 0, it shown the present of hydrocarbon seepage element. And if the value

is bigger, the bigger hydrocarbon concentrations it exist.

Figure 8: (left) show the radiance spectra of oil-polluted sand and plastic showing significant1.73 um and 2.31 (F. KuHN) while Fig XX (right) shown the enlarged 1.73 um portion of

spectral signature (radiance) of chemical constituents.

Spectral Angle Mapper (SAM) Classification

SAM is computational methods for direct comparing unknown spectra image to known

spectral end members input by the user and/or identify form the image spectra itself. In this

research, 2 inputs of spectra image will be used as known spectra represent hydrocarbon

seepage of study area. The first spectra collect from the laboratory simulation of spectral

library (refer to 3.0 Materials and method). While second spectra collect form spectral

(1)

5/25/2018 RPD15

15/24

15

libraries available on the net which represent the most spectral signature with palm oil

(tropical forest).

21

1

22

1

1

2

11cos

nb

i

nb

i

nb

i

riti

tiri

SAM computational formula (source: Hou, 2011)

The final output would be the final map showing the area of vegetation influence with

hydrocarbon influence. Data will be compiled as follows (a) Location of site, (b) Hyperspectral image

acquired. (c) spectral library development of stressed vegetation(d) GIS application including adjacent

area, geo-referencing, and last but not least is (e) ENVI imaging software image processing,

classification and spectral band analysis using Hydrocarbon Index (HI) and Spectral Angle Mapper

(SAM).

It is anticipated that the thesis will comprise the following chapters: (a) Introduction, objective

and rationales of the study (b) Overview, literature review (c) Methodology and description of the

study area, (d) Development of Spectral Library of stressed vegetation (laboratory scale analysis), (e)

Spectral Band analysis, (h) Effectiveness and applicability of the seepages detection using the

application of hyperspectral analysis, and (i) Conclusion and recommendation for further work.

(2)

5/25/2018 RPD15

16/24

16

3.2 Flow chart of study methodology

-

Figure 9: Flow chart of study methodology.

3.3 Research Activities and Key Milestone

Please refer to appendix 1

Geometric

Correction

Radiometric

Correction

Image Processing

Image

Classification

Image Post

Classification

Spectral Angle

Mapper

Hydrocarbon

Index

Map of seepage

hydrocarbon

Acquisition of

H ers ectral

Pre-processing

Development of

Spectral Library

5/25/2018 RPD15

17/24

17

4.0 Preliminary Analysis

4.1 Proof of Concept (Analysis of research history)

The developments of hyperspectral remote sensing sensor enable the end-user to analyze the

unique spectral signature of every chemical constituents in this world. The satellite image were able todetect the type of vegetation stress which been influenced by the hydrocarbon seepages. The

following Table 5 shows the result of the research conducted using hyperspectral remote sensing

analysis using satellite and airborne sensor in detecting vegetation stress.

Table 5: Past research and satellite images used in research

Satellite

images used in

the researchAuthors Papers/ Journals/Thesis

HyMap F. Kuhn, K. Oppermann andB. Horig (2003)

Hydrocarbon Index

an algorithm for hyperspectraldetection of hydrocarbons

- Hong-Yan Ren, Da-Fang

Zhuang, Jian-Jun Pan, Xue-

Zheng Shi, Hong-Jie Wang

(2008)

Hyper-spectral remote sensing to monitor vegetation stress

Hyperion EO-1 Qingjiu Tian (2012) Study on oil-gas reservoir detecting methods using

Hyperspectral remote sensing

The results in the respective areas of the similar studies are as follows (Table 6).

5/25/2018 RPD15

18/24

18

Table 6: Result and findings in research reviewed

Papers/Journals

(Authors)

Results and Findings

Hydrocarbon Indexan algorithm for

hyperspectral

detection of

hydrocarbons

F. Kuhn, K.

Oppermann and B.

Horig (2003)

Figure 10 shows the HyMap scene RGB color composite image in near-naturalcolors (16:10:3). The numbers (1) and (2) represents oil-contaminated referenceareas, (3) plastic sheets, (4) artificial grass and race track, (5) plastic roofs; insert

above left: zoomed section.

Figure 11 shows HyMap Hydrocarbon Index image (HI image) derived from thesame dataset used for figure 8 using equitation (1); red pixels indicate the presence

of hydrocarbon-bearing materials; the gray scale values were assigned using ENVI

Hue Sat Lightness 2 color table.

Therefore this study showed that a Hydrocarbon Index applied to airbornehyperspectral remote sensing data can be used to successfully detect hydrocarbons.

This includes hydrocarbon-bearing materials, and oil-contaminated ground in

particular has been able to be detected using HI images

Hyper-spectral

remote sensing to

monitor vegetation

stress

Hong-Yan Ren,Da-Fang Zhuang,

Jian-Jun Pan, Xue-

Zheng Shi, Hong-

Jie Wang (2008)

The aim was to explore the possibility of spectral reflectance of rice plants usingthe estimation of Pb concentration at suitable growth stage.

Authors using paddy plants by grown in the barrels contains a mix of 10.0 kg soilwith 0, 2.5 x 207.2 and 5.0 x 207.2 mg Pb per 1 000 g soil named it as PB0, PB1

and PB2 planted in a greenhouse. The result show that the spectral reflectance in the region of visible-to-near-

infrared light (VNIR) increased toward days, because Pb concentration cause the

canopy chlorophyll content decrease (figure 12).

Even though the initial laboratory experiment is still lacking in the total of bothmetals and plants, in which to come out with a good prediction of models and etc,

it's proven that ground remote sensing readings (through a handheld spectrometer)

would provide reliable information for the estimation of Pb concentration in riceplants at the early tillering stage when appropriate features (such as DS and Den)

of reflectance spectra are applied.

Figure 10 (left): Airborne HyMap scene in band (16:10:3)

Figure 11 (right):HyMap image derived from hydrocarbon index equation

indicate the red pixels of the presence of hydrocarbon materials.

5/25/2018 RPD15

19/24

19

Figure 12: Normalized spectral absorption depth (Dn) curves of Pb0, Pb1 and PB2

in the 50th

(a), 65th

(b), and 80th

day(c).

Study on oil-gas

reservoir detectingmethods using

Hyperspectral

remote sensing

Quingjiu Tian(2012)

The author highlighted that the absorption characteristics of oil and gas reservoirmicro leakage to the soil was mainly in 1720-1750nm and 2310-2350nm. At 2310-2350 nm the spectral image is stronger but there is possible the mineral were

confused with other mineral absorption such as calcite.

While at absorption 1720-1750, the signal is weaker which means a carefulprocessing and analysis is required.

Field Spec ASD instrument (Spectral range: 350 to 2500nm, spectral resolution of3nm and 10nm) was used to measure the spectrum reflectance of soil with different

content of oil in the lab. The method was by putting the crude oil sample ranging

from 0.5 ml to 9ml into dry Qaidam soil samples (100g) into a bottle and shake

hardly until the combination is fully mixed. The result as shown in Figure 13.

Figure 13: Comparison of different amount of crude oil in soil spectrums

The results show that the hydrocarbon peaks are located at 1670nm and 1748nm.While in 2330, 2348nm the hydrocarbon peak already existed before adding crude

oil and with increasing of crude oil/ml, the absorption peak at 2308nm and 2349nm

is equal. Later, to extract the hydrocarbon information, the Hyperion data were

used to combine with HI (refer 3.0 Materials and method).

As to conclude, in developing the spectral library of stressed plants, the laboratory conducted by Ren

(2008) and Tian (2012) will be used as a guide. Even though experiment by Smith (2002) using

methane gas were more promising and taking less duration rather than using crude oil, but due to the

none availability of the gas and it is pricy, crude oil is been chosen represent hydrocarbon. It is

5/25/2018 RPD15

20/24

20

expected that the simulation will take about 3 month and above depends on the absorption of the

plants towards the crude.

4.2 Preliminary Result

The part of the strip from the EO-1 Hyperion scene is shown in Figure 15. The left scene shownradiance in RGB while the right scene shown the preliminary result of FLAASH (image pre-

processing method) in band combination (50:23:16) for RGB. Due to the lack of data of the study

area, the region of interest (ROI) still not yet been delineated. Figure 15 shown the central pixel

spectral which mainly vegetation. It is noticeable that atmospheric correction using FLAASH has

restored the reflectance value well. The input of pre-processing detail of atmospheric correction using

FLAASH is shown below (table 7). All the detail was gathered from the metadata of Hyperion EO-1

(. met) file and from previous research (Shih, 2004).

Table 7: FLAASH input details

Latitude 4.19999981

Longitude 114.02999878

Sensor Type HYPERION

Sensor Altitude (km) 705.00

Ground Elevation

Pixel Size (m) 30.000

Flight Date (Julian Day) 2009234 (Aug 22 2009)

Flight Time GMT (HH:MM:SS) 02:30:01

Atmospheric Model TropicalWater Retrieval No

Water Column Multiplier 1.00

Aerosol Model Rural

Aerosol Retrieval 2-Band (K-T)

Initial Visibility (km) 40.0

Spectral Polishing Yes

Width (number of bands) 9

Wavelength Recalibration No

5/25/2018 RPD15

21/24

21

855:580:509nm(50:23:16)

VNIR Vegetation RGB

855:580:509nm(50:23:16)

VNIR Vegetation RGB

Figure 15: RGB (50:23:16) left in radiance and RGB (50:23:16) rightin FLAASH viewing offEO-1 Hyperion Data, Miri, Serawak, 22

thAugust 2009

855:580:509nm(50:23:16)

VNIR Vegetation in radiance RGB

855:580:509nm

(50:23:16)VNIR Vegetation with FLAASH RGB

Figure 14: Spectral of central pixel (vegetation) in radiance (left) after applying FLAASH (right)

The left spectral picture has shown the radiance RGB of the image while the right spectral shows

the reflectance spectral after atmospheric correction with FLAASH as they would appear to the

hyperspectral Hyperion sensor. The reflectance spectra of vegetative material had shown some gaps

between the spectra. The gaps represent the wavelength ranges at which the atmosphere absorbs so

much light that no reliable signal is received from the surface.

5/25/2018 RPD15

22/24

22

5.0 ConclusionIn a nutshell, it is expected that the findings made herein further contribute to the following:

Provide a relatively new onshore exploration in detecting seeps through hyperspectral remotesensing analysis

Provide a guideline for future onshore petroleum exploration especially for the SoutheastAsian region, due to the same vegetation stress availability.

Provide a scientific approach on the data analysis of hydrocarbon seeps which will beavailable for future research.

In addition, this study will demonstrate that hyperspectral remote sensing technology would be

effective to better understand the probability of oil seeps across the remote geologic area.

5.1 Limitation of research

1. Due to the required data needed is the back dated image, means from 5 to 10 years before, theonly sensor available is Hyperion EO-1. Based on the LR the Hyperion sensor is an

experimental sensor well-past its design lifetime, so the quality is nowhere near modern

airborne hyperspectral system as for example AVIRIS, CASI or HyMap (refer table 1). It is

afraid that the required result couldntbe proved.

2. The development of spectral library will take the most time of the research; therefore, adetailed procedure should be done before doing the simulation.

3. Due to the crude oil resource is limited, the usage of the will be controlled. Due to this, it isafraid that the needed result could not be achieved at the required time.

4. The main instrument, handheld spectrometer is needed during the development of the spectrallibrary. Due to the instrument is not available in the university and need to outsource from

outside, it is afraid that it would prolong the duration of the simulation.

5/25/2018 RPD15

23/24

23

6.0 List of ReferencesMcGraw-Hill Science & Technology Encyclopedia. Seismic exploration for oil and gas Volume, DOI:

"Technology & Process." Hydrocarbon exploration Retrieved 2013, 27 june from

http://www.petro.in/ic/tp/tp.html.

Baolin, Y., X. Chenwei, et al. (2010). "Spectral Feature Extraction and Analysis Based on Hyperspectral

Remote Sensing Data."Department, M. R. (2013). Petronas announces oil and gas discovery onshore Sarawak.

Ellis, J. (2002). Detecting Onshore Oil Seeps with Hyperspectral Imagery. Airborne Remote Sensing

Conference Proceedings.

Ellis, J. (2003). "Hyperspectral imaging technologies for oil seep/oil-impacted soil detection and environmental

baselines." Environmental Science & Engineering.

Etiope, G. (2009). "A Global Dataset of Onshore Gas and Oil Seeps: A New Tool for Hydrocarbon

Exploration." Oil and Gas Business.

FAO, G. F. R. A. (2010). Malaysia: Forestry & Environment (Facts & Figures). M. T. Council.

Freeman, H. (2003). Evaluation of the use of hyperspectral imagery for identification of microseeps near Santa

Barbara, California. Virginia, West Virginia University Master of Science in Geology 26.

Govender, M., K. Chetty, et al. (2006). "A Review of Hyperspectral Remote Sensing and its Application inVegetation and Water Resource Studies." 33(No. 2).

Govender, M., K. Chetty, et al. (2007). "A review of hyperspectral remote sensing and its application in

vegetation and water resource studies." Water SA 33(2): 145 - 152.

Hou, F., D. Wang, et al. (2011). The Abnormal Information Extraction of the Hydrocarbon Micro-seepage

Based on the Hyperspectral Image, IEEE.

Jusoff, K. (2007). "Advanced Processing of UPM-APSB's AISA Airborne Hyperspectral Images for Individual

Timber Species Identification and Mapping." International Journal of Systems Applications, Engineering

& Development 2(1,2007): 21-26.

Jusoff, K. (2007). "Mapping Bamboo in Berangkat Forest Reserve, Kelantan, Malaysia using Airborne

Hyperspectral Imaging Sensor." International Journal of Energy and Environment 1(1): 1-6.

Khan, S. D. (2006). Mapping Alteration Caused by Hydrocarbon Microseepages in Patrick Draw areaSouthwest Wyoming Using Image Spectroscopy and Hyperspectral Remote Sensing. Houston,

University of Houston: 19.

Korgaonkar, G. S. and R. R. Sedamkar (2012). "Hyperspectral Image Classification." International Journal of

Engineering Research 2(5): 2177-2184.

Kuhn, F., K. Oppermann, et al. (2002). "Hydrocarbon Index - an algorithm for hyperspectral detection of

hydrocarbons." Int. J. Remote Sensing 25(12): 2467-2473.

L.Liu, G. H. Huang, et al. (1999). "A GIS-Supported Remote Sensing Technology for Petroleum Exploration

and Exploitation." Petroleum Society of CIM 99(101).

Meer, F. v. d., P. v. Dijk, et al. (2000). "Hyperspectral Hydrocarbon Microseepage Detection and Minitoring :

Potentials and Limitations."

Meijde, M. v. d., H. M. A. v. d. Werff, et al. (2008). "A Spatial-Spectral Approach for Visualization of

Vegetation Stress Resulting from Pipeline Leakage."

Muda, J. (2010). Oil seepages at Kampung Minyak, Kudat Peninsula, Northern Sabah: Potential for geotourism

development. Bulletin of the Geological Society of Malaysia. 56: 49-52.

Noomen, M. F., A. K. Skidmore, et al. (2003). Detecting the Influence of Gas Seepage on Vegetation, using

Hyperspectral Remote Sensing. 3rd EARSel Workshop on Imaging Spectroscopy. Herrsching.

Oyundari, B. (2008). Spectral indicators for assessing the effect of hydrocarbon leakage on vegetation,

International Institute for Geo-Information Science and Earth Observation. Master of Science

Ren, H.-Y., D.-F. Zhuang, et al. (2008). "Hyper-spectral remote sensing to monitor vegetation stress." J Soils

Sediments(8): 323-326.

Riebeek, H. (2010). "Earth Observing-1 : Ten Years of Innovation." Retrieved 17 Sept, 2013.

Roper, W. E. and S. Dutta Oil Spill and Pipeline Condition Assessment Using Remote Sensing and Data

Visualization Management Systems

Schumacher (1999). A General Overview of Hydrocarbon Microseepage.

5/25/2018 RPD15

24/24

24

Shih, T.-Y. (2004). On the Atmospheric Correction for Hyperion Scene. ACRS 2004.

Shippert, P. (2004). Why Use Hyperspectral Imagery. Photogrammetric Engineering & Remote Sensing.

Shu-Fang, T., C. Jian-Ping, et al. (2008). "The Information of Oil and Gas Micro-seepage in Dongsheng Region

on Inner Mongolia Extration Based on the Airborne Hyperspectral Remote Sensing Image." Remote

Sensing of the Environment: 16th National Symposium on Remote Sensing of China 7123(1230K-1): 8

Singh, B. and J. Dowerah (2010). "Hyperpectral Imaging: New Generation Remote Sensing." e-Journal EarthScience India 3(III).

Smith, K. L. (2002). Remote Sensing of Leaf Responses to Leaking Underground Natural Gas, University of

Nottingham. Doctor of Philosophy.

Spencer, D. (2009). "Technical Announcement: USGS Adds NASA's EO-1 Satellite Imagery to Web-enabled

Archive."Retrieved 18 October, 2012

Tian, Q. (2012). "Study on Oil-Gas Reservoir Detecting Methods using Hyperspectral Remote Sensing."

International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences

XXXIX-B7.

Ungar, S. G., J. S. Pearlman, et al. (2003). "Overview of Earth Observing One (EO-1) Mission." IEEE

Transcation of Geoscience and Remote Sensing 41(6).

Werff, H. M. A. v. d. (2006). Knowledge-based remote sensing of complex objects:recognition of spectral and

spatial patterns resulting from natural hydrocarbon seepages.

Xiao, J., Q. Tian, et al. (2010). "Extraction of Hydrocarbon Content Information by using Hyperspectral Image

at Liaodang Bay, China."

Yang, H., F. D. V. d. Meer, et al. (2000). "Aerospace Detection of Hydrocarbon-Induced Alteration." Handbook

of Exploration Geochemistry 7: 13.