Southeast Asian peatland constitutes about one-tenth of the entire extent of
the global peat resources and occupies about 60% of the tropical peatland (Global
Environment Centre Secretariat, 2005). The largest deposit of peat soil
in Southeast Asia is found in Indonesia. Malaysia also has significant deposit
of peatland which is more extensive in the low lying poorly drained depression
basin of the coastal areas. Peatland in Malaysia covers a total area of approximately
2.4 million hectares, about 8% of the total land mass of the country (Mamit,
2009). The distribution of the deposit according the countrys three
major regions is shown in Fig. 1.
The state of Johor has the largest deposit of the resource in peninsular Malaysia
(Wetland International, 2010). The peatland of western
coast of Johor is described as highly extensive with a thickness up to about
6 m mainly underlain with marine clay and silt (ASEAN/US
Peat is a product of partial and gradual decomposition of plant materials in
marshy areas under waterlogged condition (Huat et al.,
2009). It is an organic soil with excessively high proportion of organic
matter. The formation process involves consequences of the development of an
ecosystem where accumulation rate of organic matter exceeds decomposition rate
(Huat et al., 2011).
|| Regional distribution of peat soil in Malaysia
Thus, peat production is a continuous process so long as bog plants continue to grow and die. It can therefore be distinguished from organic soil by its higher organic content.
Owing to the variation in decompositional resistance of plant materials, there
is high degree of spatial variability in the rate of decomposition of peat.
This variation in degree of decomposition of the accumulated organic matter
is represented as stratigraphic layers in the peat core (Xuehui
and Jinnming, 2009). The stratigraphic sequence is a representation of distinctive
physical and chemical properties within the deposit. Although climatic factors
such as temperature and precipitation also play a significant role in decomposition
process (Ratnayake et al., 2011), the microbial
activities is principally a physicochemical transformation process that determine
the spatial and temporal variation in the nutrient content of the deposit.
Various efforts were made to classify peat based on the variation in the physical
and chemical composition of its constituent deposit. The most generally acceptable
classification technique is based on the degree of humification known as Von
post classification scale (Klavins et al., 2008).
The Von post scale classified peat into ten successive degrees of decomposition
ranging from H1 as undecomposed to H10 as completely decomposed. The degree
of decomposition usually increases with depth up to about H7 (Huat
et al., 2011). The American Society for testing and Material (ASTM)
however, approaches peat on two bases: Degree of humification and degree of
acidity. On the bases of the degree of humification, ASTM narrowed down the
peat classification of the Von post scale into three categories: least decomposed
fabric with more than 67% fibre content, moderately decomposed hemic with fibre
content between 33-67% and highly decomposed sapric with more than 33% fibre
content (ASTM D4427-07, 2007).
Fibrous peat is characterized by high organic and fibre content, low degree
of humification (undecomposed) and highly acidic. The sapric peat on the other
hand is highly decomposed with comparatively low water retention capacity. Hemic
peat is characterized by moderate organic and fibre content (ASTM
D4427-07, 2007). ASTM D4427-07 (2007) also classified
peat according to pH scale into highly acidic (pH <4.5), moderately acidic
(pH between 4.5 and 5.5), slightly acidic (pH between 5.5 and 7.0) and basic
There is however, great difference between humid temperate peat deposit upon
which the above classifications are based and the tropical peat deposit. Wust
et al. (2003) observed that various existing peat classification
schemes failed to adequately characterize tropical peat deposit due to the great
variability in both texture and composition of the deposits. A classification
scheme uniquely for tropical organic soil was therefore developed by Wust
et al. (2003) based on ash content with particular reference to Tasek
Bera Basin peatland in the central part of Peninsular Malaysia. According to
ash based classification scheme, peat is defined as organic soil with ash content
below 35% and is classified into five classes as follows: Very low ash (ash
content between 0-5%), low ash (ash content between 5-15%), medium ash (ash
content between 15-25%), high ash (ash content 25-40%) and very high ash (ash
content between 40-55%) These can correspondingly be described in terms of organic
content as very high organic content (organic content between 95-100%), high
organic content (organic content 85-95%), medium ash content (organic content
75-85%) and low organic content (organic content 45-60%).
In view of the relevance of peat deposit to the socioeconomic development of
the communities within the study area, this study is designed to develop a model
for noninvasive delineation of the stratigraphic sequence of the deposit by
calibrating geophysical radar image with geotechnical data. The study involves
field GPR survey and laboratory analysis for the purpose of charactering detected
layers on the bases of organic content in accordance with Wusts classification.
The overall goal of the study was to develop a model for the assessment of fertility
level of the deposit for effective utilization and sustainable management of
MATERIALS AND METHODS
Theoretical background: Ground Penetrating Radar (GPR) is a near surface
geophysical tool that record the back scattered signal of the subsurface reflected
due to contrast in the electrical properties of the earth composition. The suitability
of GPR as a geophysical survey tool is strongly influenced by the electrical
and hydrogeological properties of the subsurface. Peat is characterized by relatively
low magnitude of electrical conductivity due to the presence of highly concentrated
inactive and strongly bound organic compounds. This property enables GPR to
be suitable as a subsurface surveying tool on peat deposit. The electrical and
hydrogeological properties of peat are functions of the effective properties
of various components of the aggregate deposit. There are four main components
of Southeast Asian peat namely water, air, mineral and organic contents (Wetland
International, 2010). Each of these exercises significant influence on the
electrical properties of the aggregate deposit. For instance radar signal velocity
through a material is directly related to the apparent (measured) component
of dielectric permittivity ε0 of the material according to the
where, C is the radar velocity in free space. This complex frequency-dependent electrical property of materials is a measure of polarizability of the molecular structure of materials due to the influence of external electromagnetic fields.
Being a dipolar in nature, water molecules are highly polarizable. Dielectric
permittivity of water within GPR frequency range is about 81 whereas dielectric
permittivity of most soils and materials within the same frequency range is
between 4 and 7 (Daniels, 2004). The dielectric permittivity
of air is 1. Thus the presence of water highly influenced the magnitude of this
property. Based on Eq. 1 above, higher magnitude of dielectric
permittivity implies lower radar signal velocity (Idi and
Kamarudin, 2011). On the other hand, the presence of free-phase biogenic
gas, a product of anaerobic decomposition of organic materials that are mostly
trapped within the peat deposit, having same dielectric properties as air, reduces
the overall magnitude of the dielectric permittivity and therefore enhances
the radar signal velocity (Benedetto, 2010). Thus variation
in signal velocity could be used as diagnostic for characterizing various components
of the deposit.
The range of radar signal through a material medium is governed by the total
path loss which is a function of the material loss, the spreading loss and the
target reflection or scattering loss (Daniels, 2004).
These are functions of characteristic impedance of the medium. The radar energy
attenuation rate α through a medium is related to the electrical conductivity
of the medium by the equation (Annan, 2001):
where, ω is the radar signal frequency, μ is the magnetic permeability and σ the electric conductivity of the medium.
In most materials with the exception of metallic (ferromagnetic) materials, the magnitude of the magnetic permeability is close to that of free space. Thus the attenuation rate of electromagnetic waves in a medium at a given frequency depends on the electrical conductivity of the medium.
Electrical conductivity of peat deposit is greatly influenced by the chemical
properties of the deposit such as the chemical composition, Cation Exchange
Capacity (CEC) and acidity. These properties vary with the degree of humification.
The CEC of a peat increases with increase in pH value and fibrous peat has the
largest CEC (Huat et al., 2011). The most common
exchange cations in the peat are Ca2+, Mg3+, Al3+,
K+, Na+ (NH4)+. The concentration
of these cations provides pathways for electrical conduction. Electrical conductivity
is directly related to the concentration and mobility of the cations on the
peat particles. Hence microbial activities enhance electrical conductivity which
in turn increases radar signal attenuation (McGlashan et
The chemistry of the water content due to the presence of mineral ions boosts
the conductivity of the pore water, causing the radar signal to be more strongly
attenuated based on Eq. 2. High concentration of these exchangeable
cations in fresh peat deposit enhances the dissipation of electrical energy.
Thus, signal attenuation in peat deposit decreases with increase in the level
of decomposition. Variation in EM wave parameters are therefore associated with
changes in the electrical properties and bulk density of the medium at the stratigraphic
interfaces (Gomez-Ortiz et al., 2010). Radar
parameter variation with changes in subsurface formation was used by Pati
et al. (2011) in interpreting GPR radargram for the sedimentary study
of the Indo-Gangetic plain of the Himalayas in which concealed thrust sheet
beneath the thick sediment cover of the middle Gangetic plain of India was revealed
to a maximum depth of 15 m. The radar image was acquired with GSSI model 620
GPR shielded antenna at a central frequency of 100 MHz.
Data acquisition and interpretation: The study area is a plot of peatland
located along Pontian-Pekan Nanas highway, near Kampung Batu Dua Puluh Sambian,
at Pontian district, in the state of Johor, Malaysia. The area is geographically
located at longitude 103°2749.94"E- 103°2738.88"E and latitude
1°3515.16"N-1°3508.14"N (Fig. 2). The
area is a portion of the coastal plain of southwestern Johor described by ASEAN/US
CRMP (1991) as largely underlain with marine clay, silt and the paludal
peat deposit of Holocene age.
|| Study area with the four scanned transects shown
A common offset single fold reflection profiling was used to obtain GPR cross sections along four equidistance profiles of lengths 20 m shown in Fig. 2. The profiles are 4 m apart and run in the east-west direction. The equipment used for the survey is IDS DAD fast wave radar acquisition unit at a central frequency of 200 MHz. a perpendicular polarized antenna orientation was used to scan the profile at a trace increment of 0.025.
Undisturbed peat samples were collected at depth intervals of 0.5 m from the surface to a maximum depth of 2.5 m using a 30 cm diameter cylindrical sampler. The sampled core was collected at a 10 m horizontal distance of profile 2 so that it coincides with the centre of the profile as shown in Fig. 2. The samples were carefully, sealed and parked in an appropriate container that minimize the effect of vibration during transportation.
The acquired radar image was processed with Reflexw radar image processing
software (Sandmeier, 2010). The software has the advantage
of compatibility with different GPR radar image formats and user-friendly interface
for the import, display, processing and interpretation of radar and seismic
images. The following processing steps were applied in order to enhance the
signal-to-noise ratio and improve the image quality: Subtract mean (dewow),
static correction, amplitude gain and background removal.
The performances of the applied processing techniques were evaluated by computing
the normalized root mean square error NRMSE between the raw and the processed
images. The NRMSE between a raw data f (k) and its corresponding processed data
r (k) is given by Baili et al. (2009):
where, μf is the mean value of f (k) (amplitude of the raw data at a given
scale level from k to N) and r (k) is the corresponding amplitude of the de-noised
image. The amplitudes of the raw and correspondent processed radargrams were
recorded with respect to each of the four profiles at grid intervals over a
selected point using the wiggle plot window of the processing tool. Figure
3 is the plots of the raw and processed amplitudes with respect to profile
|| Plots of the raw and correspondent processed radargrams of
|| Statistics and computed normalized root-means square error
of the processed radargrams
Thus, with relatively lower values of NRMSE in all the profiles, the processing techniques are effective in preserving the quality of information in the images. The performance is however relatively higher with respect to profile 1 in which the lowest value of NRMSE is recorded. Based on the descriptive statistics of the raw and processed images shown in Table 1, the mean of the raw and corresponding processed images are fairly closed while the standard deviations are relatively small in magnitudes. These imply that the images were fairly recovered after processing. There is however slight variation in both cases with the standard deviations recording greater variation. Based on the obtained statistics however, the overall performance of the processing techniques is adequately satisfactory.
The processed radar images are shown in Fig. 4a-d.
Physical observation of the collected samples shows that the top layer to a
depth of 0.7 m consist of water saturated high fibre content peat overlaying
the water table. The water table was encountered at a depth of 0.7 m. At 2.5
m depth, a highly saturated whitish clay, identified as kaolinite clay was encountered.
Comparison of the sampled core data collected with the control radargram (Fig.
4b) shows that the top layer above the water table corresponds the region
of strong signal activity that remains after time-shifting the data to ground
level. Water table encountered at the depth of 0.7 m corresponds to the end
of this strong signal on the radargram. This implies that the signals encountered
strong attenuation at the water interface due to higher electrical conductivity
of the water: soil mixture. The whitish clay layer at 2.5 m depth corresponds
to a strong nearly-horizontal signal activity in the radargram. It is therefore
believed that the near-horizontal activity at this point, which roughly appears
in all the radargrams around the central portion, is associated with the kaolinite
Radar signal velocities were estimated using hyperbolic velocity analysis.
Various reflection hyperbolas present within the radargram were fit with a mathematical
velocity model and the value of the best fitting hyperbola was recorded as the
respective layer velocity. Figure 5 is shows the adopted hyperbolas
and velocity values with respect to profile 1.
||Processed radar image of GPR cross section acquired along
profile (a) 1, (b) 2 with sampled core data interpolated, (c) 3 and (d)
Variation in the subsurface stratigraphy of the peat is detected from the processed radargrams based on the strength of the phase and velocity changes at the layer interfaces. The layer interfaces in the subsurface are detectable by the virtue of significant contrast in velocity and attenuation rate of the radar signal. The interfaces are therefore picked using layer picking procedure which leads to the extraction of the depth and topography of the layer interface. Arrivals are picked based on significant variation in layer velocity and attenuation rate. The picking option utilizes the velocity information saved on the 2-D velocity adaptation file to create a cross sectional profile.
There are three layer picking options: manual, continuous and semi automatic (phase follower) pickings. After several trials with both options, it has been observed that the three could be used to complement each other for optimality. Thus semi automatic picking was initially used to trace all possible layers. Manual picking was then used to trace the most significant layer(s). Figure 6 is a window snapshot of the layer picking interface with respect to profile 1. Frequency-wave number (f-k) migration was next applied to the picked layers on the bases of the velocity information. f-k migration collapses the reflection energies to the location of their sources. These reposition events to their true spatial locations.
Laboratory determination of sampled organic content: Loss on Ignition
(LOI) experiment was conducted to determine the ash content (and hence organic
content) of the representative samples in order to estimate the organic content
of the various detected layers.
|| Adapted reflection hyperbolas of profile 1 showing the adopted
velocities (m nsec-1)
|| Profile 1 layer picking interface
The LOI technique was used by Handayani et al. (2010)
in estimating the organic content of soil in a study for the evaluation of the
influence of tall fescue management on soil organic matter fraction. The technique
was also used by Gasim et al. (2011) in estimating
the organic content of four selected soil series in Tasik Chini, Pahang, Malaysia
with the aim of analyzing their physicochemical properties.
American Society for Testing and Materials (ASTM) standard procedure for the
LOI determination was used (ASTM D2974-07, 2007). According
to the standard, the fractional organic content OC of peat sample is given as:
where, Ma is the mass of ash remains after subjecting the sample to excessive heating and Md is the mass of oven dried specimen. The ratio Ma/Md is the measure of the samples ash content. The experimental procedure involves oven drying and measurement of the mass of the oven dried specimen. The specimen was then ashed in a furnace at a temperature of 440°C to a constant mass. Two sets of specimens from each of the six depth samples were prepared for the experiment.
Based on the layer picking output, four stratigraphic layers were detected in all the profiles within the depth of coverage. Figure 7 shows the detected layers picked with respect to profile 1 after the application of f-k migration. The layer cross sections are given in terms of velocity variation obtained from the hyperbolic fitting as shown in the scale bar. It has been observed that the top layer is detected as high velocity layer compared to underlying layer in all the cross sections. This could not be unrelated to the fact that the layer is associated with relatively low water content owing to its position. The layer is overlaying a relatively lower velocity zones whose upper level coincides with the water table.
The zone is made up of two layers of slightly varying signal velocity as shown.
The upper layer has relatively higher signal velocity than the underlying layer
in all the profiles. Below this horizon is a continuous layer of maximum velocity
at all the profiles. This implies that the stratigraphic sequences are nearly
consistent in all the scanned profiles. Table 2 gives the
computed values of the ash and organic contents obtained from the six sampled
||Cross sectional transect showing the stratigraphic reflection
events of profile 1. Four layers of varying reflection strengths are detected
as shown. The second and third layers have minimal variation in signal strength
|| Experimental results of the obtained ash and organic contents
|| Comparison of the measured organic content with detected
The results show that the top soil has the least organic content. The two underlying
layers under waterlogged condition have nearly the same organic content with
values ranging from 0.9097-0.9878. Thus there is no significant variation in
the level of decomposition between the two layers. Underlying these layers is
however, a significantly lower organic content layer (0.7973) which corresponds
to highest velocity layer (Fig. 8) from a depth of 2.5 m at
the point of coring. The layer appears in all the profiles at varying depths
and extents to the maximum depth limit of the signal. The layer is also characterized
by relatively lower organic content. Thus, the whitish kaolinite clay which
begins at a depth of 2.5 m is associated with less fibre and water content.
Comparison of the velocity-based stratigraphic layers in all the profiles with
the results of the laboratory measurement of organic content shows that the
top soil, having the least organic is characterized by high ash content. The
two underlying layers have nearly the same constituents they are therefore interpreted
as single stratigraphic unit. The underlying layer that appears in all the profiles
as high velocity layer characterized by relatively lower organic content corresponds
to the kaolinite clay deposit, which is abundant in tropical peat soil (Wust
et al., 2002). Even though it is also organic soil, the kaolinite
clay at this point is considered to be at a transition point between the peat
deposit and a kaolinite clay because of its low organic and fiber content and
comparatively higher signal velocity. A numerical organic content of 0.797311
(79.7311%) was recorded at this interface. This implies that the peat deposit
in the entre study area has a maximum thickness of about 2.5 m.
Peatland deposit of western state of Johor was generally categorized in terms
of depth, into three categories (Wetland International, 2010):
Shallow (less than 1.5 m deep), moderate (1.5- 3.0 m deep) and deep (more than
3.0 m deep). Although, the deep deposit has the largest land coverage (about
62%) according to Wetland International, a maximum depth of 2.5 m recorded in
this study showed that the area under study is within the moderate depth category.
The findings of this study is also fairly within the thickness range of Johor
state peat deposit given by ASEAN/US CRMP (1991) where
the variation in the deposits thickness was given as 6-2.5 m from the
west to the south-eastern coast of the state, respectively.
According to Wusts ash content-based classification system, the top layer
with an average ash content of 33.34% is classified as high ash content. The
ash content reduces to about 9.0256% at a depth level of 0.5 m probably on transition
to lesser ash content peat. This is in line with Wust et
al. (2003)s observation of upward trend increase in ash content
toward the surface in a laboratory analysis of peat samples collected at Tasek
Bera basin, central part of Peninsula Malaysia. A mean ash content of 0.0374
(3.74%) was recorded between the depth of 0.5-2.0 m with a standard deviation
of 0.035643. The two layers within this depth range are therefore considered
to be of the same constituent and interpreted as very low ash content peat.
Underlain this deposit is the kaolinite organic clay soil with ash content of
Analyzing the ash classification scale relative to ASTM humification in accordance
to Wust et al. (2003)s comparison indicates
that the top layer corresponds to fabric-to-hemic decomposition levels. This
implies that the deposit at this horizon ranges from very slight to moderately
decompose with a von post scale ranges from H1-H6. The results of the organic
content analysis indicates that the two underlying layers within the depth range
of 0.5-2.0 m are fairly homogeneous since the variation in both water content
(signal velocity) and the organic content do not exceed 7.9%. The ash content
range of 1.22-9.03% on the Wusts scale also corresponds to the range of
fibric-to-hemic with a von post scale range of H1-H6. Based on the relative
positions of the layers however, the top layer is considered fibric while the
underlain layer is hemic under waterlogged condition (Wust
et al., 2002). The lower layer ranges from 2.5 m depths downward
interpreted as kaolinite clay may not continue as peat because of the abrupt
loss of organic and fiber contents with depth. The layer interface is therefore
considered to have marked the end of peat deposit at the study area. This finding
is fairly similar to the result of several tests conducted by Zainorabidin
and Bakar (2003) which revealed that the entire Johor peat deposit is generally
hemic with organic content range of 80-96% (ash content range of 4-20%). The
variation between the two results could be attributed to the great spatial variability
in physical properties of peat. It has been observed that peat soil is highly
variable on its physical properties not only from region to region but also
from point to point within a region.
The interpreted stratigraphic sequences for the four profiles are depicted
in three 3-D fence plot of profiles shown in Fig. 9. Three
major stratigraphic layers are identified within the depth range of the study.
The first two are interpreted as fibrous and hemic peats. The layers are found
to be fairly undulating but they all approximate horizontal beddings. Horizontal
bedding peat deposit usually occur when peat is accumulated in still water environment
(Xuehui and Jinnming, 2009). This implies that the hydrology
of the environment was initially characterized by still water. Being least decomposed,
the upper layer also has the highest concentration of cation exchange capacity
(Huat et al., 2011) and is therefore the richest
in acidity and nutrient content.
||3-D fence plot of the stratigraphic sequence of the four profiles
based on the interpretation of the reflection events. All the four profiles
show high ash content fabric peat overlaying lower ash content hemic peat.
Beneath is the kaolinite clay
Thus stratigraphic and humification data provide vital information needed for
effective utilization and management of peat resources.
Stratigraphic sequences of peat deposit is controlled by a number of factors such as climate and availability of nutrients. Availability of nutrients on one hand depends on the type and part of the peat-forming plant material at given location. Plant species differ in decomposition rate depending on the physiological properties of the species. On the other hand, different parts of plant materials also decompose at different rate due to the variation in fiber content. These factors inhibit linear decomposition of undisturbed peat deposit with depth especially in a larger peatland. Thus each of the stratigraphic unit is considered a unique horizon of nearly the same constituent and interpreted as stratigraphic layers of varying levels of humification and nutrient content based on the laboratory analysis of core samples collected within the depth range.
The stratigraphic sequence of peat is similar to sedimentary series of earth layers in which each layer provides information about the prevailing condition of the deposit at the time of its formation. Qualitative interpretation of the detected layers in terms of subsurface composition however requires experimental calibration. In this study, we developed the calibration data experimentally and used it to interpret GPR radar images. The study therefore provides a model for the qualitative interpretation of the southwest Malaysian peatland on the bases of the organic content of the detected layers. Thus, GPR could serve as a tool for the monitoring and assessment of nutrient level and hydrological history of the peat deposit. It is therefore hoped that the results of this study will serve as a reference for surface radar surveying of the peatland for the effective monitoring of the fertility level and hydrological pattern of the deposit.
The authors acknowledged with thanks the contribution of Jurukur Abadi of No. 06-01, Jalan Padi Emas 4/5, Pusat Bandar Tampoi, 81200, Johor Bahru for the test data acquisition.