Content uploaded by Shijo Joseph
Author content
All content in this area was uploaded by Shijo Joseph on Dec 26, 2013
Content may be subject to copyright.
PLEASE SCROLL DOWN FOR ARTICLE
This article was downloaded by:
[Joseph, Shijo]
On:
24 November 2010
Access details:
Access Details: [subscription number 930113869]
Publisher
Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-
41 Mortimer Street, London W1T 3JH, UK
International Journal of Sustainable Development & World Ecology
Publication details, including instructions for authors and subscription information:
http://www.informaworld.com/smpp/title~content=t908394088
Spatial interpolation of carbon stock: a case study from the Western Ghats
biodiversity hotspot, India
Shijo Josephab; Ch. Sudhakar Reddyc; A. P. Thomasa; S. K. Srivastavad; V. K. Srivastavae
a School of Environmental Sciences, Mahatma Gandhi University, Kottayam, Kerala, India b
Department of Natural Resources, International Institute for Geo-Information Science and Earth
Observation (ITC), Enschede, The Netherlands c Forestry and Ecology Division, National Remote
Sensing Centre, Indian Space Research Organization, Hyderabad, Andhra Pradesh, India d Tamil Nadu
Forest Department, Geomatics Centre, Chennai, Tamil Nadu, India e Land Resources Group, National
Remote Sensing Centre, Indian Space Research Organization, Hyderabad, Andhra Pradesh, India
Online publication date: 24 November 2010
To cite this Article Joseph, Shijo , Sudhakar Reddy, Ch. , Thomas, A. P. , Srivastava, S. K. and Srivastava, V. K.(2010)
'Spatial interpolation of carbon stock: a case study from the Western Ghats biodiversity hotspot, India', International
Journal of Sustainable Development & World Ecology, 17: 6, 481 — 486
To link to this Article: DOI: 10.1080/13504509.2010.516107
URL: http://dx.doi.org/10.1080/13504509.2010.516107
Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf
This article may be used for research, teaching and private study purposes. Any substantial or
systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or
distribution in any form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses
should be independently verified with primary sources. The publisher shall not be liable for any loss,
actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly
or indirectly in connection with or arising out of the use of this material.
International Journal of Sustainable Development & World Ecology
Vol. 17, No. 6, December 2010, 481–486
Spatial interpolation of carbon stock: a case study from the Western Ghats biodiversity
hotspot, India
Shijo Josepha,b∗, Ch. Sudhakar Reddyc, A.P. Thomasa,S.K.Srivastava
dand V.K. Srivastavae
aSchool of Environmental Sciences, Mahatma Gandhi University, Kottayam, Kerala, India; bDepartment of Natural Resources,
International Institute for Geo-Information Science and Earth Observation (ITC), Enschede, The Netherlands; cForestry and Ecology
Division, National Remote Sensing Centre, Indian Space Research Organization, Hyderabad, Andhra Pradesh, India; dTamil Nadu Forest
Department, Geomatics Centre, Chennai, Tamil Nadu, India; eLand Resources Group, National Remote Sensing Centre, Indian Space
Research Organization, Hyderabad, Andhra Pradesh, India
Carbon stock distribution in different tropical forest types in India is rarely studied although India is a country with
mega-diversity. The present study estimates the biomass and carbon stock of major tropical forest types in India, and
attempts to identify suitable interpolation techniques for mapping carbon stock. Empirically derived allometric equations
and carbon conversion coefficients were used to estimate the aboveground biomass and carbon stock, respectively. The point
estimates were interpolated to spatial surface using different interpolation techniques. Two main modelling approaches were
implemented: deterministic modelling and stochastic modelling. Deterministic modelling was to interpolate point informa-
tion using similarities between measured points (inverse distance weighted (IDW) interpolation), and fitting a smoothing
curve along the measured points (polynomial interpolation). In stochastic modelling, ordinary kriging (OK) was employed
using parameters derived from semivariograms. The results showed that the average carbon stock in the study area was
84 t/ha. The highest carbon stock was in evergreen forest and the lowest in thorny scrub forest. Validation of the model
using the mean and RMS errors indicated that ordinary kriging performs better than IDW and polynomial interpolations.
Keywords: aboveground biomass; carbon stock; geostatistics; Anamalai; India
Introduction
Forest ecosystems are the major biological ‘scrubbers’
of atmospheric carbon dioxide among terrestrial biomes.
They remove nearly 3 billion tons of anthropogenic carbon
every year (3 Pg C/year) through net growth, absorb-
ing about 30% of all emissions from fossil fuel burning
and net deforestation (Canadell et al. 2007; Canadell and
Raupach 2008). The main carbon pools in tropical forest
ecosystems are the living biomass of trees, understorey
vegetation, dead mass of litter, woody debris and soil
organic matter. The carbon stored in the aboveground liv-
ing biomass of trees is typically the largest pool, more than
double the amount of carbon in the atmosphere (Sabine
et al. 2004; FAO 2006), and the most directly impacted
by degradation and deforestation. Thus, estimating above-
ground forest biomass carbon is a critical step in the global
carbon cycle.
The most direct way to quantify the carbon stored in
aboveground living forest biomass is to harvest all trees
in a known area, weigh the biomass and then multiply
by a standard value of carbon concentration to produce
an estimate of carbon stock (Kauppi et al. 1992; Goodale
et al. 2002). While this method is accurate for a particu-
lar location, it is prohibitively time-consuming, expensive
and destructive. Tropical forests often contain 300 or more
species, and research has shown that species-specific allo-
metric relationships are not needed to generate reliable
∗Corresponding author. Email: shijonrsa@gmail.com
estimates of forest carbon stocks. Instead, generalised allo-
metric equations, stratified by broad forest types are found
to be highly effective for the tropics because the DBH
(diameter at breast height) alone explains more than 95%
of the variation in aboveground tropical forest carbon stock
(Brown 2002). A remaining problem with this approach is
that stock is measured at single sites, and is not readily
scaled up.
The methods used for scaling up forest biomass can
be classified into two groups: estimating biomass from
remote sensing data (Foody et al. 2003; Broadbent et al.
2008; Anaya et al. 2009; Fuchs et al. 2009), and interpo-
lation of biomass estimates obtained in the field (Cheng
et al. 2007; Sales et al. 2007; Lufafa et al. 2008). Although
there are several studies in the second category, very few of
them have compared the relative advantages of the various
interpolation techniques.
India has a geographical area of 328 Mha, of which
51 Mha (million hectare) are under tropical forest (FSI
2003). Based on a limited number of studies, it is esti-
mated that carbon storage capacity of tropical forests is
between 1.9 and 4.1 Gt C. Two main approaches have been
taken to reach this conclusion. Using phytomass carbon
densities based on ecological studies and remote sensing
of forest areas, estimates forest phytomass C pool is 2.5 to
4.1 Gt C (Ravindranath et al. 1997). Using field inventory
of growing stock volume and biomass expansion factors
ISSN 1350-4509 print/ISSN 1745-2627 online
© 2010 Taylor & Francis
DOI: 10.1080/13504509.2010.516107
http://www.informaworld.com
Downloaded By: [Joseph, Shijo] At: 04:41 24 November 2010
482 S. Joseph et al.
relating wood volume to biomass, forest phytomass C pool
was estimated as 1.9–4.0 Gt C (Dadhwal and Nayak 1993;
Dadhwal and Shah 1997). The estimated carbon densities
per hectare of forest phytomass are mostly in the range
of 50–68 tons (Chhabra and Dadhwal 2004). Apart from
these gross estimates, there are no published data on carbon
stocks in tropical forest types in India.
India is presently a non-Annex I country according
to the Kyoto Protocol of the United Nations Framework
Convention on Climate Change (UNFCCC) and is exempt
from binding greenhouse gas (GHG) emissions targets.
However, India’s status could be revised as it makes the
transition into a developed economy. Therefore, it is essen-
tial to develop a comprehensive database on biomass
carbon stock distribution in different tropical forest types
in India. Hence the objectives of the present study are
twofold: to estimate biomass and carbon stock density
of major tropical forest types in India and to identify
the best interpolation technique for mapping the car-
bon stock under diverse climatic, topographic and biotic
gradients.
Spatial interpolation
Interpolation is a method to estimate the value of an
unknown point from a number of known observations
around that point (Myers 1994). There are two main groups
of interpolation techniques: deterministic and geostatisti-
cal. Deterministic interpolation techniques create surfaces
from measured points, based on either the extent of simi-
larity (e.g. inverse distance weighted average) or the degree
of smoothing (e.g. polynomial interpolation), while geosta-
tistical interpolation utilises the statistical properties of the
measured points (ordinary kriging).
Inverse distance weighted (IDW) interpolation
IDW interpolation explicitly implements the assumption
that things that are close to one another are more alike than
those that are farther apart. It assumes that each measured
point has a local influence that diminishes with distance. It
weights the points closer to the prediction location higher
than those farther away. The formula for IDW is:
ˆ
Z(s0)=
N
i=1
λiZ(si), (1)
where ˆ
Z(s0) is the value to be predicted for location S0,
Nis the number of measured sample points surrounding
the prediction location that will be used in the prediction,
λiare the weights assigned to each measured point that are
going to be used, Z(si)is the observed value at location Si.
The formula to determine the weight is:
λi=d−p
io
N
i=1
d−p
io
N
i=1
λi=1. (2)
As the distance becomes larger, the weight is reduced
by a factor of p. The quantity dio is the distance between
the prediction location Soand each of the measured loca-
tions Si.
Polynomial interpolation
Polynomial interpolation fits a smooth surface that is
defined by a mathematical function (a polynomial) to the
input sample points. For the first order trend, the model is:
Z(xi,yi)=β0+β1xi+β2yi+ε(xi,yi), (3)
where z(xi,yi) is the datum at location (xi,yi), βiare param-
eters and ε(xi,yi) is a random error.
Ordinary kriging
Ordinary kriging is a geostatistical technique similar to
IDW in that it weights the surrounding measured values to
derive a prediction for each location. However, the weights
are based not only on the distance between the measured
points and the prediction location but also on the over-
all spatial arrangement among the measured points. The
unknown value Z(s) at any location is typically decom-
posed into a mean (drift) component μand a residual
component ε(s) (Equation 4):
Z(s)=μ+ε(s). (4)
The predictor is formed as a weighted sum of the data
(Equation 5):
ˆ
Z(s0)=
N
i=1
λiZ(si). (5)
This equation is very similar to IDW interpolation.
However, in IDW, the weight, γidepends solely on the
distance to the prediction location. In ordinary kriging,
the weight, γi, depends on the semivariogram, the dis-
tance to the prediction location and the spatial relationships
among the measured values around the prediction location.
The semi-variance is defined as a function of the distances
among observations.
Materials and methods
Study area
The area selected for the study is the Anamalai Hills in the
Western Ghats biodiversity hotspot, where the ecological
setting is a representation of the diverse climatic and topo-
graphic gradients existing in peninsular India (Figure 1).
The annual rainfall varies from 500 mm in the rain shadow
eastern slopes to 5000 mm in the west. Mean daily tem-
perature varies from <5◦C in winter at elevations above
2000 m to nearly 40◦C in the eastern plain in summer. The
overall terrain is hilly, with altitudes ranging from 250 m
Downloaded By: [Joseph, Shijo] At: 04:41 24 November 2010
International Journal of Sustainable Development & World Ecology 483
10°35′0′′N
76°50′0′′E76°55′0′′E77°0′0′′E77°5′0′′E77°10′0′′E77°15′0′′E77°20′0′′E
76°50′0′′E
036 12
Km
76°55′0′′E77°0′0′′E77°5′0′′E77°10′0′′E77°15′0′′E77°20′0′′E
W
N
E
S
10°30′0′′N
10°25′0′′N
10°20′0′′N
10°15′0′′N
10°35′0′′N
10°30′0′′N
10°25′0′′N
10°20′0′′N
10°15′0′′N
Anamalai Wildlife Sanctuary
Figure 1. Location map of Anamalai Wildlife Sanctuary in Western Ghats, India.
in the foothills of the northeast to 2500 m in the Grass
Hills area in the southwest. Due to varied topography and
microclimatic regimes, this area is considered as one of 25
microcentres of diversity in the Indian subcontinent (Nayar
1996). Floristically, the western side of the hills is occupied
by luxuriant rain forest (humid forest), while the eastern
side is dominated by dry forests.
Methods
A stratified transect survey was conducted in the Anamalai
Hills during 2005–2006. The strata were delineated tak-
ing into consideration long-term climatic observations and
altitudinal variation within the study area. Circular plots of
10-m radius (plot size – 314 m2) were laid every 200 m
along transects. Care was taken to include all vegetation
types in the sampling. In each plot, all woody plants with
≥10 cm DBH were identified to species level, individ-
uals counted and DBH measured with a tape. Biomass
was calculated using allometric equations developed for
the Western Ghats (Murali et al. 2005). Two allometric
equations were used: one for evergreen forests (Equation 6)
and the other was for deciduous forests (Equation 7). The
basal area required for these equations was calculated from
DBH using Equation (8). Carbon conversion coefficients
differ, considering species, age, formation and commu-
nity structure of vegetation types, and range from 0.45 to
0.55 (Kauppi et al. 1992; Goodale et al. 2002; Xia et al.
2005; Ramachandran et al. 2008). Since such coefficients
were not available for the study area, a carbon conversion
coefficient of 0.5 was used in the present study. Forest
carbon storage of each vegetation type was estimated by
multiplying the forest carbon density per hectare by the
extent of forest area.
Forest area statistics are derived from the supervised
classification of IRS P6 LISS III satellite data (Indian
remote sensing satellite). Based on knowledge of the data
and ground truth information, nine different land-cover
classes were identified. Parametric signatures were used to
train a statistically based (e.g. mean and covariance matrix)
classifier to define the classes. Maximum likelihood para-
metric rule were implemented for classifying the data.
An accuracy assessment was performed by comparing the
classified image with the ground truth reference data.
Biomass evergreen
=(−2.81 +6.78 ∗Basal area) (r2=0.53)
Biomass deciduous
=(−73.55 +10.73 ∗Basal area) (r2=0.82)
Basal area =(DBH)2
4π
.(6)
Spatial interpolation was conducted on the data to
quantify the patterns of spatial variation in carbon stock.
From the total sample points (206 sample points), a sub-
set of 155 sample points (75%) was taken randomly for
model generation (training datasets) and the remaining
Downloaded By: [Joseph, Shijo] At: 04:41 24 November 2010
484 S. Joseph et al.
51 sample points (25%) were used for model valida-
tion (testing datasets). Three interpolation techniques were
applied. Inverse distance weighting was the first technique
implemented. The power value for IDW is optimised by
considering the root mean square (RMS) error. At least
10 neighbouring points were included in the prediction of
an unknown point. The second method involved fitting a
first order polynomial equation (local polynomial interpo-
lation) to the sample points. The selection of the first order
equation was based on the least RMS error. At least 10
neighbouring points were included, as in the case of IDW.
Ordinary kriging was the third method implemented for
interpolation of training datasets. The changes in semivari-
ance with distance was analysed using different models,
such as circular, spherical, exponential and Gaussian, and
the spherical model was found to be the best choice since
it gave the lowest RMS error. The other inputs (nugget,
sill and range) were derived from this semivariogram. The
nugget represents measurement/independent error and is
the deviation from zero on the y-axis. The sill is the height
that the semivariogram reaches when it levels off, located
on the y-axis, and the range is the distance where the model
first flattens out on the x-axis. The interpolated surface
and error surface were derived from the analysis. The test
data taken from the total sample plots was used for valida-
tion. The selection of best interpolation method was based
on the mean prediction error (Equation 9) and root mean
square prediction error (Equation 10).
Mean error (ME) =
n
i=1
(ˆ
Z(si)−z(si))
n
Root mean square error (RMSE)
=
n
i=1
[ˆ
Z(si)−z(si)] 2
n,(7)
where ˆ
Z(si) is the predicted value and Z(si) is the observed
value.
Results
The average values of the aboveground biomass and car-
bon stock in the whole of Anamalai Wildlife Sanctuary are
167 t/ha and 84 t/ha, respectively. Vegetation and land-
cover type mapping using the IRS P6 LISS III data showed
Table 1. Area statistics of major land-cover types and their
classification accuracy in Indira Gandhi Wildlife Sanctuary, India
(area statistics derived from the IRS P6 LISS III data of 28 March
2006).
Sl.
No. Land-cover type
Area
(km2)
Area
(%)
Producer
accuracy
(%)
User
accuracy
(%)
1 Tropical evergreen
forest
230.2 23.8 87.5 89.4
2 Montane wet
temperate forest
21.8 2.2 66.7 83.3
3 Tropical deciduous
forest
487.6 50.3 90.0 75.9
4 Thorn scrub forest 34.3 3.5 58.8 76.9
5 Grassland 85.8 8.9 70.0 95.5
6 Teak plantation 31.3 3.2 75.0 69.2
7 Tea plantation 30.2 3.1 100.0 84.6
8 Agriculture and
fallow lands
36.5 3.8 84.6 91.7
9 Water bodies 10.9 1.1 100.0 90.0
Total area 968.6 100.0
the area is dominated by deciduous forest (487.6 km2)
and evergreen forest (230.2 km2). The nine land-cover
classes, their area statistics and estimated classification
accuracy are given in Table 1. Pristine grasslands, a par-
ticular feature of the hills, covered 86 km2. In addition
to natural vegetation types, the other land-cover types in
the area are plantations, agriculture areas and reservoirs,
which together contribute 108.9 km2(11.2%). Overall
classification accuracy was 83% and the Kappa statistic
was 0.79. The lowest producer accuracy is observed for
tropical thorny scrub forest (58.8%), followed by montane
wet temperate forest (66.7%). In the case of user accu-
racy, the lowest is reported for teak plantations (69.2%),
followed by tropical deciduous forest (75.9%).
Distribution of aboveground biomass and carbon stock
in different forest types is given in Table 2. Evergreen for-
est had a high amount of biomass (236.8 t/ha) whereas
thorny scrub had a low biomass (32.23 t/ha). Summation
of carbon stock in different forest types yielded 6.44 Mt of
carbon. These forest types covered 80% of the total sanc-
tuary area. The remaining area is not considered in the
estimation of carbon stock due to the absence of sample
points in those land-cover types.
Kriging was found to be the best method for spatial
interpolation of carbon stock, in comparison with IDW and
polynomial interpolation. The lowest mean error and root
Table 2. Distribution of aboveground biomass (AGB) and carbon stock in different forest types in Anamalai Hills,
Western Ghats, India.
Forest type
Average
AGB (t /ha)
Average
aboveground
carbon stock
Area
(ha)
Tot a l AGB
(Mt)
Total aboveground
carbon stock (Mt)
Tropical evergreen forest 236.80 118.4 23020 5.45 2.73
Montane wet temperate forest 187.78 93.89 2180 0.41 0.20
Tropical deciduous forest 141.69 70.85 48760 6.91 3.45
Thorn scrub forest 32.23 16.12 3430 0.11 0.06
Downloaded By: [Joseph, Shijo] At: 04:41 24 November 2010
International Journal of Sustainable Development & World Ecology 485
Table 3. Mean error and root mean square error of interpola-
tions methods.
Validation
Interpolation method Mean error RMS error
IDW −0.194 1.31
Local polynomial −0.184 1.33
Kriging −0.144 1.29
mean square error was observed for kriging in the valida-
tion (Table 3). The predicted surface using kriging is given
in Figure 2. The use of locally fitted first-order polynomial
equations was found to be the second best method.
Discussion
The tropical evergreen forest had the high carbon stock per
hectare, which could be attributed to the high stand density
of this forest. The stand density of montane wet temperate
forest (shola) was higher than that of wet evergreen forest,
but its stunted growth in the extreme environmental condi-
tions (high elevation and high wind) resulted in a low DBH.
Therefore, the allometric equation entirely based on DBH
calculated a lower biomass and carbon stock per hectare for
shola forest. The stand density in the thorny scrub forest
was extremely low, which resulted in the lowest carbon
stock contribution to the total carbon stock. Although the
carbon stock per hectare of deciduous forest was lower in
comparison with evergreen forest, most of the area (∼50%)
is covered by deciduous forest, which resulted in a high
carbon stock contribution to the total carbon stock in the
sanctuary. Overall, the average carbon stock in all these
forest types was 84 t/ha, which is slightly higher than lit-
erature values. For example, Chhabra and Dadhwal (2004)
estimated carbon stock density of 50–68 t/ha.
The performance of kriging was better than IDW and
first order polynomial equation methods, however, the
maximum value reported in kriging was 5.29 t/ha, which is
lower than some of the sample plot values. This is because
kriging is not an exact interpolator, and the surface gen-
erated will be a function of spatial relationships among
the measured values (semivariogram). In the case of IDW
and local polynomial interpolation, the values varied from
a minimum sample plot value of 0.13 t/ha to a maxi-
mum sample plot value of 7.86 t/ha, since both techniques
are exact interpolators; however, the prediction error was
higher in both of these methods. Another problem noticed
in IDW was that it created ‘bulls eyes’ around the mea-
sured locations. Although there are few examples of direct
comparisons of deterministic and geostatistical models in
the literature, many studies have used kriging for spatial
interpolation. For example, Cheng et al. (2007) used krig-
ing for spatial mapping of aboveground biomass, nitrogen
and phosphorus in the degraded Ordos Plateau grasslands
of northwest China. Askin and Kizilkaya (2007) conducted
a similar study for assessing the spatial distribution pattern
of soil microbial biomass carbon in pasture ecosystems
in Turkey.
Conclusion
Assessment of aboveground biomass and carbon stock
yielded valuable information on their densities in differ-
ent tropical forest types in India. The highest biomass
was in tropical wet evergreen forest, while the lowest
048 16 Km
Legend
Ordinary kriging
WE
N
S
High : 5.29
Low : 0
Figure 2. Spatially interpolated surface of carbon stock in Anamalai Hills using ordinary kriging.
Downloaded By: [Joseph, Shijo] At: 04:41 24 November 2010
486 S. Joseph et al.
was in thorny scrub forest. The spatial interpolation of
point estimates using different techniques showed that
ordinary kriging performed better than polynomial and
inverse distance weighted interpolations. The error anal-
ysis using the test data showed that mean and root mean
square errors are lower for kriging interpolation. Spherical
search model in ordinary kriging was found to be more
suitable to the varied climatic and topographic gradients
than models such as circular, spherical, exponential and
Gaussian.
Acknowledgements
The authors sincerely thank M.S.R. Murthy, Head of the Forestry
and Ecology Division; R.S. Dwivedi, Group Head of the Land
Resources Group; P.S. Roy, Deputy Director (RS&GIS-AA)
and Director of the National Remote Sensing Centre; and
V.B. Mathur, Dean and Principal Investigator of the WII-NNRMS
Project, Wildlife Institute of India for their encouragement;
and the Ministry of Environment and Forests, Government of
India, for funding support. Tamil Nadu Forest Department kindly
granted permission to carry out fieldwork.
References
Anaya JA, Chuvieco E, Palacios-Orueta A. 2009. Aboveground
biomass assessment in Colombia: A remote sensing
approach. For Ecol Manage. 257:1237–1246.
Askin T, Kizilkaya R. 2007. Spatial distribution patterns of soil
microbial biomass carbon within pasture. Agri Conspec Sci.
72:75–79.
Broadbent EN, Asner GP, Peña-Claros M, Palace M, Soriano M.
2008. Spatial partitioning of biomass and diversity in a
lowland Bolivian forest: linking field and remote sensing
measurements. For Ecol Manage. 255:2602–2616.
Brown S. 2002. Measuring carbon in forests: current status and
future challenges. Environ Pollut. 116:363–372.
Canadell JG, Le QC, Raupach MR, Field CB, Buitenhuis
ET, Ciais P, Conway TJ, Gillett NP, Houghton RA,
Marland G. 2007. Contributions to accelerating atmospheric
CO2growth from economic activity, carbon intensity, and
efficiency of natural sinks. Proc Natl Acad Sci USA.
104:18866–18870.
Canadell JG, Raupach MR. 2008. Managing forests for climate
change mitigation. Science. 320:1456–1457.
Cheng X, An S, Chen J, Li B, Liu Y, Liu S. 2007. Spatial rela-
tionships among species, above-ground biomass, N, and P in
degraded grasslands in Ordos Plateau, northwestern China. J
Arid Environ. 68:652–667.
Chhabra A, Dadhwal VK. 2004. Assessment of major pools
and fluxes of carbon in Indian forests. Clim Change. 64:
341–360.
Dadhwal VK, Nayak SR. 1993. A preliminary estimate of the
biogeochemical cycle of carbon for India. Sci Cult. 59:9–13.
Dadhwal VK, Shah AK. 1997. Recent changes (1982–1991) in
forest phytomass carbon pool in India, estimated using grow-
ing stock and remote sensing-based forest inventories. J Trop
For. 13:182–188.
Food and Agriculture Organization. 2006. Global forest resource
assessment 2005. Rome: Food and Agriculture Organization
of the United Nations.
Foody, GM, Boyd DS, Cutler MEJ. 2003. Predictive relations
of tropical forest biomass from Landsat TM data and
their transferability between regions. Remote Sens Environ.
85:463–474.
Forest Survey of India. 2003. State of forests report 2003. Forest
survey of India. Dehradun, India: Government of India.
Fuchs H, Magdon P, Kleinn C, Flessa H. 2009. Estimating above-
ground carbon in a catchment of the Siberian forest tundra:
combining satellite imagery and field inventory. Remote Sens
Environ. 113:518–531.
Goodale CL, Apps MJ, Birdsey RA, Field CB, Heath LS,
Houghton RA, Jenkins JC, Kohlmaier GH, Kurz W, Liu S,
et al. 2002. Forest carbon sinks in the northern hemisphere.
Ecol Appl. 12:891–899.
Kauppi PE, Mielikainen K, Kuusela K. 1992. Biomass and carbon
budget of European forests, 1971–1990. Science. 256:70–74.
Lufafa A, Diedhiou I, Samba SAN, Sene M, Khouma M,
Kizito F, Dick RP, Dossa E, Noller JS. 2008. Carbon stocks
and patterns in native shrub communities of Senegal’s Peanut
Basin. Geoderma. 146:75–82.
Murali KS, Bhat DM, Ravindranath NH. 2005. Biomass estima-
tion equations for tropical deciduous and evergreen forests.
Int J Agric Res Governance Ecol. 4:81–92.
Myers DE. 1994. Spatial interpolation: an overview. Geoderma.
62:17–28.
Nayar MP. 1996. Hotspots of endemic plants of India. India:
Tropical Botanic Garden Research Institute.
Ramachandran A, Jayakumar S, Haroon RM, Bhaskaran A,
Arockiasamy DI. 2007. Carbon sequestration: estimation of
carbon stock in natural forests using geospatial technol-
ogy in the Eastern Ghats of Tamil Nadu, India. Curr Sci.
92:323–331.
Ravindranath NH, Somashekhar BS, Gadgil M. 1997. Carbon
flow in Indian forests. Clim Change. 35:297–320.
Sabine CL, Heiman M, Artaxo P, Bakker DCE, Chen CTA, Field
CB, Gruber N, LeQuéré C, Prinn RG, Richey JE, et al. 2004.
Current status and past trends of the carbon cycle. In: Field
CB, Raupach MR, editors. The global carbon cycle: inte-
grating humans, climate, and the natural world. Washington
(DC): Island Press. p. 17–44.
Sales MH, Souza CM, Kyriakidis PC, Roberts DA, Vidal E. 2007.
Improving spatial distribution estimation of forest biomass
with geostatistics: a case study for Rondonia, Brazil. Ecol
Modell. 205:221–230.
Xia C, Guosheng H, Aihui H. 2005. MODIS-based estimation of
biomass and carbon stock of forest ecosystems in Northeast
China. Geoscience and Remote Sensing Symposium, 2005.
IGARSS ’05. Proceedings. 2005 IEEE Int. 4:3016–3019.
Downloaded By: [Joseph, Shijo] At: 04:41 24 November 2010
