Korean Journal of Remote Sensing, Vol.29, No.6, 2013, pp.621~629 http://dx.doi.org/10.7780/kjrs.2013.29.6.5
ISSN 1225-6161 ( Print ) ISSN 2287-9307 (Online)
Feasibility of Vegetation Temperature Condition Index for monitoring desertification in Bulgan, Mongolia Hangnan Yu*, Jong-Yeol Lee*, Woo-Kyun Lee*†, Munkhnasan Lamchin*, Dejee Tserendorj*, Sole Choi*, Yongho Song* and Ho Duck Kang**
*Department of Environmental Science and Ecological Engineering, Korea University, Seoul 136-713, Republic of Korea **Department of Biological and Environmental Science, Dongguk University, Seoul, Korea
Abstract : Desertification monitoring as a main portion for understand desertification, have been conducted by many scientists. However, the stage of research remains still in the level of comparison of the past and current situation. In other words, monitoring need to focus on finding methods of how to take precautions against desertification. In this study, Vegetation Temperature Condition Index (VTCI), derived from Normalized Difference Vegetation Index (NDVI) and Land Surface Temperature (LST), was utilized to observe the distribution change of vegetation. The index can be used to monitor drought occurrences at a regional level for a special period of a year, and it can also be used to study the spatial distribution of drought within the region. Techniques of remote sensing and Geographic Information System (GIS) were combined to detect the distribution change of vegetation with VTCI. As a result, assuming that the moisture condition is the only main factor that affects desertification, we found that the distribution of vegetation in Bulgan, Mongolia could be predicted in a certain degree, using VTCI. Although desertification is a complicated process and many factors could affect the result. This study is helpful to provide a strategic guidance for combating desertification and allocating the use of the labor force. Key Words : Desertification monitoring, Remote sensing, GIS, VTCI
1. Introduction Desertification is often seen as one of the most
serious environmental problems confronting the world
from at least the 1980s onwards (Geist, 2005). The most authoritative definition of desertification remains
that of the United Nations convention to Combat
Desertification (UNCCD): ‘land degradation in arid, semi-arid and dry sub-humid areas resulting from
various factors, including climatic variations and
human activities’ (UNCCD, 1994). In other words,
desertification is a type of land degradation in which a
relatively dry land region becomes increasingly arid, typically losing its bodies of water as well as vegetation
Received August 26, 2013; Revised September 28, 2013, Revised December 10, 2013 Revised December 17, 2013; Accepted December 17, 2013. † Corresponding Author: Woo-Kyun Lee (
[email protected])
This is an Open-Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons. org/licenses/by-nc/3.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited
–6 21 –
Korean Journal of Remote Sensing, Vol.29, No.6, 2013
and wildlife. It is caused by a variety of factors, such
as climate change (such as global warming) and human
activities (such as growing population in fragile
semiarid ecosystems and irrational or unwise land
management by nomadic pastoralists) (Tsujimura et al.,
2010).
Mongolia is one of the largest landlocked countries
in the world extending between the latitude of 41˚35’ 52˚09’ N, the longitudes of 87˚44’ - 119˚56’ E, and
covering 1,564,116 km . The longest distance from 2
west to east is 2,392 km, and from north to south 1,259
temperature. Moisture condition is strongly affected by
temperature, in other words, following temperature
rising, moisture condition will become worse.
Vegetation, on the other hand, with insufficient
moisture and once beyond the endurance cannot live too long time. Which means monitoring moisture
condition, in certain degree, may make it possible to predict desertification.
Goetz (1997) reported that the negative correlation
between Land Surface Temperature (LST) and
Normalized Difference Vegetation Index (NDVI),
km. About 90% of the territory was defined as dry
observed at several scales (25 m2 to 1.2 km2), was
occupying less than 10% (Altanbagana et al., 2010). A
temperature can rise rapidly with water stress.
lands, with hyper-arid, arid and semi-arid zones
occupying more than 80%, and dry sub-humid zone recent assessment of desertification in Mongolia shows
that 5% is very severely, 18% severely, 26% moderately
and 23% slightly degraded (Banzragch and Enkhbold,
2010). This means that roughly 72% of the total territory is affected by desertification to some extent.
Climate change poses great challenges to Mongolia,
largely due to changes in vegetation cover and soil
moisture, and was indicated that the surface
Remotely retrieved NDVI and LST data may provide a valuable source of information for monitoring
drought, since meteorological data, such as
precipitation and land air surface temperature, collected
by surface observation stations often possess poor
spatial resolution, especially in remote regions with
the region which is most vulnerable to have adverse
difficult access and in some developing countries (Wan
specific features of socio-economic development
et al. (2001) proposed for monitoring drought
impacts of climate change, because of its geographical
location, weather and climate conditions as well as
(Dagvadorj, 2010). In fact, in case of Mongolia,
evapotranspiration has increased by 3.2%-10% in the desert zones and by 10%-15% in forest steppe and high
mountain zones between 1940 and 2006 (Natsagdorj,
2004). The mean annual temperature have risen 2.1℃ between 1940 and 2007 with an accelerating trend in
the recent years (Tsogoo et al., 2010). In addition, the
frequencies of rainfall covering large areas have decreased and precipitations have fallen increasingly in
form of torrential rains on small land surface (Chuluun,
2010a). The Hydro-meteorological Institute estimates an increase of 20% since 1980 in torrential rains
(Natsagdorj, 2006). These phenomena illustrate that
drought situation in Mongolia become worse.
Global warming has caused continuous rising of
et al., 2004). The Vegetation Temperature Condition Index (VTCI) approach which was developed by Wang
occurrence at regional level on the basis of the
triangular space of LST and NDVI. This study
attempted to predict change of vegetated area with
VTCI using Landsat Thematic Mapper (TM). Wan et al. (2004) suggest that VTCI is not only closely related
to recent rainfall events but also related to past rainfall
amount, and indicate that VTCI might be a better index for drought condition. Considering that we cannot
predict what the human do next, but environment could
be predicted in certain degree based on the trend of
phenomena (Such as global warming). According to this kind of reason and above paragraph logical
thinking, we assumed that the moisture condition is the
only main factor that affects desertification. Currently,
although many studies about desertification were
– 6 22–
Feasibility of Vegetation Temperature Condition Index for monitoring desertification in Bulgan, Mongolia
published, however, there are still in the level of
comparing past and current situation (Wu and Ci, 2001;
Collado et al., 2002; Yang et al., 2005; Zhang et al.,
2008). The purpose of this study is to find out the
feasibility of VTCI to predict desertification, and based on this kind of approach to assist desertification
monitoring.
decision-making (Woodcock et al., 2001; Cohen and Goward, 2004; Masek et al., 2008). In this study, 5
mosaic images (Landsat 5) were used to define the
study area. All these images (quality 9) were taken in
September and October, and the longest interval not
exceeds 25 days. Both ERDAS IMAGINE and ArcGIS software were used for image processing and
analyzing.
2) Methods
2. Materials and Methods
(1) Derivation of LST
1) Study area and Materials
Bulgan is one of the northern provinces of Mongolia,
located between latitude 47˚14’ - 50˚23’ N and
longitude 101˚37’-104˚45’ E (Fig. 1). The northern part of this province is characterized by alpine forests,
gradually blending in the arid steppe plains of the
central Mongolian highland. Temperature fluctuate between +38℃ in summer and -49℃ in winter. An
average annual temperature is -2.4℃, and an average precipitation range from 200 to 350 mm with discontinuous permafrost.
The Landsat series of satellites provides the longest
continuous record of satellite-based observations. As
such, Landsat is an invaluable resource for monitoring
global change, a primary source of medium spatial
resolution Earth observations, and a key factor in
There are three types of methods which have been
developed to retrieve LST from at-sensor and auxiliary
dada: single-channel method, split-window technique,
and multi-angle method. Because of the last two
methods it requires at least two channels, and single-
channel method is the only method that can be applied
to the Landsat platform, with one thermal channel
(Xiao et al., 2007). The thermal band data (band 6) can
be converted from at-sensor spectral radiance to
effective at-sensor brightness temperature (Chander et
al., 2009). The formula from the at-sensor’s spectral
radiance to at-senor brightness temperature is:
K2 (1) K1 ln( + 1) Lλ Where T is effective at-sensor brightness T=
temperature, K1 and K2 are calibration constants in
Fig. 1. Geographical location and elevation distribution of study area.
–6 23 –
Korean Journal of Remote Sensing, Vol.29, No.6, 2013
Kelvin, and Lλ is spectral radiance at the sensor’s
a specific NDVI value. It can be physically explained
aperture. For TM, K1 = 607.76 and K2 = 1260.56 W /
as the ratio of temperature differences among the pixels.
referenced to a back body. Therefore, corrections for
LST_max can be regarded as the ‘warm edge’ where
(m sr μm). 2
The temperature values obtained above are
spectral emissivity (ε) became necessary according to
the nature of land cover. The emissivity corrected LST were computed as follows (Artis and Carnahan, 1982): LST =
T 1 + (λ× T ρ ) ln ε
(2)
In general, the lower the value of VTCI is, the heavier
the drought occurrence is.
there is less soil moisture availability and plants are
under dry conditions; LST_min can be regarded as the
‘cold edge’ where there is no water restriction for plant
growth (Gillies et al., 1997).
Where λ is the wavelength of emitted radiance (for
which the peak response and the average of the limiting
wavelengths (λ =11.5 μm) (Markham and Barker, c 1985) will be used), ρ = h × (1.438×10-2 m K), σ is σ the Boltzmann constant (1.38×10-23 J/K), h is Planck’s
constant (6.626×10-34 J s), and c the velocity of light (2.998×108 m/s). LST results are comparable, because
the five images were acquired at the same season
(August and September) of the year. (2) Definition of VTCI
LSTNDVIi.max = a + b NDVIi
LSTNDVIi.min = a’ + b’NDVIi
Fig. 2 is the scatter plot of the NDVI and LST data
of 1990 and 2010 in the Bulgan province. After
estimating a, b, a’ and b’, we get equations as follows: For 1990
_ LSTNDVIi.max = 301.04 5.3922NDVIi LSTNDVIi.min = 274.72 + 8.1463NDVIi
For 2011
_ LSTNDVIi.max = 312.1 7.1838NDVIi
Vegetation temperature condition index is defined as: LSTNDVIi.max _ LSTNDVIi VTCI = (3) LSTNDVIi.max _ LSTNDVIi.min Where:
3. Results and discussion
LSTNDVIi.min = 284.58 + 3.2634NDVIi
(5)
(6)
As the two scatter plots shown in Fig. 2, triangles are
not obvious. This may be because samples which are
(4)
Where LSTNDVIi.max and LSTNDVIi.min are maximum and
minimum LSTs of pixels which have same NDVIi value in a study region, respectively, and LSTNDVIi
derived from the study area used to make scatter plots
just for specific day, not for several days’ accumulation
like previous study (Wang et al. 2001, Wan et al. 2004). However, study area in this study is satisfied one
important condition in using VTCI method for
denotes LST of one pixel whose NDVI value is NDVIi.
monitoring drought, which area should large enough to
the scatter plot is normally triangular at a regional scale
VTCI has been estimated for the study area for each
Coefficient a, b, a’ and b’ can be estimated from the
scatter plot of LST and NDVI in the area. The shape of
(Price, 1990; Carlson et al., 1994) if the study area is
provide a wide range of NDVI and surface soil
moisture condition.
pixel by using equation (3) and (5) (or (6)). The VTCI
large enough to provide a wide range of NDVI and
images of the study area were shown in Fig. 3 and Fig.
region, but also related to LST changes of pixels with
affected by cloud, therefore, small portion of error
surface moisture conditions.
VTCI is not only related to NDVI changes in the
4. In general, the value of VTCI ranges from 0 to 1.
However, in this study, because images were somewhat
– 6 24–
Feasibility of Vegetation Temperature Condition Index for monitoring desertification in Bulgan, Mongolia
Fig. 2. 1990 (left) and 2011 (right) scatterplots of NDVI and LST in Bulgan.
1990 NDVI
1990 VTCI
2011 NDVI
Fig. 3. Comparison of vegetation distribution between 1990 and 2011 with 1990 VTCI.
could not be avoid and made VTCI values abnormal.
What is more, since the equations (5) and (6) were derived from scatter plot, several points may not belong
in the range of control. Even though this should not be
considered as error, but it is indeed make the value out
of the range.
The category of drought situation could define to
heavy drought (0~0.2), medium drought (0.21~0.4),
light drought (0.41~0.6), normal (0.61~0.8), and wet
(0.81~1) (Wang et al. 2001). According to this
definition, the drought condition in 1990, most parts
are in the range of 0~0.8 showed in Fig. 3, and within
the bounds of this range, value from 0 to 0.6 in
majority. In other words, the soil moisture conditions
are not good. We defined the NDVI value in three
classes and observed distribution change of each area:
0.01~0.2 as soil, 0.21~ 0.7 as sparsely distributed
vegetation and 0.71~1 as full vegetated area. The
distribution changes of each classes is easy to catch out,
if comparing three images that NDVI, VTCI in 1990
and NDVI in 2011 (Fig. 3). Comparing they are area
which marked with circles showed in Fig. 3, the trend
may obviously. The full vegetated area, area 2 which
VTCI defined as combination of medium and light
drought, changed to sparsely distributed vegetation
area. The NDVI for area 1 and 3 are degraded slightly
–6 25 –
Korean Journal of Remote Sensing, Vol.29, No.6, 2013
The results shown in Fig. 3, illustrated VTCI as a
under the combination of heavy and medium drought
thematic map for efficiency desertification monitoring
that was defined by VTCI. Moisture condition of area
is helpful. Nonetheless, considering that desertification
1 and 3 are more severe than the area 2. However, there
is a complicated process, large uncertainties still exist
is more vegetation degradation at area 2, but not at area
since we did not consider natural disaster and human
1 and 3. This result shows that high latitude area is more
activities. Human activates can be separated into two
sensitive than lower altitude area to drought condition.
parts, positive and negative. Preventive measures,
According to the estimation in this study, the southern part area was analyzed in situation of highly drought in
afforestation or educations are included in positive way.
in severely drought, and vegetation may highly adapt.
brightened in many countries. However, in many
Particularly, since importance of forest is raised as a
1990 (Fig. 3). However, there was no obvious change
major carbon pool in land, these activities are
of vegetated area, this is because the area was already
(a)
(b)
Fig. 4. (a) 1990 (left) and 2011 (right) VTCI. (b) Differences of drought condition between 1990 and 2011(left), and 2011 NDVI (right).
– 6 26–
Feasibility of Vegetation Temperature Condition Index for monitoring desertification in Bulgan, Mongolia
developing countries, due to economic necessity,
negative side is difficult to avoid. Actually, in Mongolia, livestock density exceeds the carrying
a temporal melioration.
capacity due to herders’ economic well-being has
generally improved with increased livestock numbers. Due to a rise in the price of cashmere, goat numbers
have tripled since 1990 (Chuluun, 2010b). Human
impacted deforestation and degradation would come
next, and include unwise forest management,
abandoned croplands, mining activities and unpaved
multi-track roads. In the other hand, we have no choice to select two years because of the scene quality and
absence of some scenes, this make lack evidence of
continuous trend. What is more, this study did not
perform field survey. Although, the soil surface
moisture was simulated by using the soil real thermal
inertia model (Wang et al., 2001), and VTCI can be applied to monitor drought occurrence at a regional
level and to capture the distribution of drought.
However, the accuracy of VTCI faces to challenge,
since it is derived from NDVI and LST. Because of the man-made creation like asphalt and concrete which
could affect surface temperature much and make it
abnormally high, even in vegetation flourishing area.
Also, something such as lake and reservoir can make surface temperature low in the vegetation sparseness
area. Therefore, spatial resolution will play an
4. Conclusion In this study, we attempted to find out the feasibility
of VTCI for desertification monitoring. According to
analyzing, we found that most parts of Bulgan province were in situation of drought (VTCI value 0~0.6) in
1990. NDVI value was classified three classes in this
study, and based on this range full vegetated area
(NDVI 0.71~1) was much degraded from 1990 to
2010. Similar trend can be seen at the area that
vegetation sparsely distributed (NDVI 0.21~0.7).
Comparing the VTCI thematic map with NDVI map both in1990 and 2010, we find out that there is
somewhat relationship between images and mark out
3 areas to make the phenomena obviously. According
to this phenomenon, we got the area that most in danger to be degraded.
However, there are still many areas could not be
illustrated by VTCI thematic map. This should be
contributed by assumption we made. That means this
study did not consider the natural disaster and human
influence, however, there are powerful impact on
desertification process. That might makes the
prediction in this study in not enough to cover the
important role for VTCI accuracy. For improve
whole study area. In addition, the accuracy of VTCI
2011 with NDVI image in 2011, and tried to find the
regard as one single value.
accuracy of VTCI, field survey will come to necessary.
In addition, we compared VTCI images of 1990 and
area which has a risk to be degraded (Fig. 4). Area 4,
marked with circle has the most danger. Although the prediction may somewhat be rough, due to the
problems that were mentioned above, that area worth to note and need to observe. Moisture condition in
several portions got better in 2011 than 1990.
Considering global warming as an undeniable trend and
can be influenced much by spatial resolution; due to
the pixel size decide how many properties would be Although all these uncertainties make the prediction
rough, however, this study is still helpful as a material
to provide a possible approach to prediction and
strategic guidance for combat desertification and to
allocate use of labour force.
under the assumption, this phenomenon may be
influenced by positive affect of human activity or just
–6 27 –
Korean Journal of Remote Sensing, Vol.29, No.6, 2013
Environment, 113: 893-903.
Acknowledgements This study was carried out with the support of
“Forest Science Technologies Development Project
(Project Number: S211213L030320)” provided by the Korea Forest service.
Chuluun, T., 2010a. Climate Change and Sustainable Development of Mongolia: Summary of
Papers, Institute for Global Environmental
Strategies (IGES), June 17-18.
Chuluun, T., 2010b. Opportunities of Synergies
between Climate Change, Desertification,
Conservation, and Human Development at
Multiple Scales, International Conference on
References Artis, D.A. and W.H. Carnahan, 1982. Survey of
emissivity variability in thermography of urban areas, Remote Sensing of Environment, 12: 313329.
Altanbagana, M., T. Chuluun, and D. Ojima, 2010.
Vulnerability Assessment of the Mongolia
Climate, Sustainability and Development,
Fortaleza, Brazil, August 16-20.
Dagvadorj, D., 2010. The Climate Change Policy and
Measures in Monglia. Institute for Global
Environmental Strategies (IGES), June 17-18.
Gillies, R.R., T.N. Carlson, J. Cui, W.P. Kustas, and
K.S. Humes, 1997. A verification of the
‘triangle’ method for obtaining surface soil
Rangeland Ecosystems, Institute for Global
water content and energy fluxes from remote
Environmental Strategies (IGES), June 17-18.
measurement of the Normalized Difference
Banzragch, Ts. and S. Enkhbold, 2010. New Policy on
Vegetation Index (NDVI) and surface radiant
Combating Desertification in Mongolia, Institute
for Global Environmental Strategies (IGES), June 17-18.
Carlson, T.N., R.R. Gillies, and E.M. Perrry, 1994. A method to make use of thermal infrared
temperature and NDVI measurements to infer soil water content and fractional vegetation
cover, Remote Sensing Reviews, 9: 16-173.
Collado, A.D., E. Chuvieco, and A. Camarasa, 2002.
temperature, International Journal of Remote Sensing, 18: 3145-3166.
Goetz, S.J., 1997. Multi-sensor analysis of NDVI, surface temperature and biophysical variables
at a mixed grassland site, International Journal of Remote Sensing, 18: 71-94.
Geist, H., 2005. The Causes and Progression of Desertification, Ashgate Publishing, Ltd., London, UK.
Satellite remote sensing analysis to monitor
Markham, B.L., and J.K. Barker, 1985. Spectral
boundary of Argentina, Journal of Arid
Mapper sensor[J], International Journal of
desertification processes in the crop-rangeland Environments, 52: 121-133.
Cohen, B.C. and S.N. Goward, 2004. Landsat’s role in ecological application of remote sensing,
characteristics of the LANDSAT Thematic
Remote Sensing, 6: 697-716.
Masek, J.G., C. Huang, R. Wolfe, W. Cohen, F. Hall, J.
Kutler, and P. Nelson, 2008. North American
BioScience, 54(6): 535-545.
forest disturbance mapped from a decadal
Chander, G., B.L. Markham, and D.L. Helder, 2009.
Landsat record, Remote Sensing of Environment,
coefficients for Landsat MSS, TM, ETM+, and
Natsagdorj, L., 2004. Total evaporation change issues
Summary of current radiometric calibration
EO-1 ALI sensors, Remote Sensing of
– 6 28–
112: 2914-2926.
in the Mongolian territory, WMI, Scientific
Feasibility of Vegetation Temperature Condition Index for monitoring desertification in Bulgan, Mongolia
Journal, 26: 42-55.
forest change using Landsat: Generalization
across space, time and Landsat sensors, Remote
Natsagdorj, L., 2006. Climate change factors issues on
pastoral degradation of Mongolia, Improvement
issues on pastoral management, 125-140.
Price, J.C., 1990. Using spatial context in satellite data
Sensing of Environment, 78: 194-203.
Wu, B. and L. Ci, 2002. Landscape change and desertification development in the Mu Us
Sandland, Northern China, Journal of Arid
to infer regional scale evapotranspiration, IEEE Transactions on Geoscience and Remote Sensing, 28: 940-948.
Tsujimura, M., N.Wakasugi, X. Sun, and T. Endo,
Environments, 50: 429-444.
Wan, Z., P. Wang, and X. Li, 2004. Using MODIS Land Surface
2010. Completion Report of International
Internship in Mongolia, October 25 - November
and
Normalized
monitoring drought in the southern Great
Plains, USA, International Journal of Remote
1, University of Tsukuba, Japan.
Tsogoo B., D. Borchuluun, J. Dorjpure, and D.
Temperature
Difference Vegetation Index products for
Sensing, 25(1): 61-72.
Sengelmaa, 2010. Technology Transfer Role in
Xiao R.B., Z.Y. Ouyang, H. Zheng, W.F. Li, E.W.
Mongolia, Institute for Global Environmental
of impervious surfaces and their impacts on
Alleviation of Climate Change Impact in
Schienke, and X.K. Wang, 2007. Spatial pattern
Strategies (IGES), June 17-18.
United Nations Convention to Combat Desertification (UNCCD). 1994. National Plan of Action to
Combat Desertification in Mongolia, Ulaanbaatar,
land surface temperature in Beijing, China,
Journal of Environmental Sciences, 19: 250-256.
Yang X., K. Zhang, B. Jia, and L. Ci, 2005. Desertification assessment in China: An
overview, Journal of Arid Environments, 63:
226.
Wang, P.X., X.W. Li, J.Y. Gong, and G.H. Song, 2001.
Vegetation Temperature Condition Index and Its Application for Drought Monitoring, Proc.
517-531.
Zhang Y., Z. Chen, B. Zhu, X. Luo, Y. Guan, S. Guo,
of International Geoscience and Remote Sensing Symposium, 9-14 July 2001. Sydney,
Australia (Piscataway, NJ: IEEE), 141-143.
Woodcock, C.E., S.A. Macomber, M. Pax-Lenney, and
W.C. Cohen, 2001. Monitoring large areas for
–6 29 –
and Y. Nie, 2008. Land desertification
monitoring and assessment in Yulin of
Northwest China using remote sensing and
geographic
information
systems
(GIS),
Environmental Monitoring and Assessment,
147: 327-337.
