Remote Sensing Applications: Society and Environment 11 (2018) 70–82

Contents lists available at ScienceDirect

Remote Sensing Applications: Society and Environment

journal homepage: www.elsevier.com/locate/rsase

Desertification risk analysis and assessment in Northern Nigeria T

a b,⁎ c
Oloukoi Joseph , Adagbasa Efosa Gbenga , Dasat Gloria Langyit
a
Department of Cartography, African Regional Institute for Geospatial Information Science and Technology (AFRIGIST), Nigeria
b
Department of Photogrammetry and Remote Sensing, African Regional Institute for Geospatial Information Science and Technology (AFRIGIST), Off Road 1, Obafemi
Awolowo University Campus, P. M. B. 5545 Ile-Ife, Osun State, Nigeria
c
Department of Cartography; African Regional Institute for Geospatial Information Science and Technology (AFRIGIST), Nigeria

A R T I C LE I N FO A B S T R A C T

Keywords: Changes in climatic conditions and human activities have led to increase in surface temperature, potential
Desertification evapotranspiration and decrease in rainfall, thereby increasing desertification risk in the study area. Data used
Climatic variations included Landsat images and meteorological data (1984–2015). Land use land cover change (LULCC) was es-
Land surface temperature timated using maximum likelihood supervised classification. Land surface temperature (LST) was computed
Encroachment
from thermal bands while desertification risk index was calculated using meteorological data from 5 weather
Environmental degradation
stations and Normalized Differential Index (NDVI) values. The results from the LULCC revealed a significant
Land use
Remote sensing reductions in rangeland by 13,418 Km2; while water body and rock out-crop were also affected by 1255 Km2 and
2592 Km2 respectively. The result also showed built-up and barren land with increase estimated at 5755 Km2 and
2016 11,512 Km2 respectively. LST increased by 2.8 °C from 1984 to 1999 and 2.7 °C from 1999 to 2014 with
values at 43.8 °C in 1984, 46.6 °C in 1999 and 49.3 °C in 2015. A similar trend was observed from climatic data
obtained while reduction in amount of rainfall was recorded over the years. Desertification Risk Index (DRI) was
used to identify areas at risk with notable increase in 2014 encroaching south-ward. Desertification risk between
1984 and 1999 recorded the least threat with 1.32 and 8.53 as the lowest and highest risk rate while in 2014 it
increased with the highest at 9.1 and the least value of 1.97. The results of the study show that desertification in
Bauchi State is evident due to the rapid changes in climatic conditions and anthropogenic activities.

1. Introduction This implies that, desertification whether exacerbated by the dete-

Desertification is an environmental problem occurring in dry lands with
its effects experienced across the globe (Liu et al., 2014). Desertification is a
form of land degradation (Escadafal et al., 2015) in arid, semi-arid and dry
sub-humid areas resulting from various factors, including climatic varia-
tions and human activities (UNCCD, 2011). Anthropogenic activities such
as deforestation can alter climatic conditions thereby affecting surface
temperature, solar radiation immersion, evaporation rates leading to de-
sertification (Javed et al., 2008). United Nations’ report stated that, more
than one billion people across the globe were affected by drought and
desertification (UNCCD, 2004). About 30 million km2 (19 percent) of the
land surface are vulnerable to desertification or are threatened by it (Nahal,
1987). It has been estimated that 12 million hectares of land are trans-
formed into new man-made deserts every year (UNCCD, 2011) and one
quarter of the world's agricultural land has been extremely degraded (FAO,
2011). However, the problem of desertification is not only confined to arid
or semi-arid regions in the world, as its impact is experienced in virtually
all facets of life worldwide (Wang et al., 2016; Dragan et al., 2005).
rioration in the condition of topsoil, particularly in the drylands, is a
consequence of temperature variations, rainfall, soil humidity and
human actions (Du et al., 2016; D’Odorico et al., 2013). This results in
enormous loss or damage to natural resources and consequently leads to
decrease in the value of soil, water, flora and fauna resources in the
society (Gao et al., 2016).
In Nigeria, the situation is not different from the global trend as
there is a general consensus that desertification is by far the most
pressing environmental problem in the dry-land parts of the country.
Three major zones exist within the northern region of the country
which are; arid, semi-arid and dry sub-humid while moist sub-humid
and humid zones are found within the southern region (UNDP, 2007). It
has been estimated that between 50% and 75% of Bauchi, Borno,
Gombe, Jigawa, Kano, Katsina, Kebbi, Sokoto, Yobe, and Zamfara
States in Nigeria are being affected by desertification (Kabara, 2013).
The visible sign of this phenomenon is the gradual shift in vegeta-
tion from grasses, bushes and occasional trees, to grass and bushes, and
in the final stages, expansive areas of desert-like sand (National

⁎
Corresponding author at: Department of Photogrammetry and Remote Sensing, African Regional Institute for Geospatial Information Science and Technology (AFRIGIST), Nigeria.
E-mail addresses: oloukoi@rectas.org.ng, chabijos@gmail.com (O. Joseph), adagbasa@rectas.org.ng, efe_ben2000@yahoo.com (A.E. Gbenga),
dasat_gloria@yahoo.com (D.G. Langyit).

https://doi.org/10.1016/j.rsase.2018.04.012
Received 11 March 2017; Received in revised form 8 February 2018; Accepted 25 April 2018
Available online 02 May 2018
2352-9385/ © 2018 Elsevier B.V. All rights reserved.
O. Joseph et al.
Fig. 1. The study area, Bauchi State.
Strategic Action Plan, NSAP, 2012). Sand dunes increased by approxi-
mately 17% from 820 km2 to 4830 km2 between the period of
1976–1978 and 1993–1995 (Federal Ministry of Environment, 2008).
Furthermore, over grazing and over-exploitation of marginal lands have
temperature variation, changes in rainfall pattern, population pressure
resulting in over grazing and over-exploitation of marginal lands have
aggravated desertification in these frontline states (National Strategic
Action Plan, NSAP, 2012). Entire villages and major access roads have
been buried under sand dunes in the extreme northern parts of Katsina,
Sokoto, Jigawa, Borno, and Yobe States (Nwafor, 2006). Nonetheless,
the overall consequence of the increasing rate of desertification is es-
timated to be progressing at a rate of about 0.6 km per year (Nwafor,
2006). Desertification has already affected about 50% and 75% of land
within the frontline states (Kabara, 2013), which amounts to about 43%
of the total land in Nigeria with an estimated population of over 50
million (Olasupo, 2013). Similarly, Odioghor (2010) stated that thou-
sands of hectares of land mass were lost to desert conditions that were
less arable for agriculture every year. As a consequence, the eleven
Northern states are losing an estimated $5.1 billion every year through
the negative effects of desertification. Furthermore, Nwafor (2006)
stated that, Nigeria has lost about 351,000 ha of landmass to desert
encroachment and this trend has currently been estimated to be ad-
vancing southwards at a rate of about 600 m per year.
However, it has been difficult to estimate how fast desertification is
advancing in Nigeria (National Strategic Action Plan, NSAP, 2012). This
is because various state governments, until recently, had not seen de-
sertification as a problem requiring high priority attention (Olasupo,
2013) and this, has therefore resulted in poor monitoring of the de-
gradation rate of arid and semi-arid areas of the country (UNCCD,
2013). On the other hand, scientists noted that desertification is not
moving along a linear front but occurs in patches, which expand in
various directions and at rates that vary considerably (Oladipo, 1993).
Therefore, the multi-dimensional implication of desertification as it
71
Remote Sensing Applications: Society and Environment 11 (2018) 70–82
affects biodiversity, education, health, social and economic activities
amongst others (Akbari et al., 2016), call for holistic approach to
monitoring, mitigation and prevention. Consequently, in order to
combat desertification, continuous monitoring is crucial. The use of
remote sensing technology to monitor desertification has enormous
potentials to survey ground condition over a large area (Grainger, 1990;
and UNCCD, 2013).
The aim of this study is to assess the risk of desertification in
Northern Nigeria especially in Bauchi State towards effective preven-
tion, mitigation and /or reversal of encroachment. The objectives are to
analyze selected indicators of desertification such as vegetation cover,
rainfall and temperature; and provide a Desertification Risk Index (DRI)
of the study area.
1.1. The study area
Bauchi State lies between latitude 9°30' and 12°30' North and
longitudes 8°45' and 11°00'
East. The State is bordered by Kano, Jigawa and Yobe States to the
north, Plateau and Taraba States to the south, Kaduna State to the west
while Gombe State is at the north eastern boundary (Fig. 1). The study
area comprises of 20 Local Government Areas with total land mass of
about 48,329 km2 which is estimated to account for 7% of land mass in
Nigeria (NPC, 1991).
Temperatures in Bauchi State are generally high with mean daily
maximum temperatures ranging from 29.2 °C in July and August to
37.6 °C in March and April. Similarly, daily evaporation in the state
ranges from 2.4 mm in July and August to about 15.7 mm in March, the
months of January and April being the period of greatest evaporation
(Nwankwo and Agboeze, 2013). The rainy season months are March to
October, with humidity ranges of about 37–68% and rainfall amounts
to about 0.0 mm in December and January to about 343 mm in July
with annual rainfall ranging between 700 mm in the north to 1300 mm
O. Joseph et al.
in the south (Nwankwo and Agboeze, 2013). Bauchi State lies at an
altitude of about 600 m above sea level, being part of the central Ni-
geria highlands and Jos Plateau complex. It spans within two distinct
vegetation zones, namely, the Guinea Savannah in the southern part
and the Sudan Savannah in the northern part that are generally char-
acterized by undifferentiated (mixed) woodlands, particularly mixed
acacia. The state is drained by several river systems. The dominant one
is Gongola River which originates from the Jos Plateau, southwest of
Bauchi State. The western and northern parts of the state are drained by
the Rivers Bunga and Jama'are systems that constitute part of the River
Yobe system. Within the northeastern part of the state is the Dingaya
(Dingaiye) River system. Similarly, the extreme northern part of the
state is a considerable stretch of the Katagum River system. Thus, the
State has considerable surface drainage systems that could be harnessed
for development purposes with much emphasis on prevention and mi-
tigation of ecological problems like desertification.
Bauchi State has a population of 4,676,465 inhabitants which ac-
counts for 3.34% of Nigeria's total population in 2006 with population
density of 72 per sq. Km and Bauchi the capital is the most populated
estimated at 493, 810 inhabitants (National Population Commission,
2006). The study area over the years has experienced about 42%
growth in population from 3,295,337 (National Population Commis-
sion, 1991) with crude density of 57.3 per sq. Km compared to esti-
mates obtained from 2006 census. Thus, as population increases more
threat is posed to the environment.
The major occupation in Bauchi State is farming employing tradi-
tional methods and implements. About 90% of the population engage in
rained agriculture, irrigation and mix farming (Udo, 1970).
2. Materials and methods
The methodological approach is based on the use of satellite images
combined with meteorological and field information. Desertification
Risk Index was determined to depict vulnerability of Bauchi State to
desertification and its extent between 1984 and 2014.
2.1. Characteristics of data
Secondary data were also utilized for this study with primary
sources from field observations and GPS measurements. Landsat sa-
tellite images of Bauchi State were acquired for 1984, 1999 and 2014
obtained during the dry season from United States Geological Survey
(USGS) Earth Explorer platform. Similarly, digital elevation data was
obtained from the USGS. In addition, rainfall and temperature data
were acquired from Nigerian Meteorological Agency (NIMET).
Summary of data used and their characteristics are highlighted in
Table 1.
Table 1
Data sources and their characteristics.
S/N Data type
1. Landsat TM (WRS-2, Path 186–188
Row 051–053)
2. Landsat ETM+ (WRS-2, Path 186
− 188, Row 051–053)
3. Landsat OLI – TIRS (WRS-2, Path
186–188, Row 051–053)
4. Rainfall and Temperature
5. Administrative Map of Nigeria
6. GPS measurement
72
Remote Sensing Applications: Society and Environment 11 (2018) 70–82
3. Methods
3.1. Analysis of the Land Surface Temperature (LST)
The temperature recorded at the margin between the earth surface
and the immediate atmosphere is Land Surface Temperature (LST)
(Valiente et al., 2014). LST has been found to be an important element
in monitoring drought, desertification and other environmental pro-
blems (Adebayo and Zemba, 2003; Anejionu et al., 2013; and
Ifatimehin et al., 2010).
LST can be calculated using different Land Use Land Cover (LU/LC)
classes. The basis for using LU/LC in LST calculation is that climatic
conditions around urbanized settlement is affected by the changes in
LU/LC and anthropogenic activities (Javed et al., 2008). In this ap-
proach, the land surface emissivity must be taken into consideration
because retrieval of LST from multispectral sensors data requires an
accurate measurement of emissivity values of the surface (Mallick et al.,
2008, Caselles et al., 1995).
3.2. LST extraction
The images for this study were re-sampled using the Nearest
Neighbor algorithm. The thermal band 6 of 1984 and 1999 which is
acquired at 60 m resolution, thermal bands 10 of 2014 acquired at
100 m resolution, are useful in providing more accurate surface tem-
peratures. The images were re-sampled to a resolution of 30 m. Idrisi
Selva software was used to carry out the data processing.
The following steps were taken to obtain Land surface temperature
information from Landsat products using Idrisi Selva software:
Conversion of Digital Numbers DNs to top of the atmosphere (TOA)
radiance values, Conversion of TOA radiance values to TOA brightness
temperature in Kelvin, land cover classification of the study area, re-
classification of the land cover classification to emissivity values, con-
version from TOA brightness temperature in Kelvin to Land surface
temperature and conversion of degrees Kelvin into degrees Celsius.
(Appendix A).
Transects were made from South - North and West-East to observe
variation in temperature in relation to change in LULC types. Data from
NIMET weather stations were obtained and used for verification of
results for the period under consideration. Hereafter, data from Bauchi,
Jos, Kaduna, Gombe and Kano weather stations of 1984 – 2014 were
interpolated using Inverse Distance Weight (IDW) due to the limited
number of weather stations and low RMSE value of about 0.00051
compared to other methods of interpolation like kriging and thin-plate
splines.
3.3. Analysis of the vegetation cover
Vegetation cover is an important and sensitive indicator of de-
sertification in arid and semiarid regions and can be used to study these
processes (Purevdorj, 1998).
Spatial resolution / Scale Date of
Production
60 m for band 6 January 1984
60 m for band 6 February 1999
100 m for band 10 January 2014
N/A 1984–2014
1: 300,000 2005
3 m accuracy February 2015
O. Joseph et al.
Fig. 2. Sample points randomly selected for validating vegetation cover and temperature.
Fensham and Holman (1999) stated that NDVI being a guide in
evaluating vegetation was the ratio between the maximum absorption
of radiation in the red (R) spectral band against the maximum reflection
of radiation in the near infrared (NIR) spectral band. However, several
authors had recommended the adoption of Soil Adjusted Vegetation
Index (SAVI) for sparsely vegetated areas such as drylands (Huete,
1988; Terrill, 1994). But for this study, the Normalized Difference Ve-
getation index (NDVI) approach was used as it provides a simple so-
lution to the estimation of vegetation cover (Hurcom and Harrison,
1998; Eastman, 2009). (Appendix B)
3.4. Determination of the desertification risk index
Desertification risk analysis of the study area was achieved through
the adoption of the Desertification Risk Index used by Dragan et al.
(2005).
The DRI is based on the assumption that desertification risk is
highest in areas with high temperatures, low precipitation, low NDVI
and when the rainfall is not spread equally over the year (Appendix C).
Furthermore, DRI was derived through several procedures
(Appendix C). Pielou's evenness index J’ (P) was an adoption of
Shannon evenness index but in this context rainfall was used rather
than species of vegetation. It ranges between 0 and 1, with 1 re-
presenting a situation in which all the average monthly precipitations
are equally abundant (maximum evenness) which was implemented
with continuous spatial data such as NDVI from low resolution satellite
imagery. However, in implementing DRI, all other parameters involved
must be available in a spatial context as quantitative map layers. Thus,
DRI was best suited to operate within a GIS environment where map
layers can be easily combined through the spatial analyst tool.
73
Remote Sensing Applications: Society and Environment 11 (2018) 70–82
Subsequently, IDW spatial interpolation technique was applied to
derive temperature and precipitation maps with values obtained from
six meteorological stations. However, due to the fact that the Inverse
Distance Weighted method of interpolation used put into consideration
distance, Sokoto weather station was omitted since estimated values are
more influenced by nearby points than those farther away and more
accurate result is obtained using fewer control points (Kang-tsung,
2012; Zimmerman, 1999).
3.5. Validation and accuracy assessment
In reality, there is simply not enough time, energy or equipment
suitable to measure every single item within a population or sampling
frame. Thus, sampling of sites is very important. In this study, a re-
connaissance survey was conducted in February 2015 to identify the
various land use/land cover classes and vegetation distribution present
in the study area.
Primarily a numbered gridded plan of the study area to be sampled
was laid over the study area map achieved using Data Reviewer of
ArcMap 10.1. Subsequently, a computer generated random number
table was used to select which grid to sample (Fig. 2).
Meteorological measurements consisting of temperature and rainfall
measurements recorded for the last 30 years obtained from five weather
stations (NIMET) were utilized for validation of temperature distribu-
tion across the study area. Data were plotted using the Inverse Distance
Weighted interpolation method due to low Root Means Square (RMS)
error and trend noted from remotely sensed data.
Results for DRI ranging from 1 to 10 were then classified into four
qualitative classes of desertification risk ranging from very low to very
high.
O. Joseph et al. Remote Sensing Applications: Society and Environment 11 (2018) 70–82

Fig. 3. Land Use Land Cover types.

temperature values obtained from NIMET between 1984 and 2014,

while an increase experienced in 2014 to mean maximum temperature
of 35.5 °C (Table 2).
Land Surface Temperatures of the study area were recorded with
variation across different LULC types (Figs. 6 and 7). Observations re-
vealed that LST had increased over the years with variations across
different LULC types. Higher temperatures were recorded for barren
land and built-up areas while rock out-crop witnessed lower tempera-
tures.
Fig. 4. LULC dynamics between 1984 and 2014. The cross section profiles gradually changes with increase in dis-
tance across LULC types. The result obtained showed that high surface
temperature values were recorded for built-up areas and barren land
However, the DRI values obtained could not be verified using
while a decrease is observed as cross section gradually changes to rock
Desertification Prone Areas (DPA) as utilized by Dragan et al. (2005)
out-crop, waterbody or vegetated areas. This obviously shows that the
due to unavailability of DPA maps for Nigeria.
presence of vegetation helps in regulating surface temperature of an
Rather it was compared with the aridity and moisture indices
area.
(Appendix D).
Consequently, over the years under consideration, a considerable
increase in temperature was observed. This is not different with in-
4. Results and discussion
creasing temperature noted globally and confirmed by temperature
surveys (Water Civilization International Centre, WCIC, 2012). Global
4.1. Land use land cover change
warming is eminent everywhere (UNDP, 2007) as depicted in the study
of Lake Chad disappearance due to rising temperature of the ground
The classified land use maps (Fig. 3) revealed that barren land and
and atmosphere and has led to increase in desertification (Miller and
built-up areas had increased tremendously by 11, 512.38 km2 and
Spoolman, 2009; Water Civilization International Centre, WCIC, 2012).
5755.12 km2 respectively, while vegetation had immensely decreased
Thus, the situation is the same in Nigeria as general increase in
of 13, 418.97 km2. Water body and rock out-crop also recorded de-
temperature is being experienced in every part (Odjugo, 2010; Olusina
crease from about 1350 Km2 and 6200 Km2 to 100 Km2 and 3600 Km2
and Odumade, 2012) as was reported in this study. IPCC (2007) noted
respectively.
that this increase in temperature may not be leveled on climate or
Therefore, built-up, rock out-crop and range land witnessed higher
meteorological conditions only, rather the role played by human ac-
magnitude of change between 1999 and 2014 while barren land and
tivities that have brought about increase in greenhouse gases.
water experienced higher magnitude of change between 1984 and 1999
Subsequently, human activities have great influence on the tem-
(Fig. 4).
perature of the environment and changes in LULC types which results in
LST dynamics (Zhang et al., 2016). Likewise, as a result of expansion of
4.2. Land surface temperatures
built-up areas over time, green spaces are lost thereby higher tem-
peratures are experienced within the built-up areas. Correspondingly,
It was observed that highest surface temperature were at 43.8 °C,
IPCC (2007) buttressed the point that agricultural practices, soil use
46.6 °C and 49.3 °C for 1984, 1999 and 2014 respectively (Fig. 5). The
and deforestation account for 13.5–18.2% of greenhouse gases emitted
highest temperature value was recorded in 2014.
respectively. Even though most human activities may be small, the
Similarly, it was observed that the lowest land surface temperature
contributory factors to greenhouse gases are harmful.
values recorded were 18.3 °C, 14.3 °C and 10.4 °C in 1984, 1999 and
Also, considering the relevance of vegetation which serves as earth's
2014 respectively with similar trend when compared to mean air

74
O. Joseph et al. Remote Sensing Applications: Society and Environment 11 (2018) 70–82

cooler in controlling rising temperatures, loss of this part of the eco- Table 2
system poses increase in desertification threat. Summary of LST between 1984 and 2014.
Therefore, since vegetation cools the land by shading soil from di- Year LST (°C) Air temperature (°C)
rect sun rays, releasing water vapor into air through transpiration and
finally through photosynthesis, carbon dioxide is removed (Miller and Max Min Max Min
Spoolman, 2009). This suggests that temperature is reduced in vege-
1984 43.82 18.33 32.1 15.3
tated areas compared to non- vegetated areas. 1999 46.67 14.27 32.4 22.1
2014 49.37 10.43 35.5 23.2
4.3. Analysis of the vegetation cover

Considering the relevance of vegetation cover in assessment of de- Built-up & Barren land

sertification, it is pertinent that this biological indicator is given the Built-up

Built-up
required attention. Thus, in this study the Normalized Difference Built-up & Barren land Built-up Rock
Rock
Vegetation Index (NDVI) was carried out alongside reconnaissance
Rock Rock &
survey to verify the values obtained (Fig. 8). Vegetation Vegetation

It was observed that NDVI values were higher in 1984 with a mean Vegetation Vegetation

value of 0.072 (ranging between –0.46 and 0.51). Areas along river Water Bodies
Vegetation & Water Bodies
channels were noted to have higher NDVI values than other LULC types.
Year 1999 recorded lower values with a mean value of −0.249 (ran- Fig. 6. LST Variation across various LULC Types on West – East transect.
ging between −0.27 and 0.27).
The mean NDVI values from 1984 to 2014 were between 0.072 to
−0.27 (ranging from −0.47 to 0.51) with 2014 recording the least Built-up & Barren land Built-up & Barren land Built-up

maximum value of 0.22 while 1984 had the highest value at 0.51. Rock
Barren
land
Anyamba and Tucker (2005) suggested that values less than 0.1 will be Rock
considered as desert, between 0.1 and 0.2 as semi-desert while areas Vegetation Vegetation
Vegetation
with 0.2 – 0.4 as shrub and grassland, and finally pixels with more than Water Bodies
Vegetation
& Rock
0.5 will be assumed having forest land cover. Likewise, Hellden and Vegetation & Water Bodies

Tottrup (2008) classified regions that had a long-term mean monthly

NDVI value between 0.1 and 0.5 as subject to desertification.
Therefore, the relationship among vegetation density, soil moisture
and surface temperature coupled with the role of these parameters in
the desertification have shown that they are strongly related and can
explain the dynamics of desertification (Anyamba and Tucker, 2005).
Equally, vegetation cover is indeed one of the most visible sign of
desertification. Decrease vegetation cover and variation in the species
composition are sensitive indicators of desertification as observed
during field survey where sparsely distributed mango, neem, eu-
calyptus, date palm and cacti were predominantly noticed with vast
coverage of dried grassland.
The variation in temperature and rainfall (Fig. 9) between 1984 and
Fig. 7. LST Variation across various LULC Types on South - North transect.
2014 revealed that the southern part of Bauchi had experienced sig-
nificant decrease as the extent of areas with low temperature and high
amount of rainfall to about 1200 mm were within the zone. At the
northern part were high temperature values of above 34 °C and 36 °C
witnessed low amount of rainfall to below 800 mm between 1984 and
2014.
Between 1984 and 2014, Kano station recorded annual mean rain-
fall of 447.1 mm and 418 mm; Kaduna station recorded 484.9 mm and
454.2 mm; Gombe station recorded 377.5 mm and 394.9 mm; and Jos

1984 1999
2014

Fig. 5. Land Surface Temperature distribution for 1984, 1999 and 2014.

75
O. Joseph et al. Remote Sensing Applications: Society and Environment 11 (2018) 70–82

Fig. 8. NDVI distribution for the study area.

recorded 1215.8 mm and 1432.4 mm. Hence, results obtained for precipitation are very sensitive to soil moisture content as a reduction
Bauchi State were influenced by various rainfall and temperature values in either of the parameters or both leads to increase in desertification
of surrounding weather stations. risk.
Variation in rainfall connotes IPCC (2007) opinion that areas within
dry lands may experience changing rainfall patterns, intense increase
4.4. Analysis of the aridity and moisture indices
temperature that could result to loss of soil moisture. Likewise,
Bounova and Krishnamurti (1993) discovered that temperature and
Kano station recorded the lowest aridity index at 0.29 with

Fig. 9. Mean annual rainfall between 1984 and 2014.

76
O. Joseph et al.
Fig. 10. Aridity Index and Moisture Index of Bauchi State.
corresponding record for moisture content at −70.66% while Kaduna
station recorded 0.72 and −28.35 for aridity and moisture indices re-
spectively. Moisture index for Bauchi State recorded −56.5% and
−35.0% as minimum and maximum moisture content after interpola-
tion was carried out. Higher aridity and moisture indices reflected
wetter areas (Fig. 10). Similarly, the aridity index revealed that Bauchi
State lies within 2 distinct zones of Semi-arid and Dry sub-humid with
below 700–1200 mm rainfall per year and aridity index less than 0.5
and 0.5–0.65 respectively (Thornthwaite and Mather, 1962) as in
Fig. 10. Therefore, compounding effects on soil and vegetation quality
(Benabderrahmane and Chenchouni, 2010) could be experienced.
Aridity Index allows definition of water availability (deficiency and
excess) within a specific period. It is a determinant of ecosystem
characteristics affecting soil health capacity (Miller and Spoolman,
2009). Likewise, moisture index is centered on the amount of water and
wilting point (Hunt et al., 2008). The amount of moisture in the soil at a
given period influences the distribution and growth pattern of vegeta-
tion, soil microbial activity, soil erosion, wind erosion, nutrient move-
ments and other soil properties.
Furthermore, Land becomes irreversibly desertified when the root-
able soil depth is not able to sustain a certain minimum vegetation
cover as such soil moisture can be used to monitor water availability for
plant growth (Kosmas et al., 2013) especially within dry lands. So,
improvement in rainfall and temperature would increase soil moisture
content thereby causing increase in vegetation cover which would in-
fluence mitigation of desertification (Adakayi, 2012).
4.5. Analysis of the relationship between vegetation cover and temperature
The relationship between land surface temperature and vegetation
cover was verified using regression analysis in this study. It was ob-
served that land surface temperature and vegetation had a negative
relationship (Yue et al., 2007) with correlation coefficient ‘r’ at
−0.462417 (Fig. 11). Similarly, the coefficient of determination was
21.38%. This implies that the presence of vegetation can strongly de-
termine or predict land surface temperature to 21.38% if other factors
remain constant. Notwithstanding, other variables such as rainfall, soil
quality, socio-economic factors could play vital roles (Akinola, 2014).
77
Remote Sensing Applications: Society and Environment 11 (2018) 70–82
Thus, in general, and especially in semi-arid regions, strong correlations
between precipitation and NDVI can be found. Therefore, NDVI can be
used as an indicator for vegetation status and its response to pre-
cipitation variability (Barbosa et al., 2013) with emphasis on the spe-
cies composition of vegetation cover (Enete et al., 2012).
Hence, the role of vegetation in evapotranspiration (Ifatimehin
et al., 2010) which explains reduction in temperature over vegetated
areas is crucial. Thus, the overall cooling effect to the environment is
reduced due to depletion of vegetation quality thereby allowing in-
crease in surface temperature (Benabderrahmane and Chenchouni,
2010; Ifatimehin et al., 2010; Enete et al., 2012).
Hence, desertification will advance, in a certain landscape, when
the soil is not able to provide the plants with rooting space and/or
water and nutrients (Kosmas et al., 2013). Water retention capabilities
of vegetation species are necessary for their survival amidst harsh cli-
matic conditions. However, their root depth and height, even though
plays principal role in desertification risk assessment, could not be as-
certained in the course of this study.
Thus, field survey revealed that the level of vegetation density could
decrease the level of temperature. Areas with more dense vegetation
such as Yankari game reserve and riparian vegetated areas felt much
cooler. However, due to the sparse distribution of these trees and ve-
getation generally with narrow leaves, moderately high temperature
were witnessed especially in built-up areas and barren land.
4.6. Desertification Risk Index (DRI)
Desertification risk between 1984 and 1999 recorded the least
threat with 1.32 and 8.53 as the lowest and highest risk index while in
2014 it increased with values between 1.97 and 9.1 (Fig. 12).
Values recorded for DRI were further categorized into low, mod-
erate, high and very high to depict severity of desertification in the
study area (Fig. 13).
Severity of desertification over the years has intensified. There is no
doubt that there exists increase in desertification risk especially to areas
that had low risk (southern Bauchi). It is of concern that areas as at
1999 with low risk were seen to have deteriorated to high risk or very
high risk. Therefore, large spatial extent is currently at very high risk of
O. Joseph et al. Remote Sensing Applications: Society and Environment 11 (2018) 70–82

Fig. 11. Relationship between change in LST and Vegetation (NDVI).

experiencing this excruciating ecological problem. This could be as- soil moisture content thereby causing increase in vegetation cover
cribed to variation in temperature and rainfall which are vital compo- (Adakayi, 2012) as observed in 1999 where areas with vegetation
nents of DRI and decrease in vegetation across the State over the years. especially riparian vegetation witnessed low risk.
Therefore, improvement in rainfall and temperature could increase Consequently, with the northern parts showing higher risk as a

Fig. 12. DRI distribution between 1984 and 2014.

78
O. Joseph et al. Remote Sensing Applications: Society and Environment 11 (2018) 70–82

Fig. 13. Severity of desertification risk in Bauchi State.

result of high temperature and low rainfall which expresses conditions
assumed in computing DRI (Dragan et al., 2005), these conditions
certainly are not favourable for vegetation growth (Adakayi, 2012) due
to fluctuations in climatic parameters that have influenced the rate of
desertification (UNCCD, 2011).
5. Conclusion
This paper demonstrates the resourcefulness of remote sensing and
GIS as a means of analyzing desertification risk in a large area such as
Bauchi State in Northern Nigeria. Landsat TM, ETM+ and OLI-TIRS
were utilized in the assessment of desertification dynamics with focus
on Land Surface Temperature (LST), Vegetation cover and
Desertification Risk Index (DRI). Over the years, there have been
significant modifications in LULC types with rangeland experiencing
the most drastic change. The result also reflected that built-up and
barren land witnessed substantial increase at the detriment of
vegetation and other features. This increase is as a result of anthro-
pogenic activities due to changes and transformation of vegetated
surfaces to impervious surfaces. LST depicted increase over the years.
Expansion of built-up areas influences trend observed in temperature
and reduction in rainfall. NDVI was applied in analyzing vegetation
quality which revealed that areas along river channels had higher NDVI
values which corresponded with thicker vegetation observed during
field survey. LST and NDVI dynamics were related to rainfall, aridity
and moisture content of the area to have better understanding of de-
sertification within climate change context. The DRI analysis revealed
that Desertification risk between 1984 and 1999 recorded the least
threat while in 2014 the desertification risk showed an increase.
In conclusion a dire need for resilience of desertification in Bauchi
and northern Nigeria is crucial as its excruciating effects cannot be
over-emphasized. Hence, more comprehensive assessment needs to be
conducted and priority given to most vulnerable parts.

Appendix A. Derivation of Land Surface Temperature (LST)

Step One: Convert Digital Numbers DNs to top of the atmosphere (TOA) radiance values.
The various Landsat products have different formula and radiometric rescaling coefficients for conversion from raw DN values to TOA as briefly
described below.
Landsat 8 OLI and TIRS band data can be converted to TOA spectral radiance using the radiance rescaling factors provided in the metadata file:
For (Landsat TM and ETM+)
Lλ = (LMAXλ – LMINλ) / (QCALMAX – QCALMIN)*(QCAL−QCALMIN) + LMINλ
For Landsat OLI-TIRS
Lλ = MLQcal + AL) (A1)
Where

Lλ= Top of Atmosphere (TOA) spectral Radiance (Watt/m2 *scrad*µm)

LMAXλ = Spectral Radiance scaled to QCALMAX in watts/ m2 *ster* µm = 1.238
LMINλ = Spectral Radiance that is scaled to QCALMIN in watts/ m2 *ster* µm = 15.303

79
O. Joseph et al. Remote Sensing Applications: Society and Environment 11 (2018) 70–82

QCALMAX = maximum quantized calibrated pixel value (corresponding to LMAXλ) in DN = 255

QCALMIN=minimum quantized calibrated pixel value (corresponding to LMINλ) in
DN= 1 ML = Band specific multiplicative rescaling factor = 0.0003342
Qcal = Quantized and calibrated standard product pixel values (DN)
AL = Band specific additive rescaling factor = 0.1

Step two: Convert TOA radiance values to TOA brightness temperature in Kelvin.
TM, ETM+ and TIRS band data can be converted from spectral radiance to brightness temperature using the thermal constants provided in the
metadata file:
T = K2 /ln ((K1 /Lλ)) + 1) (A2)
Where:

T = At-satellite brightness temperature (K)

Lλ = TOA spectral radiance (Watts/(m2 * srad * μm))
K1 =Band-specific thermal conversion constant from the metadata
(K1_CONSTANT_BAND_x, where x is the band number, band 6 for TM and ETM+, 10 or 11for OLI_TRIS)
K2 = Band-specific thermal conversion constant from the metadata K2_CONSTANT_BAND_x, where x is the band number, band 6 for TM and
ETM+ , 10 or 11for TRIS)

Step three: Land cover classification of the study area

Using Anderson classification scheme the study area was classified into five land use land cover classes listed below: water body, barren land,
range land, built up, rock outcrop
Step four Reclassification of the land cover classification to emissivity values
Weng et al. (2004) referenced temperature as black body, therefore it is necessary to correct for emissivity corresponding to each LULC type. Each
of the LULC categories was assigned an emissivity value by reference to the emissivity classification scheme by Snyder et al. (1998). Emissivity values
for various land cover types are presented in the table below
Emissivity values for various land cover types

Source: Snyder et al. (1998).

Land surface Emissivity ԑ

Soil 0.928
Vegetation 0.982
Built-up 0.942
Water 0.98
Concrete 0.937

Step five Conversions from TOA brightness temperature in Kelvin to Land surface temperature:
Land surface temperatures (LST) can be calculated using the emissivity values as follows (Artis and Carnahan, 1982):
LST = TB /(1+(λ x TB /ρ) lnε) (A3)
Where:

λ = wavelength of emitted radiance (for which the peak response and the average of the limiting wavelengths (λ = 11.5 A.m.) (Markham &
Barker, 1985) will be used),
ρ = h × c/σ (1.438 ×10−2 m K), σ = Boltzmann constant (1.38 ×10 23 J/K), h = Planck's constant (6.626 ×10–34 J s), and c = velocity of light
(2.998 ×108 m/s).
ԑ = Emissivity
The values of λ = wavelength for thermal bands of Landsat are provided in the metadata file.

Appendix B. Determination of NDVI

NIR − RED
NDVI =
NIR + RED (B1)
Thus, the Red and Near Infrared bands were used in measuring NDVI of the study area. Bands 3 and 4 of Landsat TM and ETM+ were obtained
while bands 4 and 5 of Landsat OLI were utilized. NDVI were calculated using Eq. (B1).

Appendix C. Derivation of DRI

The DRI is calculated using the following formula:

T
DRI =
[P × J ′ (P ) × NDVI ] (C1)

80
O. Joseph et al. Remote Sensing Applications: Society and Environment 11 (2018) 70–82

Where:

T = t/tmax (t = average annual temperature; tmax= average maximum temperature);

P = p/pmax (p = average precipitation value; pmax = average maximum precipitation);
J’ (P) = Pielou's Evenness Index (Pielou, 1969); and
NDVI= Normalized Difference Vegetation Index of the driest period (NDVI).
Climatic data obtained from NIMET for Bauchi, Gombe, Jos, Kaduna and Kano weather stations for 1984–2014 were utilized and Inverse Distance
Weight (IDW) spatial interpolation technique was applied to derive temperature and precipitation maps.
The Pielou's evenness index was computed which is an adoption of Shannon evenness index (Pielou, 1969) using the formula in Eq. (C2).
s
− ∑i = 1 Pi × Ln (Pi)
J′ =
Ln (S ) (C2)
Where,

Pi = Proportion of rainfall
S = average monthly precipitation over 12 months and
Ln (S) = the log of 12 months

Subsequently, in order to obtain proportion of rainfall Eq. (C3) was used for monthly precipitation.
Ni
Pi =
N (C3)
Where:

Ni = Rainfall per month (mm)

N = Rainfall per Annum (mm)

Appendix D. Aridity INDEX

Aridity index was computed as stipulated by FAO climate estimator (2000) as:
ΔP
AI =
ΔPET (D1)

MI = (AI – 1)*100
Where,

AI = Aridity index
Δ = Mean value
P = Precipitation
PET= Potential Evapotranspiration
Moisture Index

References
Adakayi, P.E., 2012. An assessment of the rainfall and temperature variations in parts of
Northern Nigeria (Ph.D. thesis). University of Jos, Nigeria.
Adebayo, A.A., Zemba, A., 2003. Analysis of micro climatic variations in Jimeta-Yola,
Nigeria. Glob. J. Soc. Sci. 2 (1), 19–88.
Akbari, M., Ownegh, M., Asgari, H.R., Sadoddin, A., H.Khosravi, H., 2016. Desertification
risk assessment and management program. Glob. J. Environ. Sci. Manag. 2 (4),
365–380.
Anejionu, O.C.D., Okeke, F.I., Moka, E.C., Uzodinma, V.N., 2013. Mapping spatio-
temporal variations of land surface temperature in South-East Nigeria with Landsat.
FIG Working Week 2013 Environment for Sustainability Abuja, Nigeria, 6–10 May
2013.
Anyamba, A., Tucker, C.J., 2005. Analysis of Sahelian vegetation dynamics using NOAA
AVHRR NDVI data from 1981–2003. J. Arid Environ. 63, 596–614.
Artis, D.A., Carnahan, W.H., 1982. Survey of emissivity variability in thermography of
urban areas. Remote Sens. Environ. 12, 313–329.
Benabderrahmane, M.C., Chenchouni, H., 2010. Assessing environmental sensitivity areas
to desertification in Eastern Algeria using mediterranean desertification and land use
“MEDALUS” model. Int. J. Sustain. Water Environ. Syst. 1 (1), 5–10.
Barbosa, H., Tote, C., Kumar, L., Bamutaze, Y., 2013. Harnessing earth observation and
satellite information for monitoring desertification. Drought Agric. Act. Dev. Ctries.
http://dx.doi.org/10.5772/55499.
Bounova, L., Krishnamurti, T.N., 1993. Influence of soil moisture on Sahelian climate
prediction I. Meter. Atm. Phys. 52, 183–203.
Caselles, V., Coll, C., Valor, E., Rubio, E., 1995. Mapping land surface emissivity using
AVHRR data: application to La Mancha, Spain. Remote Sens. Rev. 12, 311–333.
Dragan, M., Sahsuvaroglu, T., Gitas, I., Feoli, E., 2005. Application and validation of a
desertification risk index using data for Lebanon. Manag. Environ. Qual.: Int. J. 16
(4), 309–326(www.emeraldinsight.com/1477–7835.htm)(accessed 6 September
2014).
D’Odorico, P., Bhattachan, A., Davis, K.F., Ravi, S., Runyan, C.W., 2013. Global deserti-
fication: drivers and feedbacks. Adv. Water Resour. 51, 326–344. http://dx.doi.org/
10.1016/j.advwatres.2012.01.013.
Du, Z., Xu, X., Zhang, H., Wu, Z., Liu, Y., 2016. Geographical detector-based identification
of the impact of major determinants on aeolian desertification risk. PLoS One 11 (3),
e0151331. http://dx.doi.org/10.1371/journal.pone.0151331.
Eastman, R.J., 2009. IDRISI Taiga Guide on GIS and Image Processing 214. Clarks Lab,
Clarks University, pp. 212. 〈www.clarkslabs.org〉.
Enete, I.C., Alabi, M., Chukwudelunzu, V.U., 2012. Tree canopy cover variation effects on
urban heat island in Enugu city, Nigeria. Dev. Ctry. Stud. 2 (6), 2012.
Escadafal, R., Barbero-Sierra, C., Exbrayat, W., Marques, M.J., Akhtar-Schuster, M., El
Haddadi, A., Ruiz, M., 2015. First appraisal of the current structure of research on
land and soil degradation as evidenced by bibliometric analysis of publications on
desertification. Land Degrad. Dev. 26, 413–422.
FAO, 2011. Sustainable Land Management in Practice guidelines and Best Practices for
Sub-Saharan Africa. Rome, 2011. 〈http://www.fao.org/docep/016/i2845e/
i2845e00.pdf〉 (Accessed 6 September 2014).
Federal Ministry of Environment, 2008. State of Nigerian Environment Report, Nigeria.
Fensham, R.J., Holman, J.E., 1999. Temporal and spatial patterns in drought related tree
dieback in Australian savanna. J. Appl. Ecol. 36 (6), 1035–1050. http://dx.doi.org/
10.1046/j.1365-2664.1999.00460.x.
Gao, Y., Dang, X., Yu, Y., Li, Y., Liu, Y., Wang, J., 2016. Effects of tillage methods on soil
carbon and wind erosion. Land Degrad. Dev. 27, 583–591.
Grainger, A., 1990. The Threatening desert-controlling desertification. Earth Scan,
London, p. 369.
Hellden, U., Tottrup, C., 2008. Regional desertification: a global synthesis. J. Glob. Planet.

81
O. Joseph et al.
Change 64, 169–176.
Huete, A.R., 1988. A soil-adjusted vegetation index (SAVI). Remote Sens. Environ. 25,
295–309.
Hunt, E.D., Hubbard, K.G., Wilhite, D.A., Arkebauer, T.J., Dutcher, A.L., 2008. The de-
velopment and evaluation of a soil moisture index. Int. J. Climatol. 29, 747–759.
http://dx.doi.org/10.1002/joc.1749.
Hurcom, S.J., Harrison, A.R., 1998. The NDVI and spectral decomposition for semi-arid
vegetation abundance estimation. Int. J. Remote Sens. 19, 3109–3125.
Ifatimehin, O.O., Ishaya, S., Fanan, U., 2010. An analysis of temperature variations using
remote sensing approach in Lokoja Area, Nigeria. J. Prod. Agric. Technol. 6 (2),
35–44.
IPCC, 2007. Climate Change 2007: Synthesis Report, (core writing team: Pachauri R. K.
and Reisinger A.), Report no.4, Intergovernmental Panel on Climate Change, Geneva,
Switzerland.
Mallick, Javed, Kant, Yogesh, Bharath, B.D., 2008. Estimation of land surface temperature
over Delhi using Landsat-7 ETM+. J. Ind. Geophys. Union 12 (3), 131–140.
Kabara, A.A., 2013. Nigeria: Desertification in the North, all Africa.com. 13 Jul. 2013
〈http://allafrica.com/stories/201307151105.html〉. (Accessed 3 September 2014).
Kang-tsung, C., 2012. Introduction to Geographic Information System. 6th Edition. New
York.
Kosmas, C., Rirsema, R.C., Salvati, L., 2013. Evaluation and selection of indicators for
land degradation and desertification monitoring: methodological approach. Environ.
Manag. http://dx.doi.org/10.1007/s00267-013-0109-6.
Liu S., Wang T., Kang W., David M. Several challenges in monitoring and assessing de-
sertification, Environmental Earth Sciences, doi: 10.1007/s12665-014-3926-x.
Miller, G.T., Spoolman, S.E., 2009. Living in the Environment: Concepts, Connections and
Solutions, 16th edition. Brooks/Cole CENGAGE learning.
Nahal, I., 1987. Desertification in the Arab world. Arab Develpoment Institute, Beirut,
Lebanon. In Fadia, 2013, Utilizing of Remote Sensing and GIS for Desertification
Monitoring in the Agricultural Areas, part of Iraq.
National Population Commission, 2006. National Population Census, Nigeria.
National Strategic Action Plan, NSAP, 2012. Great Green Wall for the Sahara and Sahel
Initiative, Federal Republic of Nigeria Ministry of Environment, October 2012.
Nwankwo, E.A., Agboeze, M.U., 2013. Community development and tourism: a socio-
economic analysis of tourism impacts in Bauchi. J. Econ. Sustain. Dev. 4 (10),
115–123.
Nwafor, J.C., 2006. Environmental Impact Assessment for Sustainable Development.
EDPCA Publishers, Enugu.
Odiogor, H., 2010. Special Report on Desertification in Nigeria: The sun eats our land,
Vanguard Nigeria. 3 May 2010 http://www.vanguardngr.com/2010/05/
specialreportondesertificationinnigeriathesuneatsourla nd/ (Accessed 3 September
2014).
Odjugo, P.A.O., 2010. Regional evidence of climate change in Nigeria. J. Geogr. Reg.
Plan. 3 (6), 142–150.
Olasupo, F., 2013. 43.3% land area prone to desertification in Nigeria Mailafia, Vanguard
Nigeria. 9 Jul. 2013 〈http://www.vanguardngr.com/2013/07/
433landareapronetodesertificationinnigeriamailafia/〉. (Accessed 3 September
82
Remote Sensing Applications: Society and Environment 11 (2018) 70–82
2014).
Olusina, J.O., Odumade, O.M., 2012. Modeling climatic variation parameters of Nigeria
using the statistical downscaling approach. Knowing to Manage the Territory, Protect
the Environment, Evaluate the Cultural Heritage, Rome, Italy, 6–10 May 2012.
Pielou, E.C., 1969. An Introduction to Mathematical Ecology. Wiley, New York, NY.
Purevdorj, T., 1998. The Estimation of Percent Green Vegetation cover Using AVHRR
Data: Application to Mongolian Grassland. Graduate School of Science and
Technology. Chiba University.
Snyder, W.C., Wan, Z., Zhang, Y., Feng, Y.-Z., 1998. Classification-based emissivity for
land surface temperature measurement from space. Int. J. Remote Sens. 19 (14),
2753–2774.
Terrill, W.R., 1994. FAQ on Vegetation in Remote Sensing (online). Available from
Kepler.gps.caltech.edu-/pub/terrill/rsvegfaq.txt, Version 1.0: 10/13/1994.
Thornthwaite, C.W., Mather, J.R., 1962. Average climatic water balance data of the
continents. Pt. Africa. Publ. Climatol. 15 (2) (Accessed in FAO database of NEW_CLIM
dated 15th February 2015).
Udo, R., K., 1970. Geographical Regions of Nigeria. University Press, California.
UNCCD, 2013. Economic assessment of desertification, Sustainable Land Management
and Resilience of Arid, Semi-Arid and Dry Sub- Humid areas. 2nd Scientific
Conference, 9–12 April 2013- Bonn, Germany.
UNCCD, 2011. Land and soil in the context of a green economy for sustainable devel-
opment, food security and poverty eradication. Submission of the UNCCD Secretariat
to the preparatory process for the Rio+ 20 conference, 18 November 2011. 〈www.
unccd.int/publicinfo〉 (Accessed 3 September 2014).
UNCCD, 2004. The Consequences of Desertification. A Fact Sheet prepared by United
Nations Convention to Combat Desertification. 〈www.unccd.int/publicinfo/
factsheets〉, (Accessed 3 September 2014).
UNDP, 2007. Human Development Report 2007–2008. Fighting Climate Change: Human
Solidarity in a Divided World, the United Nations Development Programme.
Valiente, J.A., Niclos, R., Barbera, M.J., Estrela, M.J., 2010. 2010. Analysis of difference
between air-land surface temperature from msg data. 〈http://www.eumetsat.int/
Home/Main/AboutEUMETSAT/publications/ConferenceandworkshopProceedings/
2010/groups/cps/documents/document/pdf_conf_p57_sl_16_valiente_p.pdf〉
(Accessed 14 September 2014).
Wang, X., Hua, T., Ma, W., 2016. Responses of aeolian desertification to a range of climate
scenarios in China. Solid Earth 7, 959–964.
Weng, Q., Lu, D., Schubring, J., 2004. Estimation of land surface temperature-vegetation
abundance relationship for urban heat island studies. Remote Sens. Environ. 89,
467–483.
Water Civilization International Centre, WCIC, 2012. Venice: 〈www.civiltacqua.org〉,
(Accessed 5 January 2015).
Zhang, F., Tiyip, T., Kung, H., Johnson, V.C., Maimaitiyiming, M., Zhou, M., Wang, J.,
2016. Dynamics of land surface temperature (LST) in response to land use and land
cover (LULC) changes in the Weigan and Kuqa river oasis, Xinjiang, China. Arab J.
Geosci. 9 (499). http://dx.doi.org/10.1007/s12517-016-2521-8.
Zimmerman, D., 1999. In Kang-tsung, C. 2012. Introduction to Geographic Information
System, 6th edition. McGraw-Hill International Edition, New York, pp. 2012.