ArticlePDF Available

Land Cover Classification and Change Detection Analyzing Multi-Temporal Landsat Data: A Case Study of Gazipur Sadar, Bangladesh between 1973 and 2017



This paper analyzed land cover changes in Gazipur Sadar – an important urban fringe of expanding Dhaka City, Bangladesh, by leveraging remotely sensed imageries between 1973 and 2017. Landsat images of1973, 1991, 2006, and 2017 were classified using widely-preferred supervised classification method. Compared against groundtruth data, the reported classification accuracy ranges from 85% to 89%. Our classified land cover maps reveal that built-up areas in Gazipur Sadar increased by 312.9%, mostly replacing vegetation cover. An overall 199.7% decrease of vegetative covers highlights on the degree of urbanization process and increasing population pressure faced by Gazipur Sadar over the past decades. The rapid decrease of vegetative cover only 57 sq. km remains out of 344 sq. km throughout the region, including the Sal (Shorea robusta) forest and other floral species – invaluable resources for biodiversity and ecosystem health, should be taken as ‘alarming’ situation by the local authority responsible for promoting and managing sustainable development goals. In that light, this study emphasizes on the need for a critical assessment of future development initiatives in the Gazipur Sadar area and suggests for maintaining acceptable tradeoffs between development and environmental protection.
Land Cover CLassifiCation and
Change deteCtion anaLYZing
MuLti-teMporaL Landsat data:
a Case studY of gaZipur sadar,
BangLadesh Between 1973 and 2017
Md Arafat Hassan1*, Rehnuma Mahjabin1, Rakibul Islam1, Sakib Imtiaz1
1 Department of Geography and Environment, University of Dhaka, Dhaka,
*Corresponding author:
aBstraCt. This paper analyzed land cover changes in Gazipur Sadar – an important urban
fringe of expanding Dhaka City, Bangladesh, by leveraging remotely sensed imageries
between 1973 and 2017. Landsat images of1973, 1991, 2006, and 2017 were classified
using widely-preferred supervised classification method. Compared against ground-
truth data, the reported classification accuracy ranges from 85% to 89%. Our classified
land cover maps reveal that built-up areas in Gazipur Sadar increased by 312.9%, mostly
replacing vegetation cover. An overall 199.7% decrease of vegetative covers highlights on
the degree of urbanization process and increasing population pressure faced by Gazipur
Sadar over the past decades. The rapid decrease of vegetative cover only 57 sq. km remains
out of 344 sq. km throughout the region, including the Sal (Shorea robusta) forest and
other floral species – invaluable resources for biodiversity and ecosystem health, should
be taken as ‘alarming’ situation by the local authority responsible for promoting and
managing sustainable development goals. In that light, this study emphasizes on the need
for a critical assessment of future development initiatives in the Gazipur Sadar area and
suggests for maintaining acceptable tradeoffs between development and environmental
KeY words:
land cover (LULC) classication, change detection, geographic information
system, Gazipur Sadar (Bangladesh), urban, sustainability
Citation: Md Arafat Hassan, Rehnuma Mahjabin, Rakibul Islam, Sakib Imtiaz (2019) Land
Cover Classification and Change Detection Analyzing Multi-Temporal Landsat Data: A Case
Study of Gazipur Sadar, Bangladesh between 1973 and 2017. Geography, Environment,
Man is leaving his mark on every part of
the Earth surface and altering the physical
attributes at a rapid rate with his activities
(Lambin et al. 1999). The changing ecosystem
has been a matter of concern in the global
studies (Dixon et al. 1994; Ojima et al. 1994),
as harmful human activities are constantly
changing the natural land cover, which in
turn, aect the carbon cycle and disturb the
balance of carbon dioxide (CO2) in the atmo-
sphere (Alves and Skole 1996). In the devel-
oping countries, urban growth, coupled with
106 GES 01|2019
the industrial development and the transfor-
mation of agricultural land into built-up areas
are some of the leading causes of vegetation
loss (Shalaby and Tateishi 2007). Land cover
is supposed to be the natural and manmade
vegetation cover but at present, it includes
human settlements, infrastructure, industries
space without vegetation and water and
other natural and anthropogenic features
(Klimanova et al. 2017). Additionally, the rap-
id alteration of land cover have raised a huge
number of issues such as the adverse eect
on the relationship between biosphere and
atmosphere, extinction of a diverse range
of species and deteriorating soil condition
(Meyer and Turner 1994). Conservation of the
existing vegetation areas and regeneration
of the lost vegetation cover is now essential
for maintaining the ecological balance and
improving the health of the environment (Xin
et al. 2011). Conservation and regeneration
initiatives require the accurate measurement
and mapping of areas experiencing vegeta-
tion loss. Tropical regions are covered with
large forest areas that contribute to the pro-
tection from climate change, by absorbing
billions of tons of CO2,and can play an import-
ant role in reducing the environmental exter-
nalities due to the degradation of forest areas
(Canadell et al. 2008).
Bangladesh is a developing country and
accommodates a huge population with in
a very small geographic area. The connec-
tion between population and environment
was not well understood until recent times.
At present, the strong inter-relationship be-
tween the population and the natural envi-
ronment is known to all. This issue came into
attention especially after the RIO declaration
in Brazil (1992) and 1994 Cairo conference on
global population (World Bank 2017). Ban-
gladesh has 162,951,560 people living in an
area of 147,570 km² and with a population
growth rate of 1.1% (World Bank Data 2016).
This huge population has dierent needs and
consumption behavior, which create exces-
sive pressure on the environment. Require-
ments of food and housing facilities triggers
deforestation (Rahman 1994). Moreover, rap-
id urbanization has caused a decrease in the
forest area, agricultural land and water bodies
(Giri et al. 1996). Many agricultural and veg-
etation covered areas are being transformed
into built up and infrastructural constructs
(Quasem 2011). Hasan et al suggested that
the country has experienced a decrease of
about 1.12 million ha of vegetation area, com-
prising agricultural and mangrove forest ar-
eas, from the year 1976 to 2010.However, the
increase of non-agricultural land during this
period was 1.22 million ha (Hasan et al. 2013).
According to Food and Agriculture Organiza-
tion, the annual deforestation rate of Bangla-
desh is 0.2%, one of the high deforestation
rates in the developing countries (Fao 2015).
Gazipur district is located at the Northern part
of Dhaka city and is one of the nearest dis-
tricts from the capital city (Bangladesh Bureau
of statistics 2011). In the past, the region was a
part of a deep forest named as Vawal Pargana.
The district has an area of 1806.36 sq. km with
17.53 sq. km of wetlands and 273.42 sq. km
of forest area. Around 34,03,912 people live in
this district, having a density of 1884 people
per sq. km and a population growth rate of
5.2% (Bangladesh Bureau of Statistics 2011).
The area is most suitable for agricultural work
and the majority of the people are involved
with agro-based economic activities. But with
the increasing pressure of population and
socioeconomic changes, the area can char-
acterized of having sharp urbanization and
industrialization rates (Islam 2013).
Sal forest (also known as Madhupur forest) is
an asset of Bangladesh, which spreads across
Dhaka and Gazipur districts. Gazipur disctrict
contains around 86% of the Sal forest of the
country. Human activities like land overuse,
deforestation, urbanization, agricultural and
industrial activities are creating a threat for
this forest. Gazipur Sadar Sub-district used to
contain 20% of this forest but the forest area is
decreasing gradually (Fazal et al. 2015). Gazi-
pur Sadar Sub-district is becoming the new
industrial hub of the country due to its prox-
imity to the capital city and the enhanced
transport facilities. However, this develop-
ment is having a negative impact on the for-
est area in this region. Industries are growing
exponentially at the expense of forest land
and are causing heavy pollutions with ad-
verse environmental consequences (Dong et
al. 1997).
107 GES
Md Arafat Hassan, Rehnuma Mahjabin et al. LAND COVER CLASSIFICATION AND CHANGE ...
Md Arafat Hassan, Rehnuma Mahjabin et al. LAND COVER CLASSIFICATION AND CHANGE ...
As a developing country, the application of
land use and other relevant geographic data
is quite scarce in Bangladesh. For a develop-
ing country, remote sensing images can be
the most reliable source of an updated land
use or land cover data (Dong et al. 1997; Yang
2002). Updated and accurate data can help in
many ways to compare scenarios at dierent
temporal scales (seasonal, monthly or annual-
ly) and can also help devising plans and poli-
cies for the future development (Alpha 2003).
Therefore, this paper employed remote sens-
ing data for detecting the land cover change.
This paper aims to analyze the changing pat-
tern of vegetation cover in Gazipur Sadar Up-
azila. This analysis attempts to give a clear pic-
ture of the present day vegetation condition
and to help understand the rate and cause of
the changing patterns.
studY area
Gazipur Sadar Sub-district is located between
latitudes of 23°53’ and 24°11’ N and longi-
tudes from 90°20’ to 92°30’ E (Figure: 1). The
total area is about 446.38 sq. km, consisting
a total population of 866,540 and a popula-
tion density of about 1,941 people per sq. km.
Gazipur Sadar Sub-district is administratively
made up of eight unions: Bashan, Baria, Gach-
ha, Kasimpur, Kaultia, Konabari, Mirzapur,
Pubail (Bangladesh Bureau of Statistics 2011).
The Turag River ows past over the western
part of the city and the Balu River ows along
the eastern side. Other notable waterbodies
are the Labandanga River, the Salda River, and
the Tongi canal. The Sub-districtis surrounded
by the Sreepur Sub-district in its north, Rup-
ganj Sub-district from south to east and Savar
Sub-district from south to west.
Gazipur Sadar is a part of Madhupur tract,
which is a terrace having Dhaka in the south
and Jamalpur and Mymensingh districts
in the north. The total area of the terrace is
about 4,244 sq. km and is slightly elevated
than the nearby oodplains, the area is also
subjected to occasional tectonic activities
(Brammer 1996). There is a similarity between
Gazipur alluvium and Brahmaputra oodplain
alluvium but the clay is called Madhupur clay.
The area has slopes and low-level circular
ridges in dierent places (Rashid, 2008).Huge
range of soils can be found in this area such
as the Red-Laterite soil and Pleistocene clay.
(FAO, 1988)
The climatic condition of the Gazipur distric-
tis similar to a tropical climate. The area has
a moderate temperature and rainfall of 25.8
°C and 2036 mm respectively. The highest
temperature can be recorded in the month
of May and the lowest temperature can be
recorded in the month of January.
Fig. 1. Geographic location of the study area
108 GES 01|2019
Remote sensing data can give a perfect
image of land use change in dierent sec-
tions of the land (Klimanova et al. 2017).
This study used multi-temporal remotely
sensed images from Landsat 5, 7, and 8,
between the years 1973 and 2017. Last 40
years have been very crucial for the forest
land of Gazipur, as from 1989 to 2009 about
20.29% forest area have decreased (Yesmin
et al. 2014). We chose four years that can
represent the past decadal variations of
land cover in the area. The Landsat data-
base made for this research was construct-
ed by four sets of data, they are Landsat im-
ages from MSS (05 December 1973), Path/
Row: 147 / 043 for 1973, TM (26 November
1991) for 1991, ETM+ (21 December 2006)
for 2006 and OLI (07 December 2017) for
2017. In the Landsat image selection pro-
cess, ‘less than 10% cloud cover was a cri-
terion for ensuring the accuracy of the clas-
sied images. Therefore, it was not possible
to nd the images of the same month for
every year. In the Gazipur district, Novem-
ber to February is winter and less cloudy
(Uddin and Gurung 2010). So, Landsat data
were selected from November to February
to minimize the seasonal and cloud cover
inuences on the acquired images. ER-
DAS Imagine (Leica Geosystems 2006) and
ArcGIS (ESRI 2005) are both very import-
ant tools for land cover assessments and
hence, for this paper both of the software
were used for data processing and analysis.
First of all, geometric correction was per-
formed as the data needed to be proper-
ly coordinated to adjust for the tectonic
movements. For executing geometric cor-
rection, we selected a reference image of
Landsat TM for the year 2017. 80 ground
control points were taken at random that
were scattered over the study area for ac-
quiring the perfect geometrically corrected
image. The root means square error (RMSE)
was very low from 0.25 to 0.45 pixel. All the
data were resampled to 30 m pixel size, us-
ing the nearest neighbor method. The co-
ordinate system of the image was set as the
Bangladesh Transverse Mercator system
(BTM). Dierent sources are tested to iden-
tify correct training area and to perform the
accuracy assessment.
Atmospheric correction was also per-
formed to reduce the atmospheric dust,
solids, and liquids. Atmospheric correction
was done following López-Serrano et al.
Image classication
The classication of image is partly aected
by the method of Anderson Scheme Level
1. (Anderson et al. 1976). The image clas-
sication was based on broad categories
of land cover and spatial resolution with a
range of 30 m to 79 m. Classication was
carried out based on the four broad land
cover categories (shown in Table 1), vege-
tation cover, waterbody, built-up area and
agricultural land.
A minimum of 70-80 samples were collect-
ed to train each of the classes and if the
sample was big enough, about 200 to 400
pixels were taken. Every land use category
consisted of about 15-20 subclasses to get
the perfect accuracy. The training classes
underwent several stages like merging, de-
letion or renaming. A model is built from
the training data, which is then run to ob-
tain the nal classied image. Despite the
caution taken while devising the model,
errors could be found in the nal output.
Common errors of classication include
the diculties arising while distinguishing
between the agricultural grasslands and
healthy vegetation (Bolstad and Lillesand
The land cover change for each of the
classes was calculated using the following
formula (Islam et al. 2017):
Magnitude = magnitude of the new year -
magnitude of the previous year
Percentage change of a particular class was
calculated by dividing the change in area
by the area in the base year (primary year),
and multiplied by 100.
For calculating the rate of yearly change
for every land cover class, the percentage
change was divided by the number inter-
vening years. Calculated data are shown in
Fig. 8.
109 GES
Md Arafat Hassan, Rehnuma Mahjabin et al. LAND COVER CLASSIFICATION AND CHANGE ...
Md Arafat Hassan, Rehnuma Mahjabin et al. LAND COVER CLASSIFICATION AND CHANGE ...
Accuracy Assessment
Mixed pixel or pixel with the same color
always create a problem (Lu and Weng
2005). The sample for this training site has
been demarcated by area of interest or AOI
which also gives local knowledge. For val-
idating the classied images with the real
life features, dierent accuracy assessments
were performed. To asses 1973, 1991 and
2006 data, a total 120 pixels were devel-
oped through stratied random sampling
method. These pixels were used to com-
pare with the features in a high-resolution
topographic map of Gazipur district. For
2017, the reference data collected from the
eld were used. For this purpose, a total of
90 ground data pointes were collected and
used to calculate the classication accura-
cy. Then, the accuracy of each classication
was evaluated in terms of the overall and
producers-users accuracy as well as the
kappa coecient.
The pattern of land cover change during
1973 to 2017 is shown in Fig. 2. During this
period, Gazipur Sadar has shown a rapid
land use change. The changes in vegeta-
tion cover are illustrated in Fig. 4, 5, 6 and
7. In the year 1973, a noticeable amount of
vegetation cover (145 sq. km) can be ob-
served in the northern part of the Sadar
, which comprised of 47% of
the total land cover and extended along
the eastern border of the area. This year
was during the period after independence
Table 1. Land cover classication scheme
Table 2. Summary of land cover classication data between 1973 and 2017
(area in sq km)
Table 3. Classication accuracies in percent (Producers-Users accuracy)
Land use cover types
Water body River, permanent and temporary open water, reservoir.
Vegetation Sal forest , deciduous forest, mixed forest,bamboo
Agricultural land Crop, open eld, fallow land, mixed forest lands
Built-up Residential, commercial, industrial,road and streets
Land cover
(in %)
(in %)
(in %)
(in %)
Vegetation 145.63 46.93 84.85 37.32 24.15 6.99 57.76 16.72
Built-up area 0.0948 0.02 13.17 3.83 133.79 38.74 137.79 39.91
Water bodies 37.59 10.88 58.21 16.85 1.62 0.47 36.83 10.66
Agricultural land 162.03 42.17 189.13 42 185.81 53.8 112.99 32.71
Land cover class 1973
Water bodies
88.5 85.5 98 77 93.5 100 84.3 100
Built-up area
90.3 100 75.4 98.5 87.5 87.4 90.5 98.5
82 72.3 95.5 92.3 85.5 88.5 87.5 82.3
Agricultural land
74.4 76.5 89.5 83.5 85.3 92.5 82.3 73.5
110 GES 01|2019
when the industrial growth of the country
was slow. Gazipur Sadar at that time was a
fully agrarian society (Lesser 1888).
Fig. 2 shows that the agricultural land is
covering the central part of the Gazipur
Sadar sub-district and some patches of ag-
ricultural lands are visible in the northern
part, inside the vast vegetation cover. The
agricultural land covered an area of 162 sq.
km in 1973, which was 42% of the total land
cover. The waterbodies mainly covered the
eastern and western borders of the area.
After the independence of Bangladesh in
1971, most of the settlements in the area
were observed to be small and negligible
(only 0.02% of the land cover). The chang-
es between 1973 and 1991 is quite evident
as the amount of vegetation cover in the
northwestern part had almost vanished
completely (Fig. 5). It was the rst sign of
massive human activity in Gazipur. The
northwestern part of the region has turned
vegetation into agricultural land and thus,
vegetation cover decreased from 47% to
37%. The scattered vegetation that can be
seen in midst of the agricultural land was
mainly planted by the humans. Rapid de-
velopment leading to the uncontrolled
expansion of the built up area has resulted
in the lling up of rivers indierent parts of
the Turag and Balu river.
Fig. 2 shows that 38 % of the total areas
was the built-up area in 2006. Scattered
vegetation of the central part had been
destroyed and infrastructure was devel-
oped in that area. The natural forest of the
northeastern part was disturbed badly due
to urban sprawl and other construction
activities. Dhaka city would be expanding
and developing towards the Gazipur and
Narayangang (Dewan et al. 2009). Our re-
sults and ndings support this prediction.
A noticeable fact is that in the year 2006, a
patch of the built-up area can be observed
expanding from south to north in a linear
pattern. This was actually the newly devel-
oped industrial territory in Gazipur Sadar.
Fast-growing industries were the main
reason behind the loss of agricultural land,
wetlands, and vegetation in this area. The
change was markedly rapid from 2004 to
2010. Another interesting fact is that the
parts of the Turag River and Balu River were
signicantly lled up and turned to tem-
porary agricultural land in 2006. By 2017,
this newly developed agricultural land in
2006was destroyed by the built-up area.
The river ows and the vegetation beside
these areas, which together had formed
an ecosystem itself, was totally removed
and replaced by settlements. In 2017, the
Institution of Human Rights and Peace
for Bangladesh lled a petition against
these encroachments that comprised of
about 30 illegal structures along the coast
of Turag River. According to the lawyers,
Fig. 2. Temporal comparison between the amount of vegetation coverage and other land
cover classes
111 GES
Md Arafat Hassan, Rehnuma Mahjabin et al. LAND COVER CLASSIFICATION AND CHANGE ...
Md Arafat Hassan, Rehnuma Mahjabin et al. LAND COVER CLASSIFICATION AND CHANGE ...
government agencies had built pillars and
walkways near the banks of the river and
the land grabbers took this opportunity to
grab the surrounding land and build illegal
structures (The Daily Star 2017).
In the late 1990s, the area started to de-
velop industrially and thus, instigating a
rapid change that had mainly focused on
the economic development and not on
the health of the environment. With the ev-
er-increasing population, the built-up area
began to rise. In Fig. 3, it is evident that the
area covered by vegetation and waterbody
classes have experienced a sharp declining
pattern. According to Figure 3, the amount
of vegetation coverage was 145 sq. km in
1973, which with a decreasing rate of 72%
was reduced to 84 sq. km in 1991. Similarly,
further vegetation clearing was observed
with a 250% decrease rate in the next 15
years, which reduced the gure to just
24.15 sq. km by 2006. These rates show that
the diminishing of vegetation coverage
had proliferated throughout the years and
also increased progressively over the pass-
ing years.
People in this region had encroached ille-
gally when there was no xed demarcation
of the forests (Rahman, 2016). The national
forest policy was adopted by the govern-
ment of Bangladesh in 1979 with an em-
phasis on forest protection. However, the
policy had partially failed because of lack
of prioritization on the participation of the
local people. The present forest policy had
been continuing from 1994, which is much
broader and well-constructed than the
previous one (Millat-e-Mustafa 2002). But
the impact of this policy was evident after
the year 2000.
At the same time, there was a Forestry
sector project (1997-2004), which was im-
plemented with the concept of protected
forest area, buer zones and participatory
tree plantation (Salam et al. 2004). After the
eective stage of vegetation conservation,
it is evident that the vegetation coverage
has increased by about 315% (Fig. 8) which
resembles Salam’s work
Population explosion and the expansion of
the built-up area can be identied as the
two of the main reason behind the dimin-
ishing vegetation coverage. From 1973 to
1991, lots of construction activities were
carried out in the region, such as the gar-
ments industries. Built up area increased
from 0.02 to 13 sq. km that was a drastic
change but as mentioned earlier, there was
no observable pattern in settlement and
built up area. The houses of dwellers were
scattered all over the region. Moreover,
there was no strong base or foundation of
these types of houses. These were mainly
temporary Jhupri (shack) type of settlement
structures. In 2006, the amount of built-up
area had remained same (133 sq. km) as the
rate of development or the growth of built-
up area gets stability with stable popula-
tion increase. In the succeeding years, the
amount of built up area rose to 137,
which reiterate the fact that over the past
11 years the built-up area had reached a
stable condition or is close to reaching the
peak of its growth. During the initial stage,
Fig. 3. Land cover in Gazipur Sadar
112 GES 01|2019
Fig. 4. Land cover map of Gazipur Sadar, 1973
113 GES
Md Arafat Hassan, Rehnuma Mahjabin et al. LAND COVER CLASSIFICATION AND CHANGE ...
Md Arafat Hassan, Rehnuma Mahjabin et al. LAND COVER CLASSIFICATION AND CHANGE ...
Fig. 5. Land cover map of Gazipur Sadar, 1991
114 GES 01|2019
Fig. 6. Land cover map of Gazipur Sadar, 2006
115 GES
Md Arafat Hassan, Rehnuma Mahjabin et al. LAND COVER CLASSIFICATION AND CHANGE ...
Md Arafat Hassan, Rehnuma Mahjabin et al. LAND COVER CLASSIFICATION AND CHANGE ...
Fig. 7. Land cover map of Gazipur Sadar, 2017
116 GES 01|2019
the increase of the agricultural land was a
major reason behind the clearing of natural
vegetation. This was because the region’s
economy was totally dependent on agri-
culture and therefore, a growth from 162
sq. km to 189sq. km in the agricultural land
can be observed from 1973 to 1991.
However, with the gradual development
of industries in Gazipur, the city has now
become an important industrial hub of
Bangladesh. BSIC (Bangladesh Small and
Cottage Industries Corporation) area was
developed that further promoted indus-
trialization along the coast of the Turag
River (Sultana et al. 2012). After the devel-
opment of the industries, the agricultural
growth had slowed down in the year 2006.
The agricultural land comprised of 185sq
km and was drastically reduced to 72 sq.
km by 2017. Another reason for this drastic
change is that the lands previously used for
agriculture are now being used for sh cul-
tivation, which is presently a very popular
practice in this area. Adopting the concept
of integrated farming, many farmers are
using the agricultural lands for sh farming
and are even destroying the vegetation
coverage for the same purpose. Although,
it is a protable initiative for the economy
of the region, sometimes poor mainte-
nance poses a challenge for environmental
sustainability (Ferdous et al. 2001).
This paper analyzed the changes in land
cover in Gazipur Sadar Sub-district using
Landsat data between 1973 and 2017.
While historically Gazipur district has been
well-known for its rich forest resources,
signicant LULC changes are taking places
due to the rapid urbanization over the past
few decades. The present trend in LULC
change indicates the negative impact of
human activities on the vegetation cov-
erage. Over the past 44 years, vegetation
coverage had diminished by 199.7% - an
alarming issue for the environment of the
area. Industrial pollution, growth of built-
up areas, encroachment of rivers and wa-
terbodies along with the exploitation of
existing resources are having devastating
consequences to the vegetation coverage.
Considering the present condition, conser-
vation of vegetation coverage is necessary
and should be one of the key prioritizing
issues. Rules and regulations must be strict-
ly followed prior to the establishment of
settlements or industrial projects. Imple-
mentation of existing policies pertaining
to the reservation of forest is essential and
appropriate policies for the conservation
vegetation coverage and valuable land
must be introduced by the concerned au-
thorities. Moreover, it is the dweller’s duty
to maintain the natural environment and
the intricate balance between econom-
ic growth and environmental health, for
ensuring a sustainable development in
the region. For this reason, participatory
programs with a bottom up approach are
highly recommended. The management
of vegetation coverage and the ecacy of
any conservation initiatives will depend on
the overall role-play of the general people.
More research and the collection of ade-
quate data will help build the conscious-
fig. 8. Relative changes in land cover (%) in Gazipur Sadar
117 GES
Md Arafat Hassan, Rehnuma Mahjabin et al. LAND COVER CLASSIFICATION AND CHANGE ...
Md Arafat Hassan, Rehnuma Mahjabin et al. LAND COVER CLASSIFICATION AND CHANGE ...
ness of the public. Remote sensing data aid
in acquiring reliable images of the study
area that can be used for further analysis,
especially when there is a limited access
to data or maps. Careful analysis can help
maintaining the accuracies of classica-
tions to over 85%. Hence, satellite images
can be an eective data source for environ-
mental analysis.
Authors wish to thank the unknown refer-
ees for their valuable suggestions, which
improved the nal form of this paper.
Alphan H. (2003). Land use change and urbanization in Adana, Turkey. Land Degradationand
Development, 14(6), 575–586.
Alves D.S. and Skole D.L. (1996). Characterizing land cover dynamics using multi-temporal
imagery. International Journal of Remote Sensing, 17(4), 835–839.
Anderson R., Hardy E. E., Roach J.T., and Witmer R.E. (1976). A land use and land cover
classication system for use with remote sensor data. USGS Professional Paper 964.
Washington, DC.
Bangladesh Bureau of Statistics (BBS) and informatics division (SID) Ministry of planning
government of the people’s republic of Bangladesh (2011).
Bangladesh (1996). Int. J. Remote Sens., 17, 2749–2759.
Bolstad P.V. and Lillesand T.D. (1991). Rapid Maximum Likelihood classication.
Photogrammetric Engineering & Remote Sensing, 57, 67–74.
Center for Urban Studies (CUS), National Institute of Population, Research and Training
(NIPORT), & Measure Evaluation (2006). Slums in urban Bangladesh: Mapping and census,
2005. Dhaka, Bangladesh/Chappell Hill, USA.
Brammer H. (1996). The Geography of the Soils of Bangladesh, Dhaka Bangladesh,
University Press: 1996.
Canadell J.G and Raupach M.R. (2008). Managing Forests for Climate Change Mitigation.
Science, 5882(320), pp. 1456-1457.
Dewan A.M., Yamaguchi Y. (2009a). Using remote sensing and GIS to detect and monitor
land use and land cover change in Dhaka Metropolitan of Bangladesh during 1960–2005.
Environ. Monit. Assess., 150, 237–249, doi:
Dewan A.M., Yamaguchi Y. (2009b). Land use and land cover change in Greater Dhaka,
Bangladesh: using remote sensing to promote sustainable urbanization. Appl. Geogr. 29,
390–401, doi:
Hundreds of acres of forest, crop land, water bodies destroyed (August 6, 2013). Dhaka
Dixon R.K., Brown S. Houghton R.A., Solomon A.M., Trexler M.C., and Wisniewski J. (1994).
Carbon pools and ux of global forest ecosystems. Science, 263, pp. 185-190.
118 GES 01|2019
Dong Y., Forster B., and Ticehurst C. (1997). Radar backscatter analysis for urban
environments. International Journal of Remote Sensing, 18(6), pp. 1351–1364.
Environmental System Research Institute (2005). Using ArcGIS. Redlands, USA: ESRI.
Fazal S.A., Bhuiyan M.A.H., Chowdhury M.A.I. and Kabir M.M. (2015). Eects of Industrial
Agglomeration on Land-Use Patterns and Surface. Water Quality in Konabari, BSCIC area at
Gazipur, Bangladesh. International Research Journal of Environment Sciences, ISSN 2319–
1414, 4(11), pp. 1-10.
Ferdous A., Kenneth J.T. (2001). Current constraints and future possibilities for Bangladesh
sheries. Food Policy, 26, pp. 297–313.
Food and Agricultural Organization, Soil Portal (1988). World Reference Base.
Giri C., Shrestha S. (1996). Land cover mapping and monitoring from NOAA AVHRR data in
Bangladesh. Int. J. Remote Sens., 17, 2749–2759.
Hasan M., Hossain M., Bari M., Islam M. (2013). Agricultural Land Availability in Bangladesh;
SRDI, Ministry of Agriculture: Dhaka, Bangladesh, p. 42, ISBN 978-984-33-6141-7.
Haq S., Rahman A., Mallik D. (2017). Population and Environment in Bangladesh. World
Development Indicators database, World Bank, 15 December 2017.
Islam K., Jashimuddin M., Nath B., Nath T.K. (2018). Land use classication and change
detection by using multi-temporal remotely sensed imagery: The case of Chunati wildlife
sanctuary, Bangladesh. The Egyptian Journal of Remote Sensing and Space Science, 21,
37–47. doi:
Klimanova O., Naumov A., Greeneldt Y., Prado R.B., Tretyachenko D. (2017). Regional trends
of land use and land cover transformation in Brazil in 2001-2012.Geography, Environment,
Sustainability, 10(4), 98-116,doi:
Lambin E.F., Baulies X., Bockstael N., Fischer G., Krug T., Leemans R., Moran E.F., Rindfuss
R.R., Sato Y., Skole D., Turner B.L., Vogel C. (1999). Land-use and land-cover change (LUCC):
Implementation strategy. IGBP Report No. 48, IHDP Report No. 10, Stockholm, Bonn.
Lawrence B. Lesser. «Economic Reconstruction after Independence». A Country Study:
Bangladesh (James Heitzman and Robert Worden, editors).
Leica Geosystems (2006). Erdas Imagine Tour Guides. USA: Leica Geosystems Geospatial
López-Sánchez C.A. (2016). Evaluation of radiometric and atmospheric correction
algorithms for aboveground forest biomass estimation using Landsat 5 TM data. Remote
Sens., 8, 1–19, doi:
López-Serrano P.M., López-Sánchez C.A., Díaz-Varela R.A., Corral-Rivas J.J., Solís-Moreno R.,
Vargas-Larreta B., Álvarez-González J.G. (2015). Estimating biomass of mixed and uneven-
aged forests using spectral data and a hybrid model combining regression trees and linear
models. iForest - Biogeosciences and Forestry 9, 226, doi:
Meyer W.B. and Turner B.L. (1994). Changes in Land Use and Land Cover: A Global
Perspective. Cambridge University Press, Cambridge, UK.
119 GES
Md Arafat Hassan, Rehnuma Mahjabin et al. LAND COVER CLASSIFICATION AND CHANGE ...
Md Arafat Hassan, Rehnuma Mahjabin et al. LAND COVER CLASSIFICATION AND CHANGE ...
Millat-e-Mustafa M. (2002). A Review of Forest Policy Trends in Bangladesh -Bangladesh
Forest Policy Trends. Policy Trend Report, 2002:114-121.
Ojima D.S., Galvin K.A., and Turner B.I. (1994). The global impact of land-use change.
Bioscience 44, pp. 300-304.
Quasem M.A. (2011). Conversion of agricultural land to non-agricultural uses in Bangladesh:
Extent and determinants. Bangladesh Dev. Stud., 34, pp. 59–85.
Rahman, A.A., Huq S., Haider R., Jansen E.G. (1994). Environment and Development in
Bangladesh, The University Press Limited, Dhaka, Bangladesh.
Rahman L.M. (2000). The Sundarbans: a unique wilderness of the world. USDA Forest
Service Proceedings RMRS-P-15-VOL-2, 143-148.
Rasheed S. (2008). Bangladesh Resource and Environmental Prole, AH Development
Publishing House, Dhaka, Bangladesh.
Salam M.A., Noguchi T., and Koike M. (2005). Factors inuencing the sustained participation
of farmers in participatory forestry: a case study in central Sal forests in Bangladesh. Journal
of Environmental Management, 74(1), 43-51.
Shalaby A. and Tateishi R. (2007). Remote sensing and GIS for mapping and monitoring
land cover and land use changes in the Northwestern coastal zone of Egypt. Applied
Geography, 27(1), 28–41.
Sultana M.S., Kulsum U., Shakila A. and Islam M.S. (2012). Toxic Metal Contamination on
the River near Industrial Area of Dhaka. Universal Journal of Environmental research and
Technology, 2(2), 56-64.
30 Structures Constructed Illegally along Turag River (2017, October 09). The Daily Star.
Uddin K. and Gurung, D.R. (2010). Land cover change in Bangladesh – a knowledge based
classication approach. In Proceedings of the 10th International Symposium on High
Mountain Remote Sensing Cartography, ICIMOD, Kathmandu, Nepal, pp. 8-11.
Xin Z.B., Xu J.X., Zheng W. (2008b). Spatiotemporal variations of vegetation cover on the
Chinese Loess Plateau (1981–2006): impacts of climate changes and human activities. Sci.
China Ser. D: Earth Sci. 51 (1), 67–78.
Yang X. and Lo C.P. (2002). Using a time series of satellite imagery to detect land use and
cover changes in the Atlanta, Georgia. International Journal of Remote Sensing, 23(9),
Received on March 28th, 2018 Accepted on November 15th, 2018
120 GES 01|2019
This paper addresses the assessment of the relation of land-use dynamics and climate change based on the relationship of land surface temperature (LST), normalised difference vegetation index (NDVI) and normalised difference built up index (NDBI) using Landsat 4, 5 TM and 8 OLI/TIRS between 1989 and 2019. In situ condition was assessed using ground observations for validating satellite-derived temperature. LST was derived using NDVI-based emissivity method which uses non-linear split-window algorithm. To establish relationship among LST, NDBI and NDVI pixel-based approach was adopted. Increase in LST was observed with an increase of impervious surface and decrease of greenery. A strong positive correlation coefficient of 0.573 (P value = 0.01) was observed between LST and NDBI for the period 1989–2019 along with a strong negative correlation between LST and NDVI (r = -0.431, P value = 0.01). The correlation of NDBI–NDVI was also moderately negative (correlation coefficient = -0.284, P value = 0.01). A section of Greater Dhaka has been identified because it showed the highest rise in LST, air temperature and impervious surface and highest decrease in vegetation and rainfall. The trend of urban increase shows that in future, areas which share their boundary with Dhaka city will experience rapid urbanisation. Those areas are also expected to experience high air and surface temperature as well. Keywords. LST; NDVI; NDBI; climate change; Greater Dhaka.
Full-text available
This study reveals the scenarios and drivers of agricultural land conversion (ALC) and consequent changes in land surface temperature (LST) in four industrial areas in Bangladesh. We used remote sensing techniques to observe ALC and LST and the Driver-Pressure-State-Impact-Response (DPSIR) framework to identify major agricultural land conversion drivers. After remote sensing and field visits, DPSIR framework adopted a structured questionnaire survey to identify the perception of local respondents in identifying priorly obtained factors on a particular question. Based on the extent of industrialization, this study selected four industrial areas Gazipur Sadar, Kaliakair Upazila, Savar, and Rupganj. Analysis of Landsat TM and OIL time series data of 1999, 2009, and 2019 reveals that ALC continued to change sharply in Gazipur Sadar, Kaliakair Upazila, and Savar; the change was steady in the Rupganj area. Between 1999 and 2019, humans converted a total of 6097.14 hectares of agricultural land into the settlement. The conversion of agricultural land to barren land followed the same trend, but most noticeable in the Gazipur Sadar. The mean LST for different land use types was higher in the Gazipur Sadar and the Kaliakair Upazila than the Savar and Rupganj Upazila. Also, the mean LST in settlement and barren land areas is higher than in the areas with vegetation and water bodies. Notably, the land surface temperature is higher in the areas where conversions are quick. Approximately 71% of respondents reported that rapid industrial development is the main reason for agricultural land conversion. Moreover, many respondents (around 64%) also agreed that soil and water quality become degraded due to agricultural land conversion in their regions. Our prognosis is that if policymakers and planners do not consider these drivers, the potential impact will enhance soon. Overall, we believe that this study can help formulate effective agriculture land use and management policies of Bangladesh and other developing countries.
Full-text available
Land use and land cover (LULC) change is considered among the most discussed issues associated with development nowadays. It is necessary to provide factual and up-to-date information to policymakers to fulfil the increasing population’s food, work, and habitation needs while ensuring environmental sustainability. Geographical Information System (GIS) and Remote sensing can perform such work adequately. This study aims to assess land use and land cover changes concerning the Barapukuria coal mine and its adjacent areas in Bangladesh by applying remote sensing and GIS (geographical information system) techniques. This research work used time-series satellite images from the Landsat 7 ETM+ satellite between 1999 and 2009 and the Landsat 8 OLI/TIRS satellite for 2019. Supervised classification maximum likelihood classifier matrix was implemented using ERDAS Imagine 2018. The images were categorised into four definite classes: settlement, agricultural land, forest land, and waterbody. Analytical results clearly indicated that settlements and agricultural land had increasing and decreasing trends over the past 20 years, respectively. Settlements increased from 22% to 34% between 1999 and 2019. However, agricultural land reduced from 69% to 59% in the same period. Settlements grew by more than 50% during this period. The research had an overall accuracy of 70%, while the kappa coefficient was more than 0.60. There were land subsidence issues because of mining activities, leading to 1.003 km2 area being depressed and 1500 houses cracked. This research depicts the present LULC scenario and the impact of the coalfield area. It is expected to reduce the burden on policymakers to prepare a proper and effective mines development policy in Bangladesh and meet sustainable development goal (SDG) 15 (Life on land).
Full-text available
Regional trends of land use/land cover transformation in Brazil during 2001-2012 were analyzed in the following order: 1) identification of the types of transitions for different land use and land cover categories and aggregated groups of transformation processes based on the Global Land Cover Facility datasets, 2) analysis of national agricultural and forestry statistics to find out the principal socioeconomic drivers, 3) land cover and land use data merging to elaborate comprehensive typology of land use/land cover changes on a regional level. The study evealed 96 types of transitions between land cover categories, aggregated into 10 groups corresponding to driving processes. It was found that the main processes of land cover transformations is related to both natural and anthropogenic origins. Cropping and deforestation are anthropogenic processes, flooding and draining are the principal natural ones. Transformation of cultivated lands and reforestation are combined natural and anthropogenic. The contribution of natural factors is higher in the states of the North (Amazonia) and the Northeast macroregions; in the Center-West and the South anthropogenic factors make larger contribution. We have also detected considerable land use/land cover changes caused by agricultural development in densely populated states of the Southeast and the South. In both macroregions planted area expands due to increase of soybeans and sugar cane production, while area of pastures is shrinking. The trends of transformations of agricultural land use revealed as a result of statistical data analysis, match with transitions of land cover categories belonging to the aggregated group of cropping processes. Transformations of land cover types with predominance of shrub vegetation were the most problematic to interpret because of lack of comparable statistical data on pastures. © 2017, Lomonosov Moscow State University. All rights reserved.
Full-text available
The degraded Chunati wildlife sanctuary (CWS) has undergone various land use changes since 1980s. In this study, land use changes of CWS were assessed from 2005 to 2015 by using Landsat TM and Landsat 8 OLI/TIRS images. The ArcGIS v10.1 and ERDAS Imagine v14 were used to process satellite imageries and assessed quantitative data for land use change assessment of this study area. Maximum likelihood classification algorithm was used in order to derive supervised land use classification. It was found that about 256 ha of degraded forest area had been increased within 10 years (2005–2015) and the annual rate of change was 25.56%. Another 159 ha of naturally forested land had been changed to other land uses having an (−) annual rate of change of 15.88%. The overall supervised classification accuracy was found 92.16% for 2015, 86.15% for 2010, and 83.96% for 2005 with Kappa values of 0.89, 0.82, and 0.81 for 2015, 2010, and 2005, respectively and these were fairly satisfactory. The results of this study would be helpful to plan and implement important management decisions in order to conserve the rich biodiversity of Chunati wildlife sanctuary.
Full-text available
Readymade garments sector of Bangladesh is playing a vital role in country's economic growth for last decade but not without an intangible cost of deteriorating the environment, biological resources and self-sufficiency in agricultural sector. Industrial activity causes one of the major environmental pollution problems in Bangladesh. This study was conducted to investigate the effects of industrial agglomeration on local land-use patterns and surface water quality of Turag River and its peripheral wetlands adjacent to Konabari, BSCIC area at Gazipur district, Bangladesh. To determine the land-use patterns, image processing and digitization were carried out using the Arc GIS 10 software. The Google images were obtained from open source ''Google Earth'' software. Statistical analysis was carried out in order to process and analyze the data. The water quality parameters (pH, DO, TDS and COD) were measured by using digital calibrated instruments and the BOD value was measured by standard 5 day BOD test method as described by APHA. The accretion of industrial development was found approximately four times in the year of 2010 compared with the year of 2004. The order of increasing patterns of land-use was industries > brick fields. The decreasing patterns of land-use were water bodies >Turag River > croplands > vegetation cover during the period of 2004 to 2010. Among different land-use types, the highest percentage of grabbed area by industries was croplands (49.44%; 356 acres) and the lowest percentage of grabbed area was water body (0.14%; 1 acres). The industrial agglomeration also grabbed 103 acres (14.31%) vegetation cover of the study area. The order of completely diminished land-use patterns of the area was, water body (44.29%) > croplands (31.90%) > vegetation (13.80%) >Turag river (11.10%). The values of pH, DO, BOD, COD, and TDS ranged from of 6.25 to 9.65, 0.55 to 2.98 mg/L, 65-142 mg/L, 192-445 mg/L and 1155-2085 mg/L respectively. Except pH, all the water quality parameters exceeded the prescribed limits set by local authority which indicates that the water of Turag River and its peripheral wetlands has been polluted severely and it should not be used in any purpose regarding human and animal life without proper treatment.
Full-text available
Solar radiation is affected by absorption and emission phenomena during its downward trajectory from the Sun to the Earth’s surface and during the upward trajectory detected by satellite sensors. This leads to distortion of the ground radiometric properties (reflectance) recorded by satellite images, used in this study to estimate aboveground forest biomass (AGB). Atmospherically-corrected remote sensing data can be used to estimate AGB on a global scale and with moderate effort. The objective of this study was to evaluate four atmospheric correction algorithms (for surface reflectance), ATCOR2 (Atmospheric Correction for Flat Terrain), COST (Cosine of the Sun Zenith Angle), FLAASH (Fast Line-of-sight Atmospheric Analysis of Spectral Hypercubes) and 6S (Second Simulation of Satellite Signal in the Solar), and one radiometric correction algorithm (for reflectance at the sensor) ToA (Apparent Reflectance at the Top of Atmosphere) to estimate AGB in temperate forest in the northeast of the state of Durango, Mexico. The AGB was estimated from Landsat 5 TM imagery and ancillary information from a digital elevation model (DEM) using the non-parametric multivariate adaptive regression splines (MARS) technique. Field reference data for the model training were collected by systematic sampling of 99 permanent forest growth and soil research sites (SPIFyS) established during the winter of 2011. The following predictor variables were identified in the MARS model: Band 7, Band 5, slope (β), Wetness Index (WI), NDVI and MSAVI2. After cross-validation, 6S was found to be the optimal model for estimating AGB (R2 = 0.71 and RMSE = 33.5 Mg·ha−1; 37.61% of the average stand biomass). We conclude that atmospheric and radiometric correction of satellite images can be used along with non-parametric techniques to estimate AGB with acceptable accuracy.
Full-text available
The Sierra Madre Occidental mountain range (Durango, Mexico) is of great ecological interest because of the high degree of environmental heterogeneity in the area. The objective of the present study was to estimate the biomass of mixed and uneven-aged forests in the Sierra Madre Occidental by using Landsat-5 TM spectral data and forest inventory data. We used the ATCOR3® atmospheric and topographic correction module to convert remotely sensed imagery digital signals to surface reflectance values. The usual approach of modeling stand variables by using multiple linear regression was compared with a hybrid model developed in two steps: in the first step a regression tree was used to obtain an initial classification of homogeneous biomass groups, and multiple linear regression models were then fitted to each node of the pruned regression tree. Cross-validation of the hybrid model explained 72.96% of the observed stand biomass variation, with a reduction in the RMSE of 25.47% with respect to the estimates yielded by the linear model fitted to the complete database. The most important variables for the binary classification process in the regression tree were the albedo, the corrected readings of the short-wave infrared band of the satellite (2.08-2.35 μm) and the topographic moisture index. We used the model output to construct a map for estimating biomass in the study area, which yielded values of between 51 and 235 Mg ha-1. The use of regression trees in combination with stepwise regression of corrected satellite imagery proved a reliable method for estimating forest biomass.
Full-text available
ABSTRACT: Bangladesh is principally an agricultural country, characterized by rice paddy agriculture dominated landscapes. So, land resource is the major asset contributing wealth and livelihood in rural areas, although land-man ratio is low. Three sets of Landsat imagery for the year of 1976, 2000 and 2010 were used in this study to identify land cover types and to quantify spatial changes. The satellite imageries were digitally interpreted with unsupervised and supervised classification to quantify total land area of the country with classifications of agriculture and non-agriculture lands, and also to measure the changes of agriculture land to other land use activities. Image analysis was done with ENVI software (version 4.3) and the ArcGIS software (version 9.3) was used to digitize and analyze all the classified and other necessary maps. The land resource of the country is divided into two categories, i.e. agriculture lands and non-agriculture lands. The agriculture lands include croplands, homestead and natural forests, mangrove forests, rivers, Kaptai lake, beels and haors, aquaculture farms, tea gardens/estates and saltpans. While, non-agriculture lands include rural settlements, urban and industrial areas, and accreted lands. However, a declining trend was observed for the total agricultural lands of the country, i.e. a decrease is noted from 91.83% in 1976 to 87.69% and 83.53% over the years of 2000 and 2010 respectively. The non-agriculture lands of the country were 8.17%, 12.31% and 16.47% during 1976, 2000 and 2010 respectively. The results of the analysis produce science-based data on the availability of agricultural land including trends of land cover change since independence.
The framework of a national land use and land cover classification system is presented for use with remote sensor data. The classification system has been developed to meet the needs of Federal and State agencies for an up-to-date overview of land use and land cover throughout the country on a basis that is uniform in categorization at the more generalized first and second levels and that will be receptive to data from satellite and aircraft remote sensors. The proposed system uses the features of existing widely used classification systems that are amenable to data derived from remote sensing sources. Revision of the land use classification system are presented in U. S. Geological Survey Circular 671 was undertaken in order to incorporate the results of extensive testing and review of the categorization and definitions. Refs.
Part I. Introduction: 1. Global land-use and land-cover change: an overview Part II. Working Group Reports: 2. A wiring diagram for the study of land use/cover change: Report of Working Group A 3. Towards a typology and regionalization of land-cover and land-use change: Report of Working Group B 4. Land-use and land-cover projections: Report of Working Group C Part III. Changes in Land Use and Land Cover: 5. Forests and tree cover 6. Grasslands 7. Human settlements Part IV. Environmental Consequences: 8. Atmospheric chemistry and air quality 9. Soils 10. Hydrology and water quality Part V. Human Driving Forces: 11. Population and income 12. Technology 13. Political-economic institutions 14. Culture and cultural change Part VI. Issues In Data and Modeling: 15. Modeling land-atmosphere interactions: a short review 16. Modeling global change in an integrated framework: a view from the social sciences 17. Data on global land-cover change: acquisition, assessment, and analysis Appendices Index.
Bangladesh is a land scarce country where per capita cultivated land is only 12.5 decimals. It is claimed that every year about one per cent of farm land in the country is being converted to non-agricultural uses (such high rate of conversion will not only hamper agricultural production but will have adverse impact on food security). The present study estimates the rate of land conversion and consequent loss of agricultural production of the country besides determining the factors affecting such conversion. The study is based mainly on field survey covering 24 villages from six divisions of the country Annual Conversion of farm land is estimated to be 0.56 per cent and the country's loss of rice production is also estimated to be between 0.86 and 1.16 per cent. The converted land is predominantly used for construction of houses, followed by roads and establishment of business enterprises. The land poor records higher rate of land conversion. The two principal determining factors for such conversion are found to be land ownership size of a household and the non-agricultural occupation of household heads. To arrest the existing rate of land conversion, the surveyed households suggest for more profitable rates of return from farming activities besides imposing special sales tax for conversion of farm land.