Monitoring land use/land cover change, urban growth dynamics and landscape pattern analysis in five fastest urbanized cities in Bangladesh

Remote Sensing Applications: Society and Environment 7 (2017) 69–83

Contents lists available at ScienceDirect

Remote Sensing Applications: Society and Environment journal homepage: www.elsevier.com/locate/rsase

Monitoring land use/land cover change, urban growth dynamics and landscape pattern analysis in five fastest urbanized cities in Bangladesh

MARK

Mohammad Mehedy Hassan Department of Geography, University of Florida, 3141 Turlington Hall, P.O. Box 117315, Gainesville, FL 32611-7315, United States

A R T I C L E I N F O

A B S T R A C T

Keywords: Five cities Land use/land cover Urban growth Landscape fragmentation Sustainable development

With little known and explored urban morphology in the fastest growing countries like Bangladesh in South Asia, this study aims at exploring urban spatial signature and explaining spatiotemporal land use and land cover patterns in the five cities (Rajshahi, Rangpur, Sylhet, Khulna, and Barisal) in Bangladesh. Using time series Landsat imagery, socioeconomic data and, other geospatial information with ecological analysis tools, this study quantifies and characterize the spatial-temporal landscape patterns and urban growth trajectory across the five selected sites. The spatial representation of these five sites demonstrates a continuous increase in urban/built-up areas replacing arable agricultural land, waterbodies, vegetation cover and wetlands, which thereby substantially altering the structure and function of the ecosystem surrounding the cities. Built up areas, representing impervious surface as observed from land cover maps in these five cities, are expanding quickly. The total builtup cover within the five cities grew from 2356 ha in 1973 to 13,435 ha in 2014 with a net increase of approximately 468%, while vegetation cover and crops field within same time period declined at 27.77% and 61.91%, respectively. This dramatic urban/built-up expansion has resulted in an increasingly faster alteration in the landscape composition causing to structural complexity at both class level and landscape level. Such rapid and unplanned urban expansion further has brought an overwhelming challenge to planners and policy makers, and has put a strain on local authorities to properly manage and utilize its limited land-based resources due to lack of time series geospatial information. The resulting thematic map and spatial information from this study is, therefore, to facilitate an understanding of urban growth dynamics and land cover change pattern in the five cities in Bangladesh. The result further can aid planners, stakeholders, and other interested groups to make the best possible choices regarding limited land-based resources to achieve an economically prosperous and environmentally sustainable future.

1. Introduction As one of the fastest urbanized countries in the world, Bangladesh has experienced an unplanned and a rapid rate of urbanization in recent years (Hassan and Nazem, 2015). Currently, 34% of her population lives in urban areas and this total is expected to reach 56% in the year 2050 (UN, 2014). The estimated growth of urban population over the last two decades was 4.22% per annum, while the national population growth rate was 2.14% per year (growth calculated from 1991 to 2011 population census). Such explosive population growth with socio-economic progress led to exurbanizaiton and urbanization in Bangladesh over the past decades which has further resulted in a territorial expansion of each cities and growth centers. As a result, the total territory occupied by cities expanded to 10,712 square kilometers, accounting for 7.25% of the country's total land areas (BBS, 2001). This phenomenal urban growth has been attributed to many factors, including natural population growth and large influx of rural-urban migration

E-mail address: mehedy@ufl.edu. http://dx.doi.org/10.1016/j.rsase.2017.07.001 Received 4 December 2016; Received in revised form 5 July 2017; Accepted 7 July 2017 Available online 08 July 2017 2352-9385/ © 2017 Elsevier B.V. All rights reserved.

(Islam, 2011), which is thought to be responsible for as much as twothirds of the hike in urban population in Bangladesh (Afsar, 2003). In addition, changes in urban delimitation and the definition of an urban area over the last decade have also facilitated urban expansion and the subsequent population increases (BBS, 2011; Islam, 2011). A notable feature of this expansion is that the growth in urban areas has spilled over to local administrative boundaries, and in some cases, multiple cities or growth centers have merged into a single entity. This rapid and uncontrolled urbanization, however, is an important phenomenon in the economic development, social progress, and cultural transformation of the country because 64% of the national gross domestic product (GDP) is generated in the urban sector (UPPR, 2011; Nazem et al., 2011). Additionally, these cities hold a special place in the country as they are the cultural, industrial, business, education, innovation, entertainment, and political heart of the entire nation (Islam, 2011). Rapid urbanization in Bangladesh is exerting enormous pressure on

Remote Sensing Applications: Society and Environment 7 (2017) 69–83

M.M. Hassan

In the north, the city of Rangpur lies between 25° 41 to 25° 49 north latitudes and 89° 12 to 89° 18 east longitudes with an average elevation at 35 m above the sea level. It is a relatively juvenile divisional headquarters that received status as a city corporation in 2010 (BBS, 2011), serving a primarily agricultural region on the greater north Bengal of Bangladesh. The city's growth, however, has been powered by its recent upgrading administrative status, from a district township to a divisional headquarters, and by the foundation of a public university, Begum Rokeya University, and a Medical College. Upgradation of this previous district town to divisional headquarters has increased its importance in all respects, including a special emphasis on its economic aspects. Currently, the city of Rangpur is the home of nearly 308,000 people, with a density of 6057 persons residing in the 50.69 square kilometers of its jurisdictional area. The city of Rajshahi is the fifth largest city in the country, located to the northwest in Bangladesh, between 24° 21’ to 24° 26’ north latitude and 88° 32’ to 88° 40’ east longitude, with an average elevation of 21 m above sea level. The economy and industry are mainly based on agricultural products and factories, like silk and other cottage industries. In spite of being an important city, industrial development in Rajshahi has not taken place to any great extent. The city, however, is well connected with a developed road network to all regional and interregional cities in the country. Currently, the city has a population of 448,087 within its 48 sq. km. jurisdiction (BBS, 2014). The Sylhet in the north is the fourth largest urban agglomeration in the country and the regional headquarters in the northeastern region of Bangladesh. The city is located on the bank of the river Surma, between 24° 51’ to 24° 56’ north latitude and 91° 5’ to 91° 54’ east longitude with an average elevation of 20 m above sea level. To the north, the elevation of the city gradually increases to above 30 m due to the presence of several undulating hillocks, whilst the south is comparatively low with an average elevation of 17 m. Topographically, the landscape is flat along the river bank, where the elevation seldom exceeds 20 m. This monotonous topography, however, is broken up by hills in the north and northeast parts of the city. The maximum height of this undulating landscape is 40 m above sea level. The hills in the metropolitan, from north-south, are presumed to have influenced the precise location of the early township foundation, but due to rapid urbanization these obstacles have become minor. The other two sites in this study are in the south of the country with proximity to the sea and the world's largest mangrove forest. These two metropolises provide excellent linkages with several navigable rivers, the sea, other coastal towns and growth centers, and with the country's largest port cities- Chittagong and Mongla. These metropolises have a unique physiography and geophysical setting and their proximity to the coast has shaped their formation. In the south, Khulna is the regional capital of southwest Bangladesh and the third largest urban center in the country. It is located at 22° 46 to 22° 54 north latitudes and 89° 28 to 89° 35 east longitudes, on the west banks of the Rupsha and Bhairab rivers, with average elevation of 7 m (23 feet) above sea level. Physiographically, the area is part of the largest delta in the world. It is characterized by Ganges tidal floodplains having lower relief and crisscrossed by innumerable canals and rivers surrounded by tidal marshes and swamps. The city of Khulna is strategically important as it links the second port city of the countryMongla, and a direct railway connection with Kolkata of India, and other parts of the country including the capital city Dhaka. The major economic sectors in the city are jute, chemicals, fish and seafood packaging, food processing, sugar mills, power generation, ship building and export processing. The township first became a corporation in 1984 with an area of 46 square kilometers. This study however, conducted beyond the city corporation areas due to continuous northward expansion into the Gilatola cantonment, the Shirmoni residential area and south-west along Satkhira-Khulna highway including the area of Khulna University.

local capacities and on limited land-based resources, leading to many problems such as pollution, crime, traffic congestion, poverty, and economic and ecological challenges for the entire country. Such growth puts a strain on the ability of city authorities to provide services like energy, health care, transportation, sanitation, and physical security. Further it underscores pressing concerns for urban planners, stakeholders, and policy makers in the country due to a shortage of resources, time-bound urban information, and lack of professional planners (Hassan and Nazem, 2015). Moreover, the urban growth in Bangladesh has typically been described as either a change in the absolute area of urban space, or as an aggregate measure of the extent of cities with respect to decadal population growth based on census data. State-owned census data in Bangladesh, however, ambiguous, has often raised conflict between government and non-government sources. Additionally, it offers a vague approach in defining administrative delimitation with unpredictable spatial information on population distribution in urban areas. Hence, such aggregate measure delivers limited information regarding the spatial patterns of urbanization, its precise physical extent, and the driving elements of the urbanization process. Empirical study suggests; as urbanization continued to increase, landcover become increasingly diverse in pattern, fragmented in structure and more complex in shape leading to a number of negative consequences to biotic and abiotic resource (Qi et al., 2014), such as losses of fresh productive agricultural land (Hass et al., 2015), disturbance to rural economic stability and lifestyle, and loses of habitat and species diversity (Wu et al., 2011). Thus, periodic urban growth and its changing patterns must be monitored and characterized, and understood its resultant impact on the land, ecology, and the environment in respect to spatial and temporal context. Although RS in combination with GIS has already proven to be a powerful and most useful technique for tracking urban growth and land use/land cover change, there have been none or few systematic studies conducted on tracking urban growth and land cover change in five cites selected as the study area for this research. In addition, previous studies have shown that (Hassan et al., 2015) the urban land in Bangladesh has increased exponentially, with development being chaotic and inefficient in the last two decades. Thus, a comprehensive understanding of urban growth and land cover change across the selected cities of the study area, which may raise awareness and concern for the need of better policies and planning for urban land development, is required. Therefore, the main purpose of this study is to explore and explain spatiotemporal land use/ cover change, urban growth dynamics and landscape pattern analysis across five urban areas in Bangladesh. Using time series Landsat imagery, socioeconomic data, and other geospatial information enabling with GIS and remote sensing and ecological analysis tools, this study quantifies and characterize the temporal pattern of urbanization and varieties of spatial signature and textural form of landscapes in five urban areas. 2. Description of the study area, data source and methods 2.1. Study areas Among the five study areas, three sites are located in the north, and the other two are located in the south of the country (Fig. 1). In general, the landscape of all five cities are easily accessible from the capital cityDhaka of the country either by road, air, or water ways. The topography of these cities ranges from low, flat wetland to slightly undulating with a few small swampy areas with elevation ranges from 20 to 35 m such as in Rangpur, Rajshahi, and Sylhet. While cities in the south such as Khulna and Barisal have an average elevation of 7 m. Nearly 2.5 million and approximately 8% of the country's urban population live across these five cities (BBS, 2014). The highest population density is found in Khulna (13,134 persons) and Sylhet city (12,704 person), and the lowest in Rajshahi at 4628 persons per sq km (BBS, 2014). 70

Remote Sensing Applications: Society and Environment 7 (2017) 69–83

M.M. Hassan

Fig. 1. Location and elevation map of five cities.

study is 86.59 sq. km. that includes the city corporation area (58.97 sq. km.) and its immediate periphery to the north, northwest part along to the Dhaka-Barisal highway. Climate pattern of these selected sites follows the general trend of the rest of the country. The cool and dry winter (December-February) is followed by a hot and rainy pre-monsoon period (March-May) and then by the cooler but very wet monsoon season (June-September). This is followed by a transitional humid and showery period up to the

The other study site in the south is Barisal, which stands on the bank of the river Kirtonkhola, located in the southern coast of Bangladesh at 90°18–90°23 east longitude and 22°38–22°45 north latitude. The average elevation of the city from mean sea level is 7 m. The high elevation lands lie in the north, northwest and eastern part of the study area where major urban development has taken place. It is the sixthlargest city of Bangladesh and the population surpassed 339,308 in the 2011 census year (BBS, 2014). The areas, however, included in this 71

Remote Sensing Applications: Society and Environment 7 (2017) 69–83

M.M. Hassan

Table 1 Synoptic representation of the study areas. Source: (BBS, 2014)(Urban Area Report) City

Location

Rank in National Context

Population (2011)

Average Elevation (meter)

Jurisdiction Area (sq.km.)

Corporation Status (Year)

Literacy Rate (%)

Khulna

22° 46 to 22° 54 north latitudes 89° 28 to 89° 35 east longitudes 24° 51’ to 24° 56’ north latitude 91° 5’ to 91° 54’ east longitude 24° 21’ to 24° 26’ north latitude 88° 32’ to 88° 40’ east longitude 22°38 to 22°45 north latitude 90°18 to 90°23 east longitude 25° 41 to 25° 49 north latitudes 89° 12 to 89° 18 east longitudes

3rd

664,728

7

50.61

1990

73.6

4th

531,663

20

41.85

2001

67.4

5th

448,087

21

97.18

1991

74

6th

339,308

7

69.19

2000

74.8

7th

308,000

35

50.69

2010

72

Sylhet Rajshahi Barisal Rangpur

Imagine 14 software; other images were subsequently co-registered to this OLI-TIRS image as a reference to perform geometric correction. The results of this image-to-image registration were examined using three different methods, including examining the root mean square error (which was less than 0.5 pixels), using image overlays, and image flickering/blending (Hassan et.al, 2015). All images were registered and re-projected to local coordinates (Bangladesh Transverse Mercator with datum Everest 1830) and a first order polynomial transformation with nearest neighbor resampling was applied.

beginning of winter. From mid-November to late February the weather is typically dry and cool A brief description of these five cities are presented in (Table 1). 2.2. Data sources This study makes uses of various spatial and non-spatial data from diverse source including multiyear population census data, socio-economic census data from city level statistical year books, and first hand data from field level observation. The main spatial and temporal data used in this study such as; Landsat MSS 60 m (1973), TM 30 m (1989), ETM 30 m (2000) and OLITIRS 30 m (2014) satellite image were derived from open access Landsat imagery services at http://glovis.usgs.gov. All Landsat image were cloud free and were acquired between January and March of the respective year. The specific date and year for this study was selected because of the availability of cloud free Landsat imagery across the five sites. GIS data like; administrative boundary, road network, and other physical features were collected from particular city corporation office, office of the City Development Authority and Urban Development Directorate (UDD). The details data characteristics, sources, temporal scale and spatial resolution are described in Table 2. To determine the extent or physical delimitation of the city study area, in most case, city corporation boundary was considered. However, for cities like Sylhet and Barisal, master planning boundary was taken into account, and for Khulna, the north boundary of the city corporation area was extended due to agglomerations of continuous built-up area and the density of contiguous ambit.

2.4. Image classification Accurate information on urban growth and land cover change derived from satellite images is crucial for the monitoring of urbanization, land management and environmental change studies (Turner et al., 2007). However, acquiring accurate information for such studies can still a challenging task, particularly in urban areas, due to extremely heterogenous landscapes and often occurring problem of mixed pixels. For example, sands or bare soils are often confused with bright impervious surfaces due to their similar spectral reflectance. To deal with these problems, a wide range of parametric and non-parametric methods for analysis of airborne and satellite derived imagery are in practice and continue to be proposed. Since increasing accuracy in land cover classification is essentially important for remote sensing analysis, data mining technology and machine learning algorithms such as Support Vector Machine (SVM), Artificial Neural Network (ANN), Decision Tree (DT) and ensembles classifier such as Random Forest (RF), used in remote sensing studies have shown a quantum increase in recent times. In particular, SVM (Vapnik, 2000) and Random Forest (Breiman, 2001) have recently received considerable attention in remote sensing image classification due to their ability to successfully handle small training data sets with computational efficiency, and often producing higher classification accuracy than the traditional methods (Qian et al., 2015; Mountrakis, Im, and Ogole, 2011). Both SVM and RF are supervised, non-parametric statistical learning techniques (Cutler et al.,

2.3. Pre-processing of the input data Prior to image processing, the Landsat images were stacked to obtain multi-band composite images and later subset with respect to the study area, in order to reduce computation and image analysis time. Using road network data, a comparatively high-resolution image from OLI-TIRS was geo-referenced mostly through ArcGIS 10.2.2 and Erdas Table 2 Description of the collected geospatial and socio-economic data. Data Types Metropolitan Master Plan (Urban Development Plan, Structure Plan) and Detailed Area Plan Landsat MSS, TM, ETM and OLI-TIRS

Year

Producer

Scale

UDD, LGED 2/21/1973, 1/27/1989, 1/26/2000, 3/21/2014

Physical Features, Road Network Data Socioeconomic data, Population Census, Urban Area Report

1984,1987,1991,2001,2008 2011,2013, 2014

Topographic Map Urban Poor Settlement Survey

1956 2011

72

USGS global land cover facilities http://glovis.usgs.gov. UDD, CUS, LGED Bangladesh Bureau of Statics (BBS) US Navy Center for Urban Studies (CUS)

60 m, 30 m Spatial Resolution

1:250000

Remote Sensing Applications: Society and Environment 7 (2017) 69–83

M.M. Hassan

within each sector was calculated (Xu et al., 2007; Rafiee et al., 2009). Finally, these values were displayed as spider graph (Fig. 6), allowing the spatial distribution of built-up areas to be visualized and illustrated.

2007), which are not imbedded with assumption problems, and typically involve the identification of multiple classes (Jin, 2012). Although non-parametric based algorithms for image classification methods have shown high image classification accuracy rates, this study uses rule-based supervised image classification techniques – a maximum likelihood classification (MLC) algorithm to classify six land cover classes in the five cities of the study area. The reasons for choosing MLC are self-evident: firstly, the purpose of this study is to examine spatial extent of built-up areas, rather than focusing on shape, dimension and the internal variation of various urban structures. Secondly, classification accuracy of non-parametric based algorithms mainly depends on availability of spatial, spectral and temporal resolution of remote sensing data. Lastly, MLC for urban land cover classification and impervious surface mapping have been employed in a number of previous studies and is still used because this algorithm greatly reduces the data requirements while still providing a greater potential to extract detailed information on urban landscape (Jat et al., 2017; Ramachandra et al., 2015; Yuan et al., 2005; Yin et al., 2005). Hence, considering the research objectives, data availability, compatibility with previous work and the availability of the algorithm in mostly image processing software, maximum likelihood algorithm in Erdas Imagine platform is used in this study for Landsat image classification for the five cities. Six types of land cover classification scheme for this study was developed, partly derived from Anderson et al. (1976) first-order hierarchical classification system and based on previous knowledge and field investigations combined with additional information from previous research. Land cover classification types were determined by a combination of supervised and unsupervised (ISODATA) classification (Suribabu et al., 2012), as well as manual interpretation of satellite images (Peng et al., 2011; Bui et al., 2013) topographic maps, survey data and ground truth information. In this way, six thematic land use and land cover categories were generated (Table 3) and classified using training sites or signature files that appeared fairly homogeneous on the image. Nearly 30–60 signatures were collected for each land cover classes and were submitted for statistical analysis of similarities (Wondrade et al., 2014). Using Google Earth, and GPS field survey data in addition to an image enhancement technique, these land covers were manually distinguished and a reasonable signature was extracted from each land use class. Using these signature files, supervised classification was carried out with a maximum likelihood algorithm (Yuan et al., 2005). Finally, to reduce potential salt-and-pepper effects, a 3 × 3 majority filter was applied before the classified land covers were used for further analysis (Yin et al., 2005; Bui et al., 2013). To demonstrate the spatial configuration of urban expansion, eight transect zones, forming fan shaped areas, were created with respect to pre-determined urban center. The fan-shaped transects were drawn in GIS, referenced around the location of each study area city corporation building in the center of the old town. Then using a GIS overlay, the built-up areas that fall within each zone were aggregated and the total area of each category

3. Results and discussion 3.1. Accuracy assessment Since image heterogeneity presents a major challenge in land cover classification, accuracy assessment was carried out in combination with field observation GPS data, manual interpretation of Landsat images, topographic maps, google earth high resolution images, and random sampling. Topographic map with random sampling, for example, was used to validate the thematic map of 1973 and 1989, while GPS point as reference dataset associated with training sample extracted from high resolution google map and Landsat images were used to validate the land cover map of 2000 and 2014. Since GPS ground truth data were mostly collected in the vicinity of the road network, additional sets of validation points using random sampling for each landcover map was generated to ensure unbiased distribution of verification data across the study area (Fig. 2). In this way, nearly 200–250 (varied from year to year) stratified sampling points were generated (Liu et al., 2009; Shooshtari and Gholamalifard, 2015; Yi et al., 2016)and then used with GPS data for class-specific accuracy assessment. The number and spatial distribution of verification point were varied due to dominance of certain land cover classes. Waterbodies for example; having consistent spectral signature across the landscape and easy to distinguish from other classes. Thus, it requires small set of validation point but provided higher classification accuracy (< 97% for all periods) then other classes. On the other hand, fallow land and agricultural land are often confusing and perhaps the most challenging task in image classification, particularly in tropical region like Bangladesh where fallow typically covers with grass, given the similar spectral reflectance as agricultural land. As a result, fallow land in this study had the lowest classification accuracy rate around 62–79%. The overall accuracy assessment result was also varied with the respect of the study area, availability of auxiliary data and the spatial resolution of the thematic map. Accuracy assessment, for example, were higher with the map of 2014 and 2000, ranges between 90 and 93 with Kapa statistics 81–87 due to available of supporting documents. However, for the map 1989 and 1973, assessment result was between 81.17 and 86 with Kappa values ranges from 73 to 77, which are good but not represent strong agreement (Congalton, 1991). However, these minor drawbacks should not compromise the goal of this study to illustrate the urban dynamics and utilities of landscape metrics in landscape fragmentation studies. 3.2. Spatial and temporal patterns of urbanization across the five cities The selected cities for this study are the major centers for banking and finance, retailing, trade, transportation, tourism, real-estate, business, political, and administrative functions that house nearly 8% of the country's total urban population (BBS, 2014) and approximately 20% of its economic units (BBS, 2013). Due to a different geographical location, topographical setting, and regional factors, as well as administrative and economic functions, city growth is looked at and studied for various spatial scales and temporal dimensions. Fig. 3 and Table 4 utilize satellite images for different time periods over the past four decades to clearly illustrate the radical transformation of the landscape pattern and the typical urban growth in the five selected cities (Fig. 4). Of the five study areas, four cities i.e., Rajshahi, Sylhet, Barisal and Khulna, began and then developed along the banks of great rivers, including the Padma, Suruam, Kittonkholam and Rupsha-Vairob, respectively. The growth and expansion of these city landscapes was primarily dominated by the meandering patterns of these rivers, the topography of the river valleys in which each city stands, and the circuitous nature of their transport routes. Later development of the Assam

Table 3 Land use/land cover classification scheme. Land Cover Types

Description

Urban/ Built up area

These include all developed land, including commercial, residential, industrial and other infrastructure Land used for agriculture, paddy field, vegetables, fruits and other cultivable lands Lake, ponds, lagoon, river, aqua fishing and vast sea water Hilly forest, homestead vegetation, coastal mangrove, bush and shrubs Open hill, exposed hilly soil, landfill, barren land, bare soil, Sandy beaches, Dune and River bank Accretion Land, Deposited land, river bank

Agriculture/Crop field Water bodies Forest and Vegetation Fallow land Lowland/Wetland

73

Remote Sensing Applications: Society and Environment 7 (2017) 69–83

M.M. Hassan

Fig. 2. GPS training point and random sampling location on map (Study area Khulna).

had a major influence on the development of that township. The area where the primary township developed was the inner side of the loop of this meandering river which was thus safe from erosion and flooding. It was positioned at a comparatively higher altitude from the surrounding area and was easily serviced by its water linkage. From observing the topographic map of 1956 and the Landsat image of 1973, one can see that the city's early growth was restricted at Chawck Bazaar and surrounding the port and launch terminal where a dense built-up was found inside the oval- shaped ring road along the Sadarghat and Port strip.

Bengal Railway (which functioned from 1892 to 1942) catalyzed substantial urban development in the city's heartland. An efficient circulation network, location, and linkage between the areas with regional towns and growth centers were important elements in both the township formation and the development of these urban areas. 3.2.1. Quantifying spatial pattern and urban growth trajectory in Barisal The coastal city of Barisal, in Southern Bangladesh grew and thrived along the northwest bank of Kittonkhola, a navigable river, which indeed 74

Remote Sensing Applications: Society and Environment 7 (2017) 69–83

M.M. Hassan

Fig. 3. Spatio-temporal land cover pattern in five cities (from 1973 to 2014). Fig. 3

observed, suggesting impressive growth toward the north (15% expansion), northwest (27% expansion) along the Dhaka-Barisal highway and also toward the southwest (25.6%), which is comparatively low lying and undulating flood prone, backwash areas, thereby consuming fresh agricultural land, vegetation cover, and wetland. In the following year, the population and the built-up area continued to increase in the city, but not as rapidly as during the previous decade. For example, the urban area in the period 1989–2000 increased to 1661 ha or 1.7 fold greater (it was 2.3 fold in 1989) than in the

The built- up area throughout these early periods was around 421 ha or only 5% of the total land area. Both the population and urban expansion in the study area became impetuous between the 1974 and 1981 census periods when net population increase was 45% with an annual growth rate of 6.4% (BBS, 1987, 1998). Built-up cover during this period (1973–1989) also exhibited the highest growth (aggregate growth 131.35%) at a rate of 8.20% annually and increased its share to 11.25% (974 ha) in city landscape formation. This increase in population growth accompanied with economic progression in the area was

75

Remote Sensing Applications: Society and Environment 7 (2017) 69–83

M.M. Hassan

Table 4 Land use/ land cover statistics of the study areas (1973-2014). Rajshahi

2014 (ha)

%

2000 (ha)

%

1989 (ha)

%

1973 (ha)

%

Agriculture Built-up Fallow land Lowland Vegetation Waterbodies Total (ha) Rangpur Agriculture Built-up Fallow land Lowland Vegetation Waterbodies Total (ha) Sylhet Agriculture Built-up Fallow land Lowland Vegetation Waterbodies Total (ha) Khulna Agriculture Built-up Fallow land Lowland Vegetation Waterbodies Total (ha) Barisal Agriculture Built-up Fallow land Lowland Vegetation Waterbodies Total (ha)

978 2708 64 99 949 447 5245

18.65 51.63 1.22 1.89 18.09 8.52 100.00

1671 1864 87 193 948 482 5245

31.86 35.54 1.66 3.68 18.07 9.19 100.00

2328 966 87 268 1247 349 5245

44.39 18.42 1.66 5.11 23.78 6.65 100.00

2440 433 52 231 1780 309 5245

46.52 8.26 0.99 4.40 33.94 5.89 100.00

1999 1831 59 64 972 101 5026

39.77 36.43 1.17 1.27 19.34 2.01 100.00

2440 936 38 58 1348 206 5026

48.55 18.62 0.76 1.15 26.82 4.10 100.00

3124 392 24 42 1269 175 5026

62.16 7.80 0.48 0.84 25.25 3.48 100.00

3550 110 20 54 1169 123 5026

70.63 2.19 0.40 1.07 23.26 2.45 100.00

428 3055 163 92 425 183 4346

9.85 70.29 3.75 2.12 9.78 4.21 100.00

605 2686 52 98 716 189 4346

13.92 61.80 1.20 2.25 16.47 4.35 100.00

769 2018 38 124 1254 143 4346

17.69 46.43 0.87 2.85 28.85 3.29 100.00

1115 484 85 234 2233 195 4346

25.66 11.14 1.96 5.38 51.38 4.49 100.00

203 3772 53 487 817 148 5480

3.70 68.83 0.97 8.89 14.91 2.70 100.00

456 2798 79 538 1380 229 5480

8.32 51.06 1.44 9.82 25.18 4.18 100.00

964 1614 113 307 2143 339 5480

17.59 29.45 2.06 5.60 39.11 6.19 100.00

2477 1027 245 225 1302 204 5480

45.20 18.74 4.47 4.11 23.76 3.72 100.00

2049 2069 192 77 3548 724 8659

23.66 23.89 2.22 0.89 40.97 8.36 100.00

3277 1661 49 330 2632 710 8659

37.85 19.18 0.57 3.81 30.40 8.20 100.00

4234 974 17 198 2332 904 8659

48.90 11.25 0.20 2.29 26.93 10.44 100.00

6003 421 61 39 1431 704 8659

69.33 4.86 0.70 0.45 16.53 8.13 100.00

have contributed to the primary development of the Barisal city in an east-west alignment, later moving northwest along the Dhaka-Barisal highway. Presently, the built-up area in the city has spread mainly along three corridors; in the northwest (26.4%) to the Dhaka-Barisal Highway, in the southwest (35.5%) along Dhaka-Patukhali and westward (16%) to the Brisal-Jhalakathi highway. Together these three trajectories accounts for nearly 78% of the built-up area in the city. The elevation heights of these areas are 6–10 feet above sea level. Due to the natural barriers of the Kirtankhola River, the city of Barisal cannot extend south, southeast, or east. In addition, due to low lying land, spatial expansion of the already built-up area cannot spread further westward.

Rajshahi

Rangpur

1973 Sylhet

1989 2000 2014

Khulna

Barisal

0

500

1000

1500

2000 Hectares

2500

3000

3500

4000

3.2.2. Quantifying spatial pattern and urban growth trajectory in Khulna In the southwest, the city of Khulna, is mainly an expansion of trade centers that were situated close to the Rupsa and Bhairab Rivers. Khulna has long been the major center for trade and commerce and is the third largest economic center in the country after Dhaka and Chittagong. The city has a linear shape, stretching north and south along the west bank of the rivers, Bhairab and Rupsa, respectively. It was observed from the topographic map of 1956 that the primary growth center was on the southwest bank of the Rupsha River, surrounding the location of the present City Center and Dack Bangla in Sadar thana, where a relatively compact urban form was found. A ribbon shaped development with sparsely populated settlement was documented along the Eastern Bengal railway corridor to the northwest of the city. During this period, the estimated urban growth, as seen from the topographic map, was approximately 350 ha with very slow organic

Fig. 4. Comparative urban/built-up growth trend in five cities (1973–2014).

previous era when the annual growth rate was calculated at 6.4%. Population at this time was growing at the rate of 3.2% in the city (BBS, 2008). Still, the major urban growth trajectory was toward the north, southwest and northwest at 18%, 26.6% and 37.8%, respectively. Due to population growth and the subsequent expansion of trade and commerce, the current development of the city is radiating southwest, north, northeast, west and northwest- from the banks of the Kirtankhola River and extending into rural areas and beyond the outskirts. Such geographical expansion of the urban cover in the city amounted to 2069 hectors in 2014 and comprised 24% of the city's landscape. Topographical features of the area, together with the coast factors, 76

Remote Sensing Applications: Society and Environment 7 (2017) 69–83

M.M. Hassan

thoroughfare, Natore-Rajshahi lane at Boalia Thana. There was also a sparsely populated settlement near the eastern corridor beside the Damkura Passage that ran toward the west alongside the DhakaRajshahi Highway. The major expansion of Rajshahi began only after the liberation of Bangladesh, while the city was connected by rail and road with other regional growth centers in north Bengal. Within a few years, Rajshahi's developed area grew to 966 ha, as estimated from a thematic map showing growth between 1973 and 1989. The city's developed area steadily expanded outwards at rates which often exceeded 10% per year, in south-east, south-west and west directions, which comprised 10.3%, 14% and 42% of developed land, respectively. Due to the presence of prominent educational institutions like Rajshahi University of Engineering Technology (RUET) and Rajshahi University in the eastern edges of the city, which occupy one-third of the current city corporation land, the built-up area could only expand between the area of the river embankment road and the Dhaka-Rajshahi Highway in the east. Rajshahi city's growth however, nearly doubled in 2000, when builtup exceeded to 1864 ha with population increases to 388,811 persons in the city corporation area and 651,062 persons in the statistical metropolitan area (BBS, 2008). Still, the bulk built-up expansion took place in the west (33%) surrounding Munshipara, Haragram, Bullanpara and along the Rajshahi-Nawabganj Highway. To the north it expanded along the major thoroughfare (airport road), a new residential area at Uposhohor and industrial area at Sopura on the northern edge of the city. Furthermore, some residential (e.g. Padma, Parijat and Chayanir) and industrial development projects also contributed to the built-up development in the city over this period. In addition, several road developments, such as the Alupatti-to-Sonadidhi road through Sahebbazar, the extension of greater road from Gourhanga to Talaimari road and Kalpana cinema hall to the rail station road during the 1990s allowed substantial land development in adjoining area. The city Rajshahi has experienced recent growth in all directions, particularly in the north along the airport road, in the west parallel to the Padma embankment and city bypass corridor, and in the east along the Dhaka-Rajshahi Highway. Built-up cover diffusing outward from the city core, with leapfrog development, scattering urban settlement over the rural landscape by consuming agricultural land, vegetation cover, water bodies, low land and fallow land that are present in between city periphery and inner circle of Rajshahi bypass thoroughfare. Currently the built-up area in the city corporation area exceeds 2708 ha as calculated from thematic map of 2014 and the city population is 449,756 persons as of the 2011 census (BBS, 2014).

growth and a comparative stable population. The process of urbanization, however, accelerated between 1950 and 1960 when planned industrialization took place at Khalishpur, including the establishment of an anchorage at Chalna with the subsequent declaration of that city as a regional headquarters in 1961. Until this time, the population in the city was 128,000, but it grew more than three-fold to 468,000 persons in 1974 (BBS, 1984) when the built- up area reached 1027 ha in the Metropolitan region. Using the easy and navigable access of the rivers Rupsa and Bhairab and convenient linkages with other regional towns and growth centers, the city has become the most important center of business, commerce, industry, tourism and transportation hub in the region. As a result, both population growth and subsequent urban expansion received stimulus after the post-independence era of Bangladesh. For the 1974–1981 census, the population in Khulna city totaled 665,417 persons (BBS, 1984) while built-up expanded to 1614 ha in 1989, in where industrial installations with residential uses were the main occupants of urban land. Throughout this phase, major urban expansion took place to the northwest along the Jessore-Khulna highway (this trajectory remains the major built-up expansion at 56%) and surrounding Khalishpur where major industrial sites like jute mills and other heavy industry (e.g. Khulna newsprint mills and Khulna hardboard mils) are situated. On the other hand, residential development continues to the south, adjoining the immediate periphery of the old town, and heading northwest near the location of Boyra, Sonadnga, Daulatpur, and the Gialtola Cantonment. In the following periods, 1989–2000, the city witnessed substantial urban growth when the built-up surpassed 2800 ha, nearly double in size compared to the previous decades. The net built-up growth was 73.3% with an annual increasing rate of 6.6% over this period. This huge urban expansion may have been linked to several ambitious projects, including 9 residential development projects on 265 acres of land at Sonadanga, Nirala, Mujgunni, Daulatpur, Shiromoni, and Mirrerdanga. In addition, a planned heavy industrial estate was developed at Shiromoni on 511.29 acres of land (http://www.kda.gov. bd/City_Development.php). Moreover, the establishment of a prominent education institution like Khulna University (1991) on the southwest outskirts and later construction of a city by-pass on the west periphery of the city that connecting with the Ruphsa Bridge on the south and the Jessore highway on the north, allowed exponential development in the Western region of the city by consuming lowlands that had once been used for agriculture and shrimp farming purposes. The city's recent growth, however, has been observed as moving in a westward direction along the four major corridors, including to the northwest surrounding the KUET road, along the old Satkhira link and Satkhira-Khulna highway, the Bastuhara bypass link, Jalil Shorani, adjacent to Khulna University and near to the Zero point. All these areas have mainly witnessed residential development by filling in of lowlands and encroaching on fresh arable agricultural land, water bodies, and vegetation cover. As a result, the built-up cover in the city exceeded 3772 ha by 2014 and accounted for nearly 69% of the city's land cover, while crop fields and vegetation cover declined 92% and 37%, respectively during that same period.

3.2.4. Quantifying spatial pattern and urban growth trajectory in Rangpur In the case of Rangpur, little or no organized urban forms were documented until the post-independence era of Bangladesh. The primary settlement according to a topographic map from 1956 was to the north-east edges of the city surrounding the present Sath-matha point (Seven road conjunction) and extending towards south-west to northeast, along the Eastern Bengal Railway corridor. The city, however, later shifted towards the north-west corridor due to the construction of some major government buildings, including the cantonment, the central hospital, administrative and education facilities, and the densification of road infrastructures coupled with proximity to commercial facilities. As a result, a gravitational shift took place in the location of railway corridors to road linkage corridors, and these places later become the main urban development area. In 1973, the current city growth areas were primarily agricultural with only a few small houses was found along the major road network. This predominantly rural, agricultural landscape gradually changed to a peri-urban landscape due to increased human settlement in the 1970s and 1980s, and subsequently, built-up space spread toward southwestward form the northwest along the main road. Presently the city accounts for nearly 1831 ha of built-up area, which is nearly double than in the year 2000 and five times higher than in 1989. The city's current growth

3.2.3. Quantifying spatial pattern and urban growth trajectory in Rajshahi In northern Bangladesh, the city of Rajshahi, is the oldest and sizeable urban form in the region, dating back to seventeenth century, when it was the preeminent trading center for silk with the Dutch and later the French and English. The city's primary growth, as it has been observed from the topographic map of 1956, is a rectangular shape with approximately 350 ha urban form (calculated in GIS) grounded just between railway and river, surrounding the location of the Dargah para and Shaheb bazar parallel to the river bank, and its alignment closely touching the northern margin of the Eastern Bengal Railway. The river in the south and railway in the north served as barriers that forced the town to grow in a ribbon-like form almost exclusively along a single 77

Remote Sensing Applications: Society and Environment 7 (2017) 69–83

M.M. Hassan

Fig. 5. Net gain and loss land cover classes in five cities over 41year study period.

4000

Gain

3000 2000

Hectares

1000 0 -1000

Loss

-2000 -3000 -4000 -5000 Rajshahi Rangpur Sylhet Khulna Barisal

Agriculture -1462 -1551 -687 -2274 -3954

Builtup 2275 1721 2571 2745 1648

Fallow 12 39 78 -192 131

Lowland -132 10 -142 262 38

Vegetation -831 -197 -1808 -485 2117

Waterbodies 138 -22 -12 -56 20

remittance enabling rural people to move to the city to take advantage of the security and the upgraded civic facilities. The majority of this migrant population previously lived on the northern bank of the Surma River. In absolute numbers, the population increased to 117,398 in 1991 census, doubled from the previous census in 1981 (BBS, 1998). In the following decade (1989–2000), built-up areas have continued to grow (at the rate of 3% per year), but at a less rapid pace than before. The urban developed area reached 2686 ha in 2000, constituting 62% of the urban land cover. Although built up growth in the city and its adjoining periphery dropped to 3.0% per year between 1989 and 2000 period, the population by this time increased remarkably to 320,280 people at the rate of 17.28% (1991–2001 census period) annually in the city (BBS, 2008). Given the physical constraints in the north, further outward expansion was restricted, and as a result, subsequent urban growth was directed toward the eastward corridor (24.34%) along the Dhaka-Sylhet highway and north-west corridor (20.70%) through the Sylhet-Sunamgang highway and these axis becomes the main trajectory for the recent urban development. Currently the city enjoys higher population growth at the rate 6.59% per year as estimated from 2001 to 2011 population census (BBS, 2014) while the built up expansion in the city comparatively stable and observed slow growth at the rate of 1.0% per year as estimated from the thematic map between 2000 and 2014 period. Currently the net built up expansion in the study area is 3055 ha which accounts for 70.29% of urban land and home of nearly 531,663 population (BBS, 2014). Over the 41-year, the selected cities in this study has experienced huge urban/built-up expansion by expense of agricultural land and vegetation cover. As represented in Fig. 5, the net amount of agricultural land and vegetation cover in the five cites was lost by 9928 and 3321 ha respectively while built-up was gained by 10,960 ha. Highest built-up expansion was taken in place like Khulna, Sylhet and Rajshahi at net amount 2745, 2571 and 2275 ha respectively. Vegetation and agricultural land, on the other hand, was lost by 485 and 2274 ha in Khulna, 1808 and 687 ha in Sylhet and 831 and 1462 ha in Rajshahi respectively. Vegetation cover, however, city in Barisal appears to be increased by 2117 ha. These are due to mainly much of the study areas are rural-urban mixed and partly because of increasing number of homestead vegetation resulting from reforestation program in the coastal areas of the country.

accelerated because of its recent administrative status as well as reputed academic institutions like Begum Rokeya University, Rangpur Medical College and Cadet College. Due to the presence of several textile, jute, and pharmaceutical companies, the area has been assuming the status of a large urban agglomeration comprising these industries and other commercial development in the inner and outer belts of city areas. The recent growth trajectory as observed from the land cover map of 2014, expanding towards the south (20.3%) surrounding the Modern stoppage, towards the manmade barrier of the railway to the east (3.5%), towards the north-west (19.3%) along the RangpurDinajpur Highway and to the west (27.1%) along the City Bypass. The city's periphery has been extending into rural areas since the mostly flat or gently rolling countryside presents no obstacles to the area's development. Unauthorized and unplanned developments at the edge of the city, especially haphazard and piecemeal homesteads, commercial areas, and other non-conforming land-uses, occurring mainly along the major routes of communication, were observed during the field survey. 3.2.5. Quantifying spatial pattern and urban growth trajectory in Sylhet The growth of the north-eastern city- Sylhet was primarily influenced by the commercial cultivation of tea, the importance of the shrine of the saints Hazrat Shahajalal and Shahparan, abundant natural resources (e.g., major gas and oil reservoir), being an important tourist gateway to the country, as well as the economic impact of international remittance from millions of overseas employees especially from the United Kingdom. The topographic map of 1956 and subsequent Landsat images suggest that the early growth of the Sylhet city took place on the flat plain of the river bank (where the average elevation is 20 m) surrounding the location of Kazir bazar, Kotwali, Lama bazar, and Sheikhghat. Later urban growth extended to the north-north-west along the major thoroughfares of Hazrat Shahjala shrine and the airport road. Based on the 1973 Landsat images, it is estimated that the net built up growth by this time covered 484 ha, accounting for 11.14% of the city land area. In the following period between 1973 and 1989, city observed a substantial built-up expansion at approximately 317%, with annual growth rates of nearly 20%. The bulk amount of the built-up expansion up to this time took place to the north and west of the city, which constitute 28% and 17% of the built up respectively. Later urban expansion extended over the hillock towards the north (16.25%), north east (18.41%), and west (19.44%) of the city. These huge built up extensions can be linked to massive infrastructure development, increasing civic facilities in the city. This expansion resulted in an important migration of the rural population into the city, especially those in the rural community whose relatives live in the United Kingdom. Therefore, one can assume that the growth rate may be due to foreign

4. Compositional and configurational changes in landscape pattern Anthropogenic activities such as urbanization continued to reveal, 78

Remote Sensing Applications: Society and Environment 7 (2017) 69–83

M.M. Hassan

Rangpur

Rajshahi

North

North

500 450 400 350 300 250 200 150 100 50 0

900 800

North West

North West

North East

700 600 500 400 300 200 100

West

West

East

0

South West

North East

East

South West

South East

South East

South

South

Sylhet

Khulna

North

North

800

2500

700

North West

600

North West

North East

500

2000

North East

1500

400 1000

300 200

500

100

West

West

East

0

South West

South West

South East

East

0

South East South

South

Barisal North 700

North West

600

North East

500 400 300 200 100

West

East

0

South West

South East South

2014

2000

1989

1973

Fig. 6. Spatial and temporal urban/built-up expansion trajectory in five cities.

by multiple impervious surfaces, possesses significant ecological threats to wildlife over time through the process of habitat loss and degradation due to landscape fragmentation (Wu et al., 2011; YeTao et al., 2010). Over the last decades, post classification land cover maps with landscape metrics have been used to study landscape fragmentation analysis (Seto and Fragkias, 2005; YeTao et al., 2010; Shrestha et al., 2012). Landscape metrics provide quantitative representation of habitat

landscapes become increasingly complex in shape and structure, leading to leap-frog and discontinuous development across the city (Yetao et al., 2010). Impervious surfaces including roads, buildings, and other man-made infrastructure serve as walls to animal movement, dividing wildlife populations, disrupting gene flow, metapopulation dynamics and species dispersal, transforming contiguous landscapes into small patches with low connectivity (Liu and Yang, 2015; Seto and Fragkias, 2005; Lausch et al., 2015). Thus, urbanization, characterized 79

Remote Sensing Applications: Society and Environment 7 (2017) 69–83

M.M. Hassan

% Urban Land

NP

MPS

80

18

2000

70

16

1800

60

14

1600 1400

12

50

1200

10

40

1000 8

30

800

6

20 10 0

600

4

400

2

200 0

0

1973

1989

2000

2014

1973

1989

2000

1973

2014

1989

AWMPFD

ED

2000

2014

LPI

120

1.28

100

1.26

80

1.24

60

1.22

40

40

1.20

30

20

1.18

0

1.16

80 70 60 50

20 10

1973

1989

2000

1973

2014

1989

2000

2014

0 1973

1989

2000

2014

2000

2014

2000

2014

Fig. 7. Changes in spatial metrics at class-level across the five cities.

PD

CONTAG

LSI

80 25

70

30

60

20

25

50 20

40

15 10

30

15

20

10

10

5

5

0 0

1973 1973

1989

2000

1989

2000

2014

1973

IJI

1.60

0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00

90 80 70 60 50 40 30 20 10 0

1.40 1.20 1.00 0.80 0.60 0.40 0.20 0.00 1989

1989

SHDI

SHEI

1973

0

2014

2000

2014

1973

1989

2000

2014

1973

1989

Fig. 8. Changes in spatial metrics at landscape-level across the five cities.

FRAGSTATS (McGarigal and Marks, 1995) spatial pattern analysis program. The results and generated matrix are shown in Fig. 7 and 8. Both class level and landscape level metrics computed in this study suggest that there are diverse landscape patterns and land fragmentation processes across the five cities. For example, the number of patches (NP) for the city of Barisal, Khulna, Rajshahi and Rangpur first increased and then declined, while the city of Rangpur continues to increase towards the end of the periods. NP basically describes the degree of subdivision for a specific class of landscape by counting the total

composition and configuration within landscapes by measuring fragmentation and describing specific patterns and process for a given landscape. As this study specifically focuses on landscape fragmentation in terms of spatial connectivity, the landscape metrics that are the most relevant to achieve the research objectives and provide insights into the key ecological components and urbanization patterns are employed and explained. Thus, six landscape level metrics, i.e., PD, SHEI, IJI, SHDI, AWMFD, LSI, and CONTG, and six class level metrics including MPS, NP, ED, AWMPFD and LPI were selected and computed using the 80

Remote Sensing Applications: Society and Environment 7 (2017) 69–83

M.M. Hassan

Diversity metrics quantify landscape composition by measuring richness and evenness of patch types (Shoyama and Braimoh, 2011). Diversity metrics are influenced by the number of land cover categories, and by patch richness and evenness. Shannon's Diversity Index (SHDI) measures relative patch diversity, or the proportional abundance of each patch type within the landscape. SHDI index tends to be zero when there is only one patch in the landscape and increases as the number of patch types increases. While Shannon's Evenness Index (SHEI) measures the patch distribution and abundance within landscape. SHEI is equal to zero when the observed patch distribution is low and approaches one when the distribution of patch types becomes more even. From 1973 to 2014 the diversity metric for the city of Barisal, Rangpur and Rajshahi increased constantly indicating that the landscape composition became more even and land use types were more equally distributed. However, this trend was inversed during 1989–2014 for the Khulna and Sylhet, suggesting an increasing heterogenous in that particular landscape. In contrast, the subsequent slight decline in SHDI and SHEI between 2000 and 2014 reflected that the continual increase in urban land use truncated the heterogeneity and made landscape uneven again. CONTAG is a landscape-level metric that measures to what extent landscapes are aggregated or clumped as a percentage of the maximum possible (O'Neill et al., 1988). CONTAG approaches 100 when all patch types are maximally aggregated and the landscape consists of single patch. While CONTAG approaches 0 when the patch types are vastly disaggregated with landscape is dominated by greater number of smaller patches. The CONTAG tend to decrease over time and particularly at a higher rate in Barisal, Rangpur and Sylhet, however the values again became higher since 2000. This tendency affirmed that landscape was vastly disaggregated and less contiguous pattern during the period of 1989–2000 then it tends to be homogeneous, mainly influenced by increasing urban classes. This fact is already evident as the land cover map shows the agricultural dominated urban landscape of 1973 became spatially heterogeneous and fragmented during the expansion of anthropogenic structure and vegetation cover by 1989. Later of the study period, vegetation and agriculture becomes segmented into smaller patches, as a result, a new matrix of urban development has replaced over the vegetation matrix. The area-weighted mean patch fractal dimension (AWMPFD) describes the shape complexity of a landscape (Tian and Wu, 2015). AWMPFD approaches 1 for landscapes with simple shape of perimeter while it approaches 2 for highly complex shape of perimeters. The fragmentation of the built-up patches, expressed by the AWMPFD, shows a fluctuation and diverse trend across the five landscapes. The value of AWMPFD shows an increasing trend at the end of the period before it dropped significantly in 2000. This trend, however, was higher in Barisal and Rajshahi then Rangpur, Sylhet and Khulna. This confirmed that the landscapes for the five cities had complex shape perimeters between 1973 and 1989 then again in 2014. The last metric computed for this study is the Interspersion Juxtaposition Index (IJI), which measures patch adjacency or how patch types are interspersed or intermixed within a landscape. IJI is calculated in percentage and approaches 100% when all classes are well interspersed and approaches zero when patch types are poorly intermingled. IJI tends to decline for the city of Barisal, Khulna and Rajshahi before it reached its highest of around 60% in 2000. For the city of Rangpur, the IJI value tended to increase from 45% in 1973 to 65% in 2014. The IJI value for Sylhet increases in 2014 but at a lower level than in 1973. Overall the IJI results indicate that Sylhet and Rangpur city landscape leading to a more intermingled than Rajshahi, Khulna and Barisal. They also indicate that the highest interspersion for all cities during the period of analysis occurred in 1989–2000. This possibly indicates a trajectory of urban expansion and landscape transformation within the study area. The gradual changes in landscape metrics across the five cities suggest that the landscapes are constantly undergoing a major transformation from one dominant land cover to another resulting from

number of patches for a given area (Turner et al., 2001). While NP is best used in conjunction with other landscape metrices, increasing patches in a given landscape generally indicates greater fragmentation with less continuity. Initially, the city of Barisal showed the greater number of vegetation patches and peaking in 1989, while urban and agricultural patches both grew gradually, reaching the greatest numberaround 550 patches- at the end of the study period. This indicates a highly-fragmented city landscape due to high number of small patches. Vegetation patches across the five cities remain highest over the study period compared to a different time period. This may be a result of the expansion of homestead vegetation and urban sprawl at the rural-urban interface at the city edges. LPI quantifies landscape composition through the percent of landscape occupied by the largest patch (Mcgarigal, 2001). It is the ratio of the area covered by the larges patch in the landscape to the total area of landscape. LPI has widely been used as an indicator of landscape fragmentation (Sun et al., 2014; Zhang et al., 2004; Zhang et al., 2012). LPI approaches 0 when the largest patch of the corresponding patch type is increasingly small and 100 when the entire landscape consists of a single patch. Largest patch index (LPI), across the five cities sharply declined since 1973 but then increased from the year of 2000. This implies that the once dominant vegetation and agricultural land cover declined due to rapid expansion of urbanization and were replaced by built-up patches. LPI both in Barisal and Rangpur decline at 44% and 60% respectively, while in Khulna, Sylhet and Rajshahi LPI declined until 1989 at 40%, 16% and 57% respectively, then increased until the end of the study time period. The examination of LPI for each class allows a better understanding about the behavior of this metric than when analyzed for the entire landscape. LPI at the class level quantifies the percentage of total landscape area comprised by the largest patch. As such, it is a simple measure of dominance. For example, agricultural patch for all the cities were the dominant classes in 1973. However, it decreased with time, declining to 95% in Barisal, 97% in Rajshahi, 96% in Khulna and 78% in Rangpur due to increasing dominance of built-up land cover. The quantity and quality of the large patches have a direct impact on the total number and diversity of species in the ecosystem (Freitas et al., 2013). Patch size affects biomass, primary productivity, nutrient storage per unit area, as well as species composition and diversity (Asgarian et al., 2015; Wei and Zhang, 2012). As a general rule, the larger the patch, the better it is for wildlife conservation. Patch Size is used to measure the subdivision of habitat in a landscape and can serve as an indicator of ecosystem function as the size of patches can indicate the number of species that the area can support. Mean Patch Size (MPS) for all the cities in this study dropped sharply since the 1973–1989 and remained constant during the rest of the period. This might seem surprising; however, it can be explained by the loss of several of the smaller vegetation patches. PD and ED metrics both measure the complexity of the shape of patch and represent the spatial heterogeneity of a landscape mosaic (Wei and Zhang, 2012). ED is a measurement of landscape configuration with applications to the study of edge effects. Both PD and ED metrics grew steadily across the five cities which indicates the increasing fragmentation in the landscapes. There is a notable spike in PD for the city of Rangpur and Sylhet in 2000 and in ED for all five cities in 1989 which can be attributed to the proliferation of vegetation and agricultural patches. The highest rate of change of both metrics was observed in the period between 1989 and 2000. Examination of edge metrics at class level found both agriculture and vegetation edge density metrics was the highest in 1973 and keep growing to the rest of the period for the five cities except Sylhet and Khulna where built-up edges replaced in agricultural classes from the year of 1989. Shape metrics are as important as patch size metrics for the understanding of landscape configuration. LSI as shape metrics for this study increases over time in all cases, although it has lower value in Khulna and Sylhet and higher value in Barisal and Rangpur at the end of the period. (Fig. 8) 81

Remote Sensing Applications: Society and Environment 7 (2017) 69–83

M.M. Hassan

Government of the People’s Republic of Bangladesh, Dhaka, Bangladesh. BBS, 2014. Population and Housing Census 2011, vol. 3, (Urban Area Report). Bangladesh Bureau of Statistics. Ministry of Planning. Government of the People’s Republic of Bangladesh, Dhaka, Bangladesh. BBS, 1984. Bangladesh Population Census l98l (National Series, Analytical Findings and National Tables) Bangladesh Bureau of Statistics, Statistics Division, Ministry of Planning, Government of the People’s Republic of Bangladesh, Dhaka. BBS, 1987. Bangladesh Population Census1981 (Report on Urban Area) Bangladesh Bureau of Statistics, Statistics Division, Ministry of Planning, Government of the People’s Republic of Bangladesh, Dhaka. BBS, 1998. Bangladesh Population Census 1991, vol. 3, (Urban Area Report) Bangladesh Bureau of Statistics, Statistics Division, Ministry of Planning, Government of the People’s Republic of Bangladesh, Dhaka. BBS, 2013. Economic Census. Bangladesh Bureau of Statistics, Statistics and Informatics Division, Ministry of Planning, Government of the People’s Republic of Bangladesh, Dhaka. Breiman, L., 2001. Random forests. Mach. Learn. 45, 5–32. Bui, T.D., Maier, S.T., Austin, C.M., 2013. Land cover and land use change related to shrimp farming in coastal areas of Quang Ninh, Vietnam using remotely sensed data. Environ. Earth Sci. http://dx.doi.org/10.1007/s12665-013-2964-0. Congalton, R.G., 1991. A review of assessing the accuracy of classifications of remotely sensed data. Remote Sens. Environ. 37, 35–46. Cutler, D.R., Edwards, T.C., Beard, K.H., Cutler, A., Hess, K.T., 2007. Random forests for classification in ecology. Ecology 88, 2783–2792. Freitas, Marcos Wellausen Dias de, Santos, João Roberto dos, Alves, Diógenes Salas, 2013. Land-use and land-cover change processes in the Upper Uruguay Basin: linking environmental and socioeconomic variables. Landsc. Ecol. 28 (2), 311–327. http://dx. doi.org/10.1007/s10980-012-9838-9. Hass, J., Dorothy, F., Yifan, B., 2015. Satellite monitoring of urbanization and environmental impacts – a comparison of Stockholm and Shanghai. Int. J. Appl. Earth Obs. Geoinf. 38, 138–149. Hassan, M.M., Nazem, M.N.I., 2015. Examination of land use/land cover changes, urban growth dynamics, and environmental sustainability in Chittagong city, Bangladesh. Environ. Dev. Sustain. http://dx.doi.org/10.1007/s10668-015-9672-8. Islam N., 2011. Urbanization in Bangladesh. International Seminar on Urbanization, Asiatic Society of Bangladesh, Dhaka. Jat, M.K., Garg, P.K., Khare, D., 2017. Modelling of urban growth using spatial analysis techniques: a case study of Ajmer City (India). Int. J. Remote Sens.) J. Int. J. Remote Sens. 29 (2), 143–1161. Jin, Jiao, 2012. A random forest based method for urban land cover classification using LiDAR data and aerial imagery. 〈http://www.uwspace.uwaterloo.ca/handle/10012/ 6770〉. Lausch, Angela, Blaschke, Thomas, Haase, Dagmar, Herzog, Felix, Uwe Syrbe, Ralf, Tischendorf, Lutz, Walz, Ulrich, 2015. Understanding and quantifying landscape structure – a review on relevant process characteristics, data models and landscape metrics. Ecol. Model. 295, 31–41. http://dx.doi.org/10.1016/j.ecolmodel.2014.08. 018. Liu, Miao, et al., 2009. Land use and land cover change analysis and prediction in the upper reaches of the Minjiang River, China. Environ. Manag. 43 (5), 899–907. Liu, Ting, Yang, Xiaojun, 2015. Monitoring land changes in an urban area using satellite imagery, GIS and landscape metrics. Appl. Geogr. 56, 42–54. http://dx.doi.org/10. 1016/j.apgeog.2014.10.002. Mcgarigal, K., 2001. Landscape Metrics for Categorical Map Patterns 2001 (Chapter 5), pp. 1–77. http://dx.doi.org/10.1007/BF00162741. McGarigal, K., Marks, B.J., 1995. FRAGSTATS: spatial pattern analysis program for quantifying landscape structure. U.S.Department of Agriculture, Forest Service, Pacific Northwest. Mountrakis, G., Im, J., Ogole, C., 2011. Support vector machines in remote sensing: a review. ISPRS J. Photogramm. Remote Sens. 66 (3), 247–259 (Research Station, Portland, Oregon). Nazem, N.I., Chowdhury, A.I., Mahbub, A.Q.M., 2011. Economic Geography of Urbanization and Development, Bangladesh Urban Forum edited by Sarwar Jahan and Nurul Islam Nazem. O'Neill, R.V., Krummel, J.R., Gardner, R., Sugihara, G., Jackson, B., DeAngelis, D., Milne, B., Turner, M., Zygmunt, B., Christensen, S., Dale, V., Graham, R., 1988. Indices of landscape pattern. Landsc. Ecol. 1 (3), 153–162. Peng, J., Xu, Y., Cai, Y., Xiao, H., 2011. Climatic and anthropogenic drivers of land use/ cover change in fragile karst areas of southwest China since the early 1970s: a case study on the Maotiaohe watershed. Environ. Earth Sci. 64, 2107–2118. Qi, Zhi-Fang, Ye, Xin Yue, Zhang, Hao, Yu, Zi. Long, 2014. Land fragmentation and variation of ecosystem services in the context of rapid urbanization: the case of Taizhou city, China. Stoch. Environ. Res. Risk Assess. 28 (4), 843–855. Qian, Yuguo, et al., 2015. Comparing machine learning classifiers for object-based land cover classification using very high resolution imagery. Remote Sens. 7 (1), 153–168. Rafiee, R., Mahiny, A.S., Khorasani, N., 2009. Assessment of changes in urban green spaces of Mashad city using satellite data. Int. J. Appl. Earth Obs. Geoinf. 11, 431–438. Ramachandra, T.V., Bharath, A.H., Sowmyashree, M.V., 2015. Monitoring urbanization and its implications in a mega city from space: spatiotemporal patterns and Its indicators. J. Environ. Manag. 148, 67–81. http://dx.doi.org/10.1016/j.jenvman.2014. 02.015. Seto, Karen C., Fragkias, Michail, 2005. Quantifying spatiotemporal patterns of urban land-use change in four cities of china with time series landscape metrics. Landsc. Ecol. http://dx.doi.org/10.1007/s10980-005-5238-8. Shooshtari, Sharif Joorabian, Gholamalifard, Mehdi, 2015. Scenario-Based land cover change modeling and its implications for landscape pattern analysis in the Neka

effect of disturbances, and the growth of human activities. This dramatic land cover change stimulated by rapid urbanization in these five cities has resulted in a fundamental change of landscape patterns. Therefore, the impact of these spatial relationships on ecological process is a promising subject and important for further studies. 5. Conclusion This study quantitatively characterizes the land use/land cover pattern and dynamics of urban expansion in five divisional cities in Bangladesh using Landsat imagery enabling with GIS and Remote Sensing techniques over the period of 1973–2014. The results suggest that the selected study area has undergone rapid urban development over these 41 years. Urban/built-up areas have increased by 468%, covering about six times the urban area as in 1973, resulting in a significant decline in vegetation cover and crop fields. Urban pressure in Bangladesh, fueled by the granting of independence in 1971, resulted in gradual increases in quality of life as well as the emergence of new services and infrastructure, such as educational, administrative, health and other civic facilities, leading to progressive urban transformation in the country. As a result, both urban/built-up expansion as well as population growth in the cities were highest during the period of 1973–1989 as can be seen in this study. Highest growth among the five study areas, by this time, was documented in Rajshahi, Sylhet and Rangpur, at 12.36%, 11.02% and 16% respectively. From 1989–2000, all five cities maintained relatively stable growth, averaging 6% per year. Growth continued during the following year; however, the rate was in most cases less than 3%, as calculated from the land cover maps of 2000 and 2014. Major urban growth took place in the direction of southwest and northwest in the cities of Khulna and Barisal, while cities in the north expanded in all directions, particularly to the east, west, north and north-west. The findings also suggest a continuous increase in built-up areas replacing agricultural and vegetation areas, at a rate much faster than the urban population growth in all five cities, whose dynamics are supported by a pre-existing growth center and road infrastructure. The changes in CONTAG, ED, AWMPFD and diversity index indicate a higher spatial heterogeneity in the landscape due to increasing dominance of built-up areas. This process however, is markedly different in Rangpur, Rajshahi and Barisal where both vegetation and built-up predominately increases resulting in a highly fragmented landscape. As such expansion puts a strain on urban efficiencies and creates a wide range of management and environmental problems, GIS and RS technologies might be considered as important tools and techniques for analyzing the processes and effects of many rapidly growing urban areas in less developed countries, where alternative sources of information are limited due to lack of resources. Hence, the time series thematic map and spatial information generated from this study using remote sensing and GIS may assist local planners and stakeholders in the country as well as concerned academia to better understand urban dynamics at a local level. In addition, it will serve as a guide for determining the best use of limited land and other resources that are available at the local level, and in planning for an urban future based on a vision of sustainable urban development and its implications. References Afsar, R., 2003. Internal Migration and the Development Nexus: The Case of Bangladesh. Bangladesh Institute of Development studies, Dhaka. Anderson, J.R., Hardy, E.E., Roach, J.T., Witmer, R.E., 1976. A land use and land cover classification system for use with remote sensor data. U.S.GS. Prof. Pap. 964, 1–26. Asgarian, Ali, Amiri, Bahman Jabbarian, Sakieh, Yousef, 2015. Assessing the effect of green cover spatial patterns on urban land surface temperature using landscape metrics approach. Urban Ecosyst. http://dx.doi.org/10.1007/s11252-014-0387-7. BBS, 2008. Population and Housing Census 2001, vol. 3, (Urban Area Report). Bangladesh Bureau of Statistics. Ministry of Planning. Government of the People’s Republic of Bangladesh, Dhaka, Bangladesh. BBS, 2001. Population Census. Bangladesh Bureau of Statistics. Ministry of Planning.

82

Remote Sensing Applications: Society and Environment 7 (2017) 69–83

M.M. Hassan

417–428. http://dx.doi.org/10.1016/j.landusepol.2011.08.006. Wondrade, N., Dick, O.B., Tveite, H., 2014. GIS based mapping of land cover changes utilizing multi-temporal remotely sensed image data in Lake Hawassa Watershed, Ethiopia. Environ. Monit. Assess. 186, 1765–1780. Wu, Jianguo, Jenerette, G. Darrel, Buyantuyev, Alexander, Redman, Charles L., 2011. Quantifying spatiotemporal patterns of urbanization: the case of the two fastest growing metropolitan regions in the United States. Ecol. Complex. http://dx.doi.org/ 10.1016/j.ecocom.2010.03.002. Xu, J.G., Liao, B.G., Shen, Q., Zhang, F., Mei, A.X., 2007. Urban spatial restructuring in transitional economy—changing land use pattern in Shanghai. Chin. Geogr. Sci. 17, 19–27. YeTao, Y., Qiming, Z., Jianya, G., 2010. Gradient analysis of landscape spatial and temporal pattern changes in Beijing metropolitan area. Sci. China-Technol. Sci. 53, 94–98. Yi, Kunpeng, Zeng, Yuan, Wu, Bingfang, 2016. Mapping and evaluation the process, pattern and potential of urban growth in China. Appl. Geogr. 71, 44–55. http://dx. doi.org/10.1016/j.apgeog.2016.04.011. Yin, Z.-Y., Stewart, D.J., Bullard, S., MacLachlan, J.T., 2005. Changes in urban built-up surface and population distribution patterns during 1986–1999: a case study of Cairo, Egypt. Comput. Environ. Urban Syst. 29, 595–616. Yuan, F., Sawaya, K., Loeffelholz, B., Bauer, M., 2005. Land cover classification and change analysis of the Twin Cities (Minnesota) metropolitan area by multi-temporal Landsat remote sensing. Remote Sens. Environ. 98, 317–328. Zhang, Liquan, Wu, Jianping, Zhen, Yu, Shu, Jiong, 2004. A GIS-based gradient analysis of urban landscape pattern of Shanghai Metropolitan Area, China. Landsc. Urban Plan. 69 (1), 1–16. http://dx.doi.org/10.1016/j.landurbplan.2003.08.006. Zhang, Sainan, York, Abigail M., Boone, Christopher G., Shrestha, Milan, 2012. Methodological advances in the spatial analysis of land fragmentation. Prof. Geogr.. http://dx.doi.org/10.1080/00330124.2012.700501. (no. June 2015: 120730022224005).

Watershed, Iran. Remote Sens. Appl.: Soc. Environ. 1, 1–19. http://dx.doi.org/10. 1016/j.rsase.2015.05.001. Shoyama, Kikuko, Braimoh, Ademola K., 2011. Analyzing about sixty years of land-cover change and associated landscape fragmentation in Shiretoko Peninsula, Northern Japan. Landsc. Urban Plan. 101, 22–29. http://dx.doi.org/10.1016/j.landurbplan. 2010.12.016. Shrestha, M.K., York, A.M., Boone, C.G., Zhang, S., 2012. Land fragmentation due to rapid urbanization in the Phoenix Metropolitan Area: analyzing the spatiotemporal patterns and drivers. Appl. Geogr. 32 (2), 522–531. Sun, Yan, Zhao, Shuqing, Qu, Wenyuan, 2014. Quantifying spatiotemporal patterns of urban expansion in three capital cities in Northeast China over the past three decades using satellite data sets. Environ. Earth Sci. 73 (11), 7221–7235. http://dx.doi.org/ 10.1007/s12665-014-3901-6. Suribabu, C.R., Bhaskar, J., Neelakantan, T.R., 2012. Land use/cover change detection of Tiruchirapalli City, India, using integrated remote sensing and GIS tools. J. Indian Soc. Remote Sens. 40, 699–708. Tian, Guangjin, Wu, Jianguo, 2015. Comparing urbanization patterns in Guangzhou of China and Phoenix of the USA: The Influences of Roads and Rivers. Ecol. Indic. 52, 23–30. http://dx.doi.org/10.1016/j.ecolind.2014.11.024. Turner II, B.L., Lambin, E.F., Reenberg, A., 2007. Theeme rgenceof land change science for global environmental change and sustainability. Proc. Natl. Acad. Sci. USA 104, 20666–20671. Turner, M.G., Gardner, R.H., O’Neill, R.V., 2001. Landscape Ecology in Theory and Practice: Pattern and Process. Springer-Verlag, New York. United Nations, 2014. Department of Economic and Social Affairs, Population Division. World Urbanization Prospects: The 2014 Revision, Highlights (ST/ESA/SER.A/352). UPPR, 2011. Poor settlement in Bangladesh, urban partnership for poverty reduction. Local government engineering department. Vapnik, V.N., 2000. The Nature of Statistical Learning Theory. Springer, New York. Wei, Yaping, Zhang, Zongyi, 2012. Assessing the fragmentation of construction land in urban areas: an index method and case study in Shunde, China. Land Use Policy 29,

83