Towards a Remote Sensing and GIS-Based Technique to Study Population and Urban Growth : A Case Study of Multan

In recent decades, the migration rates of the large cities of Punjab have been risen up to a considerable level due to the lack of employment opportunities as well as lack of facilities in the rural areas of the province. It has caused an unprecedented and unplanned urbanization across the urban areas of the province. This study has been undertaken to perform fractal analysis about the sprawl in rapidly growing city. GIS and remote sensing data have been used in this study as an emerging technology which is cost effective as well as accurate at the same time. Landsat images have been taken for the study and the sprawl has been calculated with the analysis of the data of each decade for past more than 40 years. It has been observed that the built up area is 47.8 to 141.12 Sq. Km whereas the pattern of urban settlement has been classified as clustered and linear, following the roads network. The temporal population growth also seconded these results. The population growth rate and population density increase, are based on the pixel based extraction of the data from satellite imagery for the period of 2000 to 2014, which is taken as a decision support tool. In 2000, the population of the district increased from 2,071,694 (1981 census) to 2,939,907 and then in 2013, it became 4,384,191 at a rate to 2.93% upturn per annum. Moreover, the study also reveals the extent of the growth of other land uses as well which may be taken as a reference that in an agricultural country like Pakistan, the natural resources are being wasted (by urbanization of the fertile land). There must be some master planning to avoid such things in the other cities as well.


Introduction
At present, 50% of the world's inhabitants live in urban areas.By 2050, that ratio will have increased to more than 70%.The major challenge to the developing nations in this era is urban growth or urban sprawl spreading rapidly because of hasty growth of populace, rapid financial advancements and basic substratum advancements [1] [2].
Cities exist round 2% of the Earth's surface yet consume about 75% of the world's resources, and given that more of the world's inhabitants now live in urban lands than in rural lands.According to comprehensive analysis the population is growing rapidly worldwide and is increased from 13% to 49% within fifteen years and it is expected to redouble to 60% by 2030 [3].
The main reason of urban sprawl reprehend is due to the inefficient utilization of energy, lands, resources and extensive scale infringement on agricultural lands [4].So sustainable urban planning is inevitably a major challenge for the nations.
What needs to be done?It's all about well-informed management of resources, proper congruous management, measurements, mapping and integration of energy from renewable sources is vital for all nations.Also the monitoring of leveled urban sprawl is essential for government authorities and planners in any area.
Further for sustainable urban development the reliable and efficient information about spatio-temporal pattern and built up is essential for urban sprawl measurement [3].Globally all the researchers integrate GIS in urban planning for the development of sustainable environment sprawl analysis and mapping.
The use of GIS and RS is a distinct option for viable mapping and monitoring for urban sprawl [5].GIS based technology provides advanced analytical and management tools to spatially analyze and study patterns and spatial variations and correlations to make more informed decisions [6].
Remote sensing analysis helps in assessing the damage that urban growth has on the environment and GIS techniques are valuable in determining analytical and management tools to spatially analyze and study patterns, growth, natural extent and spatial variations and correlations to make more informed decisions [7].These technologies emerged as a feasible substitute because of cost effectiveness, high-tech reliability, and offers interminable and reliable record of spatial patterns.Also it is very suitable for change detection analysis.The repeated data acquisition, synoptic view and formats processed by PCs have made remotely sensed products appropriate for the analysis of change detection applications.

Objectives
1) To identify the urban sprawl through Spatio-Temporal analysis of Multan City.
2) To predict direction of urban growth in Multan City.
3) To compare escalation in Population Growth and Urban Sprawl.

Study Area
The fifth largest city of Pakistan is Multan which is situated between 29'-22' and 30'-45' North latitude, 71'-4' and 72'-4'55 East longitude [8].It consists of four tehsils Multan Cantonment, Multan Sadar, Shujabad and Jalalpur Pirwala, having a total area of 3721 square kilometers (District office, Multan).It is situated in a bent formed by coalescent of five rivers (see Figure 1).It is separated from District Bahawalpur by river Sutlej and separated from District Muzaffargarh by river Chenab.Its average altitude is around 215 meters above sea level (MCCI).It has topography of plain grounds with a vast network of rivers, canals and water channels, making it a highly fertile land.The north side of the district is highly populated.
To carry out this research work, it was thought initially to consider the entire districts as the study area, but after observing population growth trends and keen consideration about urban extent, the Multan city along with its peri-urban area has been selected for this research work having a total area of 1013.

Methodology
The previous studies for the chosen area were all based on paper maps.This

Data Collection from Primary Sources
The data from the primary sources was collected through visiting the district offices.The obtained maps were then digitized and brought into GIS environment using ArcGIS 10.2 software.
The served areas of water supply and sewerage was acquired to give a clearer picture of the lack of facilities in the city due to unforeseen urbanization.These layers include: 1) Administrative layers (District, Tehsil, Town and UC boundary see Figure 2).
2) MC and Ex-MCL boundary extracted from old master plans.
3) Sewerage served areas & water supply areas to demonstrate that due to haphazard spread in the urban fringes of the city, some areas lack in the basic facilities and these population pockets would keep on increasing in case of absence of

Data Collection from Secondary Source
The data of phase II was obtained from satellite imagery (see Figure 3) i.e.
Landsat MSS, TM, ETM and ETM+.The acquired images details are shown in Table 1.
The satellite images were acquired keeping in consideration that the satellite images should be free from cloud cover.The layers of the acquired data was then stacked and a subset of the study area (Multan city) was extracted and some image enhancements were applied for the clarification of the data.The pre-processing also includes radiometric and geometric correction so that the classified images must be free from errors.This spatial resolution along with spectral bands was

Data Generation and Post Processing
The data acquired from phase-I was brought into GIS environment by image rectification and then digitizing is applied on that data.All the layers were mapped and the built up area along-with and without water supply and sewerage facilities is demonstrate.Image classification of the satellite imageries is performed to get the change detection, direction and pattern analysis.For this purpose, supervised classification technique is utilized using maximum likelihood algorithm, non-parametric parallelepiped classifier in ERDAS Imagine and the whole image is categorized into different type of land uses i.e. built up, vegetation, water and Soil.The third type of data set, i.e. population data was acquired from landScan.

Results and Discussions
After all the above subjected processing, the all data was mapped using ArcGIS 10.2 software to visually figure out the results.Moreover, the areas of all the land uses were calculated for each to support the argument and results.

Land Cover
Satellite image classification is emerging as an efficient method to extract geospatial data and information from accessible as well as remote areas and is applied in multi-disciplinary fields [9].In order to produce a land use map of the city, the Landsat imagery has been processed in ERDAS Imagine 2014 software using supervised classification technique with maximum likelihood algorithm.141.12 Sq. km (see Figure 4).and it has been observed that the built up area has increased three times over the period of four decades (see Figure 5).Moreover, it is significant that this growth was more rapid during the last decade and is justified with the rapidly increasing population.

Sprawl Analysis
The city experienced a rapid but uneven urbanization during the past few decades.A considerable agriculture area has been converted into built up and hence a very rich fertile land has been wasted.Initially, the sprawl was taking place in the south-west direction and no ribbon development took place but as the construction of roads started (nearly during 1950s to 1970s), the city started growing  it is expected that the city will keep on growing even on a greater rate.The haphazard growth in south direction has been observed to remain limited due the railway line, the railway station and their respective ancillary installations.The linear growth resulted in a total increment in 23.05 sq.km built up area.This area does not include the expansion yielded in radial growth rather it is confined to linear growth only (see Figure 6).

Directional Change
The directional change of the built up has been designed using a combination of the past master plans as well as the latest satellite imagery based calculations.
The Map below illustrates the results starting from pre-Muslim period to 2015.
The results show a maximum expansion towards south-west direction towards the airport.The reason being the colonization on the area is near the airport.
Another signification increase in the sprawl has been observed in the south as well as in the north-east direction.Type of sprawl is termed as linear sprawl and  is taking place along the roads (see Figure 7).This type of sprawl is a pure depiction of haphazard expansion in the urban limits which normally leads to poor management and a result of bad administration.The roads taking part in the linear development of the city are, Bahawalpur Road, Surajkund Road, Vehari Road, and Khanewal Road.A total increase of 23.05 Sq.Km area is along the edges of these roads.The individual measurements of the increment of built up along these roads are demonstrated in Table 2.

Population Growth vs. Areal Growth
To further support the above analysis, it was considered to verify it with the population growth trends in the city.Keeping in view the technologies and techniques used in the research, it was thought to be a better idea to rely solely on the satellite based approach for the mapping of population density.Multan is outstripping the rate of demographic expansion.For this purpose, the landScan was opted considering the authenticity and reliability of its data.

A.
Ali et al.DOI: 10.4236/ars.2018.73017247 Advances in Remote Sensing 4 square kilometers.It has 129 UCs with 64 rural and 65 urban UCs (District office, Multan).Out of these 65 urban UCs, 60 urban UCs fall in our study area.Multan has extreme climate with an average temperature of 49˚C in summers and 1˚C in winters.The average rainfall is 127 mm (Pakistan Weather Portal).It has a population of 1.197 million as per census of 1998 and has risen to in 2015 [8].With an increase in population, the migration rate from rural areas to urban areas has been increased drastically and resulting in disorganized urbanization of arable land (Report on Vision 2030 by Planning Commission).
good enough to provide accurate information regarding expansion of the Multan city.Moreover, another type of dataset, LandScan, has been used in this study to visualize the population growth trends from 1998 to 2015 (subjected to data availability).

Figure 3 .
Figure 3. Satellite image of the study area.
The classification process was divided into the following steps: 1) Satellite Images as input, 2) Subset the AOI, 3) Image Enhancement, 4) Selection of training samples of all land covers, 5) Run classifier using nearest neighbor algorithm, 6) Repetition of the whole process to gain the maximum possible accuracy, 7) Final Classification.The training polygons were obtained on the basis of spectral response of the bands and the image if the entire areas have been divided into four major classes i.e. 1) Water, 2) Vegetation, 3) Bare Soil, & 4) Built up.The entire process has been reiterated to get the classification done for the four years to analyze the sprawl.The figure below depicts the mapping of the results of land use land cover classification performed.The areas for each land cover type has been calculated which illustrates the change over the decades.In 1972, a total of 47.8 Sq. km area was under built-up land whereas, in 2015, this area increased to

A
wide range of techniques are available including different algorithms and formulas for the detection of change in land use and land cover utilizing the satellite imagery.These techniques are broadly divided into two categories: 1) Spectral classification of satellite imagery, and 2) Radiometric change between the range of acquisition dates of satellite imagery.The second technique includes A. Ali et al.DOI: 10.4236/ars.2018.73017251 Advances in Remote Sensing

Figure 4 .
Figure 4. Land-use classification of the study area.

Figure 5 .
Figure 5. Growth pattern in the study area.

Figure 6 .
Figure 6.Sprawl analysis of the study area.

Table 2 .
Increment of built up along the roads.

Table 3 .
4,e population amplified from 2,939,907 in 2000 to4,384,191in 2013.This implies that the population grew by about 50% which is quite synchronized with the rate of increase in built up area.This refers that either the land is consumed at normal rates or normal amounts required for growth of this amount of population or the land consumption is under rated.With the above mentioned argu- ment, even if a discrete conclusion is drawn, still it does not depict the type of growth that has proved itself to be a somehow linear (through the above analysis maps are generated see Figures8(a)-(d)). A. Ali et al.DOI: 10.4236/ars.2018.73017255 Advances in Remote Sensing

Table 3 .
Population forecasts and implicit growth rate (Source: integrated master plan of Multan 2008-2028).