3.1. Built-Up Pattern
As the dominant land use in Batticaloa municipality, the built-up area has experienced rapid growth in recent decades. The built pattern is mainly considered the parameter for identifying the urban sprawl in a city. Hence, it was achieved for the years 1990, 2000, 2010, and 2020 (see
Figure 4). The built-up area was 1030.23 hectares in 1990, 1237.86 hectares in 2000, 1367.10 hectares in 2010, and 1774.44 hectares in 2020 (see
Table 3). The accuracy level was 93% in 1990, 94% in 2000, 93% in 2010, and 94% in 2020. The built-up pattern showed a continuous gradual increase in the built-up area during this period. However, the municipality has faced sprawling growth since 1990 that was establish based on these built-up maps. As a central area in the eastern province, Batticaloa experienced the effects of civil war from the 1980s to 2000s. This context was one of the factors that caused human migration to and from other districts to access security, food, health, education, and other services. For instance, the International Organization for Migration reported in 2007 that the civil war caused more than 150,000 people to flee to eastern Batticaloa, especially to Batticaloa municipality. A total of 750 shelters were supplied to the residents of Batticaloa municipality and the Chenkalady area at the same time as construction on 700 more shelters began in 2007 [
56]. Moreover, the increase in residential and commercial buildings caused the growth of urban built-up areas during these periods. Therefore, the residents contributed significantly to this rapid urban growth in the municipality.
The gains and losses of built-up areas between 1990 to 2020 were identified based on the built-up maps (see
Figure 5). The built-up area gained 906 hectares and lost 162 hectares in these periods. Replacement of residents due to the tsunami and flood in the municipality was a notable factor for the losses of the built-up in coastal areas.
3.2. Selection of Drivers’ Variables
A total of 26 biophysical and socioeconomic factors related to urban sprawl were identified from the literature of developed and developing countries (see
Figure 6). First, it was determined whether these variables in this study were appropriate to use in the Batticaloa area. Batticaloa is a coastal region with inland water bodies; hence, physical factors such as elevation, slope, and water bodies are important factors in built-up development in the municipality. Land value and population growth have a significant relationship with urban sprawl in Batticaloa as it is a growing city. Policy regulations, people’s housing preferences, income inequality, and living standards all have a connection to urban sprawl. Education and employment are important factors for the sprawling growth of the city. Regional accessibility, transport accessibility, and service buildings, including commercial, CBD, administration, and religious buildings, are primarily related to residential growth. Therefore, these factors were focused on in this study. Then, these factors were tested with Cramer’s V, available in Land Change Modeler, IDRISI 17.0. The potential factors of urban sprawl were identified with the built-up patterns in the Batticaloa municipality. These factors were used to develop the transition potential for urban sprawl prediction.
Factors with a Cramer’s V value greater than 0.15 were considered for the prediction of urban sprawl in 2030. In this case, 20 variables (factors) met this requirement that was added to the logistic regression model. Six socioeconomic variables, such as housing preferences, income inequality, education, living standard, employment, and distance to the air runway, did not meet Cramer’s V requirements (see
Table 4). These variables were not taken into account for predicting urban sprawl patterns. Furthermore, several factors were considered in this study, which many authors ignored due to data availability. For example, Mustafa and Teller [
57] indicated that their study was limited to several urban sprawl factors in the model. They did not include accessibility to roads, trains, and jobs, which is a hidden factor that should obviously affect urban sprawl over time. Simultaneously, Saqui [
58] and Grigorescu et al. [
30] mentioned that several explanatory variables identified in previous studies were not included in their research studies due to a lack of data. Therefore, the factors used in this study can support the urban sprawl prediction in all aspects.
The Cramer’s V and coefficient values for the factors from the logistic regression analysis are presented in
Table 4. The results indicated that the spatial associations between the explanatory variables and urban sprawl differed in relation to the biophysical and socioeconomic factors. The coefficient value of the logistic regression denoted the individual influence of each factor on the transition variable with direct or inverse associations [
49]. The coefficients specified which biophysical and socioeconomic variables had the most vital contribution to describe the spatial pattern of built-up areas. The positive coefficient value for the variables indicated the direct association with urban sprawl, while the negative values indicated the inverse association with urban sprawl. However, a positive association of the variables indicated the probability of a decrease in the urban sprawl, and a negative association of the variables indicated the probability of increasing the urban sprawl.
In this case, 13 variables showed a positive association and 7 variables showed a negative association in the analysis. Policy regulations and regional accessibility had a strong positive relationship of 0.90 and 0.72, respectively, with the probability of urban sprawl. Policies in the city seem to be loose, which leads to illegal housing development, illegal land subdivision, and illegal land ownership in the city. The maximum extent of living space is not defined in the policy guidelines that lead to low-density sprawling development. In addition, the lack of rigor in taxation causes the illegal occupancy of land in the municipality, which further increases sprawling growth. Meanwhile, the regional availability in the city is relatively high compared to the other areas in the Batticaloa district. The cost of accessibility in rural areas is relatively high due to inadequate infrastructure facilities that cause migration to the city. People chose affordable areas of the city where home values were low. For example, the Thiraimadu, Saththurukondan, and Thiruperunthurai areas have a quite low land value compared to other areas of the city. Furthermore, the poor accessibility in the current land use patterns in the Batticaloa district causes people to settle in the city or urban areas. This inefficient land use pattern leads to more increased urban sprawl in the municipality.
Furthermore, the distance to a highway, distance to a secondary road, distance to a railway, and distance to a commercial area have a very weak effect on urban sprawl. The roads in the municipality were well laid out in most areas of the city. The municipality’s transportation network is comparatively accessible to all, which can be one of the reasons for the very weak relationship with urban sprawl. Moreover, land value, demographic dynamics, and population by GN division have a strong negative relationship, with –0.7519, –0.5519, and –0.5766 coefficient values, respectively. It means that these factors have a very strong effect on urban sprawl. The land value is not reasonable for all classes of people to buy land in the city. Rising house or land prices in core cities such as Puliyanthivu, Arasady, Kottamunai, and Thamaraikeny are forcing people to buy land in affordable areas such as Saththurukondan, Thiraimadu, Paalameenmadu, Kokkuvil, and Thiruperunthurai. Thus, these areas are developed as low-density, dispersed, and discontinuous sprawl in the municipality. Due to insufficient housing programs, people have owned the individual land in the city, which is increasing land prices by multiple magnitudes. Simultaneously, increasing demand for land is driving future real investments in property in the city, and land value expectations among owners are delaying land sales until a satisfactory price is obtained. Hence, the vacant spaces in the city remain without sales. Furthermore, demographic changes are significantly associated with urban sprawl. Migration from the rural areas to the city has increased the dynamics of population change in recent decades. Migrants to the city during the civil war tend to live in urban housing. It forms a sprawling growth pattern in the city under the demands of social status. At the same time, the standard of living of people from the districts encourages them to migrate to the Batticaloa municipality. Additionally, proximity to CBD, distance to educational resources, and distance to religious resources have a very weak effect on urban sprawl. These areas are mainly located in the core city, which is the reason for the very weak effect on urban sprawl. In contrast, the model does not show a significant relationship between the built-up areas and some socioeconomic factors such as housing preferences, income inequality, education, living standards, employment, and distance to air runways in the Batticaloa municipality. Simultaneously, these six factors did not meet the requirement of Cramer’s V of 0.15. However, these factors may influence urban sprawl in the future depending on people’s interferences and urban development. However, this study integrated the significant factors only in the prediction of urban sprawl.
3.4. Urban Sprawl Prediction in 2030
The built-up areas are the primary parameters to understand urban sprawl. This built-up class includes residential buildings, commercial and shopping centers, highways and major streets, industrial areas, and other related properties [
59]. The urban sprawl pattern for 2030 was simulated based on the built-up images (1990–2020) and several explanatory factors that included biophysical and socioeconomic factors. The built-up pattern was predicted in three scenarios. Scenario 1 indicated the built-up prediction for 2030 only with biophysical factors. Scenario 2 specified the built-up prediction for 2030 only with socioeconomic factors. Scenario 3 showed the built-up prediction for 2030 with biophysical and socioeconomic factors (see
Figure 6). Furthermore, the probability of changes in the built-up areas in 2030 is presented in
Table 5. The built-up area will probably change to non-built-up at around 7.97%, while the non-built-up area to built-up area can be 3.97% in 2030.
The level of kappa agreement was assessed for the simulated and existing built-up maps in 2020 using the validate tool in IDRISI 17.0. The kappa statistics for K_no (0.8550), K_standard (0.8459), and K_location (0.8572) were identified above 0.80, indicating an excellent agreement between the datasets [
49]. Thus, the transition probability matrix can be considered to apply in the prediction of built-up patterns in 2030. Furthermore, the ROC statistic was 0.9683, which designated a very strong value, and the soft prediction was precise. The pseudo R2 value was 0.6061, which specified a relatively good fit of the model because a pseudo R2 value greater than 0.2 is considered a relatively good fit.
Then, the prediction of urban sprawl was simulated in three scenarios for Batticaloa municipality. The result revealed that these three scenarios will have different extents of the built-up area in 2030. Scenario 1 indicated that the built-up area can be 1904.28 hectares in 2030. However, scenario 2 assumed that the built-up area could be 2118.24 hectares, which is relatively higher than scenario 1. In addition, the built-up area in scenario 3 showed approximately 2133.63 hectares, which was more than in scenarios 1 and 2 (see
Table 6 and
Figure 8). Scenario 1 was simulated with only biophysical factors (elevation, slope, and distance to water bodies), and scenario 2 was simulated only with socioeconomic factors (land value, demographic dynamics, policy regulations, regional accessibility, population by GN division, land value by GN division, distance to highway, distance to secondary road, distance to tertiary road, distance to railway, services buildings, proximity to CBD, distance to commercial, distance to educational, proximity to administration, distance to religious, proximity to services buildings, built-up cubic trend). This comparison confirmed that socioeconomic factors have a more significant influence on urban sprawl than biophysical factors in the municipality based on the scenario-based findings. Thus, it can be concluded that urban sprawl is a socioeconomic phenomenon, which has a superior reflection on the built-up development of the city. Additionally, biophysical factors are also crucial in predicting urban sprawl, which has a significant effect on built-up growth. Therefore, the prediction of urban sprawl with biophysical and socioeconomic factors is more appropriate to display for future urban planning and development.
In addition, the expected built-up growth in these three scenarios was identified between current (2020) and future (2030) patterns that showed different growth rates. When the built-up pattern is the same as in scenario 1 in 2030, the growth rate can be 0.73%; while the built-up pattern will be the same as in scenario 2 in 2030, the built-up growth rate can be 1.94%. The built-up pattern is the same as in scenario 3 in 2030, and the growth rate could be 2.02%. The sprawling growth moved in the direction of the northern and western parts of the municipality. However, the lagoon in the western part of the city limits the sprawling growth. Overall, scenario 3 indicated high growth among these predicted built-up patterns. Based on the historical pattern of built-up areas (1990, 2000, 2010, 2020), it confirmed that the built-up pattern in 2030 is also relatively the same as in recent years. Therefore, when the potential factors can be similar to the current ones during 2030, urban sprawl can be expected to continue and be one of the predicted scenarios. However, the factors are supposed to be limited by the influence of institutional and political measures on built-up development, such as land management, land use planning, and land regulation. In the meantime, unpredictable drivers may sometimes affect the accuracy and the patterns of the model in the future.
The findings indicate that urban sprawl in the future is typically expected in the marginal areas in the city. Urban sprawl occurs due to the undeveloped, abandoned, and agricultural land in the municipality. This finding is similar to that of Grigorescu et al. [
30] in Romania. This undeveloped land formed a leapfrog and scattered development with fewer houses and a small population. Ahrens and Lyons [
8] stated that income growth, accessibility, and population were some of the significant determinants of urban sprawl in Ireland. However, income was not identified as a significant factor of urban sprawl in Batticaloa municipality. Furthermore, Batticaloa municipality has been experiencing a residential-based sprawling in core and marginal areas, which was similarly found in Hangzhou, China [
60]. The demand of society and the socioeconomic conditions related to residential development are reasons for the urban growth and sprawling development in the city. Increasing land consumption is one of the causes of the urban sprawl growth in the municipality. Simultaneously, inadequate housing programs and increasing land value induce the people to settle in the cheaper land areas such as Kokkuvil, Sathurukondan, and Thiruperunthurai.
Furthermore, urban sprawl generally occurs due to the rapid urbanization or de-urbanization in the cities. Based on the Sri Lankan context, the largest cities are in Western Province, such as Colombo and Sri Jayawardenepura Kotte. At the same time, Kandy, Galle, and Kurunegala are some of the major cities in Sri Lanka. According to Amarawickrama [
61], nearly 25% of Sri Lankans live in cities, which is expected to rise to nearly 65% by 2030; as a result of this, additional metropolitan areas will be required to accommodate the anticipated population growth. Thus, the Eastern Metro Region has chosen Batticaloa and Ampara as metro areas with 1 million people or more [
62]. According to recent trends, most small- and medium-sized cities are growing fast as multi-functional centers. Several cities in Sri Lanka have strategies and plans for development, which pay little attention to the impacts of urban sprawl when implementing development projects [
63].
However, the most pressing challenge confronting cities is urban sprawl [
64], which has not been a focus of many urban studies in Sri Lanka [
42,
43,
61,
63]. According to the government’s plans, Western Province is to be developed into an urban center in South Asia, such as Megapolis. This province in Sri Lanka has demonstrated well the extent of urban sprawl in the country [
61,
64]. The population in nine provincial capital cities grew by 6.42% annually between 1995 and 2017. It totaled roughly 7.39 million people in 2017, according to an analysis of satellite images undertaken for the State of Sri Lankan Cities’ project in 2018. Sri Lanka’s urbanization statistics show that the country has been de-urbanizing over the past 50 years. In 1987, after removing the unit of Town Councils, the country modified its municipal boundaries, resulting in an immediate drop in the population of the cities. Since then, the administratively designated urban population has grown modestly, with no changes to the city limits. However, areas that exhibit metropolitan spatial characteristics have grown rapidly, particularly on the outskirts of big cities [
65]. Therefore, urbanization and de-urbanization are considerably influenced by urban sprawl growth.