Introduction

Urbanization is the process of urban population increase and population migration from the rural areas, which results in infrastructure development and built-up area expansion. Nowadays, over half of the world’s population resides in urban areas, and it is predicted that by 2050 about 65% of the total population will live in cities (Worldometers 2018a). Urbanization is a complex process that is always accompanied by demographic change (Yaukey 1985; Sato and Yamamoto 2005; Dyson 2011), socioeconomic change (Roberts 1978; Jense and Cowen 1999; Huang and Chen 2009), land use and land cover change (LULCC) (Carlson and Authur 2000; Lambin et al. 2001), and institutional change (Zhao 2001; White et al. 2008). Rapid urbanization worldwide has put much stress on the environment at both regional and global scales, such as increased average surface temperature (Kalnay and Cai 2003), loss of natural vegetation cover (Drummond and Loveland 2010), deteriorated air quality (Cole and Neumayer 2004; Chao and Yao 2008) etc., which has triggered environmental risks and human health issues (Phillips 1993). The magnitude of population increase is considered as the directly factor affecting the urban environment (Brennan 1999) because it influences the spatial concentration of industrial emission, vehicles, commerce, energy consumption, water use, waste management, and other stresses in an urban context (Bartone et al. 1992). It is therefore generally considered that urban size has a negative impact on the environment (Brennan 1999).

Urbanization has been a major trend in European and North American countries for almost a century, but the shift has just taken place in Asia and Africa, with the most remarkable changes found in Southeast Asia. Southeast Asia has become one of the most economically dynamic regions in the world during the past few decades. Industrialization, economic development, and urbanization have been rapidly accelerated as the region has abundant natural and human resources. It is predicted that by 2050 the total urban population in Southeast Asia will grow by another 192 million, rising from 316 million in 2017 to 508 million (Worldometers 2018b). Myanmar is the second largest country in Southeast Asia, but its total population ranks fifth and the urban population percentage only ranks ninth in this region (Table 1). Myanmar began its economic liberalization in the 1990s under military rule, but its economy did not take off until 2011, when a new government was installed. Drastic urbanization accompanied the economic surge all across the country. Although still behind many other countries in Southeast Asia in terms of economic development and urbanization, Myanmar’s urbanization process has been highly accelerated in the past few decades. Its urban population percentage increased from 24% in 1980 to 34.7% in 2016 (The World Bank 2017b), and the built-up areas experienced a 24% increase between 1992 and 2010 (Ouyang et al. 2016), reaching a total of 1044.65 km2, or 0.15% of the country’s total land area. The projected mean annual urbanization rate for Myanmar between 2015 and 2020 is 2.29%, which ranks fourth in all the Southeast Asia countries and ties Indonesia (Table 1). Among all the cities in Myanmar, no other cities have experienced more dramatic urbanization processes than Yangon, the former capital and the largest city in Myanmar, and Nay Pyi Taw, the newly established national capital since 2006. The national context made Myanmar an interesting case to examine and compare the patterns, processes, and consequences of transforming urban landscapes in rapidly urbanizing nations.

Table 1 Total population (column 2) and urban population percentage (column 3) for 2016, and predicted annual rate of urbanization for 2015–2020 (column 4) for Southeast Asia countries.
Full size table

Despite the rapid urbanization growth in Myanmar, limited literature has shed light on the environmental changes due to urbanization in Myanmar during its transitional economy. Urbanization is a complex socioeconomic process. This study will be focused on one specific form of urbanization, which is the physical expansion of built-up area in an urban context (referred to as “urban expansion” hereafter), and to examine its environmental consequences. In order to fill this gap in the literature, this study has three objectives. The first is to discover the spatio-temporal patterns of urban expansion from non-built-up (NBU) to built-up (BU) areas between 2000 and 2013 using Yangon and Nay Pyi Taw as two study cases. The second objective is to examine the environmental consequences of urban expansion at a regional scale in terms of surface temperature, forest cover, evapotranspiration, terrestrial ecosystem net primary productivity, and air quality using remotely sensed satellite images. The third is to discover the distinct driving forces of urban expansion in two cities unique to Myanmar’s transitional economy.

Study area, data, and methods

Study area

Yangon is located in Southern Myanmar at the convergence of the Yangon and Bago rivers (Fig. 1). Yangon was Myanmar’s capital between 1948 and 2005. The Yangon metropolitan area had a total population of 7.36 million in 2014 (Ministry of Immigration and Population 2015a). It remains the country’s largest city, and it is the national commercial, financial, and industrial center. Yangon has a tropical monsoon climate that features a long rainy season from May through October, followed by a dry hot season from November to April. The annual mean temperature of Yangon is 27.3 °C, and the average annual precipitation is 2378 mm (Climate-Data.org 2016a).

Fig. 1
figure 1

Map of study area. The basemap is the World Terrain Base that was compiled from a variety of sources from several data providers, including the U.S. Geological Survey, NOAA, and Esri. Please find more detailed information at http://www.arcgis.com/home/item.html?id=c61ad8ab017d49e1a82f580ee1298931

Full size image

In March 2006, the military government designated Nay Pyi Taw as the new administrative capital. The city is located in Central Myanmar between the Bago Yoma and Shan Yoma mountain ranges, approximately 320 km north of Yangon. Construction and infrastructural development began in 2002, and it is still in progress. The Nay Pyi Taw metropolitan area covers more than 4600 km2 and, in 2014, had an approximate population of 1.16 million (Ministry of Immigration and Population 2015b). Nay Pyi Taw has the same climate as Yangon. The annual mean temperature is 26.8 °C, and the annual precipitation averages 1167 mm (Climate-Data.org 2016b).

Remotely sensed data and image processing

Remote sensing technology offers great advantages over the traditional data collection methods to unobstructively observe the earth surface and phenomenon over a large geographic area. Remote sensing not only acquires up-to-date ground information, but also provides historical data, and it is the only practical way to obtain data from inaccessible regions. Remotely sensed satellite images, such as Landsat, MODIS, AVHRR, and ASTER, are readily available online and easy to manipulate using computer software. Remotely sensed data therefore become more and more popular among urban managers and planners.

In this study, Landsat images were used to examine the spatio-temporal patterns of urban expansion for Yangon (WRS-2 Path 132 Row 48) and Nay Pyi Taw (WRS-2 Path 133 Row 46). For 2000–2003, Landsat 7 Enhanced Thematic Mapper Plus (ETM+) data were used. For 2004–2011, Landsat 5 Thematic Mapper (TM) data were used because of the failure of the scan line corrector in the Landsat 7 ETM+ instrument in May, 2003. Landsat 8 Operational Land Imager (OLI) data were used for 2013. Landsat 5 data for 2012 currently are not available. All the Landsat images used in this study were acquired in the dry, hot season between January and April, which is mostly cloud free. Only one cloud-free image was used for each city per year.

A 60 × 60 km mask that covers the entire urban and surrounding rural areas was used first to extract the study area from each Landsat image. An unsupervised classification was applied to classify each image into 200 clusters using the ISODATA algorithm, and then, each cluster was identified visually as a NBU or BU area with the use of very high spatial resolution satellite images in Google Earth software. Based on historical satellite images, 100 ground truth BU points and 100 ground truth NBU points were visually detected and randomly selected from Google Earth for each year and each city. Ground truth points were then overlaid with the classified Landsat image to determine its classification accuracy. The overall classification accuracy is greater than 82% for all the images. The total BU area was calculated from classified images for each city and each year and, then, plotted against the year sequence to discover the temporal pattern of urban expansion.

The spatial pattern of urban expansion was evaluated using landscape metrics that include patch density, largest patch index, landscape shape index, and patch cohesion index for the BU area only. All the classified Landsat images were brought into FRAGSTATS (v.4) software (McGarigal et al. 2012) for the calculation of selected landscape metrics.

This study selects five standard Moderate-Resolution Imaging Spectroradiometer (MODIS) land products that are produced, validated, and distributed by National Aeronautics and Space Administration (NASA) and United States Geological Survey (USGS) to evaluate the impacts of urban expansion on the regional environment for Yangon and Nay Pyi Taw, which are land surface temperature (LST), percent tree cover (PTC), terrestrial ecosystem net primary productivity (NPP), evapotranspiration (ET), and aerosol optical depth (AOD). The detailed information of each MODIS land product used in this study is listed in Table 2. The MODIS LST image was validated within 1 K accuracy using in situ measurements in the range of 263–322 K at an atmospheric column water vapor range of 0.4–3.0 cm (Wan et al. 2002). For the MODIS annual PTC product, a standard error of 15.6% was reported from training data and 11.5% from field test data (Hanse et al. 2003). The terrestrial ecosystem NPP product has a mean error obtained within 9% of averaged published errors (Zhao et al. 2006). The correlation coefficient between flux tower ET measurements and MODIS ET predictions was reported 0.86 in Mu et al. (2011). The MODIS AOD product used in this study is retrieved at 0.55 µm wavelength (Levy et al. 2009). On a global basis, about 68% of the retrievals fall within expected error, which is considered to meet the expected accuracy levels defined by Remer et al. (2005). Although errors and uncertainties from satellites, atmosphere, and modeling procedure are inevitable, all the MODIS products used in this study meet general standards and are of high scientific quality. These five environmental parameters are selected because they are easily evaluable using readily available MODIS products, and these products have been widely used for both regional and global environmental studies (Hansen et al. 2005; Wan et al. 2004; Turner et al. 2006; Mu et al. 2007; Lee et al. 2011). The original MODIS products were downloaded and processed to create images for annual mean daytime/nighttime LST, annual mean NPP, annual mean ET, annual mean PTC, and monthly/annual mean AOD from 2000 to 2013.

Table 2 Detailed information about MODIS land products used in this study
Full size table

Time-series trend analysis

A time-series trend analysis technique (Ishtiaque et al. 2016; Wang and Myint 2016; Wang et al. 2016a) was applied to the annual maps of selected environmental parameters to identify areas that have statistically significant changes in daytime/nighttime LST, NPP, ET, and PTC. In more detail, for each particular pixel from each annual map stack, the 14-year time-series pixel values were extracted and analyzed against the year sequence using ordinary least squares (OLS) regression to examine if a pixel location had experienced statistically significant changes of LST, NPP, ET, and PTC. The regression equation is formulated as:

$$y_{dLST,nLST,NPP,ET,PTC} = \beta_{0} + \beta_{1} \times year + \varepsilon ,$$
(1)

where ydLST, ynLST, yNPP, yET, and yPTC represent the 14-year time-series daytime LST, nighttime LST, NPP, ET, and PTC values from each pixel in each map, β1 is the slope coefficient, β0 is the intercept, year is the variable of year sequence from 2000 to 2013, and ε is the error term. Only pixels that showed statistically significant changes (p value ≤ 0.05) were retained. This procedure generated a slope coefficient map and an R2 value map for each parameter.

Since AOD is an atmospheric environmental parameter, and the pixel size of the MODIS AOD product (3-km) is much larger than other products (Table 2), this parameter may only be used to assess air quality for a region rather than at a particular pixel because of the liquidity and instability of the atmosphere. Therefore, we calculated monthly and yearly mean AOD within the 60 × 60 km study boundary for both cities and plotted them against the time sequence to examine the trend.

Discovering environmental consequences of urban expansion

All the classified BU area maps were resampled from 30 m resolution to 250 m and 1000 m resolution, respectively, with each pixel now representing a percent built-up (PBU) area. A PBU difference map, therefore, was generated by subtracting a 2000 PBU image from a 2013 PBU image (PBU2013 − PBU2000), and the pixel value represented the PBU difference between 2000 and 2013.

For the pixels that have statistically significant and synchronized changes in PBU and LST, in PBU and NPP, in PBU and ET, and in PBU and PTC, the PBU difference values were analyzed vis-à-vis the LST, NPP, ET, and PTC slope coefficient values obtained from the trend analysis with the use of OLS regression to determine their relationships. The generic model was formulated as below:

$$a_{dLST,nLST,NPP,ET,PTC} = \beta_{0} + \beta_{1} \times d_{PBU} + \varepsilon ,$$
(2)

where adLST, anLST, aNPP, aET, and aPTC are the slope coefficients from the daytime LST, nighttime LST, NPP, ET, and PTC trend analyses from Eq. (1), and dPBU is the PBU difference between 2000 and 2013.

Results

Spatio-temporal patterns of urban expansion

Figure 2 shows the temporal trend of urban expansion for Yangon and Nay Pyi Taw. Both cities underwent exceptionally fast urban expansion during the study period but showed different temporal trends. The increase of BU area in Yangon exhibited a cubic polynomial relationship against the year sequence (Fig. 2a), while Nay Pyi Taw showed an exponential increase of BU area (Fig. 2b). The mean annual urban expansion rate is 8.2% for Yangon and 30.8% for Nay Pyi Taw. Yangon experienced rapid urban expansion between 2000 and 2005, but the pace slowed between 2005 and 2009 because of the relocation of the nation’s capital. Yangon’s expansion, then, picked up speed again after 2009, and the total BU area nearly tripled from 43.0 km2 in 2000 to 119.5 km2 in 2013. The total BU area in Nay Pyi Taw was only 5.2 km2 in 2000, and it increased slowly between 2000 and 2005. It rocketed from 16.6 km2 in 2005 to 170.5 km2 in 2013 due to intensive construction and infrastructure developments for the new capital. The total BU area in Nay Pyi Taw has exceeded Yangon since 2011, and it was 51 km2 larger in 2013.

Fig. 2
figure 2

Temporal trend of urbanization for Yangon (a) and Nay Pyi Taw (b)

Full size image

Figures 3 and 4 show the spatial distribution of urban expansion for both cities. Urban expansion activities in Yangon mainly focused on the north banks of the Yangon and Bago rivers, while very little changes were found on the south banks (Fig. 3). The city has been expanding deep into the north part of the metropolitan area. The Nay Pyi Taw region in 2000 was composed of many isolated small villages without major infrastructures (Fig. 4). All the old villages have been expanding and many new cities and towns gradually emerged. Road construction has become the most prominent urban expansion characteristic, and infrastructure development mainly has focused on the east side of the Yangon–Mandalay Expressway.

Fig. 3
figure 3

Urbanization trend for Yangon: 2000–2013. The basemap is the Esri World Imagery layer produced from 15-m spatial resolution TerraColor satellite imagery

Full size image
Fig. 4
figure 4

Urbanization trend for Nay Pyi Taw: 2000–2013. The basemap is the Esri World Imagery layer produced from 15-m spatial resolution TerraColor satellite imagery

Full size image

Figure 5 presents the temporal dynamics of landscape metrics for Nay Pyi Taw and Yangon. BU area in Nay Pyi Taw experienced significant increases in patch density (Fig. 5a), largest patch index (Fig. 5b), landscape shape index (Fig. 5c), and patch cohesion index (Fig. 5d), which means that urban expansion in Nay Pyi Taw started with small, regularly-shaped, and isolated patches in 2000 and ended up with large, irregular, and more physically connected patches in 2013. For the Yangon region, the largest patch index (Fig. 5b) and landscape shape index (Fig. 5c) did not change significantly, which means the largest BU patch and the patch shape did not change too much. The patch density index has decreased (Fig. 5a) and the patch cohesion index has increased (Fig. 5d), which implies that BU areas in Yangon generally becomes larger and more physically connected during the study period.

Fig. 5
figure 5

Temporal dynamics of landscape metrics for the BU area in Yangon and Nay Pyi Taw

Full size image

Temporal trends of environmental parameters

Figure 6 shows maps of slope coefficient values for the time-series trend analysis of all the environmental parameters. The slope coefficient is of great importance because it is interpreted as the rate of change against the year sequence. A positive slope coefficient value implies an increasing trend, while a negative value represents a decreasing trend. The absolute value of a slope coefficient indicates the magnitude of a trend.

Fig. 6
figure 6

Slope coefficient maps for the time-series trend analysis of daytime LST (a), nighttime LST (b), PTC (c), NPP (d), and ET (e) for Yangon, and daytime LST (f), nighttime LST (g), PTC (h), NPP (i), and ET (j) for Nay Pyi Taw, respectively. Only pixels that have experienced statistically significant (p value ≤ 0.05) changes are shown. The basemap is the Esri World Imagery layer produced from 15-m spatial resolution TerraColor satellite imagery

Full size image

The highest daytime LST increase per year was 0.13 °C for Yangon and 0.24 °C for Nay Pyi Taw, and the highest daytime LST decrease was 0.35 °C/year for both cities (Fig. 5a, f). The area of decreased daytime LST was much larger than the area of increased daytime LST in Yangon, and the decreasing magnitude of daytime LST was also greater than the increasing magnitude. However, it was the rural areas, rather than urbanized areas, that experienced decreased daytime LST. Downtown Yangon and some newly urbanized areas to the northwest of downtown experienced increased daytime LST. Nay Pyi Taw had a larger area of increased daytime LST than decreased, although the decreasing trend is still greater than the increasing. Increased daytime LST mainly was found in newly established urban areas, while decreased daytime LST was found in large bodies of water, such as lakes and reservoirs. For the nighttime LST, both cities experienced an increasing trend without a significant decreasing trend found anywhere. The highest increase of nighttime LST for Yangon was 0.19 °C/year (Fig. 6b), which was found on the city outskirts in bodies of water, while no urbanized area was found to have an increasing trend of nighttime LST. On the contrary, all the newly urbanized areas in Nay Pyi Taw have experienced an increasing trend of nighttime LST, but the increasing magnitude was small. The highest nighttime LST increase for Nay Pyi Taw was 0.43 °C/year (Fig. 6g), which also was found on the outskirts in a reservoir.

Both cities underwent afforestation and deforestation to some extent. The highest afforestation rate was 2.4%/year for Yangon and 2.2%/year for Nay Pyi Taw, and the highest deforestation rate was 4 and 4.8%/year for Yangon and Nay Pyi Taw, respectively. Although the total area of deforestation is much smaller than afforestation area in Yangon, the deforestation rate is high (Fig. 6c). Afforestation mainly was found on the city outskirts, and deforestation was found in both urbanized and rural areas. Nay Pyi Taw had a larger area of deforestation than afforestation, and the deforestation rate was also much higher than the afforestation rate. The most prominent deforestation hotspots mainly were found in newly urbanized areas and along major streets and highways. It is also worth noting that large areas of forests had been cleared around Nay Pyi Taw for dam constructions.

NPP change had different spatial patterns between the two cities. Terrestrial ecosystem NPP increase and decrease were seen in about the same magnitude for Yangon, and the total area of NPP increase was larger than the area of NPP decrease (Fig. 6d). NPP increase in Yangon mainly was found on the city outskirts, with a large hotspot found to the northeast of the downtown area, which mainly was caused by afforestation activities. Decreases in NPP were generally found everywhere in Yangon without showing a distinct spatial pattern. Nay Pyi Taw had a larger area of decreasing NPP than increasing NPP, and the NPP increase rate was much higher than the decrease rate (Fig. 6i). The highest NPP decrease rate in Nay Pyi Taw was 52.6 metric tons/year. The spatial distribution of decreasing NPP shows a clustering pattern, which was found in the newly established Nay Pyi Taw downtown and in two large reservoirs—one to the west and the other to the northeast of the downtown area. Nay Pyi Taw generally showed a larger area of decreasing NPP, a much higher NPP decreasing rate, a smaller area of increasing NPP, and a lower NPP increasing rate than Yangon during the entire study period.

Similar spatial patterns also were found for ET changes in the two cities. A larger area experienced increased ET than decreased ET in Yangon, although the increasing magnitude was slightly lower than the decreasing magnitude. The highest ET increase rate was 13.9 mm/year, and the highest ET decrease rate was 14.5 mm/year (Fig. 6e). Two large hotspots of increased ET were found to the north and northeast of the downtown area, which was caused by afforestation activities. Major ET-decreasing areas were found in downtown and some newly developed suburban areas, which were caused by deforestation and conversion from agricultural to BU areas. Nay Pyi Taw experienced a larger scale and a higher magnitude of decreased ET. The highest ET decrease rate was 34.9 mm/year (Fig. 6j). The clustering pattern of decreased ET was evident in newly developed urban areas and along major highways. Scattered small hotspots of decreased ET in surrounding rural areas also were detected and were caused by deforestation. Increased ET areas mainly were found in some agricultural lands and mountainous regions to the east and northeast of the city. Agricultural expansion and afforestation were major causes of ET increase in these areas.

Figure 7a, c show monthly mean AOD values for Yangon and Nay Pyi Taw, respectively. AOD normally increases from January and peaks in the middle of a year in June or July and, then, drops to the lowest level in December. This distinct seasonal pattern of AOD variations was consistent across the entire study period for both cities. Although there is an inconspicuous increasing trend of annual mean AOD for both cities (Fig. 7b, d), the relationship between the year sequence and the annual mean AOD values is not statistically significant.

Fig. 7
figure 7

Variation of monthly mean AOD for Yangon (a) and Nay Pyi Taw (c), and annual mean AOD for Yangon (b) and Nay Pyi Taw (d)

Full size image

The relationship between the PBU difference and slope coefficient values from the time-series trend analysis of all the environmental parameters is shown in Fig. 8 for Yangon and Fig. 9 for Nay Pyi Taw. It was discovered that the PBU changes have a highly significant (p value < 0.01) positive relationship with daytime LST changes and negative relationships with PTC, NPP, and ET changes. A significant positive relationship also was found between PBU differences and nighttime LST changes for Nay Pyi Taw, but this relationship is not statistically significant for Yangon, therefore not reported. Quantitatively speaking, on average, a 1% increase in PBU could potentially cause an increase of 0.7 °C in daytime LST (Fig. 8a), a loss of 2.3% in PTC (Fig. 8b), a decrease in NPP of 34.3 kg/m2 (Fig. 8c), and a decrease of 42.2 mm (Fig. 8d) in ET for Yangon. Similarly, a 1% increase in PBU could potentially cause an increase of 0.3 °C in daytime LST (Fig. 9a), an increase of 0.06 °C in nighttime LST (Fig. 9b), a loss of 2.5% in PTC (Fig. 9c), a decrease in NPP of 15.1 kg/m2 (Fig. 9d), and a decrease of 19.2 mm in ET (Fig. 9e) for Nay Pyi Taw.

Fig. 8
figure 8

The relationships between PBU differences and slope coefficient values from time-series trend analysis of daytime LST (a), PTC (b), NPP (c), and ET (d) for Yangon. No statistically significant relationship was found between nighttime LST slope coefficient values and PBU differences, therefore not reported here

Full size image
Fig. 9
figure 9

The relationships between PBU differences and the slope coefficient values from time-series trend analysis of daytime LST (a), nighttime LST (b), PTC (c), NPP (d), and ET (e) for Nay Pyi Taw

Full size image

Discussion

Urbanization and population

Yangon and Nay Pyi Taw have undergone extensive urban expansion processes during the study period, with distinct spatio-temporal patterns. Results show that, between 2000 and 2013, the BU area in Nay Pyi Taw spatially expanded at an astonishing rate of 30.8%/year, and Yangon expanded at a mean annual rate of 8.2%. The calculated urban expansion rates are not only much higher than the predicted national average of 2.29% for 2015–2020 (Central Intelligence Agency 2017), but also much greater than any other countries in Southeast Asia (Table 1).

Urban development for Nay Pyi Taw started from scratch to over 170 km2 of BU area in 2013. The total BU area in Nay Pyi Taw has far exceeded Yangon since 2011 and continued increasing at a high rate. It is worth noting that, despite the fact that the total BU area and the urban expansion rate of Nay Pyi Taw is much higher than Yangon, the total population in Nay Pyi Taw was only 1.16 million or 2.3% of the national population in 2014 (Ministry of Immigration and Population 2015b), compared to 7.36 million or 14.3% of the national population in Yangon (Ministry of Immigration and Population 2015a). It is apparent that the relocation of the national capital and urban expansion did not attract large-scale population migration to Nay Pyi Taw. Therefore, unlike most of the large cities around the world, population growth or migration was not the main cause of rapid urban expansion for Nay Pyi Taw, but it is the other way around—government-induced urban expansion forced a small wave of population migration to this newly constructed capital city. Nay Pyi Taw remains the least populated state/region, while Yangon remains the most populated state/region, in Myanmar nowadays (Ministry of Immigration and Population 2014).

Figure 10a shows that the total population in Yangon experienced a monotonic linear growth during the study period, and a strong statistically significant linear correlation has been discovered between population change and BU area change (Fig. 10b). Rapid urban expansion in Yangon is therefore a result of population pressure and migration from rural areas. These findings are contradictory to what was reported by the World Bank Group (2015), which claimed that Myanmar’s urban population increased between 2000 and 2010 without large-scale constructions of new urban areas. A negative relationship is also found between population and BU patch density (Fig. 10c), while the relationship between population and BU landscape shape index is significantly positive. This finding proves that the general urban expansion pattern in Yangon shows larger, more physically connected, but more disaggregated and more irregularly shaped built-up patches over time and space.

Fig. 10
figure 10

Population increase (a), the relationships between population and BU area (b), between population and BU patch density (c), and between population and BU landscape shape index (d) for Yangon region between 2000 and 2013

Full size image

The factors that created urban expansion in Yangon are mainly caused by population increase and migration from rural areas into the city. Lack of adequate land and modern resources, unequal land distribution, frequent natural disasters, shortage of clear water, religious conflicts, and economic declines are all major push factors that drive people around Yangon to move to the city. Higher income, better employment opportunities, higher living standards, access to better healthcare and education, and a rich entertainment life are factors that attract rural people to the urban area. Although Yangon’s urban expansion process was generally slow during the study period, it remains the country’s largest city and the national center for finance, trade, industry, and tourism. It was reported that the city represented about 23% of the national GDP in 2011 (Hsu and Htun 2011).

The national capital was officially moved to Nay Pyi Taw in March 2006, but Figs. 2 and 4 both show evidence that urban development was initiated two years before the announcement. Urban expansion within and around Nay Pyi Taw mainly was municipal infrastructure development, including basic physical and organizational structures and facilities, such as road construction and development of power and water supplies, while Yangon was well developed and organized a long time ago; therefore, the urban expansion in Yangon was mainly urban reconstruction and renovation during the study period. The urban planning of Nay Pyi Taw constituted many functional zones, such as residential, military, ministry, hotel, shopping center, recreational center, and international zones. However, as of today, the city still lacks the amenities that one would expect to find in a capital. Many facilities, buildings, and roads remain vacant much of the time, and the city sometimes has been referred to as a “ghost town” (Nakamura 2014). Landscape metrics analysis shows that urban expansion in Nay Pyi Taw started with small, disconnected, and regularly-shaped BU patches, and ended up with large, irregularly-shaped, but more physically connected BU patches. In addition, most changes took place after 2006 (Fig. 5). This finding proves that urban expansion processes and activities in Nay Pyi Taw is much more intense than Yangon. Unfortunately, so far the only population data for Nay Pyi Taw was reported in 2014 in Myanmar Census Report, which is not within our study period. Detailed quantitative analysis between population change and urban expansion is therefore not able to be conducted.

Comparing to other large cities in Southeast Asia, the city size and population of Yangon and Nay Pyi Taw remain relatively small today. One reason is the lag in economic development. The other reason is that some mega-urban regions in Southeast Asia, such as Singapore, Bangkok, Kuala Lumpur, Jakarta, and Manila, grow through envelopment of surrounding densely populated small urban centers and net international in-migration (Robinson 1995), which are both rarely found in Myanmar cities.

Environmental consequences of urban expansion

A time-series trend analysis of MODIS PTC images shows evidence that both cities experienced some extent of deforestation and afforestation, but the deforestation rate was generally higher than the afforestation rate. Nay Pyi Taw had a larger deforestation area and higher deforestation rate than Yangon. Yangon also had deforestation activities within and around the city, but its afforestation area was much larger than its deforestation area.

Deforestation has been increasingly serious in Myanmar over the past few decades (Leimgruber et al. 2005; Songer et al. 2009; Wang and Myint 2016). It has been discovered that deforestation has caused a significant amount of carbon release, decreased ET, and increased LST in Myanmar (Wang and Myint 2016). Although Nay Pyi Taw and Yangon are highly urbanized areas, the mean annual deforestation rate is still higher than the national average. It was found that PTC change was highly correlated with PBU change over the study period (Figs. 7b, 8c). Figure 8b shows more data points that have negative PBU change but positive PTC change, which means the Yangon region had effective afforestation activities around and within the city that converted unmanaged land or degraded land back into forested areas. Figure 9c, however, shows more data points with positive PBU change but negative PTC change, which demonstrates that deforestation in Nay Pyi Taw mainly was caused by urban expansion and urban development. It is also important to point out that some large deforestation hotspots within and around Nay Pyi Taw (Fig. 6h) were caused by dam constructions and large continuous forested areas are now submerged under the reservoir.

Surface temperature change is another consequence of urban expansion. It was found that, for both cities, the urban heating effect was much stronger during the daytime than the nighttime. The total area of increased daytime LST also was much larger than the nighttime. A strong positive relationship was found between daytime LST change and PBU change, which indicates that urban expansion caused a significant daytime LST increase for both cities. It is because manmade materials, such as asphalt and concrete, absorb shortwave radiation and store energy during the daytime. The energy is then released in the form of longwave radiation at night, heating up the lower atmosphere (Oke 1982; Bouyer et al. 2009). In addition, anthropogenic activities release heat and moisture into the atmospheric boundary layer through fuel burning, air conditioning, automobiles, and machinery that cannot be quickly dispersed (Golden 2004; Bohnenstengel et al. 2011; Pal et al. 2012). As rural and natural landscapes are continuously being converted to built-up areas, larger areas suffers the urban heat island (UHI) effect. It is interesting to note that the Yangon region had many data points with a positive PBU change but negative daytime LST change (Fig. 8a). This is because afforestation activities increased green spaces in Yangon, and the cooling effect of green spaces contributes to the mitigation of the UHI effect (Gill et al. 2007; Bowler et al. 2010; Hamada and Ohta 2010; Oliveira et al. 2011; Wang et al. 2016a). A positive relationship also was found between the nighttime LST change and PBU change for Nay Pyi Taw (Fig. 9b), although the heating effect was weaker than the daytime because the slope coefficient was much smaller. It is important to notice that some nighttime heating hotspots around Nay Pyi Taw (Fig. 6g) actually were cooling spots during daytime (Fig. 6f). This also was due to dam constructions and the creation of artificial reservoirs. Water has a cooling effect during the daytime but a strong heating effect at nighttime (Oke 1982; Kim 1991), which causes a body of water to have a higher surface temperature than its surroundings. The same phenomenon also was found in the Yangon region, with two large heating hotspots discovered to the east of the city. Using very high spatial resolution satellite imagery from Google Earth, these two hotspots were identified as large rice paddies.

Results indicate that urban expansion processes reduced the urban ET rate (Figs. 7d, 8e) for both cities, and the effect was more prominent in Yangon than Nay Pyi Taw. This finding is consistent with those reported in Owen et al. (1998) for Centre County in Pennsylvania, United States, in Kondoh and Nishiyama (2000) for Tokyo, Japan, and in Rim (2009) for Seoul, South Korea. Water bodies and vegetation cover contribute to the hydrological cycle through evaporation and transpiration, respectively. Open soil and some other types of natural land can soak up water and runoff, which also contribute to the hydrologic system. All these surfaces are now replaced by impervious surface due to urban expansion processes, which can considerably reduce regional ET rate. However, all these studied cities are located in areas where vegetation cover is abundant and water resource is sufficient. On the contrary, urban expansion has been found to increase urban ET for desert cities where both vegetation cover and precipitation are naturally scarce, such as Phoenix, Arizona, United States (Balling and Brazel 1987; Wang et al. 2016b). In order to understand the effect of urban expansion on ET, one has to understand local ecological and climatological characteristics.

The urban expansion process has caused a significant decrease of terrestrial ecosystem NPP in both Yangon and Nay Pyi Taw, but in different ways. Decrease of NPP in the Nay Pyi Taw region mainly was found in the downtown area and two dam construction sites (Fig. 6i). Urban expansion and development removed natural vegetation cover, and a large contiguous area of forests was submerged under the reservoir due to dam constructions. Therefore, decreased NPP around Nay Pyi Taw was mainly caused by deforestation and forest degradation processes. Yangon has been highly urbanized and industrialized. The decrease in NPP in Yangon region was partially due to deforestation for urban expansion. This finding is also consistent with an urbanization study in China (Xu et al. 2007) and a land-use change study in Central Germany (Schaldach and Alcamo 2007).

Although the urban expansion process has been accelerated through large-scale constructions and infrastructure developments in both cities, no evidence, through the analysis of monthly and annual mean AOD values, has yet shown to have impacts on the atmospheric environment (Fig. 7). Atmospheric aerosols in the urban area are mainly a result of anthropogenic emissions from factories, households, vehicles, and biomass burning (Elsom 2013). Increasingly, a higher concentration of aerosols has become an intractable issue confronting many major cities around the world, in both developed and developing countries (Santamouris 2011; Hoek et al. 2013; Lenoble et al. 2013). Therefore, it can be said that, although urban expansion and population growth have been prominent in Myanmar, the industrialization process has halted and the country’s economy still mainly relies on agriculture and light industry. There is no indication that heavy industry was vitalized and promoted in Myanmar during the study period. Figure 6b, d show an overall slight increasing trend in annual mean AOD values, but the relationships are not statistically significant.

Other implications

Some methods used in this study can be potentially useful for others. First, instead of examining just one case city, this study used two cases whose urban expansion processes are tightly intertwined. This method should be considered by more researchers, as urban expansion examined in isolation can lose many insights on drivers and impacts. Second, this study relied on remotely sensed satellite images as the main data source to examine urban BU areas and environmental parameters. This demonstrates an example for those study areas without having high-quality LULC maps or lack of reliable environmental and socio-economic data, especially in developing countries.

In addition, this research has direct environmental and socio-demographic implications. Nay Pyi Taw became Myanmar’s new capital in 2006, and its urban expansion process has been dramatic. What accompanied the highly accelerated urban expansion process was environmental deterioration within and around the city and the newly constructed capital, and it, in fact, did not attract large population migration. It remains the least populated region in Myanmar today. Yangon, on the contrary, was the old capital city, and its population increased simultaneously with urban expansion. It remains the most populated region and the economic center of the country. It, therefore, can be argued that the relocation of the national capital was purely from the perspective of politics. This political measure did not stimulate regional and national economic development, nor did it alleviate the population pressure in Yangon, but instead put more pressure on the environment for both cities.

The study of the former and current capitals of Myanmar demonstrates two distinct driving forces existing in transitional context: the market force and the institutional force. Although the institutional force can be powerful to initiate the large-scale urban expansion as illustrated in Nay Pyi Taw, its long-term effectiveness has yet to be evaluated. In contrast, market-force-based urban expansion seems to have a long-lasting effect on Yangon, despite the short-term stagnation due to the relocation of the capital to Nay Pyi Taw. Although other transitional economies seldom experienced such dramatic urban expansion cases as Yangon and Nay Pyi Taw, this study reminds us that we need to consider both driving forces seriously for urban expansion. In order for the country to achieve the goal of sustainability, one should note that population, cities, economy, and the environment need to be in harmony with each other, and the development has to be planned accordingly and strategically.

Conclusions

The process of urban expansion has played a significant role on both regional and global environmental change. This research studied urban expansion process and the impacts on the regional environment for two capital cities of Myanmar (Yangon and Nay Pyi Taw) between 2000 and 2013 using remotely sensed satellite images. Both cities have experienced extensive urban expansion but with different spatial and temporal characteristics. A steady increase of BU area with an increasing rate of 8.2%/year was found for Yangon, but Nay Pyi Taw had an exponential increase of urban area with a mean annual rate of 30.8%. The total BU area in Nay Pyi Taw has far exceeded Yangon since 2011. Urban expansion has caused a significant increase of LST for both daytime and nighttime, decreased PTC, decreased ET rate, and decreased terrestrial ecosystem NPP for both cities. No evidence, however, shows that urban expansion also has a negative impact on the atmospheric environment. This study underlined that two distinct driving forces co-existed in transitional economies—the market force and the institutional force—and that planning will address the environmental consequences of rapid urban expansion.