remote sensing

Article Measuring and Predicting Urban Expansion in the Region of

Jie Liu 1,2 , Hongge Ren 1,2,*, Xinyuan Wang 1, Zeeshan Shirazi 1 and Bin Quan 3

1 Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, No. 9 Dengzhuang South Road, Beijing 100094, 2 University of Chinese Academy of Sciences, Beijing 100049, China 3 College of City and , Hengyang Normal University, Hengyang 421002, China * Correspondence: [email protected]; Tel.: +86-185-7320-3680

 Received: 31 May 2019; Accepted: 4 July 2019; Published: 2 September 2019 

Abstract: Recent increases in urbanization and tourism threaten the viability of UNESCO world heritage sites across the globe. The Angkor world heritage site located in southern Cambodia is now facing such a challenge. Over the past two decades, Angkor has seen over 300,000% growth in international tourist arrivals, which has led to uncontrolled development of the nearby city of . This study uses remote sensing and GIS to comprehend the process of urban expansion during the past 14 years, and has applied the CA-Markov model to predict future urban expansion. This paper analyzes the urban pressure on the Angkor site at different scales. The results reveal that the urban area of Siem Reap city increased from 28.23 km2 in 2004 to 73.56 km2 in 2017, an increase of 160%. Urban growth mainly represented a transit-oriented pattern of expansion, and it was also observed that land surfaces, such as arable land, forests, and grasslands, were transformed into urban residential land. The total constructed land area in the core and buffer zones increased by 12.99 km2 from 2004 to 2017, and 72% of the total increase was in the buffer zone. It is predicted that the built-up area in Siem Reap is expected to cover 135.09 km2 by 2025 and 159.14 km2 by 2030. The number of monuments that are most likely be affected by urban expansion is expected to increase from 9 in 2017 to 14 in 2025 and 17 in 2030. The urban area in Siem Reap has increased dramatically over the past decade and monuments continue to be decimated by urban expansion. This paper urges closer attention and urgent actions to minimize the urban pressure on the Angkor site in the future.

Keywords: Angkor World Heritage; urban expansion; photointerpretation; remote sensing; CA-Markov model

1. Introduction Urbanization is the most irreversible and human-dominated form of land use, resulting in changes in land-cover, hydrological systems, biogeochemistry, climate, and biodiversity [1]. City boundaries continue to expand, consuming rural areas, forests, heritage sites, and other important non-urban areas [2]. The spatiotemporal analysis of urban expansion patterns and dynamics is important not only in urban geography, but also in sustainability, urban planning, and environmental studies [3]. Southeast Asia, at the center of a significant economic transformation, is now undergoing rapid growth and urbanization. The mostly rural countries of Southeast Asia had the highest spatial expansion rates, with at 7.3 percent and Cambodia at 4.3 percent [4]. However, environmental impacts and risks are a major concern in these countries. Southeast Asia is a culturally rich and diverse region containing numerous world heritage sites as classified by UNESCO. Urban expansion, understandably, has adverse impacts on these sites in terms of land use change, air pollution, traffic, invasive species, and foreign cultures [5]. Urbanization alters the surrounding natural landscape, making these sites

Remote Sens. 2019, 11, 2064; doi:10.3390/rs11172064 www.mdpi.com/journal/remotesensing Remote Sens. 2019, 11, 2064 2 of 21 susceptible to anthropogenic agents, while at the same time, urbanization improves accessibility to the sites, making them more vulnerable to tourism [6]. Satellite imagery and geospatial mapping have been widely applied to better understand urbanization for several decades. Many previous studies used satellite images to drive land cover data and employed landscape metrics to analyze the spatial patterns of land expansion in cities and regions [7,8]. In addition, a large number of studies analyzed the driving factors for urban land expansion in different cities and regions [9]. Remote sensing and GIS have also been used for heritage management [10,11]. As most heritage sites are tourist destinations located in cities, tourism creates growth and economic benefits for local communities and cities, and simultaneously makes both tangible and intangible cultural heritage vulnerable to irrevocable damage and loss [11]. If the heritage site is destroyed or seriously damaged, these urban areas cease their functional priority. It is, therefore, essential to develop planning methods to protect heritage sites located on the urban fringe. Increasing availability of remotely sensed observations and a variety of other geospatial information facilitates the development of new tools and approaches for understanding urban environments [12]. Many studies have given special attention to mapping urban growth and assessing the risk in the vicinity of cultural heritage sites. Agapiou et al. [2] used time series multispectral satellite images and predictive models to understand future urban settlement patterning, with special focus on the Paphos area in Cyprus. Angkor is a symbol of pride for the Cambodian nation and Khmer culture. It is an ancient archaeological landscape with extraordinary temples, monuments and layered remains of cities and settlements from the ninth to the fifteenth centuries [13]. In 1992, Angkor was inscribed as a United Nations Educational, Scientific and Cultural Organization (UNESCO) designated world heritage site. At Angkor, the environment is inseparable from the monuments, including forests, natural features, and agricultural lands [14,15]. Tourists have been coming to Angkor for almost a century. Siem Reap city, located 5 km south of , is the traditional gateway for visiting the Angkor world heritage site [16]. In recent years, tourism to Angkor has risen exponentially and is a key developing sector of the Cambodian economy [17]. The number of international tourists to the Angkor world heritage site has increased dramatically from 599,675 in 2004 to more than 2.45 million in 2017. The revenue from tourism in Cambodia has made a substantial contribution to growth in Cambodia’s gross domestic product (GDP). However, the rapid growth in visitation to Angkor has created processes that threaten the environment, local communities, and the value of the heritage (both tangible and intangible) [13]. The Angkor site is now affected by many factors, including proliferation of housing and road development, rapid consumption of resources, reduction of forest land, and the destruction of the ecological environment [18,19]. The rapid expansion of Siem Reap city has increased the risk of further degradation [20,21]. Mackay et al. revealed a clear disconnect between economic beneficiaries and owners that prized traditional cultural values in the Angkor world heritage site [22]. Stubbs et al. discussed applications of remote sensing to the understanding and management of Angkor [5]. Liao et al. [18] explored the land use/cover change in Angkor by using time series remote sensing images during the period from 1985 to 2013, showing that during the past 30 years, the urban land increased rapidly, mainly taking over agriculture and forest. Living with heritage is a joint Cambodian and international program focused on developing spatial methods for monitoring and analyzing landscape change in Angkor [10]. In the context of Angkor’s preservation goals, sustainable development requires integration of the conservation practices of these sites with the sustainable use and management of natural resources in the broader landscape [23]. The International Coordinating Committee for the Safeguarding and Development of the Historic Site of Angkor (ICC-Angkor) viewed the impending arrival of an international tourism industry as a force that would threaten Angkor’s long-term survival [24]. Fearful of rampant and uncontrolled development, UNESCO stated in 1996 that tourism threatens to damage this Khmer cultural legacy far more swiftly and decisively than did any ancient invaders, or even the clandestine raiders of today [25]. Therefore, this study aims to investigate and model the present urban expansion of Siem Reap city, Remote Sens. 2019, 11, x FOR PEER REVIEW 3 of 21

Remote Sens. 2019, 11, 2064 3 of 21 raiders of today [25]. Therefore, this study aims to investigate and model the present urban expansion of Siem Reap city, explore the driving factors and assess the urban pressure on Angkor at exploredifferent the driving scales. factorsIn addition, and assess this study the urban applied pressure a CA-Markov on Angkor model at di tofferent predict scales. the growth In addition, of urban thisexpansion study applied over a CA-Markovthe next 5 to model 10 years, to predict so as theto provide growth ofsome urban insights expansion for policy-makers over the next 5to to take 10 years,effective so as planning to provide and some management insights for actions policy-makers in the future. to take effective planning and management actions in the future. 2. Study Area 2. Study Area The Angkor world heritage site includes the archaeological parks of Angkor, Rolous, and BanteayThe Angkor Srei worldon the heritage Cambodian site includes plains between the archaeological the Kulen parks Mountains of Angkor, to the Rolous, north and and Banteay SreiLake on the to Cambodian the south [13]. plains In the between early 1990s, the Kulen UNESCO, Mountains in conjunction to the north with and the Tonle Cambodian Sap Lake government, to the southestablished [13]. In the the early Angkor 1990s, Zoning UNESCO, and inEnvironment conjunction withManagement the Cambodian Plan (ZEMP) government, team to established designate the the Angkorcultural Zoningsignificance and Environmentof central Angkor Management and its Planvicinity. (ZEMP) These team zoning to designate regulations the were cultural legally significanceenforced, of and central approved Angkor through and its national vicinity. legislat These zoningion [26]. regulations The ZEMP were prompted legally the enforced, government and to approvedestablish through the national national system legislation for the [26 protection]. The ZEMP of cultural prompted heritage the government sites, which to is establishstill used the to this nationalday (Figure system for1). theThe protection Angkor site of culturalis divided heritage into five sites, planning which iszones, still used namely to this Zone day 1: (Figure Monumental1). TheSites Angkor (core site zone); is divided Zone into 2: fiveProtected planning Archaeological zones, namely Reserves Zone 1: Monumental(buffer zone); Sites Zone (core 3: zone);Protected ZoneCultural 2: Protected Landscapes Archaeological (along rivers), Reserves including (buffer zone); historic Zone buildings, 3: Protected land Culturaluse practices, Landscapes Siem Reap (along river, rivers),and including river; historic Zone buildings, 4: Sites land of useArchaeological, practices, Siem Anthropological Reap river, and Roluosor Historic river; Interest Zone 4: (sites Sites not of Archaeological,included in Zone Anthropological 1 or 2); and Zone or Historic 5: Siem Interest Reap (sitesProvince, not included designated inZone in its 1 entirety or 2); and as Zonethe largest 5: Siemprotected Reap Province, area because designated of the in its abundant entirety asarchaeological the largest protected sites located area because in the ofprovince the abundant [26]. The archaeologicalCambodian sites government located in thedecree province stipulates [26]. Thethat Cambodian the core zone government is essential decree for stipulatesthe conservation that the and coredisplay zone is essentialof the Angkor for the world conservation heritage and site, display and the of bu theffer Angkor zone is world a development heritage site, zone and that the is bu essentialffer zonefor is athe development protection of zone local that residents. is essential for the protection of local residents.

FigureFigure 1. ZEMP 1. ZEMP zones zones in the in Angkor the Angko worldr world heritage heritage site. site.

The study area is located in the Angkor region with a total area of 810 km2, and lies between The study area is located in the Angkor region with a total area of 810 km2, and lies between 103◦45103°45'E0E~104 ◦~00 104°00'E0E and 13and◦36 13°36'N0N~13◦57 ~ 013°57'NN (Figure (Figure2). The 2). eastern, The eastern, western, western, and northern and northern boundaries boundaries of Angkor’sof Angkor’s buffer buffer zone were zone used were to used limit to thelimit spatial the spat scopeial scope of the of study the study area. area. In the In south, the south, the butheff erbuffer zonezone extends extends to the to the southernmost southernmost protected protected cultural cultural landscape. landscape. The The study study area area includesincludes the core and andbuffer buffer zoneszones ofof AngkorAngkor and and Roluos, Roluos, which which were were mainly mainly aff ectedaffected by by the the urban urban expansion expansion of Siem of Siem ReapReap city. city. Siem Siem Reap Reap city city is locatedis located in in relatively relatively flat flat terrain, terrain, which which provides provides a a natural natural advantage advantage for for urban expansion. The core zones in Angkor and Roluos cover approximately 161.55 km2 and urban expansion. The core zones in Angkor and Roluos cover approximately 161.55 km2 and 28.31 28.31 km2, respectively. The buffer zones occupy 189.45 km2 and 1.69 km2, respectively. There were

Remote Sens. 2019, 11, x FOR PEER REVIEW 4 of 21

km2, respectively. The buffer zones occupy 189.45 km2 and 1.69 km2, respectively. There were 48 monuments in the study area according to field investigations and literature records [14]. The study area is in a low-latitude area and features a tropical monsoon climate with high temperature and precipitation, ranging from 29°C to 30°C [27]. There is a large variation in precipitation with time Remoteand with Sens. 2019 geographic, 11, 2064 location, mainly a consequence of the terrain and monsoon activity. Generally4 of 21 speaking, the rainy season lasts from May to November while the dry season prevails from December to April. The average temperature ranges from 20°C in winter to 31°C in summer. 48 monuments in the study area according to field investigations and literature records [14]. The study Usually, tourists prefer to visit Angkor in the dry season [19]. area is in a low-latitude area and features a tropical monsoon climate with high temperature and precipitation,This study ranging mainly from focused 29◦C on to 30using◦C[ 27remote]. There sens ising a large and GIS variation to assess in precipitation the urban expansion with time of andSiem with Reap geographic city in the location, core and mainly buffer azones consequence of Angkor of theand terrain Roluos. and Angkor monsoon and activity.Roluos are Generally the two speaking,popular tourist the rainy destinations season lasts in from the Angkor May to Novemberworld heritage while site, the dryand season all land prevails in the fromcore Decemberand buffer tozones April. are The considered average temperature by the Cambodian ranges from government 20◦C in winter as topublic 31◦C land in summer. and is Usually,inalienable tourists and preferimprescriptible to visit Angkor [28]. in the dry season [19].

Figure 2. Map of the study area. Figure 2. Map of the study area. This study mainly focused on using remote sensing and GIS to assess the urban expansion of Siem 3. Data and Methodology Reap city in the core and buffer zones of Angkor and Roluos. Angkor and Roluos are the two popular tourist destinations in the Angkor world heritage site, and all land in the core and buffer zones are 3.1. Data Used considered by the Cambodian government as public land and is inalienable and imprescriptible [28]. The GIS layers of the Angkor monument sites with demarcated core and buffer zones were 3.provided Data and by Methodology the APSARA National Authority. The built-up areas in Siem Reap and the Angkor site were analyzed using multi-source high-resolution remote sensing data acquired for 2004, 2007, 2010, 3.1. Data Used 2013, and 2017, which constituted the individual study periods and collectively defined the total studyThe duration. GIS layers All of remote the Angkor sensing monument images collected sites with for demarcated this study corewere and obtained buffer from zones optical were providedsensors from by the 6 January APSARA to National 12 March, Authority. representing The the built-up dry season. areas in The Siem detailed Reap andcharacteristics the Angkor of site the weredata analyzedare provided using in multi-source Table 1. high-resolution remote sensing data acquired for 2004, 2007, 2010, 2013, and 2017, which constituted the individual study periods and collectively defined the total study Table 1. The detailed characteristics of the remote sensing data. duration. All remote sensing images collected for this study were obtained from optical sensors from 6 January to 12 March, representing Data the Source dry season. The Spatial detailed Reso characteristicslution ofAcquisition the data are Date provided in Table1. Tourism population data was derived from the tourism statistics report obtained from the official website of the Cambodian Tourism Administration. The data on land use and land cover, traffic lines, and rivers was obtained from Open Street Map [29] and then verified and corrected based on high-resolution images and Google Earth. The slope data was obtained from a digital elevation model (DEM) with 30 m resolution, downloaded from the USGS website and processed in ArcGIS software. Remote Sens. 2019, 11, 2064 5 of 21

Table 1. The detailed characteristics of the remote sensing data.

Data Source Spatial Resolution Acquisition Date 1 QuickBird 0.6 m 6 January 2004 2 ALOS 2.5 m 15 February 2007 3 ALOS 2.5 m 12 March 2010 4 SPOT5 2.5 m 12 January 2013 5 GF1 2 m 16 February 2017

3.2. Data Processing and Methodology The coordinate system of all images was WGS 1984 UTM Zone 48N and the projection was Transverse Mercator. In remote sensing, images contain geometric distortion relative to the surface target due to the influence of the imaging projection mode, changes in the orientation elements outside the sensor, inhomogeneity of the sensing medium, and the curvature of Earth. In this study, the geometric figure on the image was different from the object in the selected map projection. The issue was solved using geometric correction. The 2004 Quick Bird data was pre-processed and additional data was corrected based on this product. ENVI 5.3 remote sensing processing software was used for image pre-processing, including geometric correction and registration, cropping, image enhancement, and other operations. Geometric correction reduced the error of control points to be within 0.5 pixels. The remote sensing data from each period was cropped using vector boundary data of the study area to improve work efficiency. The color synthesis method was used to enhance the images from different phases. Following pre-processing of the remote sensing images, the next step was to map the successive land use changes in Siem Reap during the study period. Primarily, the built-up areas for each individual study period were digitized through meticulous visual interpretation of the remote sensing image from that particular year. The digitization process was completed using the ArcGIS 10.2 platform. Following this, the land use change maps for four time intervals, 2004–2007, 2007–2010, 2010–2013, and 2013–2017, were created by overlaying the vector boundary of the built-up area for each individual year with the vector boundary of the year preceding it. This was done successively for all years, with the exception of the year 2004 as this was the baseline. Simultaneously, the change in the land use/land cover category attributes (wetland, crop, forest, water, grass, barren) were also recorded from one study period to another. The urban land use map for 2017 was created, by thorough visual interpretation, to study the urban use of new built-up areas in 2017. A total of 11 main categories of urban land use were identified in the built-up areas of 2017, which included commercial, farm, forest, grass, industry, meadow, military, orchard, park, residential, and retail. The patches of land for which clear urban use could not be determined were marked as null in the land use attributes. The classification results were verified from land use and land cover maps obtained from Open Street Map.

3.2.1. Method of Quantifying Urban Expansion and Land Use Change To determine the predominant direction of urbanization in the city, a circle with a radius of 14,700 m was created, centered on the 2017 built-up area vector file and covering the entire periphery of the city in 2017. From the center of this circle, a 45-degree segmented angle was used to divide the study area into eight azimuth areas in the eight geographical directions (N, NE, E, SE, S, SW, W, NW). The built-up area in each azimuth area was then calculated for each individual study period. The expansion intensity of the built-up area, within an azimuth area, was calculated by dividing the built-up area in individual azimuth areas by the total built-up area in Siem Reap city at that point in time. The land use change maps, calculated for the four time intervals, were also overlaid on top of the urban land use map of 2017 to determine the prior uses of the new built-up areas at each point in time before 2017. As the land use change records also had null values, only the effective patches with meaningful attributes were sampled. The numbers of effective patches for each progressive time Remote Sens. 2019, 11, 2064 6 of 21 interval were 591, 50, 161, and 162 respectively. They collectively accounted for 95% of the new built-up area. The use of urban expansion was then quantified by counting the percentage of each urban land use type in the effective patches of the four land cover change time intervals.

3.2.2. Methodology to Simulate Urban Expansion

Markov Model The Markov model is a theory based on the process of the formation of Markov random process systems for the prediction and optimal control theory method [30]. Changes in urban and non-urban lands can be analyzed and summarized by the number of transition area probabilities from one status to various other statuses during a certain period of time using this model [31,32]. The application of the Markov model in land-use change modeling is promising because of its ability to quantify, not only the states of conversion between land-use types, but also the rate of conversion among the land use types [33]. A homogenous Markov model for predicting land-use change can be represented mathematically as: L = P L (1) (t+1) ij ∗ (t) where L(t+1) and L(t) are the land-use status at time t+1 and t, respectively. Pij is the transition probability matrix in a state which is calculated as follows:    P11 P1m   ···  X  . . .  m L(t+1) =  . .. . , (0 Pij 1 and p = 1, (i, j = 1, 2, ... , m)) (2)  . .  ≤ ≤ j=1 ij   P Pmm m1 ··· Cellular Automata (CA) Model Cellular automata (CA) are spatially dynamic models frequently used for land-use and land-cover change studies [34,35]. In a CA model, the transition of a cell from one land-cover to another depends on the state of the neighborhood cells [36]. The CA model has an open structure and can be integrated with other models to simulate and predict urban growth patterns in GIS [37]; thus, the CA model has a high spatial resolution and computational efficiency [38]. The CA model can be expressed as follows:

S(t, t + 1) = f(S(t),N) (3) where S is the set of limited and discrete cellular states, N is the cellular field, t and t + 1 indicate the different times, and f is the transformation rule of cellular states in local space.

CA-Markov Model Both the Markov model and the CA model are time-discrete, state-discrete dynamic models. The CA-Markov model is widely used in the simulation and prediction of land use change [39]. The CA model provides strong spatial computing power and the Markov model has the advantages of long-term prediction [2,40]. In this paper, the future range of built-up areas were predicted using the CA-Markov model applied using the IDRISI software platform. The map was based on the reality of traffic-oriented urban sprawl and expansion patterns in Siem Reap. The main factors affecting the conversion of the built-up area were summarized to be: The impact of protection and restriction policies; the impact of existing construction areas on new land development; the influence of traffic and distance between cities and towns; and the impact of other land types on the suitability of land development [41–43]. Therefore, based on the above principles, the study selected two constraint factors and five driving factors to develop a suitability map for predicting future urban expansion. The existing built-up areas and water bodies were classified as two constraint factors because they prohibit land construction. Figure3 shows the framework of urban expansion simulation. The five driving factors included the Remote Sens. 2019, 11, x FOR PEER REVIEW 7 of 21 and distance between cities and towns; and the impact of other land types on the suitability of land development [41–43]. Therefore, based on the above principles, the study selected two constraint factors and five driving factors to develop a suitability map for predicting future urban expansion. The existing Remotebuilt-up Sens. areas2019, 11and, 2064 water bodies were classified as two constraint factors because they prohibit7 land of 21 construction. Figure 3 shows the framework of urban expansion simulation. The five driving factors included the slope, the distance from the center of the existing built-up area, the distance to the slope, the distance from the center of the existing built-up area, the distance to the built-up area, the built-up area, the distance to traffic lines, and the distance to rivers. The values for the two distance to traffic lines, and the distance to rivers. The values for the two constraint factors were constraint factors were obtained from Open Street Map, and then verified and corrected based on obtained from Open Street Map, and then verified and corrected based on high-resolution images. high-resolution images. The data obtained from the study area was statistically verified to The data obtained from the study area was statistically verified to determine the range of influences of determine the range of influences of each individual driving factor on land development. The each individual driving factor on land development. The highest development suitability values were highest development suitability values were found to be decreasing outward at 300 m from the found to be decreasing outward at 300 m from the center of Siem Reap. There was no increase in land center of Siem Reap. There was no increase in land development beyond 1,000 m [44]. The development beyond 1000 m [44]. The development suitability was found to be optimal within a range development suitability was found to be optimal within a range of 200 m from the existing traffic of 200 m from the existing traffic lines, with the suitability gradually decreasing outward. No influence lines, with the suitability gradually decreasing outward. No influence of traffic lines was observed of traffic lines was observed beyond 500 m on development suitability [45]. Likewise, the most suitable beyond 500 m on development suitability [45]. Likewise, the most suitable areas for development areas for development were located within a range of 250 m from the existing built-up area, with the were located within a range of 250 m from the existing built-up area, with the suitability gradually suitability gradually decreasing outward. There was no impact on development suitability beyond decreasing outward. There was no impact on development suitability beyond 700 m from the 700 m from the built-up area [46]. The development suitability for areas within a range of 100 m built-up area [46]. The development suitability for areas within a range of 100 m from rivers was from rivers was low and gradually increased outward and remained unchanged outside 800 m [47]. low and gradually increased outward and remained unchanged outside 800 m [47]. In the case of In the case of slopes, the development suitability of an area understandably decreases as the slope slopes, the development suitability of an area understandably decreases as the slope increases. increases. Therefore, an area with a slope of less than 8% was considered the most suitable, 8%–15% Therefore, an area with a slope of less than 8% was considered the most suitable, 8%–15% was was considered more suitable, 15%–25% was generally suitable, and a slope of more than 25% was not considered more suitable, 15%–25% was generally suitable, and a slope of more than 25% was not suitable [48]. The possibility of each pixel belonging to a fuzzy set was evaluated by the membership suitable [48]. The possibility of each pixel belonging to a fuzzy set was evaluated by the function of the fuzzy set. All factor layers were classified into different suitability grades according to membership function of the fuzzy set. All factor layers were classified into different suitability the range of parameters. Figure4 shows the driving factors evaluated by the membership function of grades according to the range of parameters. Figure 4 shows the driving factors evaluated by the the fuzzy set. membership function of the fuzzy set.

Figure 3. The framework of urban expansion simulation.

Remote Sens. 2019, 11, x FOR PEER REVIEW 8 of 21 Remote Sens. 2019, 11, x FOR PEER REVIEW 8 of 21 Figure 3. The framework of urban expansion simulation. Remote Sens. 2019, 11, 2064 Figure 3. The framework of urban expansion simulation. 8 of 21

FigureFigure 4. 4. TheThe driving driving factors factors evaluated evaluated by by the the me membershipmbership function function of of the fuzzy fuzzy set. Figure 4. The driving factors evaluated by the membership function of the fuzzy set. TheThe standardizedstandardized constraintconstraint and and driving driving factors factors were were added added to the to multi-criteria the multi-criteria evaluation evaluation (MCE) (MCE)module.The module. Thestandardized weighted The weighted linearconstraintcombination linear and combination driving (WLC) factors method(WLC) were method was added selected was to selectedthe to providemulti-criteria to provide different evaluation different weights weightsof(MCE) 0.3, 0.3, module. of 0.2,0.3, 0.2,0.3, The and 0.2, weighted 0.10.2, to and binding linear 0.1 to factors.combination binding Using factors. (WLC) the collectionUsing method the editorwascollection selected tool, editor suitability to provide tool, layers suitability different were layerscombinedweights were of into 0.3,combined a0.3, suitability 0.2, into 0.2, a map. suitabilityand The0.1 to results map.binding areThe presentedfactors. results areUsing in presented Figure the 5collection. in Figure editor 5. tool, suitability layers were combined into a suitability map. The results are presented in Figure 5.

Figure 5. Suitability map. Figure 5. Suitability map. Figure 5. Suitability map.

Remote Sens. 2019, 11, 2064 9 of 21

Markov operations were performed using the data from 2013 to 2017 to obtain the transition probability. For the CA-Markov analysis, the built-up area in 2017 was used as the base map, the transition probability layer was used as the Markov transition areas file, and the suitability map was used as the transition suitability image collection, while the number of cellular automata iterations were selected to be 8. The results from the CA-Markov analysis predicted the built-up areas in 2025 and 2030.

Model Validation In order to validate the proposed approach, the kappa coefficient was employed to quantify the accuracy of the model by comparing the actual built-up areas with the simulated maps. First, the built-up areas were simulated using the proposed methodology for years 2010, 2013, and 2017, respectively. Then, the simulated results were compared with the actual maps. Table2 shows the comparisons between actual built-up areas and simulated built-up areas in 2010, 2013, and 2017. The kappa coefficients were 83%, 85%, and 87%, all of which were highly consistent. The standard accuracy of the models, which was not less than 80%, determined that they were potent predictive tools [49,50]. According to Landis and Koch [51], this task verifies and approves the simulation process. Therefore, the approach can also be executed for future years (i.e., 2025, 2030).

Table 2. The comparisons between actual areas and simulated areas in 2010, 2013, and 2017 (unit: m2).

Year Actual Built-Up Areas Simulated Built-Up Areas Kappa Coefficient 2010 48,768 50,110 83% 2013 59,602 58,115 85% 2017 81,843 79,575 87%

The methodology for prediction used in this study has some limitations and is based on certain assumptions. This study assumed that politics would remain constant, and no new regulations would be passed till 2030; existing regulations would be enforced at the same level; the natural resources needed to support continued growth would be available (water, electricity, building materials), and tourism was expected to grow at a similar rate. These factors should be further considered in future studies in order to conduct a more in-depth analysis of the urban expansion in Siem Reap.

4. Results

4.1. Model and Process of Urban Expansion of Siem Reap City The first objective of this study was to record the change in the built-up area. The results showed that in Siem Reap, the built-up area increased from 28.23 km2 in 2004 to 73.56 km2 in 2017. The progression of urban expansion is displayed in Figure6. Generally, the expansion has been recorded in all directions; however, the predominant expansion can be observed along National Highway 6 and Highway 52 in the east-west direction. In the south, the expansion occurred along Highway 63. In the north, the urban development expanded along the Siem Reap River in the periphery of the original built-up area. Remote Sens. 2019, 11, 2064 10 of 21 Remote Sens. 2019, 11, x FOR PEER REVIEW 10 of 21

Figure 6. Urban expansion of Siem Reap from 2004 to 2017. Figure 6. Urban expansion of Siem Reap from 2004 to 2017. From Figure7, it can be observed that in 2004, 2007, 2010, and 2013, the proportion of developed landFrom in the Figure northwest 7, it (NW) can wasbe observed 24%, 20%, that 23%, in and2004, 22%, 2007, respectively, 2010, and while 2013, inthe the proportion west (W) theof proportiondeveloped land of developed in the northwest land was (NW) 14%, was 18%, 24%, 17%, 20% and, 23%, 17%, and respectively. 22%, respectively, These expansionswhile in the werewest observed(W) the proportion to be along of Siem developed Reap’s mainland trawasffi 14%,c lines 18%, (e.g., 17%, National and 17%, Highway respectively. 6) in the These northwestern expansions and easternwere observed (NW-E) directions.to be along This Siem suggests Reap’s that main urban traffic development lines (e.g., in Siem National Reap wasHighway mainly driven6) in the by thenorthwestern proximity ofand roads eastern in the (NW-E) early stages directions. of its expansion.This suggests In 2017,that urban the proportion development of urban in Siem expansion Reap inwas both mainly the northwest driven by and the eastproximity (NW-E) of was roads 19%. in However,the early thestages rate of of its areas expansion. being converted In 2017, intothe built-upproportion areas of urban decreased expansion compared in both to previous the northw years,est and and east those (NW-E) in the westwas 19%. (W) andHowever, southeast the (SE)rate increasedof areas being to 13% converted and 15%, into respectively. built-up areas This decrea suggestssed compared that the areas to previous between years, the transportand those routesin the werewest developed(W) and southeast as Siem Reap(SE) increased expanded to along 13% the and tra 15%,ffic line respectively. due to agglomeration This suggests eff ects.that Thethe areas main directionsbetween the of urbantransport expansion routes werewere observeddeveloped to as be Si inem the Reap northwest expanded and eastalong (NW-E) the traffic for the line four due time to periods.agglomeration Expansion effects. trends The were main also directions observed inof the urban west (W)expansion and southeast were observed (SE). In the to northern be in (N)the direction,northwest the and area east of (NW-E) urban expansion for the four accounted time periods. for a relatively Expansion small trends part ofwere the also total observed built-up areain the in thewest whole (W) and year southeast and showed (SE). a downwardIn the northern trend. (N) This direction, was due the to the area land of useurban limitations expansion placed accounted on the protectedfor a relatively areas surroundingsmall part of the the heritage total built-up sites. area in the whole year and showed a downward trend. This was due to the land use limitations placed on the protected areas surrounding the heritage sites.

Remote Sens. 2019, 11, 2064 11 of 21 Remote Sens. 2019, 11, x FOR PEER REVIEW 11 of 21

Figure 7. An eight-direction diagram of urban expansion intensity in Siem Reap from 2004 to 2017. Figure 7. An eight-direction diagram of urban expansion intensity in Siem Reap from 2004 to 2017. According to the Angkor Heritage Management Framework (2013), the key drivers of urban developmentAccording are to population the Angkor growth Heritage and tourism.Management In 2018, Framework Siem Reap (2013), city had the a populationkey drivers of of 139,458 urban residents,development compared are population with 119,528 growth in 1998. and Currently, tourism. overIn 2018, 50% ofSiem the populationReap city had works a population in the tourism of industry.139,458 residents, The residential compared population with 119,528 growth in 1998 is a direct. Currently, response over to 50% the opportunitiesof the population for economic works in developmentthe tourism industry. through The tourism residential [13]. However,population the grow increasedth is a direct local response population to the is a opportunities relatively small for numbereconomic compared development with booming through tourists. tourism Figure [13].7 depictsHowever, the relationshipthe increased between local urbanpopulation expansion is a andrelatively changes small in tourism.number compared The number with of booming international tourists. tourists Figure visiting 7 depicts Angkor the relationship Archaeological between Park increasedurban expansion from 599,675 and inchanges 2004 to in 2,457,023 tourism. in The 2017, numb whileer during of international the same period, tourists the visiting built-up Angkor area in SiemArchaeological Reap also expandedPark increased on a yearlyfrom 599,675 basis (Figure in 20048a), to indicating 2,457,023 strong in 2017, linkages while betweenduring the the same two. Figureperiod,8 bthe provides built-up a comparisonarea in Siem between Reap also the expanded yearly growth on a yearly rate of thebasis built-up (Figure area 8a), inindicating Siem Reap strong and thelinkages number between of tourists the intwo. adjacent Figure periods. 8b provides From a 2004 comparison to 2007, visiting between tourists the yearly and the growth urban rate area of grew the bybuilt-up 49% and area 34%, in respectively.Siem Reap and From the 2007 number to 2010, of the tourists growth in rates adjacent for the periods. tourist population From 2004 and to urban2007, areavisiting decreased tourists by and 46% the and urban 16%, respectively.area grew by In 49% the periodand 34%, from respectively. 2010 to 2013, From the growth 2007 to rates 2010, for the touristgrowth and rates urban for the area tourist were population 71% and 22%, and respectively. urban area decreased The growth by rates 46% forand the 16%, tourists respectively. in the first In threethe period periods from were 2010 generally to 2013, higher the growth than urban rates areas.for the The tourist growth and rate urban for area the built-up were 71% area and in Siem22%, Reaprespectively. was observed The growth to fluctuate rates withfor the the tourists growth in rate the of first the touristthree periods population. were The generally growth higher rate for than the touristsurban areas. dropped The sharplygrowth fromrate for 2013 the to built-up 2017, while area the in growthSiem Reap rate was for urbanobserved areas to increasedfluctuate fromwith thethe previousgrowth rate period. of the These tourist results popula suggesttion. The that growth tourism rate can for drive the Siemtourists Reap’s dropped urban sharply expansion, from and2013 the to rate2017, of while urban the expansion growth continues rate for growingurban areas when increased the growth from rate the of theprevious tourist period. population These fluctuates. results suggest that tourism can drive Siem Reap’s urban expansion, and the rate of urban expansion continues growing when the growth rate of the tourist population fluctuates.

Remote Sens. 2019, 11, 2064 12 of 21 RemoteRemote Sens. Sens. 2019 2019, ,11 11, ,x x FOR FOR PEER PEER REVIEW REVIEW 12 12 of of 21 21

FigureFigure 8. 8. TheThe comparison comparison of of the the growth growthgrowth of of urban urban expansion expansion and and the the touristtourist population:population: ( (aa))) the thethe comparisoncomparisoncomparison of ofof the thethe urban urbanurban area areaarea and andand tourist touristtourist population; population;population; ( ((bbb))) the thethe comparison comparisoncomparison of ofof the thethe growth growthgrowth rates ratesrates of ofof urban urbanurban areasareasareas and andand the thethe tourist touristtourist population. population.population. 4.2. Source and Use Analysis of Land-Use Change from Urban Expansion 4.2.4.2. SourceSource andand UseUse AnalysisAnalysis ofof Land-useLand-use ChangeChange fromfrom UrbanUrban ExpansionExpansion To assess the dynamics of land cover changes, the original state of the built-up area was observed. ToTo assessassess thethe dynamicsdynamics ofof landland covercover changes,changes, thethe originaloriginal statestate ofof thethe built-upbuilt-up areaarea waswas It was found that farmland constituted a majority of land cover that was converted to urban land observed.observed. ItIt waswas foundfound thatthat farmlandfarmland constitutedconstituted aa majoritymajority ofof landland covercover thatthat waswas convertedconverted toto cover (e.g., housing and residential construction) (Figure9). The area of cultivated land occupied by urbanurban landland covercover (e.g.,(e.g., housinghousing andand residentialresidential construction)construction) (Figure(Figure 9).9). TheThe areaarea ofof cultivatedcultivated landland urban expansion in Siem Reap was 11.82 km2 from 2004 to 2007, 4.27 km2 from 2007 to 2010, 10.02 km2 occupiedoccupied byby urbanurban expansionexpansion inin SiemSiem ReapReap waswas 11.8211.82 kmkm22 fromfrom 20042004 toto 2007,2007, 4.274.27 kmkm22 fromfrom 20072007 toto from 2010 to 2013, and 4.59 km2 from 2013 to 2017. These results suggest urban areas were mostly 2010,2010, 10.0210.02 kmkm22 fromfrom 20102010 toto 2013,2013, andand 4.594.59 kmkm22 fromfrom 20132013 toto 2017.2017. ThTheseese resultsresults suggestsuggest urbanurban developed by replacing cultivable land. The area of forest and grassland occupied by urban expansion areasareas werewere mostlymostly developeddeveloped byby replacingreplacing cultivablecultivable land.land. TheThe areaarea ofof forestforest andand grasslandgrassland in Siem Reap was 1.15 km2 from 2004 to 2007, 1.58 km2 from 2007 to 2010, 3.46 km2 from 2010 to occupiedoccupied byby urbanurban expansionexpansion inin SiemSiem ReapReap waswas 1.151.15 kmkm22 fromfrom 20042004 toto 2007,2007, 1.581.58 kmkm22 fromfrom 20072007 toto 2013, and 6.25 km2 from 2013 to 2017. These figures suggest that there was also an increase in the 2010,2010, 3.463.46 kmkm22 fromfrom 20102010 toto 2013,2013, andand 6.256.25 kmkm22 fromfrom 20132013 toto 2017.2017. TheseThese figuresfigures suggestsuggest thatthat therethere amount of forest and grassland being converted into urban land. A recent report by the Bunna Realty waswas alsoalso anan increaseincrease inin thethe amountamount ofof forestforest andand grasslandgrassland beingbeing convertedconverted intointo urbanurban land.land. AA Group shows that as of the third quarter of 2015, there are 417 hotels with 17,000 rooms in Siem Reap, recentrecent reportreport byby thethe BunnaBunna RealtyRealty GroupGroup showsshows thatthat asas ofof thethe thirdthird quarterquarter ofof 2015,2015, therethere areare 417417 compared with 392 rooms in 1993. hotelshotels withwith 17,00017,000 roomsrooms inin SiemSiem Reap,Reap, comparedcompared withwith 392392 roomsrooms inin 1993.1993.

Figure 9. Statistics of urban expansion from other land cover types from 2004 to 2017. FigureFigure 9. 9. Statistics Statistics of of urban urban expansion expansion from from ot otherher land land cover cover types types from from 2004 2004 to to 2017. 2017. The proportion of built-up land designated for various purposes is shown in Figure 10. More than 70%The ofThe newly-added proportionproportion of urbanof built-upbuilt-up land land wasland used designateddesignated for residential forfor variousvarious construction, purposespurposes followed isis shownshown by inin park FigureFigure construction, 10.10. MoreMore thanwhichthan 70%70% was ofof the newly-addednewly-added case for every urbanurban time interval. landland waswas usedused foforr residentialresidential construction,construction, followedfollowed byby parkpark construction,construction, whichwhich waswas thethe casecase forfor everyevery timetime interval.interval.

Remote Sens. 2019, 11, x FOR PEER REVIEW 13 of 21

Remote Sens. 2019, 11, 2064 13 of 21 Remote Sens. 2019, 11, x FOR PEER REVIEW 13 of 21

Figure 10. Statistics of urban expansion in Siem Reap from 2004 to 2017.

4.3. Urban Encroachment on Monuments and Protected Areas Figure 10. Statistics of urban expansion in Siem Reap from 2004 to 2017. To record theFigure procession 10. Statistics of urban of urban encroach expansionments in Siemtowards Reap thefrom 48 2004 monuments to 2017. considered in this4.3. Urbanstudy, Encroachment buffers were on mapped Monuments at radial and Protected distances Areas of 200 m, 500 m, and 1,000 m from these 4.3. Urban Encroachment on Monuments and Protected Areas monuments,To record and the overlaid procession on ofthe urban built-up encroachments area map from towards 2017 the(Figure 48 monuments 11). The results considered showed in that this forstudy, 15To buof record fftheers 48 were themonuments, mappedprocession at the radialof built-upurban distances encroach area ofhad 200ments encroached m, 500towards m, andwithin the 1000 48 1,000monuments m from m of these the considered monuments,monument, in whileandthis overlaidstudy, 9 of the buffers on48 monuments the were built-up mapped had area built-up mapat radial from area distances 2017 encroachments (Figure of 200 11 ). m,within The 500 results 500m, mand and showed 1,000 8 monuments m that from for 15these had of thesethemonuments, 48 encroachments monuments, and overlaid the within built-up on the200 areabuilt-up m. had The encroachedarea monume map fromnts within 2017at the 1000 (Figure periphery m of 11). the The monument,of results Siem showedReap while were 9that of understandablythefor 15 48 of monuments the 48 foundmonuments, had to built-upbe more the areaaffected.built-up encroachments Tablearea had 3 gi vesencroached within the list 500 of within the m andaffected 1,000 8 monuments monumentsm of the monument, had in 2017. these encroachmentswhile 9 of the 48 within monuments 200 m. had The built-up monuments area atencroachments the periphery within of Siem 500 Reap m and were 8 monuments understandably had foundthese toencroachments be more affected. within Table 2003 gives m. theThe list monume of the a ffntsected at monumentsthe periphery in 2017. of Siem Reap were understandably found to be more affected. Table 3 gives the list of the affected monuments in 2017.

Figure 11. Analysis of the impact of urban expansion on Angkor monuments in 2017. Figure 11. Analysis of the impact of urban expansion on Angkor monuments in 2017.

Figure 11. Analysis of the impact of urban expansion on Angkor monuments in 2017.

Remote Sens. 2019, 11, x FOR PEER REVIEW 14 of 21

Table 3. List of affected monuments in 2017.

Newly affected Newly affected Remote Sens. 2019, 11, 2064 Affected monuments at 200 14 of 21 Time point monuments at 500 m monuments at 1,000 m m radius radius radius Prasat TaTable Noreay, 3. List Prasat of affected monuments in 2017. Prasat Pou Teng, Prei, Prasat Patri, Prasat Newly Affected Newly Affected Affected Monuments at Prasat Preah Kou, Time2017 Point Kaoh Ho, Voat Preah MonumentsPhnom Kraom. at 500 m 1 Monuments at 1000 m 200 m Radius Prasat Ou Kaek and Enteak Kosei, Tram Neak, Radius Radius Vat Chedei. 2 PrasatLoley, Ta and Noreay, Prasat Prasat Dounso. Prei, Prasat Pou Teng, Prasat Prasat Patri, Prasat Kaoh Ho, 2017 Phnom Kraom. 1 Preah Kou, Prasat Ou 1Newly affected monumentsVoat Preah at 500 Enteak m radial Kosei, distance Tram compared with that at 200 m radial distance. Kaek and Vat Chedei. 2 2Newly affected monumentsNeak, Loley, at 1,000 and m Prasat radial Dounso. distance compared with that at 500 m radial distance. 1 Newly affected monuments at 500 m radial distance compared with that at 200 m radial distance. 2 Newly affected monumentsThe spatial at 1000extent m radial of the distance built-up compared area withwithin that ea at 500ch buffer m radial of distance. all monuments was also calculated for each year in the study duration (Figure 12). The results show a linear increase in spatial extent of the built-upThe spatial area extent from 2004 of the to built-up 2017 within area withineach buffer each zone, buffer with of all 1,000 monuments m having was the alsosteepest calculated rise in forbuilt-up each yeararea infor the the study study duration duration. (Figure This is 12 follow). Theed results by the showincrease a linear in the increase built-up in area spatial within extent the of500 the m built-upbuffer, which area from was 20040.85 km to 20172 in 2004. within This each area bu increasedffer zone, to with 1.38 1000 km2 m in having 2007, and the steepestcontinued rise to inincrease built-up reaching area for 1.82 the km study2 in duration.2010, 2.12 Thiskm2 in is 2013, followed and by2.93 the km increase2 by 2017. in theThe built-upbuilt-up areaarea withinwithin thethe 500200 m m bu bufferffer, which exhibited was 0.85a negligible km2 in 2004. increase This areabut, increaseddue to the to proximity 1.38 km2 in to 2007, the monuments, and continued is toconcerning increase reachingnonetheless. 1.82 km2 in 2010, 2.12 km2 in 2013, and 2.93 km2 by 2017. The built-up area within the 200 m buffer exhibited a negligible increase but, due to the proximity to the monuments, is concerning nonetheless.

Figure 12. The area of urban expansion in buffer zones from 2004 to 2017. Figure 12. The area of urban expansion in buffer zones from 2004 to 2017. The overall built-up area in the core (Zone 1) and buffer zones (Zone 2) was also calculated for eachThe time overall period built-up (Figure 13area). Inin the 2004, core the (Zone built-up 1) and area buffer in the zones core (Zone and bu 2)ff erwas zones also hadcalculated a similar for spatialeach time extent, period covering (Figure 2.94 13). km In2 and 2004, 2.35 the km built-2, respectively.up area in Thethe constructedcore and buffer land zones area in had the corea similar and buspatialffer zones extent, further covering increased 2.94 km in2 2004, and reaching2.35 km2, 3.13 respectively. km2 and3.69 The km constructed2, respectively. land Inarea 2007, in the the ratecore ofand urbanization buffer zones in thefurther buffer increa zonesed exceeded in 2004, that reaching in the core 3.13 zone, km2 withand similar3.69 km degrees2, respectively. of linear In growth 2007, inthe both rate zones of urbanization over time. In in 2017, the buffer the built-up zone exceeded area in the that core in and the bu coreffer zone, zones with was approximatelysimilar degrees 2 toof 5linear times growth the original in both area, zones covering over5.59 time. km In2 and2017, 11.69 the kmbuilt-up2, respectively. area in the Collectively, core and buffer the built-up zones areawas 2 in both zones increased by 12.99 km from 2004 to 2017. The share of the built-up area in buffer zones amounted to 9.34 km2 alone, accounting for 72% of the total increase. Remote Sens. 2019, 11, x FOR PEER REVIEW 15 of 21 approximately 2 to 5 times the original area, covering 5.59 km2 and 11.69 km2, respectively. Collectively, the built-up area in both zones increased by 12.99 km2 from 2004 to 2017. The share of Remotethe built-up Sens. 2019 area, 11, 2064 in buffer zones amounted to 9.34 km2 alone, accounting for 72% of the15 total of 21 increase.

Figure 13. Urban expansions in Zone 1 and Zone 2 from 2004 to 2017. Figure 13. Urban expansions in Zone 1 and Zone 2 from 2004 to 2017. 4.4. Predicting the Urban Expansion and Its Impact on Monuments 4.4. Predicting the Urban Expansion and Its Impact on Monuments The results in Figures 14 and 15 reveal that the amount of built-up area in Siem Reap is expected to increaseThe results during in the Figure forecast 14 and period. Figure The 15 built-up reveal that area the is predicted amount of to coverbuilt-up 135.09 area km in 2Siemby 2025 Reap and is 159.14expected km to2 by increase 2030. The during construction the forecast may period. convert The cultivated built-up land, area forest is predicted land, grassland, to cover and135.09 barren km2 land.by 2025 In termsand of159.14 spatial km distribution,2 by 2030. The the built-upconstruction area is may predicted convert to becultivated distributed land, near forest the central land, towngrassland, and alongand barren both sides land. of In the terms main of roads. spatial These distribution, changes werethe built-up consistent area with is thosepredicted that to were be observeddistributed from near 2004 the tocentral 2017. Bytown 2025, and 14 along monuments both sides are expectedof the main to be roads. directly These affected changes by urban were expansionconsistent with at 500 those m radial that were distance, observed compared from with2004 9to monuments 2017. By 2025, in 2017.14 monuments Among 5 are newly expected affected to monuments,be directly affected 4 are most by urban likely expansion located in at the 500 core m zoneradial of distance, Angkor andcompared 1 in the with core 9 zonemonuments of Rolous. in By2017. 2030, Among 17 monuments 5 newly affected are likely monuments, to be affected 4 are by urbanmost likely expansion located at 500in the m radialcore zone distance, of Angkor compared and with1 in the 14 monumentscore zone of inRolous. 2025. By All 2030, of the 17 3monuments newly affected are likely monuments to be affected are expected by urban to beexpansion located inat the500 corem radial zone ofdistance, Rolous. Tablecompared4 gives with the list14 ofmonuments monuments in that 2025. are predictedAll of the to be3 newly a ffected affected in the nearmonuments future. are expected to be located in the core zone of Rolous. Table 4 gives the list of monuments that are predicted to be affected in the near future. Table 4. List of monuments predicted to be affected. Table 4. List of monuments predicted to be affected. Newly Affected Monuments at 500 m Radial Distance Time Point Newly affected monuments at 500 m radial distance Time point Angkor Rolous Angkor Rolous Prasat Ak Yum; ; 2025 Prasat Ak Yum; Prasat Preah Kou. 2 Prasat Baksei Cham Krong and .Phnom1 Bakheng; 2025 Prasat Preah Kou.2 Prasat Ou Kaek; Voat and 2030 – Prasat Baksei Cham Krong and Prasat Prei Monti. 3 Prasat Kravan.1 1 Monuments predicted to be affected in 2025 located in the core zone of Angkor compared with 2017. 2 Monuments predicted to be affected in 2025 located in the core zone of Rolous compared with 2017. 3 Monuments predicted to be affected in 2030 located in the core zone of Rolous compared with 2025. Prasat Ou Kaek; Voat Bakong 2030 -- and Prasat Prei Monti.3

Remote Sens. 2019, 11, x FOR PEER REVIEW 16 of 21 Remote Sens. 2019, 11, x FOR PEER REVIEW 16 of 21 1Monuments predicted to be affected in 2025 located in the core zone of Angkor compared with 2017. 1Monuments predicted to be affected in 2025 located in the core zone of Angkor compared with 2017. 2Monuments predicted to be affected in 2025 located in the core zone of Rolous compared with 2017. 2Monuments predicted to be affected in 2025 located in the core zone of Rolous compared with 2017. Remote3Monuments Sens. 2019 predicted, 11, 2064 to be affected in 2030 located in the core zone of Rolous compared with 2025. 16 of 21 3Monuments predicted to be affected in 2030 located in the core zone of Rolous compared with 2025.

Figure 14. Predicted built-up area in protected area in 2025.2025. Figure 14. Predicted built-up area in protected area in 2025.

Figure 15. Predicted built-up area in protected area in 2030. Figure 15. Predicted built-up area in protected area in 2030. Figure 15. Predicted built-up area in protected area in 2030. 5. Discussion 5. Discussion 5. DiscussionThe situation surrounding the Angkor world heritage site is a stark example of low-density urban sprawl. The growing numbers of tourists and development pressures threaten the conservation of Angkor. This paper quantified the expansion of built-up areas from 2004 to 2017, showing that urban Remote Sens. 2019, 11, 2064 17 of 21 expansion is not as prevalent in the northern core zone as in other directions in terms of area or intensity. This paper also explored the critical trends that are expected to manifest additional problems within the next few years due to urban development in Siem Reap. The study analyzed the process of urbanization at primarily two scales, one at a broader scale to map and understand past changes in the Angkor region, the second at a more local scale to evaluate the threat to individual monuments. As hypothesized, the threat of urbanization to Angkor is increasing. The study had four objectives: first, to understand and record the changes in built-up areas over four time periods in Siem Reap city, and to understand the rate of the process; second, to record changes in the land use dynamics, in particular for areas being converted for urban land use; third, to map encroachments of built-up areas near the individual sites; and lastly, to estimate future urban expansion and estimate likely threats. Regarding future changes in the Angkor region, the built-up areas are likely to expand in a linear fashion along major roadways, and are likely to cut right through the Angkor Archaeological Park. For 15 of the 48 monuments, the built-up area has been developed within 1000 m of the monuments which is concerning. Thus, the urban expansion pattern in Siem Reap is both linear and clustered. The urban expansion has profoundly impacted the sustainable development of Angkor and is expected to do so. It is understood that tourism is the main driver of this physical expansion. The fact that the need for tourism currently outweighs considerations for sustainable management to conserve the Angkor world heritage site makes solutions complicated. However, the government has introduced a series of legal measures to prohibit and restrict development and construction in planned protected areas [52], including the 1994 Zoning and Environmental Management Plan (ZEMP), the 2007 Angkor Management Plan, and the 2012 Tourism Management Plan. Various challenges complicate the process of conservation and sustainable development of Angkor [53,54]. The rate of urbanization has been shown to be gradually increasing. Importantly, our study reveals that urbanization has come at the cost of agricultural land, forests, and grasslands. Predominantly cultivable land has been converted, suggesting changes in the economics and population dynamics of the local region. To understand the underlying factors triggering this change from an agrarian community to an urban, an in-depth socio-economic analysis is required. However, what can be observed from our results at the broader scale is that the agricultural sector in the region and consequently employment is likely shifting to urban and tourism services. Urban expansion thus has affected heritage sites at multiple levels, posing a challenge to Angkor’s zoning management. The increasing populations in communities, regions, and countries require specific infrastructure and resources to sustain, and this, consequently, changes the natural and built environments. Therefore, this paper argues that the successful design and implementation of protection zones is crucial to protect heritage sites from future urban growth and environmental pressures. Maintaining the long-term integrity of Angkor poses a challenge to managers, who must strike a balance between development and preservation. It seems that this balance can often be achieved through the implementation of local land use regulations. Therefore, it is strongly recommended that ZEMP be empowered and applied in land use planning to constrain the future urban expansion of Siem Reap and to strengthen the planning and protection of the Angkor world heritage site. Protective actions need to be seriously considered. For example: Increase the size of the designated protective areas; monitor human-induced changes on the built and natural environments at regular intervals; increase local and international awareness for the protection and conservation of Angkor; and ensure compatible and sustainable development. Tourism planning and management should be an integral aspect of the site management system [55]. If this planning vacuum persists, there is a risk that future development may proceed without any coordination and might even cause serious damage to Angkor. Remote Sens. 2019, 11, 2064 18 of 21

6. Conclusions Gillespie (2012) conducted an in-depth analysis of the protected zones at the Angkor world heritage site, and concluded that the protection zone design was expected to result in a burden on local management due to the lack of research and reflection [56]. The results of this paper describe future challenges to the sites, thus providing key information crucial to protecting them from future urban growth and environmental pressures, towards a well-researched design and implementation of a protection zone management plan. The study builds a case for science-based policy and decision support. The study monitored and simulated urban expansion in the Angkor region. The results revealed that the urban area in Siem Reap increased from 28.23 km2 in 2004 to 73.56 km2 in 2017, a 160% increase. Urban growth was represented by a traffic-oriented pattern of urban sprawl. Siem Reap was revealed to be a typical city with tourism-driven urban expansion. Its urban expansion was highly consistent with the tourist population, but the change of the former lagged behind the latter. Urban expansion had the result of transforming arable land, forests, and grassland into urban residential land. Urban expansion was found to be present in the buffer zone at a radial distance of 500 m from heritage sites, and increased linearly from 0.85 km2 in 2004 to 2.93 km2 in 2017. The constructed land area in the core and buffer zones increased by 12.99 km2 between 2004 and 2017, and the land area in the buffer zone increased by 9.34 km2, accounting for 72% of the total increase. The residential construction-oriented urban expansion in the buffer zone mainly resulted in an increase in the amount of constructed land in the protected area. The forecast for urban expansion suggests that the built-up area in Siem Reap is expected to reach 135.09 km2 by 2025 and 159.14 km2 by 2030. The number of sites directly affected by urban expansion is expected to continue to increase. The core and buffer zones are still useful methods for protecting heritage sites, but they are also threatened by development-induced damage. These scientific findings of past, present, and future land-use scenarios in the Angkor region assist planners and decision-makers in formulating sustainable development and environmental protection plans. Using an integrated approach combining remote sensing and GIS to quantify the environmental forces acting on world heritage sites gives site managers the ability to calibrate conservation principles and procedures with actual urban dynamics. Based on these findings, it is clear that Angkor faces a challenge in providing a stable regulatory and sustainable development framework that can balance the inevitable tension between conservation of a much-valued heritage landscape with pressures to develop and accommodate an ever-expanding populace. The policy for managing urban development and protection of Angkor should be further improved. This paper provides insights for other heritage sites with similar challenges in developing countries.

Author Contributions: Conceptualization, J.L. and H.R.; methodology and software, H.R., J.L., and B.Q.; validation, J.L. and H.R.; formal analysis, J.L. and H.R.; investigation, J.L.; resources and data curation, J.L. and H.R.; writing—original draft preparation, J.L. and H.R.; writing—review and editing, Z.S. and X.W.; funding acquisition, X.W. Funding: This research was funded by the Strategic Priority Research Program of the Chinese Academy of Sciences (Grant No. XDA 19030504). Acknowledgments: We thank the APSARA Authority for their support and sharing of valuable data. Conflicts of Interest: The authors declare no conflicts of interest.

References

1. Grimm, N.B.; Faeth, S.H.; Golubiewski, N.E.; Redman, C.L.; Wu, J.; Bai, X.; Briggs, J.M. Global change and the ecology of cities. Science 2008, 319, 756–760. [CrossRef][PubMed] 2. Agapiou, A.; Alexakis, D.D.; Lysandrou, V.; Sarris, A.; Cuca, B.; Themistocleous, K.; Hadjimitsis, D.G. Impact of urban sprawl to cultural heritage monuments: The Case Study of Paphos Area in Cyprus. J. Cult. Herit. 2015, 16, 671–680. [CrossRef] Remote Sens. 2019, 11, 2064 19 of 21

3. Jaeger, J.A.G.; Schwick, C. Improving the measurement of urban sprawl: Weighted Urban Proliferation (WUP) and Its Application to Switzerland. Ecol. Indic. 2014, 38, 294–308. [CrossRef] 4. The World Bank. East Asia’s Changing Urban Landscape: Measuring a Decade of Spatial Growth; The World Bank: Washington, DC, USA, 2015. 5. Opschoor, H.; Tang, L.N. Growth, world heritage and sustainable development: The case of Lijiang City, China. Int. J. Sustain. Dev. World Ecol. 2011, 18, 469–473. [CrossRef] 6. Stubbs, J.H.; Mckee, K.L.R. Applications of Remote Sensing to the Understanding and Management of Cultural Heritage Sites. In Interdisciplinary Contributions to Archaeology; Springer: New York, NY, USA, 2006; pp. 515–540. 7. Seto, K.C.; Fragkias, M. Quantifying spatiotemporal patterns of urban land-use change in four cities of China with time series landscape metrics. Landsc. Ecol. 2005, 20, 871–888. [CrossRef] 8. Schneider, A.; Woodcock, C.E. Compact, dispersed, fragmented, extensive? A comparison of urban growth in twenty-five global cities using remotely sensed data, pattern metrics and census information. Urban. Stud. 2008, 45, 659–692. [CrossRef] 9. Ma, Y.; Xu, R. Remote sensing monitoring and driving force analysis ofurban expansion in Guangzhou City, China. Habitat Int. 2010, 34, 228–235. [CrossRef] 10. Fletcher, R.; Johnson, I.; Bruce, E.; Khun-Neay, K. Living with heritage: Site monitoring and heritage values in Greater Angkor and the Angkor World Heritage Site, Cambodia. World Archaeol. 2007, 39, 385–405. [CrossRef] 11. Zhao, N. Study on Conservation Planning of World Natural Heritage in the Urban Fringe. Master’s Thesis, Beijing University of Civil Engineering and Architecture, Beijing, China, 2013. 12. Miller, R.B.; Small, C. Cities from space: Potential applications of remote sensing in urban environmental research and policy. Environ. Sci. Policy 2003, 6, 129–137. [CrossRef] 13. APSARA. Angkor Heritage Management Framework. 2013. Available online: https://whc.unesco.org/en/list/ 668 (accessed on 12 April 2019). 14. Wager, J. Environmental planning for a World Heritage Site: Case study of Angkor, Cambodia. J. Environ. Plan. Manag. 1995, 38, 419–434. [CrossRef] 15. Gillespie, J. Protecting world heritage: Regulating ownership and land use at Angkor Archaeological Park, Cambodia. Int. J. Herit. Stud. 2009, 15, 338–354. [CrossRef] 16. Chandler, D. A ; Routledge: New York, NY, USA, 2000. 17. Gaughan, A.E.; Binford, M.W.; Southworth, J. Tourism, forest conversion, and land transformations in the Angkor basin, Cambodia. Appl. Geogr. 2009, 29, 212–223. [CrossRef] 18. Liao, K.T.; Wang, C.; Qi, S.H.; Khuon, K. Land Use and Land Cover Change in Angkor Heritage Site Based on Remote Sensing. Remote. Sens. Inf. 2015, 30, 119–124. 19. Hong, T.H.; Wang, M.; Huo, S.J. Remote Sensing Monitoring and Research of Angkor World Heritage Environment. China Cult. Herit. 2017, 3, 60–64. 20. Butland, R. Scaling Angkor: Perceptions of Scale in the Interpretation and Management of Cultural Heritage. Ph.D. Thesis, The University of Sydney, Sydney, , 2009. 21. Winter, T. Post-conflict Heritage and Tourism in Cambodia: The Burden of Angkor. Int. J. Herit. Stud. 2008, 14, 524–539. [CrossRef] 22. Mackay, R.; Palmer, S. The Ethics of Cultural Heritage; Springer: New York, NY, USA, 2015. 23. Liu, J.; Xu, Z.W.; Chen, F.L.; Chen, F.; Zhang, L. Flood Hazard Mapping and Assessment on the Angkor World Heritage Site, Cambodia. Remote Sens. 2019, 11, 98. [CrossRef] 24. International Coordinating Committee for the Safeguarding and Development of the Historic Site of Angkor (ICC-Angkor). Scientific Standing Secretariat Activity Report; International Coordinating Committee: Siem Reap, Cambodia, 2017. 25. APSARA; UNESCO. Angkor: Past, Present and Future; APSARA: Siem Reap, Cambodia, 1996. 26. UNESCO. Final Report of the Zoning and Environmental Management Plan for Angkor; UNESCO: Paris, , 1993. 27. Ao, H.X.; Huang, S.L. Investigation on Human and Natural Landscape of Angkor in Cambodia. Guangdong Landsc. Archit. 2016, 38, 92–95. 28. Royal Government of Cambodia. Synthesis Report on the Zoning and Environmental Management of Angkor within the Siem Reap Region; Royal Government of Cambodia (RGC): , Cambodia, 1994; pp. 71–73. Remote Sens. 2019, 11, 2064 20 of 21

29. Open Street Map Home Page. Available online: https://wiki.openstreetmap.org (accessed on 30 April 2019). 30. Jiang, G.; Zhang, F.; Kong, X. Determining conversion direction of rural residential land consolidation in Beijing mountainous areas. Nongye Gongcheng Xuebao/Trans. Chin. Soc. Agric. Eng. 2009, 25, 214–221. 31. Coppedge, B.R.; Engle, D.M.; Fuhlendorf, S.D. Markov models of land cover dynamics in a Southern Great Plains grassland region. Landsc. Ecol. 2007, 22, 1383–1393. [CrossRef] 32. Yang, X.; Zheng, X.-Q.; Chen, R. A land use change model: Integrating landscape pattern indexes and Markov-CA. Ecol. Model. 2014, 283, 1–7. [CrossRef] 33. Sang, L.L.; Chao, Z.; Yang, J.Y.; Zhu, D.H.; Yun, W.J. Simulation of land use spatial pattern of towns and villages based on CA–Markov model. Math. Comput. Model. 2011, 54, 938–943. [CrossRef] 34. White, R.; Engelen, G. Cellular automata and fractal urban form: A cellular modelling approach to the evolution of urban land-use patterns. Environ. Plan. A Econ. Space 1993, 25, 1175–1199. [CrossRef] 35. Adhikari, S.; Southworth, J. Simulating Forest Cover Changes of Bannerghatta National Park Based on a CA-Markov Model: A Remote Sensing Approach. Remote Sens. 2012, 4, 3215–3243. [CrossRef] 36. Nourqolipour, R.; Shariff, A.R.B.M.; Balasundram, S.K.; Ahmad, N.B.; Sood, A.M.; Buyong, T. Predicting the Effects of Urban Development on Land Transition and Spatial Patterns of Land Use in Western Peninsular . Appl. Spat. Anal. Policy 2016, 9, 1–19. [CrossRef] 37. Clarke, K.C.; Hoppen, S.; Gaydos, L. A Self-Modifying Cellular Automaton Model of Historical Urbanization in the San Francisco Bay Area. Environ. Plan. B 1997, 24, 247–261. [CrossRef] 38. Santé, I.; García, A.M.; Miranda, D.; Crecente, R. Cellular automata models for the simulation of real-world urban processes: A review and analysis. Landsc. Urban Plan. 2010, 96, 108–122. [CrossRef] 39. Ren, H.G.; Quan, B.; Wang, D.P.; Wang, P.; Hu, Z. Land Cover Change during 1990–2010 in Changsha and Its Prediction. Acta Agricuhurae Jiangxi 2017, 29, 93–99. 40. Rimal, B.; Zhang, L.F.; Keshtkar, H.; Haack, B.; Rijal, S.; Zhang, P. Land Use/Land Cover Dynamics and Modeling of Urban Land Expansion by the Integration of Cellular Automata and Markov Chain. ISPRS Int. J. Geo-Inf. 2018, 7, 154. [CrossRef] 41. Hong, S.J.; Zeng, Y.Q. Driving Factors of China’s Urban Spatial Growth in the Context of Institutional Change. Econ. Geogr. 2016, 36, 67–73. 42. Abdullah, A.; Lalit, K.; Khalid, A.G. Spatiotemporal Modeling of Urban Growth Predictions Based on Driving Force Factors in Five Saudi Arabian Cities. ISPRS Int. J. Geo-Inf. 2016, 5, 139–157. 43. Yin, X.F. Research on Simulation and Forecast of Construction Land-Use Change Based on CA_Markov Model. Master’s Thesis, Central China Normal University, Wuhan, China, 2014. 44. Eastman, J.R. IDRISI Kilimanjaro, Guide to GIS and Image Processing; Clark Lab, Clark University: Worcester, MA, USA, 2003. 45. Pontius, R.G., Jr.; Malanson, J. Comparison of the structure and accuracy of two land change models. Int. J. Geogr. Inf. Sci. 2005, 19, 745–748. [CrossRef] 46. Samat, N. Integrating GIS and CA-Markov model in evaluating urban spatial growth. Malays. J. Environ. Manag. 2009, 10, 83–99. 47. Li, C.; Tang, H.J.; Zhang, F.R. On Delimitation of Urban Growth Boundary Based on Construction Land Suitability Evaluation. J. Southwest China Norm. Univ. Chongqing 2017, 42, 105–111. 48. Li, D.H. Principles of Urban, Planning, 3rd ed.; China Architecture & Building Press: Beijing, China, 2001. 49. Keshtkar, H.; Voigt, W. A spatiotemporal analysis of landscape change using an integrated markov chain and cellular automata models. Model. Earth Syst. Environ. 2015, 2, 10. [CrossRef] 50. Araya, Y.H.; Cabral, P. Analysis and modeling of urban land cover change in setúbal and sesimbra, portugal. Remote Sens. 2010, 2, 1549–1563. [CrossRef] 51. Landis, J.R.; Koch, G.G. The measurement of observer agreement for categorical data. Biometrics 1977, 33, 159–174. [CrossRef][PubMed] 52. International Coordinating Committee for the Safeguarding and Development of the Historic Site of Angkor (ICC-Angkor). Fifteenth Technical Committee Report; UNESCO Office: Phnom Penh, Cambodia, 2006. 53. Save Cambodia’s Wildlife (SCW). Atlas of Cambodia: National Poverty and Environmental Maps; Save Cambodia’s Wildlife (SCW): Phnom Penh, Cambodia, 2006. 54. John, R.B.S. Democracy in Cambodia—One Decade, US$5 Billion Later: What Went Wrong? Contemp. Southeast Asia 2005, 27, 406–428. [CrossRef] Remote Sens. 2019, 11, 2064 21 of 21

55. APSARA. Angkor World Heritage Area Tourism Management Plan; APSARA: Siem Reap, Cambodia, 2012. 56. Gillespie, J. Buffering for conservation at Angkor: Questioning the spatial regulation of a World Heritage property. Int. J. Herit. Stud. 2012, 18, 194–208. [CrossRef]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).