| 113 Geoplanning Vol 7, No 2, 2020, 113-130 Journal of Geomatics and Planning E-ISSN: 2355-6544 http://ejournal.undip.ac.id/index.php/geoplanning doi: 10.14710/geoplanning.7.2.113-130 Application of Spatial Multi-Criteria Analysis and Least-Cost Path on The Highway Route Planning (Case Study of Bawen – Yogyakarta Highway) B. Marjuki a*, I. Rudiartob a Center of Data Processing – Ministry of Public Works and Housing, Indonesia b Urban and Regional Planning & Development Department – Diponegoro University, Indonesia Abstract: Infrastructure planning ideally considers the geotechnical aspects and physical conditions of the infrastructure development location and must be able to support regional development. One kind of spatial analysis technique, which has the capabilities to integrate various regional characteristics associated with its suitability for a particular use, is spatial multi-criteria analysis. By using Bawen - Yogyakarta Toll Road Plan as a case study, this research is intended to apply route planning that takes into account regional characteristics, through the involvement of Spatial Multi-Criteria Analysis, Analytic Hierarchy Process, and Least Cost Path analysis. The analysis results then compared with the government preferred route to see its advantages and disadvantages. Study results show that the generated route from the analysis has several advantages over the government preferred route while also having some shortcomings. The advantages of route analysis results compared to government preference routes include: better able to avoid earthquake and landslide-prone areas, better support to the preservation of protected areas, has more areas with flat to gentle topography, and have smaller additional construction cost as the consequences of the intersection with existing roads, rivers, and railways, In terms of affected land-use, generated route also has minimum negative impacts on the sustainability of agricultural land in the study area. The shortcomings of the analysis result are: not yet able to avoid flood and volcanic eruptions-prone areas as well as government’s preferences route, higher land acquisition cost estimation, and less support for industrial and tourism activities in the research area. Improvement of analysis methods, data, and cost assessment strategy is needed to obtain better results and more appropriate modeling and analysis, in order to support regional infrastructure planning and development. Copyright © 2020 GJGP-UNDIP This open access article is distributed under a Creative Commons Attribution (CC-BY-NC-SA) 4.0 International license. How to cite (APA 6th Style): Marjuki, B., & Rudiarto, I. (2020). Spatial Multi-Criteria Analysis and Least-Cost Path on The Highway Route Planning: A Case Study of Bawen – Yogyakarta Highway, Indonesia. Geoplanning: Journal of Geomatics and Planning, 7(2), 113-130. doi: 10.14710/geoplanning.7.2.113-130 1. INTRODUCTION Regional disparities between the western region and eastern region of Indonesia become the primary focus in the infrastructure development planning of the Indonesian Government. In this regard, the challenge of contemporary Indonesian infrastructure development is on how to reduce disparities and balance growth and development. Having to achieve this expectation, the Strategic Plan of the Ministry of Public Works and People's Housing 2015-2019 mandates that infrastructure development must integrate on the inter-regional, inter-sectoral, and inter-governmental levels. This mandate is realized in the form of the Strategic Development Area concept (SDA), which becomes the basis of infrastructure development in Indonesia. Ministry of Public Works divides the Indonesian territories into 35 SDA's, where every SDA has a specific planning policy according to the potential and problems that exist in each region. One of the planned SDA in Java Island is the Integrated Growth Center Yogyakarta-Solo-Semarang SDA (Strategic Plan Ministry of Public Works and Housing 2015-2019). Various infrastructure is planned to build in this SDA, where is one of them is Bawen – Yogyakarta Toll Road. Given that SDA-based infrastructure Article Info: Received: 7 April 2018 in revised form: Janury 2019 Accepted: January 2020 Available Online: 1 November 2020 Keywords: Spatial Multi-Criteria Analysis, Route Planning, AHP, Least Cost Path, Toll Road. *Corresponding Author: B. Marjuki Center of Data Processing – Ministry of Public Works and Housing, Jakarta, Indonesia Email: b_marjuki@pu.go.id OPEN ACCESS http://ejournal.undip.ac.id/index.php/geoplanning https://doi.org/10.14710/geoplanning.7.2.113-130 Marjuki & Rudiarto / Geoplanning: Journal of Geomatics and Planning, Vol 7, No 2, 2020, 113-130 doi: 10.14710/geoplanning.7.2.113-130 114 | development must be followed, the determination of the Bawen - Yogyakarta Toll Road route should take into account those integration aspects. Road construction should be integrated with regional development plans undertaken by provinces and districts traversed by the planned toll road. Also, toll road existence must also be able to support the existence of economic, social, and tourism activities in the region. The realization of integrated road planning requires formulation, analysis, and evaluation techniques that can involve these considerations. Multi-Criteria Evaluation is one of the analysis and evaluation techniques that can be used for that purpose. The role of Multi-Criteria evaluation in decision making has a long history, ranging from the traditional form (e.g., conventional mediation) to the form of modern automated programming through the help of computers and information technology (Köksalan, Wallenius, & Zionts, 2013). This development also includes spatial and regional planning through the development of Spatial Multi-Criteria Analysis (Malczewski, 2006). In spatial multi-criteria analysis (SMCA), regional indicators become the criteria used as the basis for determining the toll road route. Through analysis of the involved criteria, toll road alternative scenarios can be made, followed by an evaluation to determine which are the best scenarios, in the context of integration of infrastructure development and regional development. However, the factors and criteria involved in SMCA need to be formulated in advance of hierarchical urgency, since the importance of each factor is different from one to another. The urgency of each involved factor can be formulated using experts judgment techniques. Various techniques have been developed to determine the urgency of factors from the experts, but one of the most commonly used is the Analytic Hierarchy Process (AHP), developed by Wind & Saaty (1980). AHP can produce weight values and scores of involved factors and criteria that can be inputted in SMCA to find alternative solutions to the problems faced, followed by an evaluation to determine which one is the best. SMCA and AHP integration for site selection application has been conducted by many authors either in transportation route planning (Atkinson et al., 2005; Beukes, Vanderschuren, & Zuidgeest, 2011; Abdi et al., 2009; Keshkamat, Looijen, & Zuidgeest, 2009; Effat & Hassan, 2013) or other fields of study (Sánchez-Lozano & Bernal-Conesa, 2017; Mishra, Deep, & Choudhary, 2015; Bunruamkaew & Murayam, 2011). Thus, through a combination of SMCA and AHP, a toll road route plan that reflects not only the information integration of the involved criteria but also the experiences and preferences of the stakeholders involved in toll road infrastructure development can be obtained. By looking at the gap, this study combines SMCA with AHP as a tool to see an effective and efficient evacuation route based on Least Cost Path analysis. 2. DATA AND METHODS 2.1. Study Area The study is conducted at several subdistricts in Central Java Province and Yogyakarta Special Region, Indonesia, which likely will be traversed by Bawen - Yogyakarta Toll Road. These sub-districts are within the administrative area of Semarang Regency, Temanggung Regency, Magelang Regency, and Magelang City, which are part of the Central Java Province. As for Special Region of Yogyakarta, the study area covers some sub-districts of Sleman Regency and Kulonprogo Regency (Table and Figure 1). Marjuki & Rudiarto / Geoplanning: Journal of Geomatics and Planning, Vol 7, No 2, 2020, 113-130 doi: 10.14710/geoplanning.7.2.113-130 | 115 Table and Figure 1. Study Area Regency/District Sub-District Semarang Bergas Bawen Bandungan Jambu Banyubiru Tuntang Sumowono Ambarawa Pringapus Temanggung Kaloran Temanggung Kranggan Pringsurat Tembarak Selopampang Magelang Secang Grabag Windusari Kaliangkrik Bandongan Ngablak Ngluwar Tegalrejo Candimulyo Sawangan Tempuran Mertoyudan Mungkid Muntilan Pakis Borobudur Salam Sawangan Srumbung Dukun Magelang City North Magelang South Magelang Central Magelang Sleman Tempel Turi Sleman Pakem Ngaglik Mlati Minggir Seyegan Kulon Progo Kalibawang 2.2. Route Planning Criteria Criteria in SMCA is the basis for determining the evaluated object is to meet the requirements or not. Criteria, sub-criteria, and alternatives criteria for toll road route planning in this study are derived from a literature review of similar research (Atkinson et al., 2005; Effat & Hassan, 2013; Kushari, Mulyono, & Hendratno, 2015; and government regulations. The results of the criteria formulation are presented in Table 2. Table 2. Proposed Toll Road Route Planning Criteria (Analysis, 2018) Factors Criteria Sub Criteria Geotechnics Soil • Soil Texture Topography • Slope Geology • Rock Type Environmental Natural Protected Area • Natural Protected Areas in Existing Spatial Plan Cultural Protected Area • Distance from Archeological Cultural Heritage Social Land-use • Existing Land-use Land Value • Estimation of current Land Value Regional Activity System Regional Activity System • Distance from Industrial Areas • Distance from Urban Growth Centers • Distance from Tourism Destinations Safety and Additional Construction Cost Disaster Hazards • Landslide-Prone Areas • Flood-Prone Areas • Earthquake-Prone Areas • Volcanic Eruption Prone Areas Road Safety • Slope direction (aspect) Additional Construction Cost • Intersection with the road network • Intersection with the railway network • Intersection with the river network 2.3 Data Sources Various geospatial data has been collected and processed (i.e., vector data digitization, map projection conversion, vector to raster conversion) into one spatial database (geodatabase) to meet the minimum requirements for analysis. Data used in this study came from primary sources through field surveys and secondary data obtained from various agencies and institutions (Table 3). Marjuki & Rudiarto / Geoplanning: Journal of Geomatics and Planning, Vol 7, No 2, 2020, 113-130 doi: 10.14710/geoplanning.7.2.113-130 116 | Table 3. Data Sources (Analysis, 2018) Data Type Data Sources Type of Data Land-use Visual at Interpretation of 1.5 meters Orthorectified SPOT-6/7 Satellite Imagery acquired 2016 to 2017 (National Geospatial Agency – National Institute of Aeronautics and Space) Secondary Data Topography and Slope 8 meters IFSAR DEM generated from TERRASAR-X SAR Imagery (National Geospatial Agency – National Institute of Aeronautics and Space) Secondary Data Natural Protected Areas Existing Regency Spatial Plan (Regency Regional Planning Board) Secondary Data Cultural Protected Areas Temples and Archeological Site Mapping Resulted from Degroot & J Klokke (2010) study. Secondary Data Volcanic Eruption Hazard Volcanic Eruption Hazard Map from National Geological Agency Secondary Data Landslide Hazard Landslide Hazard Map from National Geological Agency Secondary Data Flood Hazard Flood Prone Map from Existing Regency Spatial Plan (Regency Regional Planning Board) Secondary Data Earthquake Hazard National Earthquake Hazard Map 2017 (Ministry of Public Works and Housing) Secondary Data Soil Texture 250 meters of global soil texture data from SOILGRIDS (ISRIC – World Soil Information) Secondary Data Land Value Land Value Map published in 2015 by the National Land Agency Secondary Data Rock Type Remote Sensing based Geologic Map at 1:50.000 Scale (National Geological Agency) Secondary Data Industrial, Tourism and Growth Center location Field Survey, Existing Regency Spatial Plan Primary and Secondary Data Road Network National, Provincial, and Regency Road Network Map published in 2015 by the Ministry of Public Works and Local Government. Secondary Data Railway Network National Topographic Map at Scale 1: 50.000 Published in 1915 by Nederland Indie Topographic Survey Agency, verified by field survey because the railway network in the study area has been closed at 1976, though there is a government plan to reactivate it in the future. Primary and Secondary Data River Network National Topographic Map at Scale 1:25.000, SPOT-6/7 Satellite Imagery Visual Interpretation Secondary Data Government’s preferred Route Plan of Bawen – Yogyakarta Toll Road Latest official Report of Bawen – Yogyakarta Toll Road Development (Ministry of Public Works and Housing) Secondary Data 2.4 Data Standardization and Normalization Determination of the optimal toll road route based on SMCA involves a series of criteria, sub-criteria, and alternatives that have a different scale and value measurement. These scale and value differences will affect the accuracy of the analysis results if it is not standardized into standard value (Drobne & Lisec, 2009). Therefore, before the analysis, any alternatives of sub-criteria were assigned and reclassified into a relative scoring scheme (from 1 to 9). The assumption used is, the higher the score of the alternative criteria, the more it is not suitable as a toll road route. Standardization parameters and references are presented in Table 4. Marjuki & Rudiarto / Geoplanning: Journal of Geomatics and Planning, Vol 7, No 2, 2020, 113-130 doi: 10.14710/geoplanning.7.2.113-130 | 117 Table 4. Cost Value Standardization Factor Criteria Sub-criteria Alternative Cost Value Cost Classification References Geotechnics Soil Soil Texture 0 -30% Sandy Soil 9 Natural Breaks Classification 30 -80% Sandy Soil 5 80 -100% Sandy Soil 1 Geology Rock Type Quartenary Deposits (alluvium, kolluvium), breccia 1 Expert Judgment Extrusive Igneous and Sediment Rock (sandstone, tuff) 5 Intrusive Igneous Rock (andesite, Granite, Dyorite) 9 Environment Natural Protected Area Natural Protected Area Type Non-Conservation Area 1 Modified from Kushari et al. (2015) and Expert Judgment Conservation Forest, National Park 9 Cultural Protected Area Distance from Temples 0 - 500 meters 9 Expert Judgment > 500 meters 1 Social Land-use Land-use Type Shrubs, Bare land, Grassland 1 Modified from Kushari et al. (2015)and Expert Judgment Mixed Garden, 3 Rice Fields, Poultry, Plantation 5 Sparse Settlements, Cemetery, 7 Industrial Area, Economic Services, Historical Places, Dense Settlements, Lake 9 Land Value Land Value 7,000 - 180,000 Rupiahs 1 Natural Breaks Classification 180,000 - 2,100,000 Rupiahs 5 > 2,100,000 Rupiahs 9 Regional Activity System Regional Activity System Distance from Industrial Area 0 - 5 kilometers 1 Equal Interval Classification 5 - 10 kilometers 5 > 10 kilometers 9 Distance from Urban Growth Centers 0 - 5 kilometers 1 5 - 10 kilometers 5 > 10 kilometers 9 Distance from Tourism Locations 0 - 5 kilometers 1 5 - 10 kilometers 5 > 10 kilometers 9 Road Safety and Additional Construction Cost Additional Construction Cost Intersection with Road Network Non Road Area 1 Expert Judgment Local Road 3 Regency Road 5 Provincial Road 7 National Road 9 Intersection with Railway Network Non-Railway Area 1 Railway Area 9 Intersection with River Network First River Order 1 Expert Judgment Second to Third River Order 3 Fourth River Order 5 Fifth River Order 7 Sixth to Seventh River Order 9 Natural Hazards Landslide Hazards Non Hazard Area 1 Construction Standards Number. 007/BM/2009, and Expert Judgment Low Hazard Area 3 Moderate Hazard Area 5 High Hazard Area 9 Flood Hazard Non Hazard Area 1 Low Hazard Area 3 Moderate Hazard Area 5 High Hazard Area 9 Earthquake Hazard PGA 0 - 0.25 g 1 PGA 0.25 - 0.3 g 3 PGA 0.3 - 0.4 g 5 Volcanic Eruption Hazard Non Hazard Area 1 Hazard Zone I 3 Hazard Zone II 5 Hazard Zone III 9 Marjuki & Rudiarto / Geoplanning: Journal of Geomatics and Planning, Vol 7, No 2, 2020, 113-130 doi: 10.14710/geoplanning.7.2.113-130 118 | 2.5 AHP Based Weighting Criteria and sub-criteria weighting in this study were conducted using the Analytic Hierarchy Process (AHP) method developed by Wind & Saaty (1980). In this study, the AHP analysis was conducted on two levels: first, AHP at the sub-criteria level (applied to sub-criteria incorporated in disaster risk criterion, regional activity system criterion, and additional construction cost criterion), and AHP at the criteria level (Figure 2). Topographic factors are not involved in AHP and SMCA because topographic factors will be used as a horizontal and vertical factor according to the modeling approach used, which is an anisotropic approach (to be described in subsequent chapters). Figure 2. Hierarchy of Criteria (Analysis, 2018) As input from AHP is the result of a questionnaire survey and interview with three experts from the Ministry of Public Works and Housing and one transportation practitioners. The result of the questionnaire was then arranged in the pairwise comparison matrix. From the obtained pairwise comparison matrix, the priority vector and its consistency index can be calculated. The value of the obtained priority vector is then used as the weight value of the criteria if the consistency index is less than 0.1. The final weight value from the obtained weight of each respondent then is calculated using the geometric mean formula. 2.6 Spatial Multi-criteria Analysis (SMCA) SMCA is implemented in Geographic Information System (GIS) through Map Algebra operation based on the raster data structure. SMCA requires the weight value of each criterion and sub-criteria to determine which criteria are more important to determine the most optimal toll road route. This weight value is derived from the AHP process described in the previous sub-chapter. The analytic method used to determine the best route of the toll road plan in this study is using the least-cost path (LCP) method. LCP is a technique to determine the shortest cost distance from one location to another based on a raster surface data called Accumulated Cost Surface (Douglas, 1994). Douglas (1994) defines Accumulated Cost Surface (ACS) as a dasymmetric representation in the form of a grid model of the Earth's surface, which refers to how much resources to spend or how much of the frictions must be passed over the model of the earth's surface. In this study, ACS is derived from the cumulative cost raster (called Cumulative Cost Surface/CCS) obtained from SMCA. CCS was obtained from the map overlay result of various baseline data (see Table 3) that have been Marjuki & Rudiarto / Geoplanning: Journal of Geomatics and Planning, Vol 7, No 2, 2020, 113-130 doi: 10.14710/geoplanning.7.2.113-130 | 119 given standardized score values and weight values in accordance with the AHP results. Several SMCA algorithms have been developed to obtain CCS. One of them is the Weighted Linear Combination (WLC) technique, which has been implemented in most of Geographic Information System software available on the market and used for this study. ACS calculation within the GIS can be based on two approaches, namely the isotropic approach and anisotropic approach (Yu, Lee, & Munro-Stasiuk, 2003). The Isotropic Approach is an approach in which the movement of calculation of cost accumulation in all directions is considered equal, while the anisotropic approach considering the movement of cost accumulation in all directions (both horizontal and vertical) has different implications to cost accumulation. In the context of road planning, some authors such as Collischonn & Pilar (2000) and Yu et al. (2003) argue that the anisotropic approach is more appropriate than the isotropic one. This opinion departs from the fact that less step and sinuous slopes are preferred than slopes that are short but steep. Either In isotropic or anisotropic approach, ACS calculations require CCS data obtained from WLC operation result. In the anisotropic approach, the difference is that in addition to requiring CCS information, anisotropic operations also require two other input data, which are a horizontal and vertical factor. The horizontal factor determines the horizontal friction, which will increase the cost of the horizontal movement direction (0 to 360 degrees), while the vertical factor determines the vertical friction, which will increase the cost in upward vertical movement (0 to 90 degrees), or downward vertical movement (0 to -90 degrees). Implementation of the horizontal and vertical factor to extracting the ACS using anisotropic approach can be done by determining the horizontal factor and vertical factor tables first. This horizontal and vertical factor table will be used as the basis to determine the cost value of the friction in horizontal movement (slope direction) and vertical movement (slope gradient). For this study, horizontal friction is considered in the analysis to facilitate the selection of road routes that avoid sunrise and sunset direction, so when the toll road is in the operational stage, the road users will not experience visibility problems due to glare disruption. Whereas, vertical friction is a representation of the increased cost due to terrain slope gradient changes, either during the road construction stage or road operational stage. The horizontal and vertical factor tables used in this study developed from the result of Yu et al. (2003) study and presented in Table 5. The INF value in the vertical factor table indicates that the cost to pass the area with a given slope is too large, so the area that has that kind of slope will be impassable. An overview of the performed analysis process is presented in Figure 3. Table 5. Vertical and Horizontal Factor Used for Anisotropic Route Extraction (Analysis, 2018) Vertical Factor Horizontal Factor Slope (Degrees) Cost Value Aspect (Degrees) Cost Value (-45) - (-90) INF 0 - 22.5 1 (-25) - (-45) 32 22.5 - 67.5 7 (-15) - (-25) 9 67.5 - 112.5 9 (-12) - (-15) 7 112.5 - 157.5 3 (-9) - (-12) 5 157.5 - 202.5 1 (-3) - (-9) 3 202.5 - 247.5 7 0 - (-3) 1 247.5 - 292.5 9 0 - 3 1 292.5 - 337.5 3 3 - 9 3 337.5 - 360 1 9 - 12 5 12 - 15 7 15 - 25 9 25 - 45 32 16 - 90 INF Marjuki & Rudiarto / Geoplanning: Journal of Geomatics and Planning, Vol 7, No 2, 2020, 113-130 doi: 10.14710/geoplanning.7.2.113-130 120 | Figure 3. Optimal Toll Road Route Analysis Workflow (Analysis, 2018) 2.7 Comparison and Evaluation of the Analysis Results Route comparison and evaluation in this study is done to find out to which extent the routes generated from SMCA and LCP is better than the government's preferred route. Comparison and evaluation are performed by looking to some indicators that affect: (1) ease of construction; (2) risk in the operational phase of the toll road; and (3) integration and support to regional development. Details of the comparison and evaluation indicators are presented in Table 6. Table 6. Evaluation Indicators (Analysis, 2018) Factors Criteria Comparison and Evaluation Indicators Ease of Construction and Level of Investment Cost Land-use - The extent of land-use type passed by the route Topography - The extent of the slope passed by the route Protected Area - The extent of the protected area passed by the route Land Value - Estimation of land acquisition cost along the route Existing Transportation Network - Number of the intersection with the road - Number of the intersection with railway Sungai - Number of the intersection with river Road Geometry - Route Length Risk in Operational Phase of Toll Road Road Safety - The extent of terrain aspect passed by the route Natural Hazards - The extent of Earthquake prone area passed by the route - The extent of volcanic eruption prone area passed by the route - The extent of Landslide prone area passed by the route - The extent of the flood-prone area passed by the route Integration with Regional Development Integration with Regional Development - Proximity to Industrial Areas - Proximity to Urban Growth Centers - Proximity to Tourism Destinations - Proximity to Archeological Protected Sites Marjuki & Rudiarto / Geoplanning: Journal of Geomatics and Planning, Vol 7, No 2, 2020, 113-130 doi: 10.14710/geoplanning.7.2.113-130 | 121 3. RESULTS AND DISCUSSION 3.1. Criteria Standardization and AHP Results Standardization of alternative criteria value in this study was conducted by reference to three sources of information, namely: (1) existing similar research and literature, (2) government rules and laws about road management, and (3) discussions with experts which be done simultaneously with AHP survey. For criteria mapped in discrete geographic objects (vector data), standardized cost value assignment is performed by converting object information (which is nominal data) to value (ratio data), in accordance with literature and expert recommendations, whereas cost value assignment for criteria mapped in continuous geographic fields is performed by classifying those values into standardized values. Complete visualization of the standardization result for each sub-criterion is presented in Figure 4. Figure 4. Standardization Results (Analysis, 2018) Marjuki & Rudiarto / Geoplanning: Journal of Geomatics and Planning, Vol 7, No 2, 2020, 113-130 doi: 10.14710/geoplanning.7.2.113-130 122 | The AHP processing in this study was conducted after the compilation of the questionnaire by experts has been completed. The AHP is performed to obtain the value of the eigenvector used as the weighting value of the criterion. The confidence level of the eigenvector value in AHP itself is measured from the Consistency Index (CI) obtained from the analysis. Based on the obtained result, the CI obtained for each respondent indicates a value below 0.1, which can be interpreted that the AHP analysis for each respondent is consistent. Thus the obtained eigenvector can be used as the criteria weight value. Eigenvalues (weights) calculation result of the criteria along with its consistency index value is presented in Table 7. Table 7. AHP Results (Analysis, 2018) Criteria Ministry of Public Works Transportation Expert Geometric Mean Weights Normalized Geometric Mean Expert 1 Expert 2 Expert 3 Expert 4 Geology 0.1949 0.1231 0.0481 0.0817 0.10 0.11 Soil 0.1839 0.1122 0.0350 0.0376 0.07 0.08 Natural Protected Area 0.1675 0.1496 0.1529 0.1596 0.16 0.18 Cultural Protected Area 0.1653 0.1415 0.1529 0.1596 0.15 0.17 Land-use 0.1341 0.1065 0.1529 0.1298 0.13 0.14 Land Value 0.0191 0.0445 0.0171 0.0210 0.02 0.03 Regional Activity System 0.0228 0.0201 0.1529 0.1386 0.06 0.06 Natural Hazards 0.0691 0.2787 0.1529 0.2523 0.17 0.18 Additional Construction Cost 0.0434 0.0239 0.1354 0.0199 0.04 0.05 Total Weights 1 1 1 1 0.90 1.00 Consistency Index 0.07 0.08 0.09 0.07 From the results of the AHP analysis described above, the final criteria and sub-criteria weights can be determined. In this study, the final weights computed from the normalized weights resulted from the geometric mean operation. Before inclusion as input for CCS generation, the decimal format's final weight values are converted to a percentage by multiplying it with 100 value. Conversion to percentage is applied to avoid the reduction effect of the final cost value (due to the multiplication operation of cost value with weight value). As for the final weight value of sub-criteria is determined from the division of the concerned criteria, in accordance with the proportion of the weight value of each sub-criterion. The final weight value of criteria and sub-criteria that has been converted to the percentage scale is presented in Table 8. Table 8. Criteria and Sub-criteria Final Weights (Analysis, 2018) Criteria Sub Criteria Criteria Weights Sub-Criteria Weights Geology 11 Soil 8 Natural Protected Area 18 Cultural Protected Area 17 Land-use 14 Land Value 3 Natural Hazards 18 Volcanic Eruption 5.61 Landslide 6.80 Earthquake 4.53 Flood 1.45 Regional Activity System 6 Distance from Urban Growth Center 2.21 Distance from Industrial Area 3.49 Distance from Tourism Destination 0.52 Additional Construction Cost 5 Intersection with Railway Network 1.23 Intersection with Road Network 1.85 Intersection with River Network 1.47 Marjuki & Rudiarto / Geoplanning: Journal of Geomatics and Planning, Vol 7, No 2, 2020, 113-130 doi: 10.14710/geoplanning.7.2.113-130 | 123 The result of consensus drawing from several experts using AHP and geometric mean in this study favored criteria of natural hazards, protected area existence, land use, and geological condition. Thus, the LCP analysis will be more sensitive to the above criteria than other criteria. This sensitivity is apparent after comparison with the government’s preferred route, where the LCP generated route cannot identify feasible and effective routes when viewed from regional activity system criterion and the land value criterion. The route generated from the LCP analysis ideally should be sensitive to the regional activity system criterion because the Bawen - Yogyakarta Toll Road is expected to support regional development and tourism activities in the study area. LCP generated route ideally should also be sensitive to the variation of land value over the study area because it will be related to the role of minimizing the complexity of land acquisition as one of the leading problems of toll road development in Indonesia. The relative value assignment strategy of cost (cost score) also plays a vital role in addition to the weighting strategy of the criteria itself. This statement can be understood from the results of the route comparison on the additional construction cost criterion. The weight value of the additional construction cost criterion is relatively smaller than the other criteria, but the route obtained from the LCP analysis generated intersection number with rivers and roads less than the government's preferred route. Thus, the cost value assignment strategy for the additional construction cost criterion is sufficient. The level details of the data used are one of the aspects that also need to be considered further. For example, this study uses the Land Use Map at Scale 1: 10,000, where is in this data, the built-up land-use is still visualized as area (e.g., residential area or industrial area). This data may not be detail and useful enough as the input in SMCA and LCP application for route planning. As can be seen from the analysis results, the route generated from LCP analysis has not shown adequate sensitivity compared to the government’s preferred route for the residential area, although the criterion weight value is quite high. This insensitivity happens because the score assignment of the residential area has been done based on its relative density, in accordance with the level of the details of those data. If detailed information that can be used as a basis for determining cost score in a more precise way is available (such as the number of buildings per block), cost models and LCP analyzes may be more sensitive to the variation in the cost value of each criterion, and then can have significant implication to the improvement of the analysis and modeling. The use of anisotropic approach (ArcGIS path distance algorithm) in this research is capable of generating a toll road route that has the cost as low as possible. As shown by the comparison results of topographic aspects in Table 8, the route of the LCP analysis can find areas with the topography as flat as possible, to minimize the cost when compared with the government's preferred route. Nevertheless, the results obtained are still can be debated further, especially in the analysis results around Bawen Sub District, Ambarawa City, and Rawa Pening Lake Area. The anisotropic approach, which considers slope gradient as a vertical cost factor, tends to favor flat topography with the slope of a small slope as the area with the best suitability for toll road routes. However, in the case of Bawen - Yogyakarta toll road, the government study route has chosen to avoid Rawa Pening Lake area, although topographically, the Rawa Pening Lake area is entirely appropriate. The government may want to avoid the complexity of construction costs associated with the soil engineering treatment in the Rawa Pening Lake area, which tends to be soft and has low engineering capacity to support road infrastructure. This fact has not been well anticipated by the model generated from this study since the soil criterion in this study is based only on the percentage of the sandy texture of the soil and has not considered other soil characteristics (due to lack of data), such as effective soil depth. If we refer to similar studies, among others by Ismail & Jusoff (2009), Beukes et al. (2011), and Chandio et al. (2012), these studies found that multi-criteria spatial analysis and LCP can provide effective and efficient route analysis results in terms of distance and time spent to travel, as well as minimal risk in terms of construction and operational cost. However, the criteria involved in those studies were not as much as the criteria used in this study. Consideration of more criteria will make the cost model have more complicated behavior, and there is a possibility to produce route analysis results that do not meet some of the criteria under consideration (as can be seen from the results of this study), although it is also there is the possibility that complex model will better represent the complexity of the cost conditions in the field. Marjuki & Rudiarto / Geoplanning: Journal of Geomatics and Planning, Vol 7, No 2, 2020, 113-130 doi: 10.14710/geoplanning.7.2.113-130 124 | The results of this study indicate the complexity of those mentioned complex behavior of the cost model above, which results in the non-fulfillment of some of the criteria. This finding confirms similar studies using relatively numerous criteria, as did by Keshkamat et al. (2009) and Kushari et al. (2015). Both studies also confirm the dynamic LCP analysis results as a result of involving more criteria in the preparation of the cost model, which resulted in some evaluation indicators are not fulfillment. Apart from the disadvantages of the analysis results, the modeling methods applied in this study along with the results that have been obtained indicate the potential that the analytical methods and techniques that have been applied can be utilized and developed further to support the transportation route planning, that can be more sensitive to various criteria, not only physical criteria but also to the sustainability of environmental functions, besides supporting the regional development. 3.2. Criteria Standardization and AHP Results CCS model generated based on input: (1) result of standardization and classification of criteria and sub- criteria; and (2 weight value of each criterion and sub-criterion. Model design and execution is performed within the ArcGIS Desktop 10.6 software using weighted sum geoprocessing operations. The weighted sum operations are carried out at two levels. First, at the sub-criteria level for disaster risk criteria, regional activity system criteria and additional construction cost criteria, and second, at criteria level that integrates the cost values of the geology, soil, protected areas, land use, land values, regional activity system, natural hazards, and additional construction costs criteria. Figure 5. CCS and ACS Results (Analysis, 2018) Marjuki & Rudiarto / Geoplanning: Journal of Geomatics and Planning, Vol 7, No 2, 2020, 113-130 doi: 10.14710/geoplanning.7.2.113-130 | 125 The ACS model is constructed using input data consisting of: (1) initial location of road plan; (2) DEM of research area; (3) CCS model that has been produced; (4) horizontal factor table; and (5) vertical factor table. Processes performed within the ArcGIS path distance algorithm (presented in Figure 5 along with CCS result), which can be described as follows: 1. Based on the initial location of the route plan and specified DEM, the algorithm calculates the surface distance from the initial location to all research areas. 2. Aspect is derived from the specified DEM, followed by a cost value assignment for each mapped direction angle in every pixel with reference to the horizontal factor table. This sub-dataset is called horizontal cost. 3. Slope is derived from the specified DEM, followed by a cost value assignment for each mapped slope angle in every pixel with reference to the vertical factor table. This sub-dataset is called vertical cost. 4. CCS Cost Model is then multiplied by the horizontal cost. 5. Through the iteration process, the accumulated sum of the cost of each pixel from the origin pixel then is summed, and the summation result of every two pixels is multiplied by surface distance value and vertical cost value between the evaluated pixels. Figure 6. LCP Routing Result (red line) Compared to Government’s Preferred Route (Green Line) Marjuki & Rudiarto / Geoplanning: Journal of Geomatics and Planning, Vol 7, No 2, 2020, 113-130 doi: 10.14710/geoplanning.7.2.113-130 126 | Least cost path (LCP) analysis is an optimal path determination analysis technique based on pixel-pixel searches across the study area, using criteria: (1) the shortest distance from the origin location to the destination location; along with (2) the lowest accumulation of cost. LCP requires input data consisting of (1) the result of the ACS model, (2) backlink data obtained from ACS generation, and (3) the destination point, which is the end location of the path to be extracted. The backlink data is a raster data indicating the direction of movement of the ACS data (quantified by the pixel value of backlink data ranging from 1 to 8 to represent the eight directions of possible horizontal movement) from the origin point to the destination point. The result of LCP analysis that has been done is presented in Figure 6. 3.3. Comparison and Evaluation Results Obtained Toll Road Route from LCP analysis then compared and evaluated versus government's preferred route. To properly evaluate the affected areas, the generated route from LCP analysis and the government's preferred route has been buffered 80 meters either on the right side or to the left side of the route. Buffering of 80 meters is assumed to be sufficient to cover the necessary road space (own road space, road benefit space, and road monitoring space). The results of the performed evaluation are presented in Table 9. Table 9. Comparison and Evaluation Results (Analysis, 2018) Criteria Sub Criteria Alternative Criteria Route Generated from LCP Government’s Preferred Route Land Use Land-use Traversed by Route Settlements 166. 59 Ha 128.1 Ha Industry and Services 3.6 Ha 2.4 Ha Government Facilities 0.2 Ha 0 Ha Forests 0 Ha 0 Ha Plantations 16 Ha 35.4 Ha Agricultural Areas 694.9 Ha 751. 2 Ha Mixed Gardens 216. 2 Ha 298.1 Ha Bare Lands 0.7 Ha 3.2 Ha Slope Route Slope 0 - 3 % 114. 41 Ha 95. 24 Ha 3 - 9 % 514. 28 Ha 386. 1 Ha 9 - 25 % 449. 21 Ha 620. 9 Ha > 25 % 31. 76 Ha 123. 3 Ha Protected Area Protected Areas Traversed by Route Natural Protected Areas 0 Ha 0 Ha Cultural Protected Areas (300-meter radius from Temples) 111.3 Ha 122 Ha Land Value Land Acquisition Estimation Land Value 4,188,176,199,328 IDR 3,581,766,068,528 IDR Intersection with Existing Transportation Network Intersection with National. Provincial and Regencies Road National Roads 6 Intersections 5 Intersections Provincial Roads 4 Intersections 5 Intersections Regencies Roads 27 Intersections 30 Intersections Local Roads 102 Intersections 112 Intersections Intersection with Railway Railway 4 Intersections 2 Intersections First River Order 45 Intersections 46 Intersections Marjuki & Rudiarto / Geoplanning: Journal of Geomatics and Planning, Vol 7, No 2, 2020, 113-130 doi: 10.14710/geoplanning.7.2.113-130 | 127 Criteria Sub Criteria Alternative Criteria Route Generated from LCP Government’s Preferred Route Intersection With River Network Intersection with River Second to Third River Order 47 Intersections 49 Intersections Fourth River Order 12 Intersections 9 Intersections Fifth River Order 0 Intersection 8 Intersections Sixth to Seventh River Order 0 Intersection 0 Intersection Road Geometry Route Length Total Route Length 69 Km + 408 m 76 Km + 720 m Risk on Toll Road Operational Phase Route Direction North. South 30.45 km 24. 36 km North West. South East 13.35 km 17. 52 km South West. North East 19.39 km 26. 37 km West. East 6.22 km 8.46 km Volcanic Eruption Hazard Hazard Zone 1 0 Ha 0 Ha Hazard Zone II 0 Ha 0 Ha Hazard Zone III 14.99 Ha 8.5 Ha Landslide Hazard Hazard Zone 1 702.12 Ha 835 Ha Hazard Zone II 89.27 Ha 179.9 Ha Hazard Zone III 18.67 Ha 20.8 Ha Flood Hazard Flood Prone Area 12.9 Ha 0 Ha Earthquake Hazard Peak Ground Acceleration 0.25g -0.3g 63.9 Ha 79.7 Ha Peak Ground Acceleration 0.3g - 0.4g 945.1 Ha 1034.6 Ha Peak Ground Acceleration > 0.4g 100.1 Ha 111.2 Ha Integration with Regional Development Accessibility from Industrial Area Proximity to Industrial Areas 73 Industry Locations 113 Industry Locations Accessibility from Urban Growth Centers Proximity to Urban Growth Centers 10 Growth Centers (Ambarawa. Grabag. Temanggung. Kranggan. Secang. Muntilan. Salam. Tempel. Sleman. Pakem) 8 Growth Centers (Temanggung. Kranggan. Secang. Kota Magelang. Mertoyudan. Mungkid. Borobudur. Dekso) Accessibility from Tourism Destinations Proximity to Tourism Destinations 129 Tourism Destinations 142 Tourism Destinations (includes Borobudur Temple) Accessibility from Cagar Budaya Proximity to Temples and Archeological Sites 57 Archeological Sites 68 Archeological Sites Note: Cells marked with green shade indicate the superior result Development and implementation of transportation infrastructure. either in central or local governments can consider the results obtained from this study to conduct further evaluation of the Bawen Marjuki & Rudiarto / Geoplanning: Journal of Geomatics and Planning, Vol 7, No 2, 2020, 113-130 doi: 10.14710/geoplanning.7.2.113-130 128 | Toll Road - Yogyakarta route plan. as well as adopting the method to similar infrastructure projects. Route obtained from the study can also be integrated with government studies on road segments that are perceived as having not met the disaster threat criteria. environmental criteria. support for food security. and construction costs criteria. The method of determining the trace used in this study. which includes SMCA. AHP. and anisotropic LCP analysis. is recommended to be incorporated in government's transportation planning activities. especially at the initial planning stage of the feasibility study. In addition to being able to provide route analysis through an automated process along with tools and evaluation techniques. this method can be extended to scenario- based route modeling. so various alternative scenarios can be proposed and further evaluated. which are most appropriate and feasible. Formalization of methods can be done in the form of legislation. technical guidance. or Standard of Procedures (SOP). A comparison between isotropic and anisotropic approaches in determining the optimal route also needs to be studied and simulated further in future studies. This assessment is needed to ascertain which approach is the best in transportation infrastructure route planning. Assessments can be conducted in different regions with different regional characteristics to see how each approach performs. The testing of different LCP algorithms also needs to be applied. LCP analysis used in this research is using LCP Djikstra algorithm (Dijkstra. 1959) with a queen pattern (eight directions of movement). On the other hand. various LCP algorithms have been developed. for example. A* (Hart. Nilsson. & Raphael. 1968). Best First Search. A* Manhattan Heuristics. A* Diagonal Shortcut Heuristics. or LCP algorithms dedicated explicitly to road-based route planning such as SmartTerrain (Yu et al.. 2003) and Baek's Cut and Fill (Baek & Choi. 2017). The above LCP approaches and algorithms can be an alternative to determine transportation routes that may generate better results. Knight movement patterns (24 directions of movement) that have been implemented in the GRASS GIS software can also be studied further along with those various algorithms. The development of a cost model based on actual cost is also recommended. Incorporation of actual cost into the cost model will have a strategic value. among others is: (1) projecting the real cost-benefit condition of the implementation of infrastructure development; (2) can be valuable information in preparing infrastructure development budget plans; (3) represents the condition of the cost that is closer to the actual conditions in the field. Real cost information for each criterion can be simulated using current cost standards. or based on experience/budget realization reports from toll road or other infrastructure projects that have been implemented. 4. CONCLUSION Based on the obtained results. we can conclude that the utilization of an anisotropic approach in the toll road route planning can give the result of route analysis. which can minimize the cost. The cost model is getting more complex along with the addition of more criteria and may negate one criterion with other criteria. resulting in the insensitivity of route analysis to specific criteria. particularly on criteria that have low weight and/or criteria that use generalized data. Nevertheless. these findings are not fixed because there is a possibility that the cost model can be more sensitive to low weight criteria. as long as it is applied a proper cost assessment strategy and/or represents the real cost condition in the field. The existence of criteria that have real cost values such as land value (which used in this study) can be useful to estimate the project budgets and costs. which is essential in the planning stage of infrastructure development programs. in order to obtain appropriate planning products.The obtained route from this study is also shorter about 8 kilometers compared to government's preferred route. allowing road users to save a certain amount of resources when the toll road has been operational. These findings indicate that there are alternative routes to government's preferred route. which is environmentally friendly and helpful to support food security sustainability. This study still has some limitations related to the implementation of SMCA. LCP. and AHP analysis to support the route planning of the Bawen - Yogyakarta Toll Road. There are several criteria for toll road planning that still use a relatively subjective value assessment and have not been involved in research. both Marjuki & Rudiarto / Geoplanning: Journal of Geomatics and Planning, Vol 7, No 2, 2020, 113-130 doi: 10.14710/geoplanning.7.2.113-130 | 129 in terms of road planning regulations. toll road planning feasibility standards. and conditions and complexity of problems in the field. This research is carried out on a regional scale and uses publicly available data with different quality in terms of scale and updates. 4. ACKNOWLEDGMENTS The authors would like to acknowledge the Ministry of Public Works and Housing of Republic Indonesia who funded the research under the Karyasiswa Kedinasan Scholarship Program. 5. REFERENCES Abdi. E.. Majnounian. B.. Darvishsefat. A.. Mashayekhi. Z.. & Sessions. J. (2009). A GIS-MCE based model for forest road planning. Journal of Forest Science. 55(4). 171–176. [Crossref] Atkinson. D. M.. Deadman. P.. Dudycha. D.. & Traynor. S. (2005). Multi-criteria evaluation and least cost path analysis for an arctic all-weather road. Applied Geography. 25(4). 287–307. [Crossref] Baek. J.. & Choi. Y. (2017). A new algorithm to find raster-based least-cost paths using cut and fill operations. International Journal of Geographical Information Science. 31(11). 2234–2254. [Crossref] Beukes. E. A.. Vanderschuren. M.. & Zuidgeest. M. H. P. (2011). Context sensitive multimodal road planning: a case study in Cape Town. South Africa. Journal of Transport Geography. 19(3). 452–460. [Crossref] Bunruamkaew. K.. & Murayam. Y. (2011). Site suitability evaluation for ecotourism using GIS & AHP: A case study of Surat Thani province. Thailand. Procedia-Social and Behavioral Sciences. 21. 269–278. [Crossref] Chandio. I. A.. Matori. A. N. B.. WanYusof. K. B.. Talpur. M. A. H.. Khahro. S. H.. & Mokhtar. M. R. M. (2012). Computer application in routing of road using least-cost path analysis in hillside development. Research Journal of Environmental and Earth Sciences. 4(10). 907–911. Collischonn. W.. & Pilar. J. V. (2000). A direction dependent least-cost-path algorithm for roads and canals. International Journal of Geographical Information Science. 14(4). 397–406. [Crossref] Degroot. V.. & J Klokke. M. (2010). Interrelationships among Central Javanese temples: the example of Asu. Lumbung and Pendem. Archipel. 80(1). 45–75. [Crossref] Dijkstra. E. W. (1959). A note on two problems in connexion with graphs. Numerische Mathematik. 1(1). 269–271. [Crossref] Douglas. D. H. (1994). Least-cost path in GIS using an accumulated cost surface and slopelines. Cartographica: The International Journal for Geographic Information and Geovisualization. 31(3). 37–51. [Crossref] Drobne. S.. & Lisec. A. (2009). Multi-attribute decision analysis in GIS: weighted linear combination and ordered weighted averaging. Informatica. 33(4). Effat. H. A.. & Hassan. O. A. (2013). Designing and evaluation of three alternatives highway routes using the Analytical Hierarchy Process and the least-cost path analysis. application in Sinai Peninsula. Egypt. The Egyptian Journal of Remote Sensing and Space Science. 16(2). 141–151. [Crossref] Hart. P. E.. Nilsson. N. J.. & Raphael. B. (1968). A formal basis for the heuristic determination of minimum cost paths. IEEE Transactions on Systems Science and Cybernetics. 4(2). 100–107. [Crossref] Ismail. M. H.. & Jusoff. H. K. (2009). Planning of access road using satellite technology and best path modeling. Modern Applied Science. 3(3). 83–95. [Crossref] Keshkamat. S. S.. Looijen. J. M.. & Zuidgeest. M. H. P. (2009). The formulation and evaluation of transport route planning alternatives: a spatial decision support system for the Via Baltica project. Poland. Journal of Transport Geography. 17(1). 54–64. [Crossref] Köksalan. M.. Wallenius. J.. & Zionts. S. (2013). An early history of multiple criteria decision making. Journal of Multi-Criteria Decision Analysis. 20(1–2). 87–94. [Crossref] Kushari. B.. Mulyono. A. T.. & Hendratno. A. (2015). Least-cost Path and Multi-criteria Decision Analyses on the Sumatera High-Grade Highway Route Plans. Malczewski. J. (2006). GIS-based multicriteria decision analysis: a survey of the literature. International Journal of Geographical Information Science. 20(7). 703–726. [Crossref] Mishra. A. K.. Deep. S.. & Choudhary. A. (2015). Identification of suitable sites for organic farming using AHP https://doi.org/10.17221/52/2008-jfs https://doi.org/10.1016/j.apgeog.2005.08.001 https://doi.org/10.1080/13658816.2017.1356463 https://doi.org/10.1016/j.jtrangeo.2010.08.014 https://doi.org/10.1016/j.sbspro.2011.07.024 https://doi.org/10.1080/13658810050024304 https://doi.org/10.3406/arch.2010.4176 https://doi.org/10.1007/bf01386390 https://doi.org/10.3138/d327-0323-2jut-016m https://doi.org/10.1016/j.ejrs.2013.08.001 https://doi.org/10.1109/tssc.1968.300136 https://doi.org/10.5539/mas.v3n3p83 https://doi.org/10.1016/j.jtrangeo.2008.04.010 https://doi.org/10.1002/mcda.1481 https://doi.org/10.1080/13658810600661508 Marjuki & Rudiarto / Geoplanning: Journal of Geomatics and Planning, Vol 7, No 2, 2020, 113-130 doi: 10.14710/geoplanning.7.2.113-130 130 | & GIS. The Egyptian Journal of Remote Sensing and Space Science. 18(2). 181–193. [Crossref] Sánchez-Lozano. J. M.. & Bernal-Conesa. J. A. (2017). Environmental management of Natura 2000 network areas through the combination of Geographic Information Systems (GIS) with Multi-Criteria Decision Making (MCDM) methods. Case study in south-eastern Spain. Land Use Policy. 63. 86–97. [Crossref] Wind. Y.. & Saaty. T. L. (1980). Marketing applications of the analytic hierarchy process. Management Science. 26(7). 641–658. [Crossref] Yu. C.. Lee. J. A. Y.. & Munro-Stasiuk. M. J. (2003). Extensions to least-cost path algorithms for roadway planning. International Journal of Geographical Information Science. 17(4). 361–376. https://doi.org/10.1016/j.ejrs.2015.06.005 https://doi.org/10.1016/j.landusepol.2017.01.021 https://doi.org/10.1287/mnsc.26.7.641