Land-use planning: Implications for transport sustainability

Land Use Policy 50 (2016) 252–261

Contents lists available at ScienceDirect

Land Use Policy journal homepage: www.elsevier.com/locate/landusepol

Land-use planning: Implications for transport sustainability Marzieh Reisi, Lu Aye ∗ , Abbas Rajabifard, Tuan Ngo Renewable Energy and Energy Efficiency Group, Department of Infrastructure Engineering, The University of Melbourne, VIC 3010, Australia

a r t i c l e

i n f o

Article history: Received 10 September 2014 Received in revised form 9 September 2015 Accepted 20 September 2015 Keywords: Transport sustainability Indicators Scenarios Land-use planning

a b s t r a c t This research study explores three urban planning scenarios for Melbourne, Australia in 2030 and their implications for transport sustainability. As part of the analyses, a transport sustainability index, derived from 10 sustainability indicators, was developed and applied to compare the scenarios. A base-case scenario, an activity-centres scenario, and a fringe-focus scenario were used to consider compact to expanded urban development patterns. The activity-centres scenario, which favours compact development patterns, had the highest transport sustainability index. In contrast, the fringe-focus scenario that significantly expands urban development in the fringe resulted in a lower transport sustainability index. The results of scenario analysis would influence decisions regarding urban development in 2030. Crown Copyright © 2015 Published by Elsevier Ltd. All rights reserved.

1. Introduction The transport sector is a major user of fossil energy, which causes air pollution and also contributes to global warming (DobranskyteNiskota et al., 2007). Negative environmental and social impacts of transport impose large costs on society. It is estimated that air pollution, noise and accident related costs are at least 5% of GDP for industrialised countries (Verhoef et al., 2001). Hence, the transport sector with its significant environmental, social and economic impacts, is an important element of urban sustainability (Haghshenas and Vaziri, 2012). The 1987 Brundtland report (from the United Nations World Commission on Environment and Development) defined sustainable transport as “transport that meets the current transport and mobility needs without compromising the ability of future generations to meet these needs” (Balack, 2004). In recent years, many studies have worked on planning for sustainable transport systems. Due to uncertainty regarding the effects of particular policies on urban transport systems, considering their effects on transport sustainability is a challenging task for policy-makers (Shiftan et al., 2003). There is a high level of interaction between transport activities and land-use planning decisions. For example, land-use patterns affect access to facilities and destinations, and consequently affect the amount and methods of travel. Therefore, understanding these

∗ Corresponding author. Fax: +61 393063389. E-mail addresses: [email protected] (M. Reisi), [email protected] (L. Aye), [email protected] (A. Rajabifard), [email protected] (T. Ngo). http://dx.doi.org/10.1016/j.landusepol.2015.09.018 0264-8377/Crown Copyright © 2015 Published by Elsevier Ltd. All rights reserved.

interactions is important for urban planning (Litman, 2014b). Over the past 20 years, benefits of increased density in urban areas have been considered for regional planning. Bartholomew (2007) reviewed 80 urban planning scenarios in more than 50 US metropolitan areas and found that an 11% increase in density decreases the vehicle miles travelled (VMT) and NOx emissions by 2.3% and 2.1%, respectively. Similarly, a study of future urban development scenarios for 11 metropolitan areas in the Midwestern US concluded that a 10% increase in density would reduce VMT and transport-related air emissions by 3.5% (Stone et al., 2007). In another study, Jakimaviˇcius and Burinskiene (2009) assessed three urban development scenarios (compact development, extensive development, and decentralised concentrated development) in terms of fuel consumption, distance travelled, and driving time in a car during morning peak hours. The results indicated that compact and mixed land-use reduce fuel consumption. Despite scenario analysis on the effect of land-use on distance travelled and related emissions, there is a lack of studies that consider the effects of urban planning scenarios on transport sustainability for Melbourne. This paper applied an indicator-based approach for sustainable transport modelling. The study considered the impacts of three urban planning scenarios on transport sustainability for Melbourne using various sustainable transport indicators. It provides insights into the future developments of the transport sector in 2030 resulting from the three urban planning strategies. First, a list of transport sustainability indicators was identified. In the next step, indicators were quantified using a landuse/transport interaction model, which were then normalised and integrated into a single index. In the final step, a transport sustainability index was quantified for three urban planning scenarios for Melbourne in 2030: a base-case scenario, an activity-centres sce-

M. Reisi et al. / Land Use Policy 50 (2016) 252–261

nario, and a fringe-focus scenario. Application of these scenarios is useful in assessing how the implementation of different policy measures may change transport sustainability in Melbourne. 2. Method In the first step of this study, an indicator-based approach was undertaken to select and quantify transport sustainability indicators in environmental, social and economic dimensions. In the next step, selected indicators were normalised, weighted and integrated into a transportation environmental impact index (TEII), a transportation social impact index (TSII), and a transportation economic impact index (TCII). In the final step, selected indicators and consequently the environmental, social and economic indices were predicted for three urban planning scenarios in 2030. The details of each step are described in this section. 2.1. Indicator selection The transport sector has a range of negative impacts. Therefore, to achieve a sustainable transport system, it is essential to assess the transport sector using indicators in decision-making when implementing solutions to these impacts. During the last two decades, measuring sustainability by indicators has been widely used by the scientific community and policy-makers (Dobranskyte-Niskota et al., 2007). Indicators as quantitative measures can be applied to simply illustrate complex phenomena (EEA, 2005). Indicator choice is a trade-off between available data sources and selection criteria (comprehensiveness, relevance, data availability, measurability, transparency) (Dur et al., 2010; Haghshenas and Vaziri, 2012; Li et al., 2009; Spiekermann and Wegener, 2004; Zito and Salvo, 2011). It should be balanced, reflecting a combination of environmental, social and economic objectives (Litman, 2005). A list of transport sustainability indicators is presented in Table 1. The selected spatial scale for this study [SLA (statistical local area) level] limited the number of indicators that can be quantified. As the objective of sustainable transport depends on the context (Castillo and Pitfield, 2010), the selected set of indicators represented Melbourne’s trend towards transport sustainability and covered environmental, social and economic dimensions. Indicators such as vehicle kilometres travelled (VKT), passenger kilometres travelled (PKT), length of railways and main roads, and proportion of residents with public transit services within 500 m, are intermediate indicators rather than final indicators. Hence, they were used to quantify the final indicators such as depletion of non-renewable resources, emissions and accessibility. It is noted that considering air pollutants and mortality effects of air pollutants in two separate indicators may risk double counting the effect. However, along with harming human health, air pollutants can cause a variety of environmental effects such as acid rain, eutrophication, and ozone depletion. Therefore, considering air pollutants as well as the mortality effects of air pollutants, attempts to consider all these impacts. 2.2. Indicator quantification Statistical local areas (SLAs) in Melbourne were selected as the study areas in this paper. Using the 2006 Australian Bureau of Statistics (ABS) database and the 2007 Victorian Integrated Survey of Travel and Activity (VISTA07) conducted by DOT (2007), the selected indicators were quantified, which is presented in the following sections in detail. 2.2.1. Transportation environmental indicators Selected environmental indicators were quantified as follows:

253

• Depletion of non-renewable resources: the inclusion of depletion of non-renewable resources is justified by the definition of a sustainable transport system as a system that minimises nonrenewable resources consumption. In this study, the amount of primary fuel (crude oil) consumed is a measure for resource depletion. To calculate the primary fuel consumed, first transport energy consumption was estimated by multiplying VKT for private transport and PKT for public transport by energy factors estimated by Rickwood (2009). To convert MJ of energy consumption in transport to litres of primary fuel, the petroleum refinery efficiency must be applied. By knowing transport energy consumption and petroleum refinery efficiency in Melbourne (90%) (Australian Government, 2010 BREE, 2010), the amount of primary fuel consumed was calculated. • Transport emissions: according to the definition of a sustainable transport system as a system that produces emissions only within the planet’s ability to absorb them, selection of greenhouse gas (GHG) emissions as one of the transport environmental indicators is justifiable. Emissions of pollutants other than GHGs from transport into the air are major sources of poor air quality. The justification for including them is the same as including GHG emissions. In this study, GHG emissions were estimated by multiplying VKT for private transport and PKT for public transport by emission factors estimated by Rickwood (2009). CO, PM10 , and NO2 were also quantified using emissions factors presented by the Australian National Pollutant Inventory (NPi, 2008). Hence, estimates of total VKT and percentage of trips by cars (modal split) in each SLA are required. According to the literature, spatial and land-use planning, economic, social, and behavioural factors are the main factors influencing transport development (Corpuz et al., 2006; Dargay and Hanly, 2003; Giuliano and Dargay, 2006; Lindsey et al., 2011; Næss, 2009; Shiftan et al., 2003). Distance from the CBD, area of SLAs, access to public transport, walkability, proportion of couples with children to other households, household annual income, and car ownership, were selected for estimating VKT; while population density, area of SLAs, access to public transport, walkability, proportion of couples with children to other households, household annual income, and car ownership, were selected for estimating modal split (Corpuz et al., 2006; Dargay and Hanly, 2003; Giuliano and Dargay, 2006; Haque et al., 2013; Kitamura et al., 1997; Kobos et al., 2003; Lindsey et al., 2011; Litman, 2012; Miller and Ibrahim, 1998; Newman and Kenworthy, 1991; Paravantis and Georgakellos, 2007; Pongthanaisawan and Sorapipatana, 2010; Soltani and Somenahalli, 2005; Whelan et al., 2010; Zhang et al., 2012). VKT and percentage of trips by cars (modal split) were quantified as functions of the above selected land-use and socio-economic factors, using an artificial neural network (ANN) model. A feed-forward neural network model is the most popular type of ANN in transport modelling studies. “This network consists of an input layer, an output layer, and one or more hidden layers between the input and output. The number of nodes in the input layer is usually identical to the number of independent variables, while the number of nodes in the output layer is the same as the number of dependent variables” (Limanond et al., 2011). A neural network model is applied to map between a set of inputs and a set of outputs, and provides appropriate responses. The process of adopting an appropriate network to the data is known as neural network training (Limanond et al., 2011). The ANN was implemented for Melbourne data in MATLAB. VKT and modal split were introduced to the MATLAB neural network fitting tool as outputs, and selected socio-economic and land-use factors were introduced as inputs. Seventy per cent of data was used for training, 15% for validation, and 15% for testing, which are the default settings in MATLAB. When training data is fed as inputs to the model, the neu-

254

M. Reisi et al. / Land Use Policy 50 (2016) 252–261

Table 1 Sustainable transport indicators selected for Melbourne. Selected indicators for the study

Unit

Transportation environmental indicators Depletion of non-renewable resources Greenhouse gas (GHG) emissions (CO2-e ) Other air pollutants (CO, NO2 , PM10 ) Land consumption for transport

L per household annually kg per household annually kg per household annually km2 per household

Transportation social indicators Accessibility Fatalities and injuries related to traffic accidents Mortality effects of air pollutants

Score between 0 and 1 persons per household annually persons per household annually

Transportation economic indicators Car ownership costs Vehicle and general costs of accidentsBenefits of active transport

$ per household annually $ per household annually$ per household annually

ral network assigns appropriate weights to each link in the network and calculates the outputs. Validation data is not directly used for training; it is used for validation purposes only. Testing data has no effect on training and provides an independent measure of network performance during and after training (Edara, 2003). The number of hidden neurons were set at 10, which is the default setting in MATLAB. Mean Square Error (MSE), R2 , and mean bias error (MBE) measures were applied to evaluate suitability of the ANN for modelling VKT and modal split. MSE is the average squared difference between the observed and estimated variables, while R2 measures the correlation between the observed and estimated variables, and MBE is the mean of the difference between the observed and estimated variables. As the ANN model provided low MSE (0.06 for the VKT model and 0.002 for the modal split model), low MBE (0.08 for the VKT model and −0.0006 for the modal split model), and high R2 value (0.75 for the VKT model and 0.93 for the modal split model) in this study, it can be concluded that the ANN can be applied for modelling VKT and modal split as functions of the socio- economic and land-use factors. As there is no prediction available for VKT and modal split in 2030, the relationship between these parameters and socio-economic and land-use factors found by the ANN can be applied to predict them as function of 2030s socio-economic and land-use factors, which were predicted by the government. Prediction of the socio-economic and land-use factors is provided in Section 2.4. It is worth noting that GHG emissions from transport are correlated with fossil fuel use and consequently depletion of nonrenewable resources. The close matching of these two indicators raises the question of why both have been chosen. Both have been chosen because of separate important impacts in relation to resources depletion and global warming. • Land consumption for transport: Melbourne’s land-use map contains information about lands devoted to roads. The Calculate Geometry tool in ArcGIS 10 was applied to estimate the area of lands devoted to roads in Melbourne. 2.2.2. Transportation social indicators • Accessibility: an equitable transport system must provide a fair distribution of transport services and equal access to facilities. Therefore, assessing accessibility can show equity issues and transport disadvantages (Pitot et al., 2006). Several types of accessibility measures have been used in transport planning studies, such as gravity measures, distance measures, cumulative opportunity measures, and utility-based measures (Apparicio and Seguin, 2006; Lotfi and Koohsari, 2009; Makri and Folkesson, 1999; Zhang et al., 2011). There is no best approach for accessibility measurement because different situations and purposes need different approaches (Geurs and Wee, 2004).

There are some issues with the accessibility indices previously developed for Melbourne. Firstly, in some accessibility indices found in the literature (DPIE and DHSH, 1997; Faulkner and French, 1983), straight-line distance between origins and destinations were measured, which does not capture all aspects of accessibility. Secondly, in some other literature (GISCA, 2011), public transport was considered solely as a service to be accessed, and not as a means of potential access. Despite using the concept of Pitot et al. (2006), the accessibility index developed in this study is the first of its kind developed for Melbourne. As the objective of the accessibility index in this study is to describe the proximity of SLAs to a series of facilities, distance measures and cumulative opportunity measures were applied to quantify accessibility by walking and public transport (trams, trains and buses). To quantify accessibility by walking, the Network Analyst extension in ArcGIS 10 was applied to find the shortest network distance between origins (centres of SLAs) and destinations (parks, education facilities, health services, business zones, and public transport stations). Then, a fuzzy linear function was applied to weight the SLAs for the level of accessibility according to the measured distance. The linear function classifies the measured distance using a scale from 0 to 1, based on the possibility of being a member of a specified set (i.e. appropriate distance to facilities). For those locations that are not accessible, a value of 0 is given, while 1 is allocated to those locations that are accessible; and values between 0 and 1 are assigned to some levels of accessibility (the closer the value to 1, the greater the accessibility) (ESRI, 1999). To quantify accessibility to public transport, public transit stops (called PTS X) on the road network that are within a specific walking distance from each destination were selected. Then, public transit stops (called PTS Y) on public transit routes that are within a given travel time from PTS X were selected as well. Following this, the shortest network distance on the road network between PTS Y and the SLA centres was found using the ArcGIS 10 Network Analyst extension. At the end, each SLA was weighted for the level of accessibility to the nearest PTS Y according to the calculated distance using the fuzzy linear function, in the same way as done for the walkability index. All accessibility measures to different destinations by different modes of public transport (bus, tram and train) were averaged into a final accessibility index by public transport for the SLA. It is noted that appropriate travel distances to facilities were extracted from Pitot et al. (2006) . In the final step, the walkability index and the accessibility index by public transport were aggregated into a single index (using the average of walkability and accessibility by public transport), which shows the overall accessibility for each SLA. • Fatalities and injuries related to traffic accidents: a sustainable transport system should find a way to minimise the number of traffic accidents, the severity and risks (Castillo and Pitfield,

M. Reisi et al. / Land Use Policy 50 (2016) 252–261

2010). The number of deaths and injuries due to accidents in Australia are 0.8 and 14.8 per 100 million VKT, respectively (ATSB, 2007). As there are no available published data for crash fatalities and injuries specific to Melbourne, the Australian aggregated data was applied in this study. • Mortality effects of air pollutants: according to the results of epidemiological studies, transport- related air pollutants cause adverse health effects. Moreover, air pollution is strongly associated with daily mortality in Melbourne (EPA, 2000). Transport-related air pollutants are a mixture of different substances, and there are high correlations between different pollutants in the mixture. Therefore, the effects of individual pollutants cannot be considered separately (EEA, 2005). Generally in epidemiological studies, one pollutant (usually PM10 ) is considered as an indicator of a complex mixture. However, there is clear evidence that some pollutants (such as ozone) are poorly correlated with PM10 and they might have independent health effects (Kunzli et al., 1999; MacDonald Gibson et al., 2013). Moreover, according to the State of the Environment Report (2001), particulate and ozone concentrations are not predicted to have a clear reduction in the future in Melbourne (Coffey, 2003). So in this study, PM10 and ozone were used to estimate mortality effects of air pollutants. Relative risk of death values used for PM10 and ozone are 1.0009 and 1.0023, respectively (EPA, 2000). Once the relative risks have been determined, the impact fraction can be calculated using Eq. (1) (Ostro, 2004): AF =

RR − 1 RR

(1)

where: AF = impact fraction of the health effects for the exposed population; RR = relative risk of the air pollutant. AF is the attributable risk of air pollutants to the population, which is expressed as a percentage of the total risk to the population. For example, if AF is 0.09% for PM10 , this means that 0.09% of air pollution-related deaths in the population can be attributed to PM10 . To estimate the expected number of deaths due to air pollution (E), the AF is applied to the total population (Ostro, 2004): E = AF

B P

(2)

where: E = expected number of deaths due to air pollution; B = population incident of mortality (i.e., deaths per 1000 people); P = relevant exposed population to the effect. 2.2.3. Transportation economic indicators • Car ownership costs: car ownership costs in Melbourne were estimated to be 72.18% per km in 2006, including both standing costs (depreciation, interest on loan, registration, licence, Royal Automobile Club of Victoria (RACV) membership, other on-road costs) and running costs (fuel, tyres, service/repairs) (RACV, 2012). • Costs of accident fatalities and injuries: estimated human losses due to accidents are approximately AU$2.4 million per fatality and AU$214,000 per injury. Moreover, vehicle costs and general costs related to accidents are 49.02% and 29.64% of human costs, respectively (BITRE, 2009). The number of fatalities and injuries related to transport was included as a social indicator in this study. Therefore, to avoid double counting, human costs were not included in this indicator. Vehicle and general costs related to transport accidents were only quantified in this indicator. • Benefits of active transport: although walking and cycling have many benefits (congestion savings, road provision savings, vehicle operating cost savings, savings related to public transport operating costs, external parking savings, road safety, environmental pollution savings, noise reduction, health cost savings (Fishman et al., 2011; PWC, 2011)), to have a balance with the

255

costs considered in this study, benefits that are in accordance with the considered costs were quantified here as: car ownership costs savings, health costs savings, and environmental pollution savings. The savings were quantified as follows: - Car ownership costs savings: multiplying the avoided VKT due to walking and cycling by car ownership costs, it was estimated that car ownership costs savings were 138% per household in 2006. - Savings in accident-related costs: accidents costs would be saved by $0.093 per km by walking and cycling (PWC, 2011). - Savings in costs of mortalities related to air pollutants: this factor is quantified by multiplying the avoided number of deaths due to emission reduction, by the value of the life, which is AU$1.3 million according to the Bureau of Infrastructure, Transport and Regional Economics (BTRE, 2005). To develop a transport sustainability index, after quantification, indicators must go through the normalisation and weighting process, which is described in the following section. 2.3. Index development Before integrating indicators into a single index, each indicator has been normalised using IN = I − Imin /Imax − Imin , where Imin and Imax are the minimum and maximum values of the indicators. The normalised indicators (IN ) have values between 0 and 1 (Maoh and Kanaroglou, 2008). The weights given to different indicators before their aggregation influence the outcomes of the composite indices, so a transparent method must be applied for weighting indicators (Freudenberg, 2003; Juwana, 2012). There are three weighting methods: equal weighting, weighting based on expert judgement, and weighting based on statistical models (Saisana, 2011). Expert judgment is usually applied to derive the weights that are subjective and differ from one expert to another (Bojkovic et al., 2010; Da Silve et al., 2008; Rasafi and Zarabadipour, 2009). When such information is not available, an equal weighting method may be applied. However, each indicator has its own characteristics. With the equal weighting approach, there is a risk that certain aspects will be double counted. In other words, combining variables with high correlations could introduce double counting into the final index (Freudenberg, 2003; OECD, 2008; Zhou et al., 2007). Based on the limitations of expert judgment and equal weighting, there is a need for a statistical weighting method. Principal component analysis/factor analysis (PCA/FA) was used in this study to overcome the problems in other weighting methods. PCA/FA has been used in the composite indices’ building processes to basically group the indicators and define the weights. In this method, weightings are only applied to correct the overlapping information of two or more correlated indicators (Fernando et al., 2012). PCA/FA clusters individual indicators (most of which are correlated to each other) to create a composite index that captures as much of the variance of the individual indicators as possible. Each factor, extracted using principal components analysis, shows a set of indicators with which it has the strongest relationship (DeCoster, 1998). The first step in the PCA/FA is to check the correlation between indicators. The second step is to select a certain number of factors (fewer than the number of individual indicators) using the PCA for factor extraction. Each factor depends on a set of coefficients (loadings), measuring the correlation between the individual indicator and the latent factor. The number of extracted factors is assigned based on different principals. Factors chosen are the ones that: (i) have eigenvalues larger than one; (ii) contribute individually to the explanation of overall variance by more than

256

M. Reisi et al. / Land Use Policy 50 (2016) 252–261

Table 2 Transport sustainability indicators in 2006 and 2030 under the different scenarios.

Depletion of non-renewable resources (L) GHG emission (kg) Other emissions (kg) Land consumption for transport (km2 ) Accessibility (−) Fatalities and injuriesrelated to accidents (person) air pollutants (person) Mortality effects of Car ownership costs ($) Vehicle and general costs of accidents ($) Benefits of walking and cycling ($)

2006

Base-casescenario

Activity-centres scenario

Fringe-focusscenario

237,561 474,425 9,122 0.020 0.34 0.2723 0.005 1,312,289 56,972 45,157

214,330 429,638 8,070 0.030 0.48 0.2496 0.01070 915,635 24,905 68,587

213,887 428,745 8,054 0.026 0.61 0.2491 0.01068 913,759 24,856 213,224

214,773 430,607 8,079 0.038 0.32 0.2499 0.01073 917,444 24,936 66,928

10%; and (iii) contribute cumulatively to the explanation of the overall variance by more than 60% (OECD, 2008; Saisana, 2011). The third step involves rotation of the factors. The rotation is used to minimise the number of indicators that have a high loading on the same factor. The idea of rotation is to obtain a simpler structure of the factors, which is ideally a structure in which each indicator is loaded exclusively on one of the factors. The last step relates to construction of the weights from the factor loadings after rotation, using the square of factor loadings which represent the proportion of the total unit variance of the indicator explained by the factor, and the eigenvalue which represents the sum of the squared of factor loadings (DeCoster, 1998). Details of the PCA/FA are described in Kline (1994). SPSS software was applied to carry out PCA/FA in this study. The variables that were set for the PCA/FA in SPSS were sustainability indicators in each SLA for the whole Melbourne area. After factor extraction, the indicators’ weights were extracted using Eq. (3) (Gomez-Limon and Riesgo, 2008):

 wkj =

factor loadingkj

2 (3)

eigen valuej

where: w = weight of indicator; j = number of principal components or factors; k = number of indicators; eigen value = sum of squares of factor loadings. Once the weights are calculated, the intermediate sustainability indicators (ISIj ) need to be calculated, corresponding to each of the principal components j (Gomez-Limon and Riesgo, 2008):

ISIj =

k=n 

wkj Ik

(4)

k=1

where: ISI = intermediate sustainability indicator; w = weight of indicator; I = normalised indicator; j = number of factors; k = number of indicators.

Finally, the sustainability index was calculated as a weighted aggregation of the intermediate sustainability indicators (GomezLimon and Riesgo, 2008): I=

m 

˛j ISIj

(5)

j=1

where: I = sustainability index; j = number of principal components; ISI = intermediate sustainability indicator.˛ = Weight of intermediate sustainability indicator =

eigen valuej j=m  eigen valuej j=1

Since the formula includes the correlation coefficient between factors and each of the indicators, the weights will be specific for each indicator. Hence, PCA/FA overcomes the weakness of assigning a common weight to all indicators (Fernando et al., 2012). Multiplying the indicators by their weights and adding them up, results in the transport sustainability indice (Gomez-Limon and Riesgo, 2008). It is worth noting that PCA/FA was conducted for environmental, social and economic indicators separately to calculate the weights for the indicators and consequently extract the TEII, TSII, and TCII. Then, the PCA/FA was repeated to calculate the weights for each sub-index and extract the final sustainability index. 2.4. Scenario development This section presents three urban planning scenarios for 2030. Any projection of the future is based on assumptions on the trends of transport sustainability indicators, which depends on uncertain demographic and economic forces. Scenarios provide visions of the future based on a specific framework and adopting specific assumptions. They do not provide planning solutions, but give a clear idea about possible planning solutions to be used in the future (Banister, 2000). Scenarios based on Melbourne 2030 planning policies ranged between compact and expansive urban development patterns. The

Table 3 Weights of indicators based on PCA/FA. 2006

Depletion of non-renewable resources (L) GHG emission (kg) Other emissions (kg) Land consumption for transport (km2 ) Accessibility Fatalities and injuriesrelated to accidents (person) air pollutants (person) Mortality effects of Car ownership costs ($) Vehicle and general costs of accidents ($) Benefits of walking and cycling ($)

Base-case scenario

Activity-centres scenario

Fringe-focus scenario

Factor 1

Factor 2

Factor 1

Factor 2

Factor 1

Factor 2

Factor 1

Factor 2

0.254 0.253 0.255 0.033 0.438 0.430 0.015 0.408 0.404 0.024

0.227 0.229 0.208 1.726 0.024 0.037 1.221 0.078 0.088 1.388

0.290 0.286 0.284 0.017 0.482 0.471 0.005 0.496 0.493 0.001

0.082 0.084 0.069 1.314 0.002 0.022 1.063 0.0004 0.006 1.014

0.291 0.287 0.286 0.017 0.473 0.468 0.006 0.472 0.471 0.006

0.079 0.083 0.062 1.303 0.008 0.018 1.087 0.013 0.015 1.088

0.291 0.286 0.285 0.017 0.517 0.459 0.001 0.493 0.489 0.002

0.080 0.084 0.063 1.307 0.027 0.085 0.922 0.001 0.008 1.025

M. Reisi et al. / Land Use Policy 50 (2016) 252–261

base-case scenario is based on projections provided in Melbourne 2030 (DTPLI, 2002), Victoria in Future (DTPLI, 2012), and Melbourne, Let’s talk about the future (State Government Victoria, 2012). The activity-centres and fringe-focus scenarios encourage urban development in the defined activity centres and urban fringe, respectively. The performance of each scenario with respect to transport sustainability was measured through selected environmental, social and economic indicators, as well as a transport sustainability index. The scenarios were defined as follows: • Scenario 1: base-case scenario The base-case scenario for 2030 is based on government plans for Melbourne in 2030 and represents a plausible future scenario against which results for other scenarios were compared. As mentioned before, to quantify some indicators, we need to quantify VKT and modal split. Hence, it is needed to predict these values for 2030. Based on our knowledge, there are no predictions available for VKT and modal split in 2030. However, the government has provided some predictions for the land-use and socio-economic factors considered in this study. So, the relationship between VKT and modal split, and the land-use and socio-economic factors found by the ANN were applied to predict VKT and modal split as functions of 2030 land-use and socio-economic factors. MATLAB is able to develop an ANN model that predicts the future values of the dependent variables (VKT and modal split) based on the relationship between current dependent and independent variables (socio-economic and land-use factors). Methods of predicting the land-use and socio-economic factors are provided as follows. Before describing the prediction process, it is worth noting that a time series forecast was applied to predict some variables in the future, using SPSS software which gives the ability to choose the best forecasting model for each time series (SPSS Inc., 2008). - Population density in 2030: the Department of Transport, Planning and Local Infrastructure (DTPLI, 2012) uses the Australian Bureau of Statistics (ABS) current published statistics on births, deaths, and migration to project the population in 2030. - DTPLI (2012) projected household type in 2030 by maintaining living arrangement probabilities (by age and sex) as of the 2006 census, and applied these to the future population. This allows only the size and age of the population to influence household formation.

257

- There is no publicly available forecast data for household income in 2030. In the absence of such projections, household annual income was projected using the household income trend between 2004 and 2007 for Melbourne’s SLAs using the ABS census data. - The DTPLI used a combination of trend analysis and ongoing consultation with local authorities to determine the most likely locations of future dwelling constructions. - In predicting public transport accessibility in 2030, the City of Melbourne (2011) referred to a study by Scheurer (2010). Scheurer developed a scenario for 2030 in Melbourne, which is a combination of different improvements in the transport system and includes providing 10-minute services (or better) on all tram routes, doubling all tracked rail lines and a range of bus routes, speeding up public transport operations by a mix of traffic priority measures, and operational and design improvements. Based on this scenario, the composite accessibility index would change from 14.5 to 19.8. The results of the 2030 scenario in Scheurer (2010) were used to predict the nearest distance to public transport and accessibility by public transport in 2030 in this study. Changes in network coverage (the proportion of all residents and jobs located within walking distance from public transport services) were used as changes to the nearest distance to public transport in this study, since reduction in the nearest distance to public transport increases the proportion of all residents and jobs located within walking distance from public transport services. Changes in the composite index were also used as changes to the accessibility index by public transport in this study. - The Victorian Government proposed a ‘20-minute city’ model for Melbourne walkability in the future (State Government Victoria, 2012). Based on this model, 95% of Melbourne residents would live within one km walking catchment of basic day-to-day services. To estimate walkability in 2030, distances from each SLA to business, education, parks, and health facilities will improve to one km, and then walkability is estimated using the method described before. Considering changes in walkability and access by public transport based on the above section, the accessibility index was quantified for 2030 using the process described. - There is no information available about land consumption for transport in the future. To predict this indicator for 2030, it was assumed that the road area per capita is constant over time. So by considering population growth in 2030, this indicator was estimated for 2030. - The Australian Transport Safety Bureau (ATSB, 2007) has provided road crash casualties between 1925 and 2005. Considering the

Fig. 1. TEII, TSII, TCII (base-case scenario, activity-centres scenario, fringe-focus scenario).

258

M. Reisi et al. / Land Use Policy 50 (2016) 252–261

changes between these years, there would be 0.28 deaths and 15.84 serious injuries per 100 million vehicle kilometres travelled in 2030. - Based on Eq. (2), the population in 2030 and the number of deaths in 2030 are needed to estimate the expected number of deaths due to air pollutants. As mentioned before, population density in 2030 was estimated based on the population projected by the DTPLI. The City of Melbourne (2013) estimated the death rate based on historical data for Melbourne published by the ABS, and then extrapolated it into the future. Using the same method, the ABS census information about the number of deaths in Melbourne between 2006 and 2011 were used in this study to estimate the number of deaths in Melbourne in 2030 and the number of deaths related to air pollutants. - The RACV provides vehicle standing and running costs for 2010–2013. Assuming the same growth trend as the past, average car ownership cost in Melbourne is estimated to be 56.70 cents per km in 2030. - The average cost of a fatal crash was $1.7 million and $2.67 million in 1996 and 2006, respectively. The average cost of a serious injury crash was $408,000 and $266,000 in 1996 and 2006, respectively (BITRE, 2009; BTE, 2000). Using the same annual growth rate, the average cost of a fatal crash will be $10.11 million and the average cost of a serious injury crash will be $113,672 in 2030. On the other hand, the average cost of a fatality was $1.5 million and $2.4 million in 1996 and 2006, respectively. The average cost of a serious injury was $325,000 and $214,000 in 1996 and 2006, respectively (BITRE, 2009; BTE, 2000). Using the same annual growth rate, the average cost of a fatality will be $9.71 million and the average cost of a serious injury will be $92,940 in 2030. Vehicle and general costs were 78.30% and 62.55% of human costs in 1996 and 2006, respectively (BTE, 2000). Assuming the same trend of changes, vehicle and general costs will be 38.37% of human costs in 2030. - Using walking and cycling as modes of transport results in savings in car ownership costs, as well as savings in costs related to accidents and air pollutants. Savings related to car ownership costs and accident costs were quantified using the methods presented in the above sections. To quantify costs of mortalities related to air pollutants, using costs of air pollution-related mortality in 2000 and 2006 ($1.3 million and $1.56 million, respectively) (BTRE, 2005) and assuming the same growth rate in the future, the costs of air pollution-related mortality was predicted to be $3.01 million in 2030. • Scenario 2: activity-centres scenario One of the directions of the Melbourne 2030 plan is building a more compact city, which encourages the concentration of new development in activity centres near current infrastructures. Activity centres will be built up as a focus for high-quality development, activity and living for the whole community. A substantial proportion of new housing in or close to activity centres and other strategic redevelopment sites will offer good access to services and transport. The activity centres are, or will be, well-served by public transport, and they offer a wide range of services and facilities benefiting the whole community (DTPLI, 2002). The activity-centre scenario assumes that the provisions from government for 2030 (i.e., Melbourne 2030 plan; Melbourne, let’s talk about the future; Victoria in Future) will be exceeded in the activity centres. In scenario planning, planners have control on some factors, while they do not have control on others. Socio-economic factors such as population, household type, and household income are not under the control of planners. Hence, the

factors that were modified in this scenario, compared to the basecase were: dwelling types, distance to the nearest public transport, walkability, and land consumption for transport. In this scenario, it was assumed that accessibility by public transport and walkability would improve by 40% compared to the base-case scenario. It was also assumed that 40% less land would be devoted to transport in 2030 compared to the base-case scenario. As a result of these improvements, VKT and proportion of car usage for travel reduce in the activity-centres scenario compared to the base-case scenario. As the analyses were undertaken at the SLA level, a 40% increase or decrease was equally applied to all SLAs. • Scenario 3: fringe-focus scenario This scenario assumes that all developments would be directed towards the green wedges (fringe) in 2030. This model may be defined by a belief in the advantages of spatially larger, more decentralised cities that are characterised by suburban, low-density residential living and extended road networks, which were rooted in the early 20th century (Alford and Whiteman, 2008). The 12 nonurban areas that surround the built-up urban areas of metropolitan Melbourne and are outside the urban growth boundary are known as green wedges. The green wedges accommodate agricultural and recreational uses, as well as a variety of important functions that support Melbourne. These include major assets such as airports, sewage plants, quarries and waste disposal sites; uses that support urban activity but cannot be located among normal urban development. This scenario assumes continued growth in the fringe until 2030, without limitations. To contrast with the activity-centres scenario, this scenario represents the case where there is 40% more: access to public transport; walkability; and lands devoted to transport in the fringe. Moreover, there is 40% less development in the activity centres compared to the base-case scenario. These changes would result in a slight increase in VKT and the proportion of car trips in the fringe-focus scenario. Same as the activity-centre scenario, a 40% increase or decrease was equally applied to all SLAs in this scenario.

3. Results In order to investigate the effects of the three developed planning scenarios, it is necessary to perform a comparison. The performance of the scenarios with regard to transport sustainability indicators in 2030 are presented in Table 2. The transport sustainability index for each scenario was evaluated by combining different transport sustainability indicators into the transportation environmental, social, and economic sustainability indices. It is worth noting that normalisation and weighting procedures were conducted for each scenario separately. Two factors were extracted using PCA/FA to calculate the indicators’ weights (factors with eigenvalues larger than one were selected, as discussed before). Weights of the indicators for each scenario are presented in Table 3. The results showed that Melbourne would have a more sustainable future from the transport perspective in 2030 compared to 2006. The transport sustainability index per household would improve from 0.61 to 0.66. TEII, TSII, and TCII would also improve compared to 2006 at the household level (Fig. 1). The transportation sustainability indices would also increase in all aspects in the activity-centres scenario compared to the basecase scenario (Fig. 1). The overall transport sustainability index would increase in the activity-centres scenario by 5.7% compared to the base-case. This increase is probably due to improvements in transport facilities in the activity centres, which are the centres

M. Reisi et al. / Land Use Policy 50 (2016) 252–261

where more population and employment growth are expected in the future. The transportation sustainability indices would decrease in all dimensions in the fringe-focus scenario compared to the basecase scenario (Fig. 1). The overall transport sustainability index would decrease in the fringe-focus scenario by 1.5% compared to the base-case. This reduction is probably due to dispersion in city development, which causes more travel.

4. Discussion This paper provided a set of transport sustainability indicators to evaluate the effects of urban planning on transport sustainability in Melbourne. For this task, a normalised transport sustainability index has been built using identified indicators. This study developed three urban planning scenarios to assess progress towards transport sustainability in 2030, which has not been considered before in the transport sustainability literature. The results suggested that more population density, better walkability, and improvements in public transport facilities in activity centres increase the transport sustainability index in the activity-centres scenario compared to the base-case scenario. However, dispersion in city development in the fringe-focus scenario results in more travel, and consequently a lower transport sustainability index in this scenario compared to the base-case scenario. Another additional advantage of this study is that converse to the methodological approach proposed in the literature giving different weights based on experts’ opinions to the indicators (Castillo and Pitfield, 2010; Krajnc and Glaviˇc, 2005) or giving similar weights to different indicators (Haghshenas and Vaziri, 2012; Ronchi et al., 2002), our weighting approach is based on PCA/FA. The main reason for choosing this method is that weights extracted by PCA/FA try to correct overlapping information of correlated indicators. Moreover, the calculated weights based on this statistical method are not subjective. However, weights of different indicators calculated by PCA/FA are only used to correct overlapping information between correlated indicators, and do not show the importance of the indicators. So, if there is no correlation, weights cannot be estimated with this method. Contrary to other studies in which the quantities of the indicators were directly observed or measured (Haghshenas and Vaziri, 2012 Zito and Salvo, 2011), this study modelled VKT and modal split as functions of the socio-economic and land-use factors, and were then used to quantify transport energy consumption, emissions, and their related social and economic impacts. There are some limitations in this study that can be improved in future research. Lack of data for the selected spatial scale limited the number of indicators which could be considered in this research. For example cost of traffic congestion and parking related costs were not available. It would be complete these indicators could be included. Moreover, the scenarios developed here are subjective projections of future possibilities, that were proposed under a range of assumptions (Buxton et al., 2011). One of the major problems in planning scenario is that scenarios are not able to identify their own assumptions and there is the potential of incorrect results (Keepin and Wynne, 1984). Trying to consider a variety of views in scenario planning is the only way to avoid this problem (Ney and Thompson, 2000). The magnitude of changes from the base-case scenario can have an important effect on the final results. Too small or too large changes might not be able to demonstrate the effect of urban planning on transport sustainability. This is confirmed by the small changes of the transport sustainability index in the proposed scenarios compared to the base-case, even with 40% changes in the

259

indicators, socio-economic and land-use factors. If selected scenarios had smaller changes (rather than 40%), there could barely be any change in the transport sustainability index. So, this percentage of changes is suitable, as it demonstrates the minimum change needed for achieving minimum sustainable transport in alternative scenarios compared to the base-case. This study provided a limited outline of alternative urban planning policies. Vision for the future of transport sustainability in Melbourne in 2030 might be limited by current assumptions and narrow consideration of future possibilities. So, there is a need for more comprehensive alternative scenario development considering a range of changes from the base-case scenario. This study is one step forwards to consideration of future urban planning effects on transport sustainability.

5. Conclusions There are substantial uncertainties regarding future urban developments, which often make future prediction inadequate. The transport sector is one of those fields with many uncertainties. Despite these uncertainties, to achieve sustainable transport, policy-makers need to have insights into future developments. This paper presented possible future developments in urban planning in Melbourne. These developments were illustrated by indicators, which express the potential expected impacts on transport sustainability. The results of this study illustrate benefits of using the transportation sustainability indices in assessing policy alternatives and identifying the optimal urban planning strategy for achieving transport sustainability goals. The results found are consistent with other analyses highlighting the benefits of compact development in urban areas (Banister, 2000; Cervero, 1996; Ewing and Hamidi, 2014; Holden and Norland, 2005; Litman, 2014a; Newman and Kenworthy, 1989; OECD, 2012), which results in less car travel and consequently more sustainable transport. It was also determined that a decentralised development in the fringe-focus scenario would lead to lower transportation sustainability indices in all dimensions compared to concentrated development in the activitycentres scenario. Decentralisation of cities results in an extensive increase in trip lengths, that in turn increases car dependence and reduces the possibilities of using public transport (Banister, 2000). In another words, compact development in designed centres, controlling urban sprawl and providing adequate public transport developments allow more sustainable mobility to be achieved.

References Alford, G., Whiteman, J., 2008. Macro-urban form, transport energy use and greenhouse gas emissions: An investigation for Melbourne, 31st Australian Transport Research Forum. Apparicio, P., Seguin, A., 2006. Measuring the accessibility of services and facilities for residents of public housing in Montreal. Urban Studies 43, 187–211. ATSB, 2007. Road Crash Casualties and Rates, Australia, 1925 to 2005. In: Bureau, A.T.S. (Ed.). Australian Transport Safety Bureau. Australian Government, 2010. National greenhouse accounts (NGA) factors. Department of climate change and energy efficiency. Balack, W., 2004. Sustainable Transport: Definitions and Responses. TRB/NRC Symposium on Sustainable Transportation, Baltimore, MD. Banister, D., 2000. Sustainable urban development and transport-a Eurovision for 2020. Transp. Rev. 20, 113–130. Bartholomew, K., 2007. Land use-transportation scenario planning: promise and reality. Transportation 34, 397–412. BITRE, 2009. Road crash costs in Australia 2006. In: Bureau of Infrastructure. Transport and Regional Economics. Bojkovic, N., Anic, I., Pejcic-Tarle, S., 2010. One solution for crosscountry transport-sustainability evaluation using a modified ELECTRE method. Ecol. Econ. 69, 1176–1186. BREE, 2010. Australian Petroleum Statistics. Bureau of Resources and Energy Economics. BTE, 2000. Road Crash Costs in Australia. Bureau of Transport Economics.

260

M. Reisi et al. / Land Use Policy 50 (2016) 252–261

BTRE, 2005. Health Impacts of Transport Emissions in Australia: Economic Costs. In: Services, D.o.T.a.R. (Ed.). Department of Transprt and Regional Services, p. 169. Buxton, M., Alvarez, A., Butt, A., Farrell, S., Densley, L., Pelikan, M., O’Neill, D., 2011. Scenario planning for Melbourne’s peri-urban region. RMIT University, Melbourne, Australia. Castillo, H., Pitfield, D.E., 2010. ELASTIC-A methodological framework for identifying and selecting sustainable transport indicators. Transp. Res. Part D 15, 179–188. Cervero, R., 1996. Mixed land- uses and commuting: evidence from the American Housing Survey. Transp. Res. Part A 30, 361–377. City of Melbourne, 2013. Population forecast, City of Melbourne. Coffey, 2003. Fuel quality and vehicle emissions standards-Cost benefit analysis. Coffey Geosciences PTY LTD. Corpuz, G., McCabe, M., Ryszawa, K., 2006. The Development of a Sydney VKT regression model, 29th Australian Transport Research Forum, Gold Coast. Da Silve, R., Da Silve, M., Macedo, M., 2008. Multiple view of sustainable urban mobility: the case of Brazil. Transp. Policy 15, 350–360. Dargay, J.M., Hanly, M., 2003. The Impact of Land Use Patterns on Travel Behaviour. European Transport Conference, France. DeCoster, J., 1998. Overview of factor analysis. Dobranskyte-Niskota, A., Perujo, A., Pregl, M., 2007. Indicators to assess sustainability of transport activities, European Comission, Joint Research Centre. JRC European Commission. DOT, 2007. Victorian Integrated Survey of Travel and Activity 2007 (VISTA 07). Department of Transport (Victoria), Melbourne. DPIE, DHSH, 1997. Rural, remote and metropolitan areas classification. Department of Primary Industries and Energy, Department of Human Services and Health. DTPLI, Infrastructure 2002 Melbourne 2030, Planning for sustainable growth. Department of Transport, Planning and Local. DTPLI, 2012. Victoria in Future 2012, population and household projections 2011-2031 for Victoria and its regions. Department of Transport, Planning and Local Infrastructure. Dur, F., Yigitcanlar, T., Bunker, J., 2010. Towards sustainable urban futures: evaluating urban sustainability performance of the Gold Coast, Australia. In: 14th IPHS Conference, Istanbul, Turkey. Edara, P., 2003. Mode Choice Modeling Using Artificial Neural Networks. Virginia Polytechnic Institute and State University. EEA, 2005. EEA core set of indicators:Guide. EEA Technical Report, Copenhagen. EPA, 2000. Melbourne mortality study: Effects of ambient air pollution on daily mortality in Melbourne 1991–1996. Environmental Protection Authority. ESRI, 1999. How fuzzy memebership works. Ewing, R., Hamidi, S., 2014. Measuring Urban Sprawl and Validating Sprawl Measures. Metropolitan Research Center at the University of Utah for the National Cancer Institute, The Brookings Institution and Smart Growth America. Faulkner, H.W., French, S., 1983. Geographic Remoteness: Conceptual and Measurement Problems. Bureau of Transport Economics. Fernando, M.A.C.S.S., Samita, S., Abeynayake, R., 2012. Modified factor analysis to construct composite indices: illustration on urbanization index. Trop. Agric. Res. 23, 327–337. Fishman, E., Ker, I., Garrard, J., Litman, T., Rissel, C., 2011. Cost and health benefit of active transport in Queensland: Stage two report - Evaluation framework and values. CATALYST for Health Promotion Queensland. Freudenberg, M., 2003. Composite Indicators of Country Performance: A Critical Assessment. OECD Publishing. Geurs, K., Wee, B., 2004. Accessibility evaluation of land-use and transport strategies: review and research directions. J. Transp. Geogr. 12, 127–140. GISCA, 2011. Accessibility remoteness index of Australia (ARIA) review. National Centre for Social Applications of Geographic Information Systems. Giuliano, G., Dargay, J., 2006. Car ownership, travel and land use: a comparison of the US and Great Britain. Transp. Res. Part A 40, 106–124. Gomez-Limon, J.A., Riesgo, L., 2008. Alternative approaches on constructing a composite indicator to measure agricultural sustainability. In: 107th Seminar, January 30–February 1, 2008, Seville, Spain. European Association of Agricultural Economists. Haghshenas, H., Vaziri, M., 2012. Urban sustainable transportation indicators for global comparison. Ecol. Indic. 15, 115–121. Haque, M., Rahman, M., Sayed Khan, A., Parve, M., 2013. Impact of land use parameters on household travel behavior. Am. J. Civil Eng. Architect. 1, 70–74. Holden, E., Norland, I.T., 2005. Three challenges for the compact city as a sustainable urban form: household consumption of energy and transport in eight residential areas in the greater Oslo region. Urban Studies 42, 2145–2166. Jakimaviˇcius, M., Burinskiene, M., 2009. Assessment of Vilnius city development scenarios based on transport system modelling and multicriteria analysis. J. Civil Eng. Manage. 15, 361–368. Juwana, I., 2012. Development of a water sustainability index for West Java, Indonesia. In: School of Engineering and Science. Victoria university. Keepin, B., Wynne, B., 1984. Technical analysis of the IIASA energy scenarios. Nature 312, 691–695. Kitamura, R., Mokhtarian, P.L., Laidet, L., 1997. A micro- analysis of land use and travel in five neighborhoods in the San Francisco Bay Area. Transportation 24, 125–158. Kline, P., 1994. An Easy Guide to Factor Analysis. Routledge. Kobos, P.H., Erickson, J.D., Drennen, T.E., 2003. Scenario analysis of Chinese passenger vehicle growth. Contemp. Econ. Policy 21, 200–217.

Krajnc, D., Glaviˇc, P., 2005. A model for integrated assessment of sustainable development. Resour. Conserv. Recycl. 43, 189–208. Kunzli, N., Kaiser, R., Medina, S., Studnicka, M., Oberfeld, G., Horak, F., 1999. Health costs due to road traffic-related air pollution. In: An Impact Assessment Project of Austria, France and Switzerland, Third Ministerial Conference for Environment and Health, London. Li, F., Liu, X., Hu, D., Wang, R., Yang, W., Li, D., Zhao, D., 2009. Measurement indicators and an evaluation approach for assessing urban sustainable development: a case study for China’s Jining City. Landscape Urban Plann. 90, 134–142. Limanond, T., Jomnonkwao, S., Srikaew, A., 2011. Projection of future transport energy demand of Thailand. Energy Policy 39, 2754–2763. Lindsey, M., Schofer, J.L., Durango-Cohen, P., Gray, K.A., 2011. The effect of residential location on vehicle miles of travel, energy consumption and greenhouse gas emissions: Chicago case study. Transp. Res. Part D 16, 1–9. Litman, T., 2005. Well- measured-developing indicators for comprehensive and sustainable transport planning. Victoria Transp. Policy Instit. Litman, T., 2012. Land use impacts on transport: How land use factors affect travel behavior Victoria Transport Policy Institute. Litman, T., 2014. Analysis of Public Policies That Unintentionally Encourage and Subsidize Urban Sprawl, in: Institute, V.T.P. (Ed.), Supporting paper commissioned by LSE Cities at the London School of Economics and Political Science, on behalf of the Global Commission on the Economy and Climate. Litman, T., 2014b. Land use impact on transport. How land use factors affect travel behaviour, Victoria Transport Policy Institute. Lotfi, S., Koohsari, M., 2009. Measuring objective accessibility to neighbourhood facilities in the city (a case study: zone 6 in Tehran, Iran). Cities 26, 133–140. MacDonald Gibson, J., Brammer, A., Davidson, C., Folley, T., Launay, F., Thomsen, J., 2013. Environmental burden of disease assessment. Springer Science & Business Media. Makri, M.-C., Folkesson, C., 1999. Accessibility measures for analyses of land use and travelling with geographical information systems. Proceedings of the 2nd KFB- Research Conference, 1–17. Maoh, H., Kanaroglou, P., 2008. Pattern of land use development and sustainable transportion: a simulation approach. 43rd Annual Conference: Shaking up Canada’s Transportation system to Meet Future Needs. Miller, E.J., Ibrahim, A., 1998. Urban form and vehicular travel: some empirical findings. Transp. Res. Rec. 1617, 18–27. Næss, P., 2009. Residential location, travel behaviour, and energy use: Hangzhou metropolitan area compared to Copenhagen. Indoor Built Environ. 18, 382–395. Newman, P., Kenworthy, J., 1989. Cities and Automobile Dependance: An International Sourcebook. Aldershot, England. Newman, P., Kenworthy, J.R., 1991. Transport and urban form in thirty two of the world’s principal cities. Transp. Rev. 11, 249–272. Ney, S., Thompson, M., 2000. Cultural discourses in the global climate change debate. In: Jochem, E., Sathaye, J., Bouille, D. (Eds.), Society, Behaviour, and Climate Change. Kluwer Academic Publishers, Dodrecht The Netherlands, pp. 65–92. NPi, 2008. Emission estimation technique manual for combustion engine. National Pollutant Inventory. OECD, 2008. Handbook on constructing composite indicators, methodology and user guide. Organisation for Economic Co-operation and Development. OECD, 2012. Compact City Policies: A Comparative Assessment. OECD Publishing. Ostro, B., 2004. Outdoor air pollution: assessing the environmental burden of disease at national and local levels. Environmental burden of disease series 5. Paravantis, J., Georgakellos, D., 2007. Trends in energy consumption and carbon dioxide emissions of passenger cars and buses. Technol. Forecasting Social Change 74, 682–707. Pitot, M., Yigitcanlar, T., Sipe, N., Evans, R., 2006. Land Use & Public Transport Accessibility Index (LUPTAI) tool: the development and pilot application of LUPTAI for the Gold Coast. Planning and Transport Research Centre (PATREC). Pongthanaisawan, J., Sorapipatana, C., 2010. Relationship between level of economic development and motorcycle and car ownerships and their impacts on fuel consumption and greenhouse gas emission in Thailand. Renew. Sustain. Energy Rev. 14, 2966–2975. PWC, 2011. A walking strategy for NSW- Assessing the economic benefits of walking. PricewaterhouseCoopers. RACV, 2012. Driving your dollars. Rasafi, A., Zarabadipour, S., 2009. Considering sustainable transport in Iran using analytical hierarchy process. Environ. Sci. Technol. 11, 33–46. Rickwood, P., 2009. The Impact of Physical Planning Policy on Household Energy Use and Greenhouse Emissions, Faculty of Design, Architecture, and Building. University of Technology, Sydney. Ronchi, E., Federico, A., Musmeci, F., 2002. A system oriented integrated indicator for sustianable development in Italy. Ecol. Indic. 2, 197–210. Saisana, M., 2011. Weighting Methods, Seminar on Composite Indicators: From Theory to Practice. Ispra, Italy. Scheurer, J., 2010. Scenarios for future land use-transport integration in the city of Melbourne (and beyond). RMIT-AHURI Research Centre. Shiftan, Y., Kaplan, S., Hakkert, S., 2003. Scenario building as a tool for planning a sustainable transportation system. Transp. Res. Part D 8, 323–342. Soltani, A., Somenahalli, S., 2005. Household vehicle ownership: does urban structure matter? In: 28th Australian Transport Research Forum, Sydney.

M. Reisi et al. / Land Use Policy 50 (2016) 252–261 Spiekermann, K., Wegener, M., 2004. Evaluating urban sustainability using land-use transport interaction models. Eur. J. Transp. Infrastruct. Res. 4, 251–272. SPSS Inc., 2008. SPSS Forecasting 17.0. SPSS Inc., Chicago. State Government Victoria, 2012. Melbourne, let’s talk about the future. Stone, B., Mednick, A.C., Holloway, T., Spak, S.N., 2007. Is compact growth good for air quality? J. Am. Plann. Assoc. 73, 404–418. Verhoef, E., Ubbels, B., Rodenburg, C., Nijkamp, P., 2001. Sustainable mobility. Res. Memorandum 14. Whelan, G., Crockett, J., Vitouladiti, S., 2010. A New Model of Car Ownership in London: Geo-Spatial Analysis of Policy Interventions, European Transport conference.

261

Zhang, L., Hong, J., Nasri, A., Shen, Q., 2012. How built environment affects travel behavior: A comparative analysis of the connections between land use and vehicle miles traveled in US cities. J. Transp. Land Use 5, 40–52. Zhang, X., Lu, H., Holt, J., 2011. Modeling spatial accessibility to parks: a national study. Int. J. Health Geogr. 10. Zhou, P., Ang, B., Poh, K., 2007. A mathematical programming approach to constructing composite indicators. Ecol. Econ. 62, 291–297. Zito, P., Salvo, G., 2011. Toward an urban transport sustainability index: an European comparison. Eur. Transp. Res. Rev. 3, 1–17.