Sie sind auf Seite 1von 12

Multiple Accessibility: Developing a Tool for Evaluating Land Use-Transport Integration

Multiple Accessibility: Developing a Tool for Evaluating Land Use-Transport Integration

Carey Curtis Department of Urban & Regional Planning, Curtin University of Technology, Western Australia, c.curtis@exchange.curtin.edu.au +61 8 9266 2061, F +61 8 9266 2711 Jan Scheurer Institute for Sustainability and Technology Policy (ISTP), Murdoch University, Western Australia, j.scheurer@murdoch.edu.au +61 8 9360 2876, F +61 8 9360 6421 Keywords: Activity Centres, Land Use Transport Integration, Public Transport, Accessibility

Abstract This paper reports on a research exercise to test the value of various accessibility measures as elements in a multi-layered diagnostic planning tool for sustainable accessibility through land use-transport integration. The paper consists of three parts. First we provide a concise overview of our literature review, which examined the practical application of a range of accessibility measures put forward by others. We have developed a seven-fold categorization of accessibility measures: spatial separation measures; contour measures; gravity measures; competition measures; time-space measures; utility measures and network measures. We then explore the relevance of these measures to a planning practice aimed at achieving more sustainable accessibility through land use-transport integration. Finally, the approach is tested in the real world of Perths metropolitan area by assessing the way in which it can be used as a diagnostic tool for determining the future role of each of Perths 125 activity centres. The paper concludes with some reflections on the benefits and practical implications of using multiple accessibility measures as a planning tool.

Introduction
In recent years Australian planners have taken a renewed interest in finding planning strategies that achieve the integration of land use and transport as a means of delivering more sustainable travel outcomes. At a national level Australia's state planning and transport Ministers adopted a National Charter on Integrated Land Use and Transport Planning (DoTARS, 2003). At a state level providing access to public transport through integration with land use activity is the key focus of most recent metropolitan planning strategies. Some states have gone further looking to find ways of harnessing the efforts of local government. For example in Western Australia the 30 metropolitan local governments have agreed to work cooperatively with the state in accordance with an Integrated Transport Planning Partnering Agreement. The notion of 'land use transport integration' is not new; indeed it has been part of planning ideology for decades (Curtis, 2005). However despite the clear intentions of planning policy, its selective implementation resulted in Australian cities taking a low density and dispersed form. Not only are cities spread out but also land use activity is scattered across these very large metropolitan areas. As a result the possibility of supplying a high frequency public transport system to serve these centres represents a significant and expensive challenge. It is not surprising that Australian cities have a high proportion of motorized travel and that car kilometres per capita (averaging 7,416 km per year) are second only to North America (averaging 10,694 km per year) (Kenworthy and Laube, 2001). In reality public transport has only offered a competitive service into CBDs during peak travel times. An examination of the new metropolitan planning strategies for Australian cities shows a great enthusiasm for creating a close relationship between activity centres and public transport. Perths Network City (WAPC, 2004)
Name of author

Multiple Accessibility: Developing a Tool for Evaluating Land Use-Transport Integration

was adopted by the State Government in 2004. It sets the framework for urban development for a 25 year period to 2029. The aim is to achieve a network of mutually dependent sub-centres linked by a high quality public transport system. The integration of land use and transport lies at the heart of the strategy. A network of activity corridors centred on either a main arterial road or suburban railway line utilising land up to 400m on either side is proposed. A large number of activity centres are proposed to be developed at intervals along the activity corridor. The intention is that these will be the focus of daily activity needs. They will include small-scale employment, shopping and services, and medium to higher density housing all placed within walking distance of the public transport stop at the centre. The aim is to provide a high quality walking and cycling network to and within centres close to home and high frequency public transport to link the activity centres. Melbournes metropolitan strategy, Melbourne 2030 (DOI, 2002) follows a similar fashion, although in this strategy 115 activity centres are proposed. These planning strategies indicate a departure from the past approach in that the number of centres proposed is much higher and the emphasis on access by public transport is stronger. Perths Network City, for example, replaced Metroplan (DPUD, 1990) which proposed a hierarchy of centres, Strategic Regional Centres (8), Regional Centres (14), District Centres and Neighbourhood Centres, each defined by the amount of retail floor space. The assumption was that access to these would be by private car. Ten years later the Metroplan ideas were further refined by the Metropolitan Centres Statement of Planning Policy (WAPC, 2000). This added new centres as defined by local government and more importantly set out to promote a wider functional role for centres by indicating the need for commercial floor space as well as retail. While retail floor space ceilings were set according to centre type, no equivalent was given for the amount of commercial floor space. The policy intention is to promote the use of public transport (WAPC, 2000) but the document is silent on how this will be achieved or to which centres and at what level of service. There has been no practice of considering accessibility to activity centres by public transport or in directly integrating service provision with land use planning. Instead the public transport agency aims to meet a minimal service level primarily as a social welfare initiative rather then a real alternative mode of travel to the car. The more recent Network City suffers the same problem. So while the aim of these new strategies is to provide a broad strategic framework to coordinate the location and development of activity within the metropolitan area they do not provide a means to determine which centres should perform which role and in particular within the goal of sustainable accessibility. This forms the core of our research. Of interest is to establish the means to determine which of the myriad of possible activity centres should be promoted based on its ability to deliver multi-modal and sustainable accessibility. This demands a multi-dimensional perspective not only establishing which centres have or could have mixed use development (rather than just retail activity) but also which have or could have more sustainable accessibility. Importantly centres must not be considered in isolation from each other. It is the relationship between centres and the way in which they are networked which forms an important component of sustainable accessibility. In Perth, for example, it is possible to identify over 100 centres that have non-home based trips as a destination. At a strategic level it is not prudent to promote every one of these centres as one in which a high frequency public transport service could be provided. While Network City gives some direction on the importance of some centres, for example the strategy argues that first the CBD should maintain primacy and second that there should be strong centres at the ends of each activity corridor in order that public transport can run efficiently in both directions along the corridor, it does not provide a strategic perspective on which centres should form the initial focus of development activity and/or public transport investment. To achieve this requires a multi-layered diagnostic planning tool for sustainable accessibility through land use-transport integration. The next section reviews the accessibility measures put forward by others and considers their suitability to meet this need.

Measuring Accessibility
The concept of accessibility has been developed, and cast into measurable indicators, in parallel with the concept of mobility. Bhat et al (2000) credit Hansen (1959) with the first significant scholarly work on the subject. While mobility is concerned with the performance of transport systems in their own right, accessibility adds the interplay of transport systems and land use patterns as a further layer of analysis. Accessibility measures are thus capable of assessing feedback effects between transport infrastructure and modal participation on the one hand, and urban form and the spatial distribution of activities on the other hand. Litman (2003) points out that accessibility explicitly takes on board the land use-transport connection and handles trip numbers and travel time as indicators. Geurs and van Eck (2001) define accessibility as: the extent to which the land use-transport system enables (groups of) individuals or goods to reach activities or destinations by means of a (combination of) transport mode(s). (p36)
Name of author

Multiple Accessibility: Developing a Tool for Evaluating Land Use-Transport Integration

In contrast, Bhat et al (2000) use the following definition: Accessibility is a measure of the ease of an individual to pursue an activity of a desired type, at a desired location, by a desired mode, and at a desired time. (p1) It is notable that Geurs and van Eck (2001) make specific reference to land use and transport, thus implying that accessibility is intricately linked to, and primarily determined or enabled by transport infrastructure and urbanisation patterns, whereas this spatial dimension is not at all emphasised, though still implicit, in Bhats et al (2000) definition. Bertolini et al (2005) discuss the balancing act that is the development of suitable indicators for accessibility with regard to their application in practical policy making. They argue that, an accessibility measure must meet two basic requirements: it must be consistent with the uses and perceptions of the residents, workers and visitors of an area, and it must be understandable to those taking part in the plan-making process (p210). They also provide a definition of sustainable accessibility as the amount and diversity of places that can be reached within a given travel time and/or cost (p209) with as little as possible use of non-renewable, or difficult to renew, resources, including land and infrastructure (p212). There are a wide range of methodological approaches for the measurement of accessibility. The work of both Bhat et al (2000) and Geurs and van Eck (2001) are particularly relevant. This paper attempts to consolidate the range of accessibility measures; Table 1 shows our seven-fold classification.

Spatial Separation Measures


The spatial separation model only uses the physical distance between infrastructure elements as input and is thus suitable for the analysis of nodes and network structures (Leake and Huzayyin 1979). It is easy to understand and calculate and requires minimal, easy-to-obtain data input (Baradaran and Ramjerdi 2001). However, there is no reference to land use patterns, spatial distribution of opportunities, or to network constraints to do with travel speed or other sources of resistance. Critically, spatial separation measures do not take into account behavioural aspects of travel choices, particularly the variable attraction of activities and the variable value of time to different groups of trip-makers (ibid). Table 1: Accessibility Measures an Overview Methodological Category Spatial Separation Model (Bhat et al, 2000) Infrastructure Measures (Geurs & van Eck, 2001) Travel Cost Approach (Baradaran & Ramjerdi, 2001) Approach / Measure Measures travel impediment or resistance between origin and destination, or between nodes. Travel impediment measures can include: Physical (Euclidean) Distance; Network Distance (by mode); Travel Time (by mode); Travel Time (by network status congestion, free-flow etc); Travel Cost (variable user cost or total social cost); Service Quality (eg. public transport frequency) Contour Measures (Geurs & van Eck, 2001) Cumulative Opportunity Model (Bhat et al, 2000) Defines catchment areas by drawing one or more travel time contours around a node, and measures the number of opportunities within each contour (jobs, employees, customers etc). Pros & Cons Data is generally easily available from digital mapping material and other public sources. No consideration of land use patterns and spatial distribution of opportunities. Euclidean distance invariably leads to a significant underestimation of actual distance travelled. Some of these problems can be resolved by employing measures other than distance to quantify travel impediment, with travel time.

1) SPATIAL SEPARATION MEASURES

2) CONTOUR MEASURES

Incorporates land use and attends to infrastructure constraints by using travel time as indicator for impediment. Definition of travel time contours may be arbitrary and does not differentiate between activities and travel purposes. Methodology cannot capture variation in accessibility between activities within the same contour.

Name of author

Multiple Accessibility: Developing a Tool for Evaluating Land Use-Transport Integration

Methodological Category 3) GRAVITY MEASURES Gravity Model (Bhat et al, 2000) Potential Accessibility Measure (Geurs & van Eck, 2001)

Approach / Measure Defines catchment areas by measuring travel impediment on a continuous scale.

Pros & Cons More accurate representation of travel resistance than in contour measure but tends to be less legible. Does not differentiate between travel purposes and individual drivers for travel. Provides a accessibility. regional perspective on

4) COMPETITION MEASURES

Competition Measures (van Wee, Hagoort and Annema 2001) Joseph & Bantock Measure (1982) Inverse Balancing Factor Model (Geurs & van Eck, 2001) Time-Space Measures (Bhat et al, 2000 and Geurs & van Eck, 2001) Person-Based Measures (Geurs & van Wee, 2004) Utility Measures (Bhat et al, 2000 and Geurs & van Eck, 2001) Utility Surplus Approach (Baradaran & Ramjerdi, 2001)

Incorporates capacity constraints of activities and users into accessibility measure. May make use of any of the preceding three models.

5) TIME-SPACE MEASURES

Measures travel opportunities within pre-defined time constraints.

Well-suited to examine trip chaining and spatial clustering of activities. Usually requires project-specific user surveys, limiting the geographical range and compatibility of data.

6) UTILITY MEASURES

Measures individual or societal benefits of accessibility. Indicators can include: Economic utility (to the individual, or to the community); Social or environmental benefits (eg. social inclusion, greenhouse effects); Individual motivations of travel (by activity or travel purpose); Option and non-user benefits of transport infrastructure. Measures centrality across entire movement networks. Networks can be represented by: the primal approach (networks are understood as intersections connected by route segments); the dual approach (networks are understood as route segments connected by intersections)

The empirical link between infrastructure provision and economic performance is tenuous and contested. The indicator can analyse existing motivations of travel, but cannot anticipate feedback effects between land use and travel patterns, or future behaviour patterns of users.

7) NETWORK MEASURES

Network Measures: Multiple Centrality Assessment (Porta et al 2006a, 2006b)

More intuitive, and allows for the incorporation of a travel impediment measure in the network analysis. Clearly captures the topological form of a network, and can be used to assess its spatial legibility

Contour Measures
The contour measure in the terminology of Geurs and van Eck (2001), or the cumulative opportunity model in Bhat et al (2000), prominently uses the element of travel time in the composition of the indicator, and defines thresholds of maximum desirable travel times for different types of activities: catchment areas of jobs, employees, customers, visitors and other members of the travelling public are mapped out as contours for each node under consideration. This approach incorporates land use patterns as well as infrastructure constraints. However, in applying a rigid boundary to the catchment area identified, this indicator is not capable of differentiating between opportunities inside this area, despite the fact that actual travel times obviously vary among activities within the same contour bracket (or isochrone). The indicator also treats activities as equal regardless of their cost or desirability for users. Critically, the contour brackets chosen for each type of activity are almost invariably arbitrary, not necessarily reflecting the real drivers of trip making from a user perspective, unless dedicated research is conducted on this component. Thus this indicator is highly sensitive to the choice of demarcation area (Baradaran and Ramjerdi 2001). Bertolini et al (2005) also recognise the weakness of a potentially arbitrary travel cost or time contour (isochrone), yet they make a strong plea for considering a 30Name of author

Multiple Accessibility: Developing a Tool for Evaluating Land Use-Transport Integration

minute time limit where journeys to work are concerned, and recommend this measure to be taken separately for each mode and for different network conditions (car-free flow, car-congestion, public transport, cycling, walking).

Gravity Measures
This sets out to overcome the shortfall of rigid and/or arbitrary contour brackets by treating opportunities differently along a continuum of time and distance. In most cases, this is done by identifying the actual travel time for each opportunity and using a relatively generic distance decay function as a proxy for the disutility experienced by transport users with increasing travel time, cost or effort (Geurs and van Wee 2004). The model, however, still treats every transport user within the study area equally and disregards variations in individual preferences in relation to the desirability of activities (Baradaran and Ramjerdi 2001).

Competition Measures
These consider the presence of competition factors in accessibility. For example the capacity of medical facilities poses an upper limit to the number of potential users, and similarly, an abundant number of potential employees may compete for a limited number of jobs available in a given location. Conversely however, competition may also (and simultaneously) occur between employers for suitably skilled workers, generating two-way constraints to theoretical accessibility. This is captured in the double-constrained spatial interaction model, credited in Geurs and Van Wee (2004) to Wilson (1971). Competition effects in accessibility measures are discussed in more detail in van Wee, Hagoort and Annema (2001) and Shen (1998). In the view of van Wee et al, contour and potential accessibility measures lead to a bias favouring centralisation of activities, that is, the locations with the highest accessibility scores are necessarily the ones with the highest degree of centrality in the transport network. However, the centralisation of activities in transport nodes can reach a point beyond which their accessibility at a regional scale actually declines: for example, once the number of jobs concentrated there exceeds the number of potential employees within a reasonable commute. To include competition effects in an accessibility measure, van Wee et al (2001) propose the introduction of an additional dimension, or extension, to the indicator. Not only is a location zone assessed for the number of activities within a given travel time (or other travel impediment factor): each of the destination zones is further assessed for its capacity of a given activity and relative to activity choices in adjoining zones, and the results factored into the measure of the original zone. This procedure can be repeated for the destination zones, adding a further extension, and so forth however, the model is subject to diminishing returns with growing distance from the original location. Testing the model in a practical application in the Netherlands, the authors conclude that accessibility values change by up to 10% if competition effects are included. However, their next argument acknowledges the weakness of the approach: what exactly does a 10% increase or reduction in accessibility mean? The geographical complexity of the model limits its legibility, and it is recommended to use index values (eg. base case = 100) to make it more understandable. Further complicating factors are links to the economic cycle, changes to land values and travel costs, and the observation that the labour market (for which the model was generated) is far from homogenous. It is also noted that employees vary in elasticity to choose their job location, depending on income, family status, housing situation and other factors.

Time-Space Measures
Time-space measures focus specifically on the time budgets, or space-time paths, of transport users. Bhat et al (2000) identify three types of time constraints in this context: capability constraints (limitations to the number of activities a person can accommodate within a given time frame); coupling constraints (the need to be in particular places at particular times); and authority constraints (the times of operation of given activities, or of components of transport infrastructure/service). This approach is highly suitable for the evaluation of tripchaining and of spatial clustering effects of activities (Burns 1979, Hall 1983, Baradaran and Ramjerdi 2001). Both Bhat et al (2000) and Geurs and van Eck (2001) point out, however, that the information required for this approach is not usually available from standardised travel surveys and therefore often needs to be collected specifically. This limits the opportunities for data aggregation over larger areas, and the compatibility of data sets collected in different surveys. Baradaran and Ramjerdi (2001), quoting Wang (1996), further note that the recognition of time constraints alone in this approach does not yet do justice to the full spectrum of motivations for individual travel choices.

Utility Measures
Utility measures are designed to capture the benefit to users from accessibility to opportunities. This can occur in monetarised form as a measure of economic utility, or as an indicator for social equity (or for other sustainability objectives). It can also be applied as a behavioural indicator, measuring the value individuals afford to the accessibility of particular activities. Geurs and van Eck (2001) point to the weakness of
Name of author

Multiple Accessibility: Developing a Tool for Evaluating Land Use-Transport Integration

empirical evidence for the link between infrastructure provision and economic activity, and the relative inability of this approach to capture feedback effects between transport patterns and land use changes over time. Bhat et al (2000) highlight the inevitable bias in defining a set of choices for activities and opportunities to be included in this approach, and its inherent conservatism it cannot predict the emergence of new choices and their effects on travel behaviour. Baradaran and Ramjerdi (2001) also mention the problematic integration of income effects in this approach. While disregarding such effects restricts the efficacy of the model, their inclusion and consequently, the allocation of a higher utility value on activities performed by higher-income earners raises concerns with equity (Geurs and van Eck 2001).

Network Measures
Porta, Crucitti and Latora (2006a, 2006b) measure accessibility across entire movement networks and this is of particular interest to our research. Two approaches are distinguished: the primal approach and the dual approach. Each approach is based on the identification of nodes and edges as the twin components of any network: in the primal approach, street segments are considered as edges and street intersections are considered as nodes. In the dual approach, it is the other way around. This indicator captures a fundamental characteristic of social networks, as popularised in the seminal work of Milgram (1967) on the average six degrees of separation between any two individuals in the US. In this example, the distance between two individuals in a social network cannot be expressed in a meaningful way by applying a quantitative measure of length to their relationship. It can, however, be measured by counting the number of direct relationships in a chain that connects any two individuals in the sample (degree of separation). The dual graph approach is derived from the space syntax methodology first developed by Hillier and Hanson (1984). The motivation here is to identify the continuity of streets over a multiplicity of intersections as a key attribute of the legibility and functionality of movement networks. From these measures a number of indexes for network nodes are derived including: Degree centrality, defined as the proportion of nodes directly connected to node i out of all nodes within the network; Closeness centrality, defined as the inverse average distance between the node and all nodes within the network; Betweenness centrality, defined as the average proportion of paths between any two nodes within the network that traverse the node out of the total number of possible paths between these two nodes. Efficiency centrality, defined as the ratio of the actual inverse average shortest path length between the node and all directly connected nodes to the theoretical average shortest path length (Euclidean distance) within that sample. Information centrality, defined as the relative drop in network efficiency where the node is removed from the network.

Scheurer and Porta (2006) attempt to apply this methodology to public transport networks and arrive at a different picture from the model suited to the analysis of urban streets. In particular, they define every transferfree link between pairs of nodes on the network as a separate edge in its own right. The degree of the node is thus equivalent to the number of other nodes that can be accessed on public transport without a transfer, and the topological distance between any two nodes (degrees of separation) is equivalent to the number of segments between transfers that make up the journey. Edges are assigned an impediment factor in the primal approach that takes in travel time and frequency of service (travel opportunities per hour) rather than physical distance (which is not very relevant to public transport users, since it bears no direct relation with travel time, cost or service frequency).

Developing a tool for evaluating land use transport integration


The review of common accessibility measures in the previous section forms a framework to assess the performance of a public transport network, and its correlation to the land use structure and geographical distribution of activities, in the Perth metropolitan area in Western Australia. It has been possible to acquire detailed land use and employment data for the activity centres associated with the network nodes, enabling a closer correlation of public transport network structure and performance with the activity density and profile of the nodal catchment areas. It is thus anticipated that the additional information can help overcome some of the quantitative weaknesses identified in the Melbourne model (Scheurer and Porta, 2006), such as the weighting of network paths according to the strength of travel demand in the nodes in

Name of author

Multiple Accessibility: Developing a Tool for Evaluating Land Use-Transport Integration

question, and the use of spatial separation or travel impediment measures before and after the opening of the railway. Map 1 shows the 2007 public transport network in metropolitan Perth. The impediment values given for each route segment are composed of the average travel time in minutes divided by the number of services per hour per direction, and multiplied by a coefficient of 8 to achieve better readability. Therefore the higher the impediment value the greater the spatial and functional resistance to travel accessibility by public transport. The service frequencies and travel times implied relate to the weekday interpeak period (roughly between 9.00 and 15.00 Mondays to Fridays). During these times, all train lines in Perth are operated at a frequency of at least four services per hour (15-minute headways), while bus routes have been differentiated into routes at or above the rail standard (four or more services per hour), and routes below this standard (two to three services per hour). Bus routes with fewer than two services per hour are not included in the map, except where several of them form a trunk route and together meet the minimum standard.

Map 1: Public transport in Perth with modes, service levels and impediment values per route segment. dij = 8tij/fij where: tij = fij = travel time service frequency in departures per hour per direction in minutes

The map clearly shows the radial nature of the rail and most parts of the bus network. In 1998/99, an earlier package of network improvements in Perth saw the introduction of an 80-km circular bus route operated every 15 minutes during the day, connecting six train stations (Fremantle, Shenton Park, Stirling, Bayswater, Oats Street, and Murdoch along the future Perth-to-Mandurah line) and significant travel generators such as shopping malls, civic centres, university campuses and hospitals. For much of its length, the route is now among the busiest bus corridors in the metropolitan area and still represents the only high-frequency public transport service available for non-radial travel relations in Perth. The impediment figures in Map 1, however, suggest that trips

Name of author

Multiple Accessibility: Developing a Tool for Evaluating Land Use-Transport Integration

between bus-train interchanges along the Circle Route are still consistently more conveniently done by using rail and transferring in the city centre (see Table 2). This is mainly due to the relatively large geographical spacing between Perths existing four rail lines, and the relatively wide radius of the Circle Route which runs at an average distance from the city centre of 15 km. This circumstance is set to change with the opening of the Perth to Mandurah railway though, when the Circle Route will provide a path with a lower travel impediment between interchanges on adjacent rail lines (namely between Fremantle and Murdoch) than the rail-only transfer trip through the city centre (Table 2). At the point of writing, Perths public transport regulator was contemplating a doubling of service frequencies on the Circle Route between Fremantle and Murdoch during peak hours after the rail opening (Cox 2007), thus providing an adequate response to the enhanced role of this route segment within the reconfigured network. Table Perths 2: Cumulative impediment public values for selected orbital transport travel paths on network

Path Fremantle Stirling or v.v. Shenton Park Stirling or v.v. Stirling Bayswater or v.v. Bayswater Oats Street or v.v. Oats Street Fremantle or v.v. Oats Street Murdoch or v.v. after network expansion Murdoch Fremantle or v.v. after network expansion

Cumulative impediment value (Circle Route) 87 43 76 53 125 78 47

Cumulative impediment value (Rail with transfer in city centre) 66 29 28 24 66 23 69

Nodal connectivity and land use integration Map 2 shows a nodal connectivity index, derived from a street connectivity indicator developed for an urban design-based context analysis of local areas (ISTP et al 2007). This indicator allocates a value to each point on the network that is equivalent to the number of physical connections into adjacent nodes, minus two. Thus a four-way intersection has a value of two, an intermediate point along a street has a value of zero, and a dead-end (cul-de-sac) has a value of minus one. The resulting connectivity value has been multiplied by the number of departures per hour on each mode, and a mode-specific coefficient derived from the ratio of seat capacity per vehicle and average load factor as identified in Kenworthy and Laube (2001). In the Perth case, this coefficient is 1 for rail and 0.2 for bus (meaning that a train in Perth carries on average about five times the number of passengers a bus does). Obviously, this ratio can be expected to change over time, as vehicles are renewed and the role of each mode within the system changes. Map 2 further highlights the transit-oriented activity centres and corridors identified in Perths Network City strategy (WAPC 2004). These corridors generally follow established train lines and arterial routes with a variety of adjacent land uses; however, it is notable that the inner section of the Armadale train line, the more recently built Clarkson line as well as the new Mandurah line do not coincide with Network City activity corridors (most of the Clarkson line and the inner section of the Mandurah line are located in the median of a freeway).

The nodal connectivity values show a steep gradient from the two major interchanges in the city centre (Perth Central and Perth Esplanade) as the most accessible nodes with very high values, to a range of suburban activity centres particularly along the rail corridors with medium values, and a large number of bus-based network nodes with extremely low values. So in this approach we start to see how the tool can give us a public transport accessibility value for each centre and enable comparisons with the whole system.

Name of author

Multiple Accessibility: Developing a Tool for Evaluating Land Use-Transport Integration

Map 2: Nodal connectivity indexes for Perths public transport with Network City activity centres and corridors highlighted in the background. CVi = ((aij-2))*(fr(i)+ 0.2fb(i)) where: aij = N(i) = fr(i) = fb(i) = links converging in node nodes adjacent (nearest number of rail departures per number of bus departures per hour per direction from node i i, with neighbours) hour per jN(i) to direction and node from node ij i i

Our interest is not just with how accessible these places are from a transport perspective but also what value each activity centre has from a place-based perspective. Bertolinis model for the balanced integration (or not) of nodal (transport hub) and place (activity centre) functions of localities highlights the issue: An unbalanced node could either increase its place-value (for instance by attracting property development) or decrease its node-value (perhaps through reduction in the level of transportation services). A reverse reasoning can be applied to an unbalanced place: either the level of connection will be increased or a different, and perhaps more locally oriented, functional mix will be developed (Bertolini, 2005, pxx). In order to include the land use (or opportunity) dimension of accessibility we are investigating ways of assigning such a value to each centre and then to consider how to weight this in relation to impediment and connectivity scores.
Name of author

Multiple Accessibility: Developing a Tool for Evaluating Land Use-Transport Integration

In order to demonstrate our approach and the issues it raises we will direct our attention to the two suburban nodes with the highest connectivity scores, Stirling and Oats Street. Both nodes are located on purpose-designed rail-bus interchanges that integrate train lines with a high service frequency (8 services per hour per direction), the aforementioned Circle Route and several feeder bus routes. Both nodes also illustrate the challenges associated with the creation of more transit-oriented development and, in fact, of a transit system that is more responsive to urban land uses and activity patterns. Stirling station is located in the median of a radial freeway, with a bus station bridging the interchange at a perpendicular angle. About 1.5 km to the south, there is a large free-standing shopping centre known as Innaloo. The City of Stirlings council offices are located some 500 metres to the north. The immediate surroundings of the train and bus interchange are used for car parking, open space and a reserve for a future grade-separated road junction with a planned arterial road to the southwest (Stephenson Highway). Table 3 shows the current land use/activity mix. It is from this that we will devise a land use indicator for each centre. While the Circle Route connects the shopping mall, train station and council offices as critical ingredients of a viable town centre, pedestrian amenity between these facilities, and in the vicinity of the train station in general, is low (DPI 2002). The City of Stirling has plans to encourage future shopping centre extensions to develop a main street environment in the direction of the train station, and to place high-density residential and office development on either side of the freeway. However, the spatial requirements of the proposed road interchange, and the reluctance of the road authority to scale it down in the interest of a compact, walkable activity centre, constrain the extent of developable sites available. In combination with the out-of-centre focus of the retail, this circumstance makes it questionable whether Stirling will be able to develop the critical mass of residents and jobs to become a vibrant and diverse, transit-oriented and pedestrian-friendly activity centre. Newman (2005) suggest a minimum of 10,000 residents or jobs within the walkable catchment of a station to achieve this, which means that Stirling will have to add at least 4,000 residents or jobs from its current count of just under 6,000 (Table 3).
Table 3: Approximate residential and employment concentration in the 800-metre catchment of Stirling and Oats Street train stations in 2001. Source: ABS 2006

Indicator Number of occupied dwellings Commercial floorspace in sq m Residential population Number of jobs Activity Proportion relative to Newmans activity scale

Stirling 690 215,754 1,670 4,309 5979 59.8%

Oats Street 1,650 242,031 3.450 2,014 5464 54.6%

Oats Streets claim to activity centre status is even more tenuous than Stirlings. To date, the only major local traffic generator is a tertiary education facility adjacent to the train station (Carlisle TAFE). The remainder of the catchment area is taken up by low-density housing with a recent trend towards sub-division of residential blocks for duplexes and triplexes, and some large-lot light industrial uses. The small bus station is conveniently located between the rail platforms and the TAFE, all at ground level. However, the bus routes cross the rail line at grade, generating delays which have been accommodated by way of generous scheduled recovery times for Circle Route buses at the interchange. The nearest more diverse activity centre functions to Oats Street station are located about 600 metres south-west around the intersection of Albany Highway and Hillview Terrace, where the Circle Route crosses another highfrequency bus route along one of Network Citys radial transit-oriented corridors. Albany Highway through the nearby suburbs of Victoria Park and St James is a relatively continuous linear activity centre developed around a former tram line in the first half of the 20th century, with several free-standing smaller shopping malls interspersed with street-oriented development that has seen some residential and employment densification in recent years. Pedestrian amenity along this strip is relatively high, with slow traffic lanes separated by a median strip, landscaping, ample footpaths, on-street parking and a multitude of crossing points. However, there is no plan to extend a similar street environment towards Oats Street train station and thus create a functional connection between the transport node and the activity centre. Given the modest level of current or future residential and employment concentration in the wider area Oats Streets 800-metre station catchment falls about 4,500 short of Newmans (2005) 10,000 activities threshold (Table 3), and given the quite established character particularly of the residential neighbourhoods, the prospects for such a transition appear neither viable in land use terms, nor realistic in political terms.
Name of author

10

Multiple Accessibility: Developing a Tool for Evaluating Land Use-Transport Integration

In the categorisation of Bertolini (2005), Oats Street can be regarded as an unbalanced node, or an element on the public transport network whose importance is disproportional to its place functions (read: urban activities in its catchment area). In this logic and in the absence of land use development that will establish a significantly greater concentration of activities, the Public Transport Authority may be well advised to focus future improvements to orbital bus routes onto transfer points such as Victoria Park or Cannington, where the rail station is located in closer proximity to the existing activity centres, rather than Oats Street. In this way we see the potential of this index as a diagnostic tool. Stirling, in contrast, is a combination of an unbalanced node and a dependent node/place (Bertolini, 2005). Located in a geographical gap within a functionally scattered activity centre, it still derives its place significance from the concentration of activities just beyond its walkable catchment, and its nodal significance from its location at the crossroads of rail and bus routes connecting other, more balanced nodes and places. Stirlings fortunes to evolve towards a greater balance in this respect itself, however, hinge on a formidable challenge to conventional approaches to activity centre planning in Perth. In order to develop a sufficient concentration of activities to complement its nodal with place functions and become a significant destination in its own right, Stirling would need to embrace a combination of some or all of the following factors: high-rise residential and office development adjacent to a train station in a low-rise environment; reductions in road access and parking supply to free up additional space for development; conversion of excess open space to urban land uses; and restrictions or clear directives on future extensions of the Innaloo shopping mall in favour of pedestrian-oriented retail closer to the station. While each of these initiatives are clearly the subject of contemporary policy debate in Perth (and elsewhere), and their spirit is reflected in planning documents such as Network City, they also remain highly contested between stakeholders and decision makers. As such, it remains uncertain whether the impetus towards urban consolidation derived from an exercise of public transport accessibility analysis can be effectively translated into a comprehensive strategy of place-making that breaks drastically with conventional planning practice in one of the worlds most dispersed cities.

Conclusions and outlook


There is clearly no agreement about the accessibility index that is most suitable for the assessment of urban and regional land use and movement systems. Bertolini et al (2005) recommend the use of a contour measure based on travel time and/or a travel cost measure that takes into account travel purposes, socio-demographic factors and the effects from measures such as road pricing and parking management. Conceding the limitations of a sharply defined isochrone for mapping individual travel decisions that are much more spatially and temporarily fluid in real life, they suggest the consideration of a gravity-based measure that can show a more gradual decline of attraction utility with increasing travel time and cost. Geurs and van Wee (2004), too, build a strong case for the incorporation of several perspectives on accessibility into common measurements or, failing that, the application of several accessibility measures in the same context. The application of the model in a Perth context attempts to combine several measures of accessibility including both public transport and a measure or value of place. The relative accessibility of activity centres and network nodes can be determined by conducting a Multiple Centrality Assessment and used to inform decisions about both public transport network configuration, and about land use intensification in the catchment areas of nodes and corridors that gain in accessibility. This tool is well-suited to inform local area planning to add detail to the strategic directions spelled out in the Network City strategy, and to identify gaps in public transport service that need to be addressed to achieve the congruence of movement and land use the document aspires to. It is also well-suited to inform priorities for the future expansion of public transport infrastructure in the Perth metropolitan region.

Acknowledgment
This research was funded by an Australian Research Council Linkage Grant (LP0562422) The impacts of transit led development in a new rail corridor.

References
[ABS] Australian Bureau of Statistics (2006) Census 2001. Available online at www.abs.gov.au, last accessed June 2007 Baradaran S, Ramjerdi F (2001) Performance of Accessibility Measures in Europe. Journal of Transportation and Statistics, September/December 2001 Bertolini L, Le Clercq F, Kapoen L (2005) Sustainable Accessibility: A Conceptual Framework to Integrate Transport and Land Use PlanMaking. Two Test- Applications in the Netherlands and a Reflection on the Way Forward. Transport Policy, Vol 12, pp 207-220 Bertolini L (2005) Cities and Transport: Exploring the Need for New Planning Approaches. Chapter 5 in Albrechts L, Mandelbaum S J (2005, Eds) The Network Society. A New Context for Planning. Routledge, Oxford, UK Bhat C, Handy S, Kockelman K, Mahmassani H, Chen Q, Weston L (2000) Development of an Urban Accessibility Index: Literature Review. Research project conducted for the Texas Department of Transportation. Center for Transportation Research, University of Texas, Austin (TX), USA

Name of author

11

Multiple Accessibility: Developing a Tool for Evaluating Land Use-Transport Integration

Burns L D (1979) Transportation, Temporal and Spatial Components of Accessibility. Lexington Books, Lexington (MA), USA Cox S, Public Transport Authority (2007) Personal Communication Curtis C (2005) The Windscreen World of Land Use Transport Integration: Experiences from Perth, a Dispersed City. Town Planning Review Vol 76 (4) pp 423-453 [DOI] Department of Infrastructure (Victoria, 2002) Melbourne 2030. Planning for Sustainable Growth. Melbourne (VIC) [DOTARS] Department of Transport and the Regions 2003, National Charter of Integrated Land Use and Transport Planning, Department of Transport and the Regions, Canberra. [DPI] Department of Planning and Infrastructure (Western Australia, 2002) Stirling Transport Interchange Integrated Access Plan. Perth (WA) [DPUD] Department of Planning and Urban Development 1990, Metroplan. A Planning Strategy for the Perth Metropolitan Region. Perth (WA) Geurs K T, van Eck J R (2001) Accessibility Measures: Review and Applications. Rijksinstituut voor Volksgezondheid en Milieu (National Institute of Public Health and the Environment, RIVM) and Urban Research Centre, Utrecht University. Bilthoven/Utrecht, Netherlands Geurs K T, van Wee B (2004) Accessibility Evaluation of Land-Use and Transport Strategies: Review and Research Directions. Journal of Transport Geography, Vol 12, pp 127-140 Hall R W (1983) Travel Outcome and Performance: The Effect of Uncertainty on Accessibility. Transportation Research B, Vol 17, No 4, pp 275-290 Hansen W G (1959) How Accessibility Shapes Land Use. Journal of the American Planning Institute, Vol 25, pp 73-76 Hillier B, Hanson J (1984) The Social Logic of Space. Cambridge, UK [ISTP] Institute of Sustainability and Technology Policy (ISTP), Department of Planning and Infrastructure (DPI), Mirvac Fini (2007) Sustainable Urban Design Practical Fieldwork Project 2007. Murdoch University, Perth (WA) Joseph A E, Bantock P R (1982) Measuring Potential Physical Accessibility to General Practitioners in Rural Areas: a Method and Case Study. Social Science and Medicine, Vol 16, pp 85-90 Kenworthy J, Laube F (2001) The Millennium Cities Database for Sustainable Cities. International Association of Public Transport (UITP), Brussels, Belgium and Institute for Sustainability and Technology Policy (ISTP), Murdoch University, Perth, Australia Leake G R, Huzayyin A S (1979) Accessibility Measures and Their Suitability for Use in Trip Generation Models. Traffic Engineering and Control, Vol 20, No 12, pp 566-572 Litman T (2003) Measuring Transportation: Traffic, Mobility and Accessibility. ITE Journal, Vol 73, No 10, pp 28-32 Newman P (2005) Transit-Oriented Development: An Australian Overview. Transit-Oriented Development Making it Happen. Fremantle (WA), July Porta S, Crucitti P, Latora V (2006a) The Network Analysis of Urban Streets: A Dual Approach. Physica A, Statistical Mechanics and its Applications, in print Porta S, Crucitti P, Latora V (2006b) The Network Analysis of Urban Streets: A Primal Approach. Environment and Planning B: Planning and Design, Vol 33, in print Scheurer J, Porta S (2006) Centrality and Connectivity in Public Transport Networks and their Significance for Transport Sustainability in Cities. World Planning Schools Congress, Mexico DF, 13-16 July Shen Q (1998) Location Characteristics of Inner-City Neighbourhoods and Employment Accessibility of Low-Wage Workers. Environment and Planning B, Vol 25, pp 345-365 Van Wee B, Hagoort M, Annema J A (2001) Accessibility Measures with Competition. Journal of Transport Geography, Vol 9, pp 199-208 [WAPC] Western Australian Planning Commission 2004, Network city: community planning strategy for Perth and Peel, WAPC, Perth. [WAPC] Western Australian Planning Commission 2000, Statement Of Planning Policy No. 9: Metropolitan Centres Policy Statement For The Perth Metropolitan Region - A Policy Statement For Metropolitan Centres, Retailing, Offices And Associated Development In The Perth Metropolitan Region Western Australian Planning Commission, Perth. Wilson A G (1971) A Family of Spatial Interaction Models, and Associated Developments. Environment and Planning, Vol 3, No 1, pp 1-32

Name of author

12