options(scipen = 999, digits = 3)
knitr::opts_chunk$set(cache = TRUE, warning = FALSE, echo = FALSE, message = FALSE, dpi = 600, fig.fullwidth=TRUE)
library(rgdal)
library(sf)
library(tmap)
library(tmaptools)
library(tidyverse)
library(tidyselect)
library(janitor)
library(grid)
library(gridExtra)
cols <- RColorBrewer::brewer.pal(9, 'Set1')
divs <- c(cols[1], cols[5], cols[6], cols[3], cols[2], cols[4])
cols <- c(cols[2], cols[3], cols[6], cols[1])
yes.no <- c(cols[4], cols[2])

Abstract

‘Park ’n’ Ride’ facilities (PnR) initially emerged to accommodate motorists that would otherwise exhaust the local supply of parking around train stations and other rapid high occupancy vehicle nodes but increasing became a planning strategy to provide commuters from auto-dependent suburbs with access to rapid high occupancy vehicle to reduce their environmental impacts and inner-city road and parking capacity requirements. Theoretically, PnR should influence modal choice by making the transfer between car and rapid transit more convenient yet this base assumption rarely matches the empirical reality. Our synthesis of the PnR literature suggests that motorists deciding whether to park-and-ride have considerations beyond minimising their travel duration and expenses, and we develop a new integrative model of PnR, multi-modalism, and modal choice to illustrate how reliability and competing transport planning strategies such as inter-city mobility, transit-oriented development, and active transport interact and inform modal choice. Upon laying these theoretical foundations, we empirically examine the extent to which developing or modifying PnR influences modal choice in our case study context, Brisbane, Australia. Our research findings suggest that it is new rather than modified PnR that influence modal choice and that new park and riders are typically drawn from nearby locations rather than peripheral and therefore auto-dependent areas. This influence is particularly evident in suburbs closer to the inner city, and is problematic given that these are not the intended users of PnR. Our synthesis and examination of multi-modalism and modal choice has important implications for researchers, planners, and policy makers attempting to influence modal choice and improve the efficiency of urban mobility.

Keywords: park ‘n’ ride; PnR; auto-dependency; suburbanisation; transit ridership; modal choice.

Acknowledgement: This research is conducted through a project funded by the Australian Research Council Linkage Project grant LP160100031 with additional support from the industry partner the Queensland Department of Transport and Main Roads. Notably, the interpretations of the analysis are solely those of the authors and do not necessarily reflect the views and opinions of the Queensland Department of Transport and Main Roads or any of its employees.

Introduction

As cities worldwide attempt to reduce auto-dependency, Park ‘n’ Ride facilities (PnR) have the capacity to smooth this transition by providing already auto-dependent commuters with the opportunity to leave their automobiles parked alongside rapid transport nodes (i.e., train stations, bus rapid transit terminals, and ferry terminals) and complete their journey by rapid High Occupancy vehicle (rapidHOV; Meek 2008; Guerra et al. 2012; Kuby et al. 2004; Lane et al. 2006; Turnbull 1995). While PnR have strong theoretical foundations, the empirical foundations are fragmented and the empirical findings incomparable between studies (Mingardo 2013). For instance, it remains unclear whether PnR are persuading motorists to park at PnR, or if they are persuading former walkers, cyclists, and high occupancy vehicle riders to drive to the PnR instead (Parkhurst 2000; Dijk and Parkhurst 2014; Mingardo 2013; Meek et al., 2009). Likewise, it remains unclear whether PnR influence modal choice or whether park and riders self-select locations near PnR (Zahabi et al 2012; Christiansen 2017). Thus whether PnR influence modal choice remains the subject of debate.

This study addresses the fragmented empirical foundations of PnR as follows. We begin by exploring the history of PnR to clarify the intended purpose of this planning tool. Then we identify the emerging types of PnR to determine whether all PnR share a universal purpose. Through a synthesis of the broad and fragmented empirical PnR literature we identify the key conceptualisations and operationalisations located throughout this scholarship, and provide the necessary foundations to develop a new analytical model for exploring the relationship between PnR and modal choice. Drawing on this new analytical model, we examine the relationship between change to PnR and modal choice across our case study context: Brisbane, Australia.

The History of PnR

To clarify the purpose of PnR, it is necessary to explore its historical emergence. While informal parking at transit stations or at the urban periphery may have existed for as long as the automobile itself, the first formal PnR were created in Detroit in the 1930s (Noel 1988). The early adopters of PnR were often motorists that had previously parked informally (Heggie and Papoulias 1976). PnR became more prevalent throughout the United States following the 1968 Federal Aid Highway Act, which enabled planning authorities to request federal reimbursement for up to half of their PnR development expenses (Noel 1988). During those early years, PnR was heralded as ‘the most important innovation in HOV transportation since the Second World War’ (Boyce et al. 1972 as cited in Chen et al. 2014: 257). Given that PnR can enable auto-dependent commuters to bypass inner-city traffic congestion onboard rapid transit, and reduce their automobile depreciation rate, insurance premiums, fuel consumption, maintenance expenses and even exposure to accidents, the purpose of PnR was initially motorist-centric (Noel 1988; van der Heijden and Molin 2002).

While the purpose of PnR was initially motorist-centric, it increasingly became network-centric and expanded internationally during the 1970s as part of the ‘new realism’ paradigm in transportation planning. Essentially, this paradigm prioritised the environment, pedestrians, HOV, cyclists, and short-stay automobiles before long-stay automobiles (Figure 1). Notably, PnR stands apart from other ‘new realism’ policies because rather than a radical overhaul of the built form, it attempts to smooth the transition for auto-dependent commuter. As such, Shaw and Walton (2001: 1031) aptly term PnR as ‘pragmatic multi-modalism’.

While network-centric PnR studies typically regard providing auto-dependent commuters with the opportunity to participate in multi-modalism as the intended purpose of PnR (Chang and Morlok 2005; Jensen 1995; de Nazelle et al. 2010; Kessler and Schroeer 1995; Olowoporoku et al. 2012), these studies may also include benefits derived from multi-modalism. For instance, multi-modalism draws parking demand away from the inner-city where space for parking is more expensive, and likewise draw shopping away from the inner-city and major shopping centres towards local shops located near PnR (Parkhurst and Richardson 2002). Further, multi-modalism has the flexibility to concentrate HOV riders at specific nodes, and thus lowering the operational cost of empty seats on trunk lines or feeder services (Karamychev and van Reeven, 2011). Last, multi-modalism expands the catchment of potential workplaces for jobseekers thus their employability becomes less contingent on owning an automobile or being centrally located (Noel 1988). In sum, the history of PnR suggests that its purpose was initially motorist-centric as it sought to reduce the travel expenses of auto-dependent commuters. Now PnR has become more network-centric by considering the collective expense of accommodating motorists and the collective savings of multi-modalism.

**Figure 1. The 'new realism' hierarchy of travel and travellers, which places 'the environment' at the apex. Based on Oxford's 1973 Balanced Transport Policy (as reported by Parkhurst and Dudley 2004).**

Figure 1. The ‘new realism’ hierarchy of travel and travellers, which places ‘the environment’ at the apex. Based on Oxford’s 1973 Balanced Transport Policy (as reported by Parkhurst and Dudley 2004).

PnR Typologies

As the intended purpose of PnR has evolved, its form, placement, and function have diversified to the point that multiple typologies have emerged to capture and describe this diversity (Table 1). While these PnR typologies are incomparable between studies and study frames, authors typically classify PnR according to similar sets of PnR characteristics. As such, identification of these defining PnR characteristics has the potential to explain how PnR influences modal choice and how this varies throughout a major metropolitan area.

The metropolitan role of each PnR is determined by characteristics that include urban centrality, informality, intended transit function, and accessibility. While delivery agencies lack complete control of how their PnR are used, these typologies suggest that the initial placement of PnR is a crucial determinant of their use.

Table 1. Proposed PnR typologies

Typically, the highest priority of PnR delivery agencies is to locate PnR where it has the potential to serve the greatest number of users within a given budget (Cervero et al., 2004; Vuchic, 2007). To this end, delivery agencies are likely to strategically place their PnR according to: (a) theoretically-informed expert criteria (American Association of State Highway and Transportation Officials, 2004); (b) population catchment-based demand analyses (Keck and Liou, 1976; Hendricks and Outwater, 1998; Norlida et al., 2007); or (c) agent-based simulations (Wang et al., 2004; Farhan and Murray, 2008).

When relying on theoretically-informed expert criteria, agencies typically choose sites that fulfil as many criteria as possible. The criteria are based on the assumption that these sites will attract more motorists and minimise community conflicts. A few authors have compiled lists of expert criteria for placing PnR that are summarised below (Table 2).

Table 2. Expert criteria for PnR placement.

In contrast to expert criteria, population catchment-based demand analyses evaluate proposed PnR sites according to the size and social composition of the population within its ‘commuter shed’ (Horner and Grubesic 2001; Bolger et al. 1992). For example, Horner and Grubesic (2001) identify two typical demographic profiles of PnR users using Principle Components Analysis, and then calculate a ‘derived demand index’ according to how closely proposed locations mirror demographic profiles. While a novel approach, this also has the potential to reinforce existing socio-spatial inequities of access to PnR.

The shapes of commuter sheds vary between studies, but typically resemble either ring buffers or parabolic shapes such as a cone, pear or ellipse with the centre of mass farther from the CBD based upon the assumption that motorists prefer to avoid backtracking (Figure 2; Horner and Grubesic 2001; Holguin-Veras et al. 2012). PnR commuter sheds in urban peripheries tend to larger-at least in theory-because access-ways here are expected to be faster and relatively uncongested (O’Sullivan and Morrall 1996).

**Figure 2. PnR commuter-shed shapes appearing in research and policy (based on Holguin-Veras et al. 2012).**

Figure 2. PnR commuter-shed shapes appearing in research and policy (based on Holguin-Veras et al. 2012).

Finally, agent-based simulations typically calculate ‘Break Even Distances’(BED) where parking-and-riding is the rational choice for motorist seeking to minimise their travel duration and expenses, and PnR are placed accordingly (Holguin-Veras et al. 2012). For instance, Holguin-Veras et al. (2012) identify (a) a drive/park BED where parking-and-riding requires too much lateral driving or backtracking to be the rational choice, and (b) the inner/outer-city BED where the expense of the modal switching (e.g. from car to train) outweighs the savings afforded by rapid transit. Other simulations introduce additional elements such PnR queues (Su and Wang 2019), departure reliability (Chen et al. 2017), PnR lot layouts (Lai and Shalabay 2007), and road toll locations and rates (Li et al. 2012; Liu and Meng 2014).

Notably, simulations assume that motorists are perfectly rational decision-makers who only seek to minimise their travel duration and expense. This assumption fails to explain why motorists routinely ignore vacant PnR bays. Presumably, they do so to maximise the portion of their journey spent within the comfort and intimacy of their own car (Parkhurst 2000; Horner and Groves 2007). Moreover, simulations assume that the number of commuter journeys is fixed whereas, in reality, PnR can induce extra trips to the CBD by making travel faster and cheaper (Meek et al. 2011; Parkhurst 2000). While human behaviours such as ‘comfort seeking’ and ‘induced travel’ could be challenging to simulate, they are well-known in practice. As such, it may be preferable to substitute inner-city PnR bays with feeder services so that these behaviours become less viable (Parkhurst and Buxton, 2005).

PnR and Modal Choice

As discussed previously, the relationship between PnR and modal choice is the subject to debate given that it is unclear whether PnR are: attracting motorists; persuading pedestrians, cyclists, HOV riders to drive; or self-selection has a role (Parkhurst 2000; Dijk and Parkhurst 2014; Mingardo 2013; Meek et al., 2009; Zahabi et al 2012; Christiansen 2017). Nonetheless, a synthesis of the empirical PnR literature can reveal the key conceptualisations and operationalisations located throughout this literature, and provide the foundations for developing a new analytical model for exploring the relationship between PnR and modal choice.

The empirical foundations of PnR are increasingly being questioned given that there remains limited evidence that PnR influence modal choice or improve transport networks. For instance, motorists routinely drive past vacant PnR parking bays and continue to their destination thus providing a plain example that PnR lack sufficient appeal and influence (Bos et al. 2004; Karamychev and van Reeven 2011). For PnR to address network-centric aims, they should be located to intercept motorists near their point of origin for the greatest reduction in travel otherwise PnR begins to operate merely as a shuttle service covering just the CBD (Parkhurst and Meek 2014). PnR located farther from the general path of motorists can increase travel since motorists must divert their path toward a station and may need to double back or continue further should they encounter a fully-occupied PnR lot (Parkhurst and Meek 2014). In itself, PnR can make auto-dependent residential development more palatable (Parkhurst 2000; Israel and Cohen-Blankshtain, 2010), and thus supporting urban sprawl (Lehrer 1994). Even where successful, PnR may be unable to permanently relieve downstream traffic congestion because a free-flowing traffic can induce travellers to shift from other modes back to their cars in a cyclical fashion (Parkhurst 2000; Dickins 1991).

Surveys at PnR suggest that park and riders have considerations beyond minimising their journey duration and expenses when choosing whether to park-and-ride or not. These considerations include:

  • Perceived safety and lighting at PnR (Bos et al. 2003);

  • Perceived social inferiority of HOV riders (de Jong et al. 2003; Hensher and Rose 2007; Ashmore et al. 2019); and

  • Anxiety increased by multi-modalism given the unreliability of HOV (Gaver 1968; Jackson and Jucker 1981; Lo and Tung 2003; Lo et al. 2006; Abdel-Aty et al. 1995).

To expand upon the last point, each additional link in the multimodal chain requires introducing further ‘slack time’ (or leeway) into the allotted travel time budget to address occasions outside the commuter’s control (Gaver 1968; e.g. late arrivals, missed connections, and already fully occupied vehicles). In the context of modal choices, ‘slack time’ could be required in four distinct situations:

  1. Travel-time reliability considers the overall variability in travel duration (Asakura and Kashiwadani 1991). This is of particular concern during the journey to work (JTW) when commuters are expected to arrive on time. While car-based commuters must also contend with travel-time reliability that is introduced by traffic congestion, the advantage of their modal choice is that it can be re-routed around obstacles while bus riders are trapped on a fixed path. On the other hand, train and bus rapid transit (BRT) may be regarded as much more reliable since these modes run on dedicated tracks.

  2. Connectivity reliability accounts for the likelihood that all network nodes will remain connected (Iida and Wakabayashi, 1989). As such, modal choices that entail trip chaining introduce the concerns of connectivity reliability, while modal choices that entail just a single mode such as driving a car or riding a bus direct avoid these concerns altogether. Further, each additional transfer along a multi-modal trip-chain introduces further uncertainty as a connecting service may arrive late or not at all.

  3. Capacity reliability refers to the likelihood that an arriving transit vehicle will accommodate everyone waiting at the station (Chen et al. 1999; Lo and Tung 2000). As such, driving a car mitigates against scenarios such as the arrival of a full bus or train, and the resultant wait for the next service that may or may not be full (Li et al. 2007; Pang and Khani 2018). Service providers seeking efficiency often reduce service frequency and operate their automobiles close to maximum capacity. This may unintentionally reduce capacity reliability, and therefore discourage transit ridership (Meek et al. 2011).

  4. Parking reliability accounts for the time spent cruising for parking or adapting one’s plans when no parking can be found (Stieffenhofer et al. 2015; Tsang et al. 2010). This is a concern both of commuters who drive directly to work and of those who park-and-ride. When parking reliability at a PnR is poor, motorists may begin to leave for work earlier in the morning in order to maximise their chances of securing a parking bay, thus detracting from their quality of life (Wang et al. 2014). For instance, PnR lots in Perth, Australia, are typically filled by 7:30 am on weekdays (Lin et al. 2014). In the San Francisco Bay Area, a study conducted more than a decade ago found that 56 percent of PnR users arrived before 6:00 am (Shirgaokar and Deakin 2005). Here, the introduction of PnR fees prompted PnR users to arrive later in the morning, and did not cause a significant drop in their numbers (Syed et al. 2009). This suggests that paying for parking improves parking reliability. Beyond empirical observations, the effect of charging parking fees at PnR lots has also been explored via simulations. For example, Yang and colleagues (2013) have identified an equilibrium between the number of reserved and unreserved parking bays, which maximises PnR usage. The equilibrium point can be made dynamic by having the reservations expire when motorists take too long to arrive (Liu et al. 2014).

While these four reliabilities and slack time have to potential to explain some of the anxieties associated with parking and riding, arguably these same concepts could be further dimensions for explaining modal choice. As such, we developed a framework of modal choice and reliability that contrasts exposure to unreliability by modal choice (Figure 3).

**Figure 3. A framework of modal choice and reliability.**

Figure 3. A framework of modal choice and reliability.

As illustrated in Figure 3:

  1. Driving directly to work requires slack for two situations that are when difficult to circumnavigate road obstructions are encountered, and when inner-city parking is difficult to locate.

  2. Parking-and-riding requires slack for at least three situations that are when PnR parking is difficult to locate, when rapid transit arrives full or late, and when connections are missed. From this understanding, motorists may choose to forfeit PnR altogether to minimise situations that require slack, and therefore minimise the uncertainty and resultant anxiety of their journey to work.

  3. Riding a direct transit service also requires slack for at least three situations given that busses may arrive full or late, connections may be missed, and travel time may vary given that direct transit services are particularly vulnerable to road obstructions as they are locked to a fixed route.

  4. Riding both feeder and rapid services require slack for at least four situations thus suggesting that this modal choice invites greater uncertainty since feeder services may arrive full or late, connections may be missed, and then these uncertainties are introduced again by the rapid service.

  5. Walking to a station and then riding transit that is the intended modal choice within TOD zones has walking to eliminate the uncertainties of travel time and parking reliability however two situations remain since rapid services may arrive full or late, and connections missed.

  6. Relying entirely on active transport (walking and/or cycling) is a mode choice that eliminates most uncertainties and therefore requires minimal slack. As such, residents that live within a range of their destination that is feasible for walking or cycling can claim the greatest control over their time, while more auto-dependent residents may find that driving direct affords the greatest control.

Notably, Mobility as a Service (MaaS) is an emerging planning paradigm that has the potential to salve some of these journey-to-work uncertainties by informing travellers of HOV arrivals and parking availability at connections and destinations (MaaS; Sipe and Pojani 2018a; 2018b). Knowledge pertaining to each reliability concern would allow commuters to make more informed modal choices on a given day, and potentially enable them to reserve parking well in advance (Chen et al., 2014; Zhang et al., 2011). For example, PATH2Go is a new smartphone application that provides motorists in the San Francisco Bay Area with personalised information on real-time traffic conditions, parking availability, and expected transit departures as they are en route to a destination (Zhang et al. 2011).

There is also evidence that some PnR were demonstrative attempts to win votes or salve public disputes about parking supply rather than motorist-centric or network-centric aims (Huntley 1993; Duncan and Christensen 2013). Given that PnR parking is typically free or priced below inner-city parking rates to appeal to motorists, the revenue typically does not cover the expenses and any inner-city parking trade-offs overlooked unless both the outer-city PnR and inner-city parking fall within the purview of a single agency (Noel 1988). Indeed, with the purview of outer-city PnR and inner-city parking split apart, the metropolitan parking stock may increase unless inner-city parking is reduced at greater rate than outer-city PnR bays (Dijk and Parkhurst 2014). Outer-city agencies may interpret their fully-occupied PnR lots as operating at peak efficiently, while inner-city agencies are left to manage the outer-city commuter parking spill-over. What is clear is that while PnR can be a politically-loaded topic, providing PnR is arguably more palatable than asking residents to alter their lifestyles (Parkhurst and Meek 2014; Willing and Pojani 2017).

The placement of PnR can present a social justice issue for local communities given PnR reduce the cost of inner-city shopping thus drawing shoppers away from their local businesses (Meek et al. 2011; Parkhurst 2000). Likewise, PnR can divert portions of the peak traffic flow into smaller-capacity suburban roads and thereby flood the suburb with thoroughfare motorists rather than visitors (Parkhurst, 2000; Horner and Groves 2007; Duncan and Cook, 2014; Noel 1988). As observed in Australia, Germany, and the Netherlands, the aesthetics, traffic noise, and exhaust associated with PnR can be detrimental to nearby property values and air quality, and reduce the appeal of reaching the rapid transit node by foot or bike (Duncan and Christensen, 2013; Hess et al., 1999; Parkhurst 2000; Dijk and Parkhurst 2014; Wiseman et al. 2012; Mingardo 2013; Meek et al. 2009).

Last, the coexistence between PnR and TOD has remained complex thus bringing into question where PnR fits within the broader new realism planning paradigm (Pojani and Stead 2015; Willson 2005; Yang and Pojani 2017). For instance, while PnR and TOD both aim to increase HOV ridership, TOD restricts the local parking supply to make driving impractical, while PnR increases parking supply next to stations in order to attract motorists from areas beyond the TOD zone (Willson 2005; Giuliano 2004). Given that it can take years to accumulate a critical user threshold to make TOD appealing, PnR is seen as a relatively inexpensive means for establishing a local customer base within a TOD zone (Cervero et al. 2004). However, if a TOD zone is ultimately successful, authorities may find that PnR parking spaces become unaffordable (Weinberger 2012; Dovey et al. 2015; Ginn 2009). As such, there is insufficient evidence to suggest that PnR are influencing modal choice and mounting evidence of PnR that are detrimental for communities or competing in the same space as TOD.

A New Integrative Model of PnR, Multi-modalism, and Modal Choice

Having explored the all-too-human anxieties associated with modal-choice yet absent from expert-criteria, demand analyses, and especially agent-based simulations, it is now necessary to draw focus to how the placement of PnR may influence modal choice. When evaluating sites for PnR, it is important to keep in mind that commuters are not randomly distributed throughout the urban form. Indeed, self-selection bias could play a role in residential location since, when deciding where to live, people generally make trade-offs amongst a range of factors that include dwelling type, community composition and preferred modal choice (Alonso 1969). A person that wishes to live a car-free lifestyle may place greater importance on features such as urban centrality, TOD environment and HOV frequency even if this means paying a housing premium. Meanwhile, someone who prefers driving could regard these as unnecessary extras. Given self-selection bias and a strained relationship between PnR and TOD, we posit that:

  • Residential choice precedes modal choice. Thus only the travel behaviour of residents predating the introduction or modification of a PnR can demonstrate whether PnR is having the intended influence over modal choice.

  • A walk/drive ‘break even distance’ exists that delineates a plausible ‘walk shed’ around a PnR, within which residents may regard walking as being less costly that driving.

By incorporating these elements into a new integrative model of PnR, multi-modalism, and modal choice (Figure 4), we eliminate the false dichotomy that residents must either drive directly to work or park-and-ride (as is typically presented in PnR scholarship). This also provides the means through which we are able to integrate the fragmented PnR literature. Last, the model may also integrate TOD principles into PnR research by expanding the set of Break Even Distances (BED) that influence modal choice from: (1) a destination BED where the expense of changing modes exceeds the expense of inner-city traffic congestion and parking; and (2) an origin BED where the expense of diverting a driving route towards a PnR exceeds the expense of inner-city traffic congestion and parking (see Holguin-Veras et al. 2012) to include: (3) an inter-city BED delineating where travel could be away from the inner-city; and (4) an walking BED where walking or cycling to a node is more efficient than driving and parking, which typifies a TOD area. These delineations in turn enclose: (5) park and ride sheds that represent the catchment areas where drivers may regard parking and riding as feasible; (6) walk sheds where commuters may regard walking or cycling as feasible; and (7) passenger sheds where commuters have a frequent (7a) feeder HOV for reaching nodes and (7b) direct HOV where the cost of reaching nodes outweighs the costs inner-city congestion on-board HOV.

**Figure 4. An integrative model of PnR, multi-modalism, and modal choice. (Adapted from Bolger et al 1992, Horner and Grubesic 2001, O'Sullivan and Morrall, 1996, Buxton and Parkhurst, 2005, Holguin-Veras et al. 2012).**

Figure 4. An integrative model of PnR, multi-modalism, and modal choice. (Adapted from Bolger et al 1992, Horner and Grubesic 2001, O’Sullivan and Morrall, 1996, Buxton and Parkhurst, 2005, Holguin-Veras et al. 2012).

In sum, while the intended purpose of PnR is to promote multi-modalism and thereby reduce the collective expense of driving, evidence that PnR influence modal choice remains limited while evidence that PnR can be detrimental continues to mount. Therefore, we start our study from the ground up by determining whether the introduction of new PnR and/or the modification of pre-existing PnR has a discernible influence upon the residents that were present both before and after these changes. Greater Brisbane, Australia, is our study area.

Greater Brisbane as a Case Study

Greater Brisbane is located in the south-eastern region of the Australian state of Queensland (Figure 5) and it is the third most populous Australian city with more than two million residents (ABS 2016). From 2006 to 2016, the Greater Brisbane population grew by 23 percent placing the city’s population growth five percent ahead of the national average.

states <- readOGR('.','STE_2016_AUST') %>%
  st_as_sf()

central.station <- st_as_sf(data.frame(id = c("Brisbane Central Station"), longitude = c(153.0261831), latitude = c(-27.466224)), coords = c("longitude", "latitude"), crs = st_crs(states))

the.lakes <- st_as_sf(data.frame(id = c("Springfield Lakes", "North Lakes"), longitude = c(152.916667, 153.020556), latitude = c(-27.683333,-27.234167)), coords = c("longitude", "latitude"), crs = st_crs(states))

sa2.suburbs <- readOGR('.','SA2_2016_AUST') %>%
  st_as_sf() %>%
  mutate(SA2_MAIN16 = varhandle::unfactor(SA2_MAIN16)) %>%
  filter(GCC_NAME16 == "Greater Brisbane") %>%
  select(SA2_MAIN16, SA2_NAME16) %>%
  cbind(st_distance(st_centroid(.),central.station, by_element = TRUE))  %>%
  rename("sa2.to.central.km" = names(.)[3]) %>%
  mutate(sa2.to.central.km = as.numeric(sa2.to.central.km) /1000 )

labels <- sa2.suburbs %>%
  filter(SA2_NAME16 == "Brisbane City" | SA2_NAME16 == "Ipswich - Central" |
           SA2_NAME16 == "Logan Central" | SA2_NAME16 == "Redcliffe" | SA2_NAME16 == "Cleveland") %>% 
  mutate(SA2_NAME16 = str_replace(SA2_NAME16,"Brisbane City", "Brisbane"),
         SA2_NAME16 = str_replace(SA2_NAME16,"Ipswich - Central", "Ipswich"),
         SA2_NAME16 = str_replace(SA2_NAME16,"Logan Central", "Logan")) %>%
  select(SA2_NAME16)

study.area <- sa2.suburbs %>%
  st_union() %>%
  st_sf() %>%
  mutate(GCC_NAME16 = "Greater Brisbane")

sa2.suburbs.2016 <- read_csv("LongSA2UAI5P_MTWP2016.csv") %>%
  clean_names() %>%
  as_tibble() %>%
  fill(sa2_ur) %>%
  filter(five_years_ago == "Same as in 2016") %>% 
  mutate(SA2_MAIN16 = sa2_ur,
         driving.2016 = car_as_driver + motorbike_scooter,
         pnring.2016 = rowSums(select(.,contains("driver"), contains("motorbike"), -car_as_driver, -motorbike_scooter)),
         toding.2016 = rowSums(select(.,-contains("driver"), -contains("motorbike"), -contains("passenger"), -sa2_ur, -five_years_ago, -bicycle, -other, -walked_only, -worked_at_home, -did_not_go_to_work, -not_stated, -not_applicable, -total, -SA2_MAIN16))) %>% 
select(SA2_MAIN16, driving.2016, pnring.2016, toding.2016)  %>%
  mutate(driving.2016 = driving.2016 / rowSums(select(.,driving.2016, pnring.2016, toding.2016)) *100,
         pnring.2016 = pnring.2016 / rowSums(select(.,driving.2016, pnring.2016, toding.2016)) *100,
         toding.2016 = toding.2016 / rowSums(select(.,driving.2016, pnring.2016, toding.2016)) *100)

sa2.suburbs <- sa2.suburbs %>%  
  left_join(sa2.suburbs.2016, by = c("SA2_MAIN16", "SA2_MAIN16"))

sa2.suburbs.2011 <- read.csv("CG_SA2_2011_SA2_2016.csv")

sa2.suburbs.2011 <- read_csv("LongSA2UAI5P_MTWP2011.csv") %>%
  clean_names() %>%
  as_tibble() %>%
  fill(sa2_ur) %>%
  select(-five_years_ago) %>% 
  group_by(sa2_ur) %>% 
  summarise_each(funs(sum)) %>% 
  mutate(SA2_MAINCODE_2011 = sa2_ur,
         driving.2011 = car_as_driver + motorbike_scooter,
         pnring.2011 = rowSums(select(.,contains("driver"), contains("motorbike"), -car_as_driver, -motorbike_scooter)),
         toding.2011 = rowSums(select(.,-contains("driver"), -contains("motorbike"), -contains("passenger"), -sa2_ur, -bicycle, -other, -walked_only, -worked_at_home, -did_not_go_to_work, -not_stated, -not_applicable, -total, -SA2_MAINCODE_2011))) %>% 
select(SA2_MAINCODE_2011, driving.2011, pnring.2011, toding.2011) %>% 
  right_join(sa2.suburbs.2011, by = c("SA2_MAINCODE_2011", "SA2_MAINCODE_2011")) %>%
      mutate(driving.2011 = driving.2011 *  PERCENTAGE / 100,
         pnring.2011 = pnring.2011 * PERCENTAGE / 100,
         toding.2011 = toding.2011 * PERCENTAGE / 100,
         SA2_MAIN16 = SA2_MAINCODE_2016) %>% 
  select(-SA2_MAINCODE_2011, -PERCENTAGE, -SA2_MAINCODE_2016) %>%
  group_by(SA2_MAIN16) %>%
  summarise_each(funs(sum)) %>% 
  mutate(driving.2011 = driving.2011 / rowSums(select(.,driving.2011, pnring.2011, toding.2011)) *100,
         pnring.2011 = pnring.2011 / rowSums(select(.,driving.2011, pnring.2011, toding.2011)) *100,
         toding.2011 = toding.2011 / rowSums(select(.,driving.2011, pnring.2011, toding.2011)) *100)

sa2.suburbs <- sa2.suburbs %>%  
  left_join(sa2.suburbs.2011, by = c("SA2_MAIN16", "SA2_MAIN16")) %>% 
 mutate(driving.change = driving.2016 - driving.2011,
         pnring.change = pnring.2016 - pnring.2011,
         toding.change = toding.2016 -toding.2011)

PnR.parking <- readxl::read_xlsx("PnR_parking_geocoded.xlsx") %>%
  clean_names() %>%
  rename(park.and.ride = site, bays.2011 = x2011, bays.2016 = x2016) %>%
  mutate(change = ifelse(bays.2011 == 0, "new", ifelse(bays.2011 < bays.2016 & bays.2011 != 0, "increase", ifelse(bays.2011 > bays.2016, "decrease", "same")))) %>%
  mutate(change = factor(change, levels = c("new", "increase",  "same", "decrease"))) %>%
  mutate(change.bays = round(bays.2016 - bays.2011, 0)) %>%
  select(park.and.ride, bays.2011, bays.2016, change.bays, change, latitude, longitude) %>%
  as_tibble() %>%
  st_as_sf(coords = c("longitude", "latitude"), crs = st_crs(sa2.suburbs)) %>%
  st_intersection(sa2.suburbs) %>%
  cbind(st_distance(.,central.station, by_element = TRUE)) %>% 
  rename("platform.to.central.km" = names(.)[18]) %>% 
  mutate(platform.to.central.km = as.numeric(platform.to.central.km)/1000,
         nearest.pnr = as.numeric(rownames(.)))

PnR.table <- select(PnR.parking, nearest.pnr, park.and.ride, change.bays, bays.2011, bays.2016, change, platform.to.central.km) %>%
  st_set_geometry(NULL)

PnR.parking2 <- filter(PnR.parking, bays.2011 != 0)

sa2.suburbs <- cbind(sa2.suburbs, 
                 st_nearest_feature(st_centroid(sa2.suburbs), PnR.parking)) %>% 
  rename("nearest.pnr" = names(.)[13]) %>%
  left_join(PnR.table, by = c("nearest.pnr", "nearest.pnr"))

sa2.suburbs <- cbind(sa2.suburbs, 
                       st_distance(st_centroid(sa2.suburbs), PnR.parking, by_element = FALSE) %>%
                                     apply(1, FUN=min)) %>% 
  rename("nearest.pnr.km" = names(.)[20]) %>% 
  mutate(nearest.pnr.km = round(nearest.pnr.km/1000, 1),
         walkable.range = ifelse(nearest.pnr.km <= 0.8, "yes", "no"))

MB2016 <- read_csv("2016 census mesh block counts.csv") %>%
  janitor::clean_names() %>%
  select(mb_code_2016, dwelling, person) %>%
  as_tibble()

MB2016 <- readOGR('.','MB_2016_QLD') %>%
  st_as_sf() %>%
  filter(GCC_NAME16 == "Greater Brisbane") %>%
  mutate(mb_code_2016 = as.numeric(as.character(MB_CODE16))) %>%
  select(mb_code_2016) %>%
  st_centroid() %>% 
  left_join(MB2016, by = c("mb_code_2016", "mb_code_2016"))

MB2016 <- cbind(MB2016,
                st_distance(MB2016, PnR.parking, by_element = FALSE) %>%
                  apply(1, FUN=min)) %>%
  rename("km.from.rtn" = names(.)[4]) %>%
 mutate(km.from.rtn = km.from.rtn/1000)

MB.WALKING2016 <- filter(MB2016, km.from.rtn <= 0.8)

MB2011 <- read_csv("censuscounts_mb_2011_aust.csv") %>%
  janitor::clean_names() %>%
  select(mb_code_2011, dwelling, person) %>%
  as_tibble()

MB2011 <- readOGR('.','MB_2011_QLD') %>%
  st_as_sf() %>%
  filter(GCC_NAME11 == "Greater Brisbane") %>%
  mutate(mb_code_2011 = as.numeric(as.character(MB_CODE11))) %>%
  select(mb_code_2011) %>%
  st_centroid() %>% 
  left_join(MB2011, TABLE, by = c("mb_code_2011", "mb_code_2011"))

MB2011 <- cbind(MB2011,
                st_distance(MB2011, PnR.parking, by_element = FALSE) %>%
                  apply(1, FUN=min)) %>%
  rename("km.from.rtn" = names(.)[4]) %>%
 mutate(km.from.rtn = km.from.rtn/1000)

MB.WALKING2011 <- filter(MB2011, km.from.rtn <= 0.8)
tmap_mode('plot')
tm_shape(states, bbox = st_buffer(study.area, dist = 14)) +
  tm_borders(col= "gray27") +
  tm_shape(filter(states, STE_NAME16 == "Queensland" | STE_NAME16 == "New South Wales" | STE_NAME16 == "Victoria")) +
  tm_text("STE_NAME16", col="gray27", size = 0.8) +
   tm_shape(study.area) +
  tm_borders(col= "black") +
  tm_text("GCC_NAME16", col= "black", size = 0.8, ymod = 0, xmod = -2.9) +
  tm_compass(position = c("left", "bottom"))+
  tm_scale_bar(position = c("right", "bottom"), width = 0.4)
**Figure 5. Greater Brisbane, Queensland, Australia [without the base map included in published version]**

Figure 5. Greater Brisbane, Queensland, Australia [without the base map included in published version]

Much of Greater Brisbane’s post-war development occurred along its highway system, and its HOV network is ranked average by international standards in terms of coverage, frequency, and reliability (ARCADIS 2017: 11). Greater Brisbane comprises at least 380 kilometres of rail and since 2000 and a 25 kilometre BRT network operates along three radial lines emanating from the inner city (Transport and Main Roads 2019). In addition, Brisbane City Council has recently proposed an underground ‘metro’ system that will employ accordion busses rather than trains (BCC 2018). Besides high investments in transit, the current state government promised during its election campaign in 2017 to add to Greater Brisbane’s PnR. Consequently, PnR appeared in the 2018 budget (with $15 million allocated), and the construction of 2,300 new PnR parking bays commenced in 2019 (Moore 2018). With both the state and local government seeking to expand Brisbane’s transit network and its PnR, it is clear that the city-region is attempting to wean itself from high auto-dependency.

Data and analysis

To determine the extent to which changes and expansion in PnR facilities are converting Greater Brisbane motorists into park-and-riders, we draw on the Australian Bureau of Statistics’ 2011 and 2016 Census of Population and Housing that includes 234 different modal choices (Methods of Travel to Workplace) aggregated to 236 suburbs (defined using the 2016 Statistical Area Two census unit ) that nest wholly within the Greater Brisbane study frame. Between censuses, the suburb boundaries can shift across the two censuses we use thus we employ a correspondence table to concord the 2011 census tables to the 2016 suburb boundaries. When only population or dwelling counts are required, we employ blocks (defined using the 2016 MeshBlock unit ) that further nest within the suburbs to provide finer-grained dwelling locations but lack modal choice data. To minimise the influence of self-selection bias, workers should to be exposed to the changes at a given PnR and thus need to be present both in 2011 and 2016. As such, using the residential mobility question only workers that have indicated that they were at the same address five years earlier are retained from the 2016 census. Furthermore, for both simplicity and reproducibility, the 234 modal choices were filtered for keywords indicative of the choice between: (a) parking and riding; (b) driving only, or (c) riding HOV only. These counts were then converted to suburb proportions. In addition to modal choice, the proximity to the nearest PnR was measured, revealing that 77 percent or 181 out of all 236 suburbs were located beyond the typically regarded walkable range (of 800 metres) and thus arguably require a car or bus to access the PnR (Australian Bureau of Statistics, 2016; Transport and Main Roads 2019; Yang and Pojani 2017; Queensland Government 2009; Figure 6).

tmap_mode("plot")
  tm_shape(states, bbox = st_buffer(study.area, dist = 0.2)) +
  tm_borders(col= "gray27") +
  tm_shape(sa2.suburbs) +
  tm_polygons(col = "walkable.range",
              title = "within 800m from a PnR", 
              palette = yes.no,
              legend.hist = T,
              showNA = FALSE,
              colorNA = NULL) +
   tm_shape(study.area) +
  tm_borders(col= "black") +
  tm_shape(filter(labels, SA2_NAME16 != "Brisbane")) +
  tm_fill(alpha = 0) +
  tm_text("SA2_NAME16", col="black", size = 0.6, bg.color = "white", bg.alpha = 0.2, fontface="bold") +
  tm_shape(central.station) +
  tm_dots(size = 0.1, col = "white", border.col = "black", shape = 24) +
  tm_text("id", col="black", size = 0.6, bg.color = "white", bg.alpha = 0.2, fontface="bold", ymod = 0.4) +
  tm_compass(position = c("left", "bottom"))+
  tm_scale_bar(position = c("center", "bottom")) +
  tm_layout(frame = T,
            legend.position = c("center", "center"),
            legend.hist.size = 0.6,
            legend.hist.width = 0.7,
            legend.outside = T,
            legend.text.size = 0.7,
            legend.title.size = 0.9)
**Figure 6. Greater Brisbane's 236 suburbs defined according to the Australian Bureau of Statistics' 2016 Statistical Area Two geographic standard [without the base map included in published version]**

Figure 6. Greater Brisbane’s 236 suburbs defined according to the Australian Bureau of Statistics’ 2016 Statistical Area Two geographic standard [without the base map included in published version]

To observe change in the capacity of the 122 Greater Brisbane PnR (over the period 2011 to 2016), parking bay counts were supplied by the Queensland Department of Transport and Main Roads for the same period. PnR were geocoded and ground truthed using remotely sensed imagery. In addition to bay counts, proximity to Central Station is measured to operationalise the ‘centrality’ of the PnR given that parking and riding may lose appeal closer towards the inner-city where it entails travelling across traffic congestion (Figure 7). Further, the coverage and regularity of HOV services increases towards the inner-city and thus potentially makes for an increasingly viable alternative to driving or parking and riding. The results are reported below, and all mentions of ‘change’ specifically refer to change between 2011 and 2016.

tmap_mode("plot")
  tm_shape(states, bbox = st_buffer(study.area, dist = 0.2)) +
  tm_borders(col= "gray27") +
  tm_shape(PnR.parking) +
  tm_dots(col = "platform.to.central.km",
          size = 0.07,
          title = "km from Central Station",
          palette = divs,
          style= "fixed",
          breaks = c(0, 10, 20, 30, 40, 50, Inf),
          legend.hist = T,
          legend.format = list(digits = 1)) +
  tm_shape(study.area) +
  tm_borders(col= "black") +
  tm_shape(filter(labels, SA2_NAME16 != "Brisbane")) +
  tm_fill(alpha = 0) +
  tm_text("SA2_NAME16", col="black", size = 0.6, bg.color = "white", bg.alpha = 0.2, fontface="bold") +
  tm_shape(central.station) +
  tm_dots(size = 0.1, col = "white", border.col = "black", shape = 24) +
  tm_text("id", col="black", size = 0.6, bg.color = "white", bg.alpha = 0.2, fontface="bold", ymod = 0.4) +
  tm_compass(position = c("left", "bottom"))+
  tm_scale_bar(position = c("center", "bottom")) +
  tm_layout(frame = T,
            legend.position = c("center", "center"),
            legend.hist.size = 0.6,
            legend.hist.width = 0.7,
            legend.outside = T,
            legend.text.size = 0.7,
            legend.title.size = 0.9)
**Figure 7. The centrality of Greater Brisbane's 122 PnR [without the base map included in published version]**

Figure 7. The centrality of Greater Brisbane’s 122 PnR [without the base map included in published version]

Results

Greater Brisbane households are typically auto-dependent with just six percent living auto-free. Further, 91 percent of the population reside within single-family, low-density housing thus more likely to require an automobile to reach their dwelling. When examined at the block-level rather than suburb-level, the estimate for ‘auto-dependent’ dwellings located further than 800 meters from a rapid transit node in 2016 increases from 77 percent to 81 percent, and residents to 82 percent. Notably, there is only minimal change between the 2011 to 2016 estimates despite the rising implementation of Transit-Oriented Development (TOD) during this period. During this same period, Greater Brisbane’s net PnR capacity increased from 20,660 to 24,863 or by 17 percent (Transport and Main Roads 2019) yet parking and riding has declined slightly since 2011 while driving directly to work has increased. Next, capacity at individual PnR is examined, which reveals that capacity decreases typically outweigh increases yet there are also multiple new, large PnR that explain why Greater Brisbane’s net PnR capacity has increased (Figure 8). While these findings suggest that increasing PnR capacity is failing to convert motorists into park-and-riders, these are metropolitan-level findings thus a suburb-level examination alongside changes to nearby PnR may yield further findings.

ggplot(data = PnR.parking, mapping = aes(x = bays.2011, y = bays.2016, label = park.and.ride)) +
  geom_point(mapping = aes(color = change)) +
  scale_color_manual("parking supply", values= cols) +
  coord_fixed(ratio = 1, expand = TRUE, xlim = 0:1200, ylim = 0:1200) +
  theme_bw() +
  theme(legend.position = "right") +
  xlab("car bays in 2011") +
  ylab("car bays in 2016")
**Figure 8. Greater Brisbane PnR capacity: 2011 to 2016 change**

Figure 8. Greater Brisbane PnR capacity: 2011 to 2016 change

Mapping change to PnR capacity reveals that capacity is decreasing towards Ipswich in the west, while new PnR are emerging towards the southwest and northeast (Figure 9). Notably the new facilities are located towards two recent large-scale master planned estates: Springfield Lakes and North Lakes thus suggesting that rather than TOD, these two master-plans are orientated around auto-dependency (Figure 10).

tmap_mode("plot")
tm_shape(states, bbox = st_buffer(study.area, dist = 0.2)) +
  tm_borders(col= "gray27") +
  tm_shape(PnR.parking) +
  tm_dots(col = "change",
          palette = cols,
          size = 0.07,
          title = "parking supply",
          legend.hist = T,
          legend.format = list(digits = 1)) +
  tm_shape(study.area) +
  tm_borders(col= "black") +
  tm_shape(filter(labels, SA2_NAME16 != "Brisbane")) +
  tm_fill(alpha = 0) +
  tm_text("SA2_NAME16", col="black", size = 0.6, bg.color = "white", bg.alpha = 0.2, fontface="bold") +
  tm_shape(central.station) +
  tm_dots(size = 0.1, col = "white", border.col = "black", shape = 24) +
  tm_text("id", col="black", size = 0.6, bg.color = "white", bg.alpha = 0.2, fontface="bold", ymod = 0.4) +
  tm_compass(position = c("left", "bottom"))+
  tm_scale_bar(position = c("center", "bottom")) +
  tm_layout(frame = T,
            legend.position = c("center", "center"),
            legend.hist.size = 1,
            legend.hist.width = 0.7,
            legend.outside = T,
            legend.text.size = 0.7,
            legend.title.size = 0.9)
**Figure 9. Greater Brisbane PnR capacity: 2011 to 2016 change [without the base map included in published version]**

Figure 9. Greater Brisbane PnR capacity: 2011 to 2016 change [without the base map included in published version] 

tmap_mode("plot")
tm_shape(states, bbox = st_buffer(study.area, dist = 0.2)) +
  tm_borders(col= "gray27") +
  tm_shape(study.area) +
  tm_borders(col= "black") +
  tm_shape(filter(PnR.parking, change == "new")) +
  tm_bubbles(col = cols[1],
          border.col = "black",
          shape = 21,
          size = 'bays.2016', title.size = "bays within new park and rides",
          legend.size.is.portrait = TRUE) +
  tm_shape(the.lakes) +
  tm_dots(col = "black", size = 0.07) +
  tm_text("id", col="black", size = 0.6, bg.color = "white", bg.alpha = 0.2,fontface="bold") +
  tm_shape(central.station) +
  tm_dots(size = 0.1, col = "white", border.col = "black", shape = 24) +
  tm_text("id", col="black", size = 0.6, bg.color = "white", bg.alpha = 0.2, fontface="bold", ymod = 0.4) +
  tm_compass(position = c("left", "bottom"))+
  tm_scale_bar(position = c("center", "bottom")) +
  tm_layout(frame = T,
            legend.position = c("center", "center"),
            legend.outside = T)
**Figure 10. Greater Brisbane PnR capacity: bays per new PnR in 2016 [without the base map included in published version]**

Figure 10. Greater Brisbane PnR capacity: bays per new PnR in 2016 [without the base map included in published version]

Next suburb-level modal choice is examined alongside change to closest park and ride facilities (Figure 11). Notably, parking and riding has generally declined throughout Greater Brisbane except within suburbs located near new PnR within 10 kilometres of Central Station. Given this general decline, it is necessary to examine changes to workers that are riding HOV only and that are driving only. Similar to the general decline for parking and riding, there is a general decline for riding HOV only and particularly amongst suburbs located closer to Central Station. Interestingly, this general decline flattens towards Central Station for suburbs located near new PnR. It is plausible that some of these new facilities could be drop offs locations for feeder HOV services or new entrances to busways. Last, driving only has a general incline that corresponds to the general decline for HOV. This suggests that HOV only riders are converting to driving only, and particularly where nearby PnR capacity has remained similar and less so where new facilities have emerged. While changes to PnR could in part explain these changes to modal choice, it is notable that the inner-city network of bypass tunnels and bridges has also expanded throughout this period and potentially made driving appealing once more. As such, it is necessary to examine where changes to modal choice are occurring.

P1 <- ggplot(data = filter(sa2.suburbs, pnring.change < 5), mapping = aes(x = sa2.to.central.km, y = pnring.change, label = park.and.ride)) + geom_point(mapping = aes(color = change, alpha = 0.4, stroke = 0, shapes=20)) +
  scale_color_manual(values= cols) +
  geom_smooth(data = filter(sa2.suburbs, pnring.change < 5 & change == "new"), method = "loess", col = cols[1], span = 2, se = FALSE) +
  geom_smooth(data = filter(sa2.suburbs, pnring.change < 5 & change == "increase"), method = "loess", col = cols[2], span = 2, se = FALSE) +
  geom_smooth(data = filter(sa2.suburbs, pnring.change < 5 & change == "same"), method = "loess", col = cols[3], span = 2, se = FALSE) +
  geom_smooth(data = filter(sa2.suburbs, pnring.change < 5 & change == "decrease"), method = "loess", col = cols[4], span = 2, se = FALSE) +
  geom_hline(yintercept=0, linetype="dotted") +
  theme_bw() +
  theme(legend.position = "none", aspect.ratio = 1, plot.title = element_text(size=12,face = "bold",hjust = 0.5)) +
  xlab("km from Central Station") +
  ylab("%change") +
  labs(title= "parking and riding")

P2 <- ggplot(data = sa2.suburbs, mapping = aes(x = sa2.to.central.km, y = toding.change, label = park.and.ride)) +
  geom_point(mapping = aes(color = change, alpha = 0.4, stroke = 0, shapes=20)) +
  scale_color_manual(values= cols) +
  geom_smooth(data = filter(sa2.suburbs, change == "new"), method = "loess", col = cols[1], span = 2, se = FALSE) +
  geom_smooth(data = filter(sa2.suburbs, change == "increase"), method = "loess", col = cols[2], span = 2, se = FALSE) +
  geom_smooth(data = filter(sa2.suburbs, change == "same"), method = "loess", col = cols[3], span = 2, se = FALSE) +
  geom_smooth(data = filter(sa2.suburbs, change == "decrease"), method = "loess", col = cols[4], span = 2, se = FALSE) +
  geom_hline(yintercept=0, linetype="dotted") +
  theme_bw() +
  theme(legend.position = "none", aspect.ratio = 1, plot.title = element_text(size=12,face = "bold",hjust = 0.5)) +
  xlab("km from Central Station") +
  ylab("%change") +
  labs(title= "public transport only")

P3 <- ggplot(data = sa2.suburbs, mapping = aes(x = sa2.to.central.km, y = driving.change, label = park.and.ride)) +
  geom_point(mapping = aes(color = change, alpha = 0.4, stroke = 0, shapes=20)) +
  scale_color_manual(values= cols) +
  geom_smooth(data = filter(sa2.suburbs, change == "new"), method = "loess", col = cols[1], span = 2, se = FALSE) +
  geom_smooth(data = filter(sa2.suburbs, change == "increase"), method = "loess", col = cols[2], span = 2, se = FALSE) +
  geom_smooth(data = filter(sa2.suburbs, change == "same"), method = "loess", col = cols[3], span = 2, se = FALSE) +
  geom_smooth(data = filter(sa2.suburbs, change == "decrease"), method = "loess", col = cols[4], span = 2, se = FALSE) +
  geom_hline(yintercept=0, linetype="dotted") +
  theme_bw() +
  theme(legend.position = "none", aspect.ratio = 1, plot.title = element_text(size=12,face = "bold",hjust = 0.5)) +
  xlab("km from Central Station") +
  ylab("%change") +
  labs(title= "driving only")

L1 <- ggplot(data = sa2.suburbs, mapping = aes(x = sa2.to.central.km, y = driving.change, label = park.and.ride)) +
  geom_point(mapping = aes(color = change)) +
  scale_color_manual(values= cols) +
  labs(colour="parking supply") +
  theme_bw() +
  theme(legend.position = "right")

L1 <- cowplot::get_legend(L1)
grid.newpage()
cowplot::plot_grid(P1, P2, P3, L1)
**Figure 11. The association between 2011 to 2016 change to Greater Brisbane PnR parking supply, centrality, and modal choice**

Figure 11. The association between 2011 to 2016 change to Greater Brisbane PnR parking supply, centrality, and modal choice

Mapping the changes to park and riding throughout Greater Brisbane does not reveal any distinct spatial concentrations of change (Figure 12) thus it is necessary to draw focus towards the inner-city (Figure 13). Interestingly, parking and riding is increasing within the suburbs closest to PnR that are not the intended park and riders given that they are located within walkable distance of the rapidHOV. In addition, parking and riding is often decreasing within suburbs near PnR located midway from Central Station, which could suggest that outer-city park and riders are exhausting midway PnR rather than closer to home outer-city PnR, which is termed ‘rail heading’ and occurs when workers intend to maximise the portion of their journey to work within the comfort of their own car rather than a public train or bus. This behaviour is problematic given that PnR are intended to keep cars parked in the outer-city where parking space is cheaper, and driving distances may be minimised (Buxton and Parkhurst, 2005).

tmap_mode("plot")
tm_shape(states, bbox = st_buffer(study.area, dist = 0.2)) +
  tm_borders(col= "gray27") +
  tm_shape(filter(sa2.suburbs, pnring.change < 10)) +
  tm_polygons(col = "pnring.change",
              title = "% change: parking and riding", 
              palette = "RdYlGn",
              n = 9,
              midpoint = 0,
              style= "quantile",
              legend.hist = T,
              legend.format = list(digits = 1),
              showNA = FALSE,
              colorNA = NULL) +
   tm_shape(study.area) +
  tm_borders(col= "black") +
  tm_shape(PnR.parking) +
  tm_dots(col = "gray20",
          size = 0.07) +
  tm_shape(filter(labels, SA2_NAME16 != "Brisbane")) +
  tm_fill(alpha = 0) +
  tm_text("SA2_NAME16", col="black", size = 0.6, bg.color = "white", bg.alpha = 0.2, fontface="bold") +
  tm_shape(central.station) +
  tm_dots(size = 0.1, col = "white", border.col = "black", shape = 24) +
  tm_text("id", col="black", size = 0.6, bg.color = "white", bg.alpha = 0.2, fontface="bold", ymod = 0.4) +
  tm_compass(position = c("left", "bottom"))+
  tm_scale_bar(position = c("center", "bottom")) +
  tm_layout(frame = T,
            legend.position = c("center", "center"),
            legend.hist.size = 0.6,
            legend.hist.width = 0.7,
            legend.outside = T,
            legend.text.size = 0.7,
            legend.title.size = 0.9)
**Figure 12. Greater Brisbane PnR: 2011 to 2016 percent change [without the base map included in published version]**

Figure 12. Greater Brisbane PnR: 2011 to 2016 percent change [without the base map included in published version]

Next, workers riding HOV only are examined, which notably reveals that the general decline is occurring along rapid transit corridors where we assume that TOD is occurring and walking to rapidHOV is more convenient (Figure 14). If TOD is indeed occurring along these corridors, then it is plausible that absolute figures for rapidHOV ridership have remained constant while the proportional figures have declined due to an expanding population.

tmap_mode("plot")
tm_shape(states, bbox = st_buffer(central.station, dist = 0.4)) +
  tm_borders(col= "gray27") +
  tm_shape(filter(sa2.suburbs, pnring.change < 10)) +
  tm_polygons(col = "pnring.change",
              title = "% change: parking and riding", 
              palette = "RdYlGn",
              n = 9,
              midpoint = 0,
              style= "quantile",
              legend.hist = T,
              legend.format = list(digits = 1),
              showNA = FALSE,
              colorNA = NULL) +
  tm_shape(study.area) +
  tm_borders(col= "black") +
  tm_shape(PnR.parking) +
  tm_dots(col = "gray20",
          size = 0.07) +
  tm_shape(filter(labels, SA2_NAME16 != "Brisbane")) +
  tm_fill(alpha = 0) +
  tm_text("SA2_NAME16", col="black", size = 0.6, bg.color = "white", bg.alpha = 0.2, fontface="bold") +
  tm_shape(central.station) +
  tm_dots(size = 0.1, col = "white", border.col = "black", shape = 24) +
  tm_text("id", col="black", size = 0.6, bg.color = "white", bg.alpha = 0.2, fontface="bold", ymod = 0.4) +
  tm_compass(position = c("left", "bottom"))+
  tm_scale_bar(position = c("center", "bottom")) +
  tm_layout(frame = T,
            legend.position = c("center", "center"),
            legend.hist.size = 0.6,
            legend.hist.width = 0.7,
            legend.outside = T,
            legend.text.size = 0.7,
            legend.title.size = 0.9)
**Figure 13. Inner-Brisbane PnR: 2011 to 2016 percent change [without the base map included in published version]**

Figure 13. Inner-Brisbane PnR: 2011 to 2016 percent change [without the base map included in published version] 

tmap_mode("plot")
  tm_shape(states, bbox = st_buffer(central.station, dist = 0.4)) +
  tm_borders(col= "gray27") +
  tm_shape(sa2.suburbs) +
  tm_polygons(col = "toding.change",
              title = "% change: HOV only", 
              palette = "RdYlGn",
              n = 9,
              midpoint = 0,
              style= "quantile",
              legend.hist = T,
              legend.format = list(digits = 1),
              showNA = FALSE,
              colorNA = NULL) +
  tm_shape(study.area) +
  tm_borders(col= "black") +
  tm_shape(PnR.parking) +
  tm_dots(col = "gray20",
          size = 0.07) +
  tm_shape(filter(labels, SA2_NAME16 != "Brisbane")) +
  tm_fill(alpha = 0) +
  tm_text("SA2_NAME16", col="black", size = 0.6, bg.color = "white", bg.alpha = 0.2, fontface="bold") +
  tm_shape(central.station) +
  tm_dots(size = 0.1, col = "white", border.col = "black", shape = 24) +
  tm_text("id", col="black", size = 0.6, bg.color = "white", bg.alpha = 0.2, fontface="bold", ymod = 0.4) +
  tm_compass(position = c("left", "bottom"))+
  tm_scale_bar(position = c("center", "bottom")) +
  tm_layout(frame = T,
            legend.position = c("center", "center"),
            legend.hist.size = 0.6,
            legend.hist.width = 0.7,
            legend.outside = T,
            legend.text.size = 0.7,
            legend.title.size = 0.9)
**Figure 14. Inner-Brisbane High Occupancy Vehicle (HOV) Use: 2011 to 2016 percent change [without the base map included in published version]**

Figure 14. Inner-Brisbane High Occupancy Vehicle (HOV) Use: 2011 to 2016 percent change [without the base map included in published version]

Last, workers driving only are examined, which again reveals a pattern that corresponds to workers that are riding HOV only thus suggesting that HOV riders along the rapidHOV corridors are converting to motorists (Figure 15). This finding is problematic given that TOD is occurring along these corridors and yet workers in these areas are converting to motorists.

tmap_mode("plot")
tm_shape(states, bbox = st_buffer(central.station, dist = 0.4)) +
  tm_borders(col= "gray27") +
  tm_shape(sa2.suburbs) +
  tm_polygons(col = "driving.change",
              title = "% change: driving", 
              palette = "RdYlGn",
              n = 9,
              midpoint = 0,
              style= "quantile",
              legend.hist = T,
              legend.format = list(digits = 1),
              showNA = FALSE,
              colorNA = NULL) +
  tm_shape(study.area) +
  tm_borders(col= "black") +
  tm_shape(PnR.parking) +
  tm_dots(col = "gray20",
          size = 0.07) +
  tm_shape(filter(labels, SA2_NAME16 != "Brisbane")) +
  tm_fill(alpha = 0) +
  tm_text("SA2_NAME16", col="black", size = 0.6, bg.color = "white", bg.alpha = 0.2, fontface="bold") +
  tm_shape(central.station) +
  tm_dots(size = 0.1, col = "white", border.col = "black", shape = 24) +
  tm_text("id", col="black", size = 0.6, bg.color = "white", bg.alpha = 0.2, fontface="bold", ymod = 0.4) +
  tm_compass(position = c("left", "bottom"))+
  tm_scale_bar(position = c("center", "bottom")) +
  tm_layout(frame = T,
            legend.position = c("center", "center"),
            legend.hist.size = 0.6,
            legend.hist.width = 0.7,
            legend.outside = T,
            legend.text.size = 0.7,
            legend.title.size = 0.9)
**Figure 15. , Inner Brisbane Driving Only: 2011 to 2016 percent change [without the base map included in published version]**

Figure 15. , Inner Brisbane Driving Only: 2011 to 2016 percent change [without the base map included in published version]

In sum, these research findings reveal that 81 percent of Greater Brisbane households in 2016 were located within auto-dependent areas, which potentially explains why only six percent of households are auto-free. The PnR net capacity increased by 17 percent from 2011 to 2016 and these increases typically concentrated near emerging master planned estates. This suggests that these new PnR are intended to reduce driving from newly created auto-dependent suburbs rather than the already created auto-dependent suburbs. In addition, increases to parking and riding are typically occurring closer to the inner-city and near new PnR, which is problematic given that these riders are generally located within walkable range of a PnR and also better serviced by other HOV services. Last, workers that formerly rode HOV only to work are converting to driving along the rapidHOV corridors thus suggesting that HOV services are losing appeal or driving is gaining appeal, which potentially may be explained by the increased number of inner-city bypass tunnels and bridges, or an induced demand problem if converting outer-city motorists to HOV has cleared the roads for inner-city motorists.

Discussion and Conclusion

Theoretically, the provision of PnR should influence modal choice by making trip-chaining between cars and rapid transit more convenient. However, the base assumptions of PnR theory and the empirical realities of PnR do not match.

Our review of the PnR literature suggests that the purpose of PnR was initially motorist-centric to reduce the expenses associated with driving but as the new realist planning paradigm emerged, the purpose of PnR increasingly became network-centric as PnR could reduce demand for inner-city road and parking capacity. Next, our examination of PnR typologies suggests that the function of PnR is determined by: proximity to the CBD; whether the PnR is formal or informal and whether the PnR connects commuters to other cities. Our review of empirical examinations of the relationship between PnR and modal choice revealed a fragmented literature of incomparable research methods and findings (e.g. expert criteria, intercept surveys, demand analyses and agent-based simulations) that typically either: examine the motivations of current park and riders (i.e. perceived safety at the PnR; the social status of PnR users and transit riders; and the comfort of driving relative to riding) or assumes that motorists are perfectly rational actors that will become park and riders upon encountering a convenient PnR (i.e. reduces travel time or costs).

While current understandings of whether and how PnR influence modal choice remain severely limited, our synthesis suggests that the following PnR considerations should include its: (a) urban centrality; (b) proximity to households; (c) proximity to traffic choke points; (d) placement within pre-existing road capacity and networks; (e) proximity to expressway intersections; (e) advantage relative to driving routes such as expressway; (f) advantage relative to walking, cycling, or riding HOV to the PnR; and (g) savings relative to inner-city parking.

In addition, our synthesis introduced reliability as a modal choice concern therefore a consideration when exploring the influence of PnR given that travel-time, connectivity, capacity and parking reliability cumulatively determine the appeal of PnR relative to the alternatives, and the amount of slack time required according to modal choice to ensure that commuters can confidently arrive at work on time. Last, our synthesis of provided the foundations for developing a new integrative model of PnR, multi-modalism and modal choice that delineates the urban form with BED that potentially influence the appeal of multi-modalism and the optimal modal choice, which include inter-city, inner/outer-city, driving/parking, parking/cycling, and parking/walking BED, and in turn enclose driving sheds, parking sheds, riding sheds, cycling sheds, and walking sheds. As such, this modal has the capacity to integrate a broad range of transport and new realist planning paradigms including parking and riding, transit orientated development, and walkability.

Upon laying these theoretical foundations, we examined whether developing or modifying PnR does indeed influence modal choice. We found that auto-dependency has increased throughout Greater Brisbane given that larger proportion of residents live further than 800 meters from rapidHOV node in 2016 than 2011. Given this increased auto-dependency, demand for PnR should have heightened yet what was observed is that park and riding is declining relative to driving. Further, while PnR bays have increased throughout Greater Brisbane, these are typically located within new PnR located near new large housing estates thus suggesting these PnR were intended to accommodate new residents rather than convert long-term residents.

Given this spatial finding, we next examined whether changes to individual PnR was influencing the modal choices of nearby neighbourhoods. This examination revealed that increasing or decreasing PnR capacity has no discernible influence over modal choice; however introducing new PnR converts nearby residents into park and riders, which is problematic given that these motorists may be living within cycling or walking distance of the PnR and thus not the intended users. This influence was particularly evident closer to the inner-city, which is particularly problematic if the new ridership is drawn from commuters that previously rode HOV directly to work and now prefer to drive park way to access rapidHOV services. Last, mapping changes to modal choice confirmed that parking and riding is typically on the rise nearer to rapid transport nodes and so too is driving direct to the workplace, while relying only on HOV is declining within these spaces. As such, these findings could suggest that nearby motorists are arriving first at the PnR and once the parking supply is exhausted, continuing on to their workplaces. If this is the case, then this may explain why motorists that live further from PnR and therefore require the greatest slack time when choosing to park and ride are instead choosing to drive direct.

These findings could have important implications for transport and land-use planning since given that it is new PnR rather modified PnR that are influencing modal choice thus suggesting motorists may be unaware of changes unless the changes are as apparent as a new facility. Further, these findings suggest that popular PnR should be restricted to more distantly located households so that there is a reliable supply of parking when they take the risk of diverting their path towards a PnR.

There are multiple avenues for further research emerging from our study. For instance, demand analyses and simulations could introduce the decision-making criteria that we have synthesised from the broader PnR literature such as perceived safety, comfort-seeking and reliability. Further, transport research more broadly could introduce the break-even distances and mode-sheds synthesised and developed within our literature review for developing a more integrative understanding of urban transportation systems. In addition, these studies could examine the spatial association between where PnR users reside and their chosen PnR rather than their nearest PnR should suitable data become available. Last, the occupancy rate of PnR parking lots could be examined given that full PnR lots and even PnR overspill parking could potentially explain why many PnR lots have no discernible influence on modal choice. By working together to deepen the theoretical and empirical understandings of PnR, researchers, planners and policy makers will be better equipped to smooth the urban transition from high auto-dependency towards alternate modes such as HOV, cycling and walking, and thereby improve the reliability, efficiency and sustainability of urban mobility and the liveability of future cities.

References

Abdel-Aty, M.A., Kitamura, R., Jovanis, P.P. 1995. Investigating Effect of Travel Time Variability on Route Choice using Repeated-Measurement Stated Preference Data. Washington, DC: Transportation Research Board.

Adnan, S., Alyia, S.A., Hamsa, K., Azeez, A., 2012. “A literature review on the parking demand of park and ride facilities.” In: Conference on Urban Planning and Management in Malaysia, Kuala Lumpur, Malaysia.

Alonso, W., 1964. Location and Land Use. Cambridge, MA: Harvard University Press.

ARCADIS, 2017. Chasing Urban Mobility: Moving towards a Connected Sustainable Future. Retrieved 27 September 2018 from https://www.arcadis.com/media/6/4/F/%7B64FC28A4-6215-4E4C-88DA-FC01AE2B849B%7DAAP_MOBILITY%20INDEX_spread.pdf

Asakura, Y., Kashiwadani, M., 1991. “Road network reliability caused by daily fluctuation of traffic flow”. In PTRC Summer Annual Meeting, 19th, University of Sussex, UK.

Ashmore, D. Pojani, D. Thoreau, R., Christie, N., Tyler, N. 2019. ‘Gauging differences in public transport symbolism across national cultures: implications for policy development and transfer.’ Journal of Transport Geography [forthcoming].

Australian Bureau of Statistics, 2011. Census of Population and Housing. Retrieved 28 March 2018 from http://www.abs.gov.au/websitedbs/censushome.nsf/home/datapacks

Australian Bureau of Statistics, 2016. Census of Population and Housing. Retrieved 28 March 2018 from http://www.abs.gov.au/websitedbs/censushome.nsf/home/datapacks.

Bolger, D., Colquhoun, D., Morrall, J., 1992. “Planning and Design of Park-and-Ride Facilities for the Calgary Light Rail Transit System” Transportation Research Record 2:141-148.

Bos, I.D.M., Van der Heijden, R.E.C.M., Molin, E.J.E., Timmermans, H.J.P., 2004. “The Choice of Park and Ride Facilities: An Analysis Using a Context-Dependent Hierarchical Choice Experiment”. Environment and Planning A 36:1673-1686.

Bos, I., Molin, E., Timmermans, H., van der Heijden, R., 2003. “Cognition and relative importance underlying consumer valuation of park-and-ride facilities”. Transportation Research Record 1835:121-127.

Boyce, D.E., Allen, B., Mudge, R.R., Slater, P.B., Isserman, A.M., 1972. Impact of Rapid Transit on Suburban Residential Property Values and Land Development. Philadelphia, PA: University of Pennsylvania

Brisbane City Council, 2018. Brisbane Metro Industry Briefing. Retrieved 28 March 2018 from https://www.brisbane.qld.gov.au/sites/default/files/20180803-brisbane-metro-industry-briefing-presentation-july-2018.pdf

Cervero, R., 2004. Transit-Oriented Development in the United States: Experiences, Challenges, and Prospects. Washington, D.C: Transportation Research Board.

Chang, D.J., Morlok, E.K., 2005. “Vehicle speed profiles to minimize work and fuel consumption”. Journal of Transportation Engineering 131:173-182.

Chen, A., Yang, H., Lo, H.K., Tang, W.H., 1999. “A capacity related reliability for transportation networks”. Journal of Advanced Transportation 33:183-200.

Chen, C., Xia, J., Smith, B., Olaru, D., Taplin, J., Han, R., 2017. “Influence of travel time variability on train station choice for park-and-rider users”. Transportation Research Procedia 25:2473-2489.

Chen, Z., Xia, J., Irawan, B., Caulfied, C., 2014. “Development of location-based services for recommending departure stations to park and ride users”. Transportation Research Part C: Emerging Technologies 48:256-268.

Christiansen, P., Engebretsen, Ø., Fearnley, N., Hanssen, J.U., 2017. “Parking facilities and the built environment: Impacts on travel behaviour”. Transportation Research Part A: Policy and Practice 95, 198-206.

de Jong, G., Daly, A., Pieters, M., Vellay, C., Bradley, M., Hofman, F., 2003. “A model for time of day and mode choice using error components logit. Transportation Research Part E: Logistics and Transportation Review 39(3), 245-268.

de Nazelle, A., Morton, B.J., Jerrett, M., Crawford-Brown, D., 2010. Short trips: An opportunity for reducing mobile-source emissions? Transportation Research Part D: Transport and Environment 15(8):451-457.

Department of Transport and Main Roads, 2019. Busways. Retrieved 14 March 2018 from https://www.tmr.qld.gov.au/busways.aspx.

Dickins, I.S.J., 1991. “Park and ride facilities on light rail transit systems”. Transportation 18, 23-36.

Dijk, M., Parkhurst, G., 2014. “Understanding the mobility-transformative qualities of urban park and ride polices in the UK and the Netherlands”. International Journal of Automotive Technology and Management 4:246-270.

Dovey, K., Woodcock, I., Pike, L., 2015. “Intensifying Melbourne: Transit-oriented urban design for resilient urban futures”. Planning News 41:10-11.

Duncan, M., Christensen, R.K., 2013. “An analysis of park-and-ride provision at light rail stations across the US”. Transport Policy 25:148-157.

Duncan, M., Cook, D., 2014. “Is the provision of park-and-ride facilities at light rail stations an effective approach to reducing vehicle kilometres travelled in a US context?”. Transportation Research Part A: Policy and Practice 66:65-74.

Farhan, B., Murray, A.T., 2008. “Siting park-and-ride facilities using a multi-objective spatial optimization model.” Computers & Operations Research 35.2: 445-456.

Gaver, D.P., 1968. “Headstart strategies for combating congestion.” Transportation Science 2(2):172-181.

Ginn, S., 2009. “The application of the park & ride and TOD concepts to develop a new framework that can maximise public transport patronage” Brisbane: Queensland University of Technology Dissertation.

Giuliano, G., 2004. “Land use impacts of transportation investments”. The Geography of Urban Transportation 3:237-273.

Guerra, E., Cervero, R., Tischler, D., 2012. “Half-mile circle: Does it best represent transit station catchments?”. Transportation Research Record 2276.1 (2012): 101-109.

Heggie, I.G., Papoulias, D.B., 1976. “Operational Performance of Park and Ride: Objectives and Achievements in Oxford”. Working Paper 23 Transport Studies Unit, University of Oxford.

Hendricks, S., Outwater, M., 1998. “Demand forecasting model for park-and-ride lots in King County, Washington”. Transportation Research Record 1623.1: 80-87.

Hensher, D.A., Rose, J.M., 2007. “Development of commuter and non-commuter mode choice models for the assessment of new public transport infrastructure projects: A case study”. Transportation Research Part A: Policy and Practice 41(5):428-443.

Hess, P.M., Vernez Moudon, A., Catherine Snyder, M., Stanilov, K., 1999. “Site Design and Pedestrian Travel”. Transportation Research Record 1674:9-19.

Holguin-Veras, J., Yushimito, W.F., Aros-Vera, F., Reilly, J., 2012. “User rationality and optimal park-and-ride location under potential demand maximization”. Transportation Research Part B: Methodological 46:949-970.

Horner, M.W., Groves, S., 2007. “Network flow-based strategies for identifying rail park-and-ride facility locations”. Socio-Economic Planning Sciences 41:255-268.

Horner, M.W., Grubesic, T.H., 2001. “A GIS-based planning approach to locating urban rail terminals”. Transportation 28:55-77.

Huntley, P., 1993. “Park and ride: strategic modal shift or cheap overflow”. Local Transport Today 22 July 12-13.

Iida, Y., Wakabayashi, H., 1989. “An approximation method of terminal reliability of road network using partial minimal path and cut sets” in: Transport Policy, Management & Technology towards 2001: Selected Proceedings of 5th World Conference on Transport Research.

Israel, E., Cohen-Blankshtain, G., 2010. “Testing the decentralization effects of rail systems: Empirical findings from Israel”. Transportation Research Part A: Policy and Practice 44:523-536.

Jackson, W.B., Jucker, J. V., 1982. “An empirical study of travel time variability and travel choice behavior”. Transportation Science 16:460-475.

Jensen, S.S., 1995. “Driving patterns and emissions from different types of roads”. Science of the Total Environment 169:123-128.

Karamychev, V., van Reeven, P., 2011. “Park-and-ride: Good for the city, good for the region?”. Regional Science and Urban Economics 41:455-464.

Keck, C.A., Liou, P.S., 1976. “Forecasting demand for peripheral park-and-ride service”. Transportation Research Record 563.

Kessler, J., Schroeer, W., 1995. “Meeting mobility and air quality goals: Strategies that work”. Transportation 22:241-272.

Kuby, M., Barranda, A., Upchurch, C., 2004. “Factors influencing light-rail station boardings in the United States”. Transportation Research Part A: Policy and Practice 38:223-247.

Lai, L., Shalaby, A.S., 2007. “Cellular automaton simulation of vehicle dynamics in park-and-ride facilities”. Transportation Research Board 86th Annual Meeting Compendium of Papers.

Lane, C., DiCarlantonio, M., Usvyat, L., 2006. “Sketch Models to Forecast Commuter and Light Rail Ridership”. Transportation Research Record 1986(1):198-210.

Lehrer, U.A., 1994. “Images of the Periphery: The Architecture of Flex Space in Switzerland”. Environment and Planning D: Society and Space 12:187-205.

Li, Z.-C., Lam, W.H.K., Wong, S.C., 2012. “Modelling intermodal equilibrium for bimodal transportation system design problems in a linear monocentric city”. Transportation Research Part B: Methodological 46:30-49.

Li, Z.-C., Lam, W., Wong, S., Zhu, D.-L., Huang, H.-J., 2007. “Modelling park-and-ride services in a multimodal transport network with elastic demand”. Transportation Research Record 1994(1):101-109.

Liu, W., Yang, H., Yin, Y., 2014. “Expirable parking reservations for managing morning commute with parking space constraints”. Transportation Research Part C: Emerging Technologies 44:185-201.

Liu, Z., Meng, Q., 2014. “Bus-based park-and-ride system: a stochastic model on multimodal network with congestion pricing schemes”. International Journal of Systems Science 45(5):994-1006.

Lo, H.K., Luo, X.W., Siu, B.W.Y., 2006. “Degradable transport network: Travel time budget of travellers with heterogeneous risk aversion”. Transportation Research Part B: Methodological 40:792-806.

Lo, H.K., Tung, Y.-K., 2003. “Network with degradable links: capacity analysis and design”. Transportation Research Part B: Methodological 37:345-363.

Meek, S., Ison, S.G., Enoch, M.P., 2008. “Park and ride: Lessons from the UK experience”. Transportation Research Board 87th Annual Meeting, 13-17 January 2008, Washington DC., Paper No. 08-0730.

Meek, S., Ison, S., Enoch, M., 2009. “Stakeholder perspectives on the current and future roles of UK bus-based Park and Ride”. Journal of Transport Geography 17:468-475.

Meek, S., Ison, S., Enoch, M., 2011. “Evaluating alternative concepts of bus-based park and ride”. Transport Policy 18:456-467.

Mingardo, G., 2013. “Transport and environmental effects of rail-based Park and Ride: evidence from the Netherlands”. Journal of Transport Geography 30:7-16. Moore, T., 2018. Where Brisbane’s New Park and Ride Stations will be Built from 2019. Retrieved 14 March 2018 from: https://www.brisbanetimes.com.au/national/queensland/where-brisbane-s-new-park-and-ride-stations-will-be-built-from-2019-20181011-p5095s.html.

Noel, E.C., 1988. “Park-and-ride: alive, well, and expanding in the United States”. Journal of Urban Planning and Development 114(1): 2-13.

Norlida, A. H., Jamilah, M., and Karim, M.R. 2007 “Parking Duration of Fringe Park-and-Ride Users and Delineation of Stations Catchment Area: Case of the Kuala Lumpur Conurbation.” In: Proceedings of the Eastern Asia Society for Transportation Studies.

Olowoporoku, D., Hayes, E., Longhurst, J., Parkhurst, G., 2012. “The rhetoric and realities of integrating air quality into the local transport planning process in English local authorities”. Journal of Environmental Management 101: 23-32.

O’Sullivan, S., Morrall, J., 1996. “Walking distances to and from light-rail transit stations”. Transportation Research Record 1538(1): 19-26. Pang, H., Khani, A., 2018. “Modelling park-and-ride location choice of heterogeneous commuters”. Transportation 45: 71-87.

Parkhurst, G., 2000. “Influence of bus-based park and ride facilities on users’ car traffic”. Transport Policy 7:159-172.

Parkhurst, G., Buxton, S., 2005. “Towards a truly strategic appraisal process for evaluating park and ride schemes: The development of an assessment framework for London”. In: Proceedings of the 3rd UK Transport Practitioners Meeting, Aston University, Birmingham, UK, 5th-6th July, 2005.

Parkhurst, G., Richardson, J., 2002. “Modal integration of bus and car in UK local transport policy: the case for strategic environmental assessment”. Journal of Transport Geography 10:195-206.

Pojani, D., Stead, D., 2015. “Transit-oriented design in the Netherlands”. Journal of Planning Education Research 35:131-144.

Queensland Government, 2009. South East Queensland Regional Plan 2009-2031. Retrieved 14 March 2018 from: https://www.ipswich.qld.gov.au/__data/assets/pdf_file/0005/7916/seq-regional-plan-2009.pdf.

Shaw, J., Walton, W., 2001. “Labour’s new trunk-roads policy for England: An emerging pragmatic multimodalism?”. Environmental Planning A 33:1031-1056.

Shirgaokar, M., Deakin, E., 2005. “Study of park-and-ride facilities and their use in the San Francisco Bay Area of California”. Transportation Research Record 1927.1: 46-54.

Sipe, N., Pojani, D. 2018a. “For Mobility as a Service (MaaS) to solve our transport woes, some things need to change.” The Conversation 19 November.

Sipe, N., Pojani, D. 2018b. “The battle to be the Amazon (or Netflix) of transport”. The Conversation 27 November.

Spillar, R.J., 1997. Park-and-Ride Planning and Design Guidelines. New York, NY: Parsons Brinckerhoff, Inc.

Stieffenhofer, K.E., Barton, M., Gayah, V. V, 2015. “Assessing park-and-ride use and user reactions to parking management strategies: A case study in Puget Sound, Washington”. In: Transportation Research Board 95th Annual Meeting, Washington DC, United States.

Su, Q., Wang, D.Z.W., 2019. “Morning commute problem with supply management considering parking and ride-sourcing”. Transportation Research Part C: Emerging Technologies [forthcoming].

Syed, S., Golub, A., Deakin, E., 2009. “Response of regional rail park-and-ride users to parking price changes: systemwide results and a detailed study of two stations”. Transportation Research Record 2110(1): 155-162.

Transport and Main Roads, 2019. PnR Parking Supply for 2011 and 2016. Brisbane: Transport and Main Roads.

Tsang, F.W.K., Shalaby, A.S., Miller, E.J., 2010. “Improved modelling of park-and-ride transfer time: Capturing the within-day dynamics”. Journal of Advance Transport Research 39:117-137.

Turnbull, K.F., 1995. Effective Use of Park-and-Ride Facilities. Washington, DC: Transportation Research Board.

Van der Heijden, R., Molin, E.J.E., 2002. “Locating P&R facilities based on travel behaviour: a Dutch case study. WIT Transactions on The Built Environment 60.

Vuchic, V.R., 2007. Urban Transit Systems and Technology. Hoboken, NJ: John Wiley & Sons.

Wang, H., Meng, Q., Zhang, X.-N., 2014. “Park-and-ride network equilibrium with heterogeneous commuters and parking space constraint”. Transportation Research Record, 2466(1): 87-97.

Weinberger, R., 2012. “Death by a thousand curb-cuts: Evidence on the effect of minimum parking requirements on the choice to drive”. Transport Policy 20: 93-102.

Willing, R., Pojani, D. 2017. “Is the suburban dream still alive in Australia? Evidence from Brisbane”. Australian Planner 54(2):67-79.

Willson, R., 2005. “Parking policy for transit-oriented development: Lessons for cities, transit agencies, and developers”. Journal of Public Transport. 8(5)79-94.

Wiseman, N., Bonham, J., Mackintosh, M., Straschko, O., Xu, H., 2012. “Park and ride: An Adelaide case study”. Road & Transport Research: A Journal of Australian and New Zealand Research and Practice 21(1):39-52.

Yang, H., Liu, W., Wang, X., Zhang, X., 2013. “On the morning commute problem with bottleneck congestion and parking space constraints”. Transportation Research Part B: Methodological 58:106-118.

Yang, K., Pojani, D., 2017. “A decade of transit oriented development policies in Brisbane, Australia: Development and land use impacts”. Urban Policy and Research 35(3), 347-362.

Zahabi, S.A.H., Miranda-Moreno, L.F., Patterson, Z., Barla, P., 2012. “Evaluating the effects of land use and strategies for parking and transit supply on mode choice of downtown commuters”. Journal of Transport and Land Use 5:103-119.

Zhang, L., Li, J.-Q., Zhou, K., Gupta, S.D., Li, M., Zhang, W.-B., Miller, M.A., Misener, J.A., 2011. “Traveller Information Tool with Integrated Real-Time Transit Information and Multimodal Trip Planning”. Transportation Research Record 2215(1): 1-10.