Multiscale accessibility and urban performance

Abstract

Accessibility is fundamentally thought to be related to functional, economic, and social performances of cities and geographical systems and, therefore, constitutes an essential aspect for spatial planning. Previous studies focused on cities or metropolitan scales, often disregarding their position within regional and national systems, which can greatly affect their performance. Although accessibility at various spatial scales has been examined, the studies focused on accessibility patterns at different scales, with no reference to the level of accessibility of cities over local, regional, and national scales simultaneously, i.e. multiscale accessibility. This study aims to elucidate the multiscale accessibility level of individual cities and examine its relationship to urban performance in the urban system of Israel. Spatial accessibility was analyzed using the space syntax methodology for the entire national road network across multiple geographic scales—from the local to the national scale. Based on three distinct spatial accessibility systems identified, a unique multiscale accessibility profile was created for individual cities in Israel. Subsequently, each city’s multiscale accessibility profiles were examined against urban performance indicators determined from urban scaling theory. We found that the superiority of cities characterized by high accessibility level plays a role not only for a specific scale but also over scales and spatial systems. Moreover, most urban performance indicators related to the multiscale accessibility profiles of cities, while some multiscale accessibility profiles can be related to over- or under-performance of cities. The findings suggest that pervasive accessibility across spatial scales is inherently connected to urban performance and may indicate on the implementation and interpretation of accessibility. These findings may assist in various aspects of spatial planning at various scales.

Keywords

Multiscale accessibility urban performance urban scaling space syntax Israel

Introduction

Spatial accessibility is a fundamental and essential term in urban, regional, and national planning. Despite many definitions proposed (Bielik et al., 2018; Chen and Yeh, 2019; Geurs and van Wee, 2004; Hansen, 1959), a short and simple one refers to the proximity of person or place to all other persons and places (Batty, 2009), where the term places also include opportunities and activities. In general terms, accessibility of a place is fundamentally related to its functional, economic, and social performance (Geurs and van Wee, 2004; Spiekermann and Neubauer, 2002) with significant potential effects of future development (Hansen, 1959; Wachs and Kumagai, 1973).

Road networks are the basic elements that allow accessibility connecting places at multiple geographic scales, from the street and the neighborhood through the city to the regional and national scales (Parham et al., 2017; Serra and Pinho, 2013). Indeed, in recent years there is a growing effort to investigate and model road networks, focusing on the importance of the network itself for a wide range of phenomena (e.g. Boeing, 2019, 2017; Marshall et al., 2018). A great part of those efforts is oriented around accessibility. Previous studies have examined accessibility in specific contexts or for defined areas, usually up to the city or the metropolitan scale (e.g. Bielik et al., 2018; Curtis, 2011; Hansen, 1959), though without explicit reference to their larger spatial context. However, cities are not isolated as they are tightly interconnected (Jiang, 2018) and are the main components of regional and national systems, while concomitantly their position within these systems can greatly affect their performance (Law and Versluis, 2015; Serra et al., 2015).

Analysis of accessibility at regional or country scales is not new, reiterating the classical locational theories from the mid-19th century, which incorporated the spatial dimension into economic models by examining settlement patterns according to economic and physical considerations. These theories emphasized that location and size of settlements in a given space are significantly affected by proximity to other places by geographic location, distance, and size of the neighboring populated centers (Portugali, 2011: 17–37), specifically, a city’s relative accessibility. However, these studies suffered from coarse definitions due to their limited information and lack of knowledge, technologies, computing power, data resolution, etc. Advantages of state-of-the-art geographic information technologies allow the rapid analysis of accessibility at a national context with greater resolution and finer granularity. Indeed, several recent studies used high-resolution analyses indicate associations of the national spatial configuration with several types of functional and socio-economic aspects (Serra et al., 2015), commuting patterns (Law and Versluis, 2015), and the growth potential of cities (Parham et al., 2017). Based on the centrality analysis of the road network, a complex relationship was exposed between the spatial structures to functional and social-economic indicators across different spatial scales. For example, spatial accessibility, based on the centrality measures of integration, was found to be highly correlated with population density (R²=0.83 at the radius of 5 km), workplace density (R²=0.81 at the radius of 5 km), the number of commuters (R²=0.51 at the radius of 10 km) (Law and Versluis, 2015), working population density (R²=0.56 at the radius of 2 km), and movement volume (R²=0.47–0.68 in the radius of 10–20 km) (Serra et al., 2015).

One of the main issues regarding accessibility relates to the scale with which accessibility measures are defined (Batty, 2009) and requires some clarification. Most studies have referred to the analysis of “scale” in two different ways. The first one describes the scale in a Euclidean manner referring to physical distance, or time-based distance (e.g. Chen and Yeh, 2019; Parham et al., 2017; Zhang et al., 2019). The second stems from the scale-size of analysis is often defined with conventional spatial units such as neighborhoods, statistical areas, cities, metropolitan areas, regions, etc. (e.g. Boeing, 2019; Kwok and Yeh, 2004; Vieira and Haddad, 2015). This ambiguity is also reflected using the term multiscale analysis (e.g. Boeing, 2020; Yamu and Frankhauser, 2015). Indeed, recently Batty (2020) noted this inconsistency between scale and size in the literature.

New perspectives may link these two aspects of size and scale, as the emergence of spatial scales through the road network can be related to spatial units. Studies indicate that the spatial patterns of accessibility at various geographic scales have exposed complex spatial structures built by increasing scales. They begin by highlighting a patchy pattern of roads aggregating to settlements or cities as well as regions with more central places which are prominent from the rest of the settlement system (Kaplan et al., 2020; Law and Versluis, 2015; Serra and Pinho, 2013). Moreover, these spatial accessibility patterns may be linked to a general spatial division (Kaplan et al., 2020; Serra and Pinho, 2013). These types of methods inherently join the scale and size of a given geographical system, which is essential for a wider and more coherent perspective on spatial systems, as well as on the definition of scale (Batty, 2020).

Most studies including those mentioned previously have referred to accessibility at various spatial scales, while only scant attention has been given to the multiscale accessibility level of cities, i.e. the accessibility level of a city over several scales. Alternatively, referring to a city through its “full” spatial context as part of local, regional, and national systems (Kaplan et al., 2019, 2020). To simplify, multiscale accessibility can be described in two ways. The first, and the most common one, is as a horizontal perspective of the accessibility level for an entity at each of several scales, while the second can be described as a vertical perspective of the combined level accessibility for an entity over those scales. The second perspective is argued to be consistent with the theory of pervasive centrality according to which centrality functions diffuse throughout a city’s network at all spatial scales with a strong spatial correlation (Hillier, 2009, 2012a, 2012b; Omer and Kaplan, 2019). Indeed, a recent study points to this hypothesis for the national level, where the correspondence of settlement performance is related to their accessibility over scales, as well as for different scales of accessibility. This is a city’s multiscale accessibility profile (MPA) (Kaplan et al., 2019). Thus, this perspective of multiscale accessibility may quantify the effect of the city’s “full” spatial context, noting that the city may be characterized by various levels of accessibility at various spatial scales. This may be manifested through different spatial systems (Kaplan et al., 2020). This knowledge may have important implications for spatial planning policy across all local, regional, and national spatial levels (Ceccato and Persson, 2003).

This study aims to clarify and reveal the unique contribution of vertical multiscale accessibility to elucidate and explain urban performance. The intention is to measure the multiscale accessibility level of individual cities and examine the relationship to urban performance in the national urban system of Israel. The remainder of this paper is structured as follows: the methodological framework including the case study, the construction of the national multiscale accessibility model and definition of urban performance indicators (UPI) are described. The research ﬁndings are presented in the next section followed by discussion and conclusions in the closing section.

Methodological framework

This research was conducted in Israel. The methodological framework implemented here includes three main steps: accessibility analysis of all settlements at various geographical scales, exposure of MAP of settlements/cities, and measuring urban performance according to city size as part of the Israeli urban system. Subsequently, the analysis of relationships among the MAP and UPI were examined. Detailed information for each step is presented below.

Data were collected from two main data sources: (i) GIS layers of the Israel road network obtained from GISrael (a geographic information database in Israel, a product of the Mapa Company), and (ii) urban indicators from the Israeli Central Bureau of Statistics (https://www.cbs.gov.il) (see section “Urban performance indicators”). The data sources were collected for the year 2012.

Case study

Israel is a small country spanning 22,072 km² with a population of approximately nine million citizens, densely populated (∼410/km² in all land area and ∼7250/km² in the built area) with a developed economy and rapid history of development. It is characterized by high urbanization levels with ∼88% of the populace located in cities or municipalities (ICBS, 2019). This study mainly focused on urban settlements, i.e. all cities and local municipalities in Israel 2012 (n = 194). From now on, the term cities contain both categories.

National road network model

Mobility modes in Israel, including the mode split of journeys to employment, are based on usage of the road networks (Bank of Israel, 2018) and emphasize the dominant role of the road network in Israel state. This dominance is reflected well in the official Peripherality Index, which is calculated by a combination of road network accessibility of local authority (taking into account city size) and her network distance to the boundary of Tel Aviv District (economic heart) (Kaplan et al., 2020). These reinforce the road network as the main infrastructure for an accessibility analysis in Israel. The national road network model was analyzed by the Space Syntax methodology, which was found suitable for applied a national model with high resolution and with respect to various spatial scales (Law and Versluis, 2015; Parham et al., 2017; Serra et al., 2015). Briefly, this is a set of theories and techniques for topo-geometric analysis of spatial configurations allowing a reliable representation of centrality and accessibility levels across all geographic scales by using different distance types (Al Sayed et al., 2014). The road network in Israel is disconnected, a kind of “island state”, without continuity to surrounding countries. The national road network model used here is based on data for the entire national paved road system due to the findings showed the importance of all road types for understanding city performance (Serra et al., 2015). The original road representation is skeletal and was transformed into Road Center-Lines based on different road types. Following this, a semi-automated simplification process reduced unnecessary complexity and transformed the road network into a road segment map, which was found to be consistent with the results of the “traditional” Space Syntax segment line model (Krenz, 2017a) and similarly to (Parham et al., 2017; Serra and Hillier, 2019). The final road network model comprises 333,303 road segments (nodes). These processing steps were done in ArcGIS (ver. 10.3).

Further, the spatial configuration of Israel’s road network was analyzed based on angular segment analysis, based on the assumption that an angular distance (i.e. the cumulative angular changes made along a route) is the most suitable distance for this purpose. The superiority of the angular distance is related to the people’s cognitive representation and preferences in process of movement route choice (Jayasinghe et al., 2016; Manley et al., 2015) but also to the road network structure. It was found that the angular structure of road networks is in correspondence with the official designations of roads (road hierarchy) when these properties cannot be captured by the metric description (Molinero et al., 2017). Furthermore, a recent study found that the structural properties of street networks channel movements to the angular structure, even when the selected shortest route is selected in metric terms (Omer and Kaplan, 2019). Indeed, the angular distance has been found the most appropriate distance type for detecting spatial patterns at the urban, regional, and national scales (Hillier et al., 2012; Serra et al., 2015; Serra and Pinho, 2013) and particularly found more suitable for accessibility analysis in Israel (Kaplan et al., 2020).

In order to examine the spatial accessibility in Israel, the integration centrality measure was selected, which corresponds to the graph-based closeness centrality measure. This measure describes how close a given node (road segment) is to all other nodes and represents the degree of accessibility for each road segment in the network at the entire road network (radius N) (Omer and Jiang, 2015):

Closeness (V_{i}) = \frac{n - 1}{\sum_{k = 1}^{n} d (V_{i,} V_{k})}

(1)

where n is the total number of segments (nodes) within a road network and d is the shortest angular distance from a given road segment (V_i) to every other road segment (node V_k) in the segment map. In order to reveal a full multiscale perspective of the urban system in Israel, the integration measures were calculated for a large set of radii representing different spatial scales. To ensure that there are no hidden scales between the chosen radii, the principle suggested by Kaplan et al. (2020) was used, where the integration value for each radius strongly correlates to the integration value in the adjacent radius (r > 0.9). The final chosen radii are (km): 0.5, 1, 1.5, 2, 3, 4, 5, 7.5, 10, 12.5, 15, 20, 25, 30, 40, 50, 75, 100, 150 and N. The calculation of Integration centralities was performed using the Depthmap software (ver. 10.3, UCL). Finally, a multiscale accessibility analysis was conducted based on the different patterns of accessibility across geographic space.

Multiscale accessibility analysis

The multiscale accessibility analysis was based on the division of Israel spatial configuration into three spatial accessibility systems: the local (radii up to 5 km), the regional (radii between 5 and 25 km) and the national (radii larger than 25 km) (Kaplan et al., 2020), which were validated for the current study in acceptable method done previously (Kaplan et al., 2020; Serra and Pinho, 2013). This method is consistent with the needs connecting the scale to geographical systems (Batty, 2020). This analysis is able to reveal the multiscale accessibility level of all settlements in the nationwide system, expressly, the accessibility level of each settlement over the three spatial systems. First, the accessibility level of all settlements for each scale is calculated by taking the mean integration value of all road segments within the official boundaries of the settlement’s built-up area (e.g. Law and Versluis, 2015; Serra et al., 2015). Second, the accessibility level of a settlement (high or low) was categorized using the classification technique of “head/tail breaks” due to the long-tail distributions of the accessibility level of settlements for local and regional scales. This then categorizes the settlements into “head” (values above mean) and “tail” (values below mean) (Jiang, 2013, 2015). Applying this algorithm for each accessibility value over all examined radii resulted in two groups for each scale: the high group containing settlements with accessibility level above the mean, and the low group containing settlements with accessibility level below the mean (Supplementary Table S1). The resulting process classified all settlements into eight MAPs according to their accessibility dominance at the three spatial accessibility systems (Kaplan et al., 2019).

Urban performance indicators

In this study, 20 UPI were chosen to represent different aspects of urban performances including functional and economic, as well as demographic and social variables (Table 1). As noted recently by Batty (2020), a variable that has an extensive base such as population should be normalized with respect to the unit of analysis. Therefore, the UPI was derived from urban scaling theory described by power-law scaling relations of macroscopic quantities (Bettencourt et al., 2007; Bettencourt and West, 2010). Empirical scaling laws describe the linear and the nonlinear relationship of urban parameters with respect to city size (Batty, 2008; Bettencourt et al., 2007). Formally, urban scaling is deﬁned by (Bettencourt et al., 2007):

Yi = Y_{0} N^{β}

(2)

where

Y_{0}

is a normalization constant, N is the population size, and β is the scaling exponent and Yi is the observed value of the metric for each specific city. Generally, the scaling of urban systems are consistent for many countries and are important for characterizing complex urban dynamics dependent on city size and population growth. It has been shown that infrastructure is a sublinear process (β < 1, economy of scale), consumption is generally a linear process (β = 1) and socioeconomic characteristics tend to be superliner (β > 1, increasing returns) (Bettencourt, 2013; Bettencourt et al., 2007). Hence, urban scaling theory is essential in order to examine the accessibility effect while controlling for the city’s size.

Table 1.

Urban indicators and their scaling relations.

Variables type	Code	Variable	R²	β	N^a
Functional and economic	1	Total employment^b	0.98	0.99	189
	2	Employees^c	0.98	1.00	189
	3	Self-employed	0.87	0.94	189
	4	Businesses	0.87	1.09	131
	5	Revenue	0.91	0.93	190
	6	Expenditure	0.91	0.93	190
	7	Wages	0.91	0.96	189
	8	Motorization^d	0.87	0.96	194
	9	Cars^e	0.87	0.97	194
Demographic	10	Families with children	0.99	1.01	194
	11	Workforce population	0.99	1.01	194
	12	Pensioners	0.83	1.16	188
	13	Children	0.95	0.98	194
	14	Inflow	0.80	0.82	194
	15	Outflow	0.91	0.91	194
Social	16	Crime (trespassing)	0.86	1.07	188
	17	Crime (moving violations)	0.89	1.25	189
	18	Crime (breaking/entering)	0.82	1.14	187
	19	Crime (total)	0.91	1.14	189
	20	Matriculation	0.79	0.87	154

^aThe total number of cities in our study is 194. For some variables a data for some cities were unavailable.

^bAll workers.

^cSalaried workers.

^dAll vehicles.

^ePrivate vehicles.

To measure UPI, two main steps were implemented. The first explored power-law patterns for all cities in Israel across several indicators (Table 1). The second step measured urban performance as the deviation between the observed value of a city relative to the expected value based on the allometric fit. This method creates a scale-independent comparison enabling the evaluation of urban performance concerning cities of comparable size, thereby, allowing unbiased comparison to other cities in the urban system (Bettencourt et al., 2010; Sobolevsky et al., 2014). Thus, the power-law scaling relationships were derived for all variables (Table 1). After examining the scaling relationships for each urban indicator, and found a reliable model with a high fit (R² > 0.8 for all variables, Table 1), the expected Y_i for all indicators calculated, for each city. Then, the urban performance deviations for each city are quantified by the residuals (Bettencourt et al., 2010):

Dev (i) = Log \frac{Y_{i}}{Y_{0} N_{i}^{β}}

(3)

Dev(i) represents a normalized UPI for a given city “i”. This is performed for each metric. Note, that the “0” value of UPI indicates that the city being examined falls on the allometric fit, while values below (or above) are interpreted as under (or over) performance, i.e. relatively low/high values for a city of specified size. Finally, the relations between MAPs and UPIs of cities were statistically examined to test two main questions: do cities with a high or low level of accessibility are differed in terms of UPI and, if so, are there systematic differences between categories of cities? The results are presented in the following section.

Results

The characterization and examination of the spatial configuration of multiscale accessibility are based on two steps. The first examines the accessibility level of settlements separately for each spatial system, namely the local, the regional and the national (Figure 1(a) to (c)), while the second focuses on the combination between the three systems, i.e. the MAP of settlements (Figure 1(d) to (f)). The investigation of urban performance here focuses on all cities in Israel (Figure 1(f)).

Figure 1.

The spatial distribution of high and low accessibility level of settlements at each spatial system: (a) local; (b) regional; and (c) national; and, the spatial pattern of multiscale accessibility profiles (MAP): (d, e) at the settlement level; and (f) at the city level.

The results of three separate T-tests for each spatial system presents a clear pattern of differences in urban performance between groups of cities distinct by accessibility level (Figure 2). Significant differences in UPI are discerned between the groups of high or low accessibility level of cities at the three scales by which accessibility was measured. Further, high accessibility level is significantly associated with increased urban performance, except for familial variables (10 and 13). This finding is positive for most indicators (Figure 2 top, variables 1–9, 11–12, and 14) but has a negative dimension in terms of several indicators, especially crime (Figure 2 bottom, variables 15–19). Additionally, it is worth noting that the largest differences between high and low levels of accessibility of cities in terms of UPI are in the domain of crime, implying that high accessibility may have a negative effect. For some standard of living variables (3, 7–9, and 18), regional but especially national high accessibility levels dominant over local accessibility. Lastly, no significant difference is found for variable 20 over all scales.

Figure 2.

Results of three independent T-tests examining the mean differences at UPI between the two groups of high and low accessibility levels of cities for each spatial system (separate T-test for each system). The local, regional, and national spatial systems are colored by blue, green, and red (respectively), corresponding to Figure 1(a) to (c). Columns with solid fill are significantly different (p < 0.05). The top figure refers to variables where a high level of performance is considered “positive”, for example, the number of employees. The bottom figure refers to variables where a high level of performance is considered as “negative”, for example, the amount of crime. Notes: At the local scale, the results of Levene’s test for variables 3, 7, and 15–20 are significant (p < 0.05) while other variables are nonsignificant (p > 0.05); at the regional scale, the results of Levene’s test for variables 1, 2, 11, and 15 are significant (p < 0.05) while other variables are non-significant (p > 0.05); at the national scale, the results of Levene’s test for variables 1, 2, 11, 15, and 18 are significant (p < 0.05) while other variables are nonsignificant (p > 0.05); in cases of significant results of Levene’s test the result of T-test taken when equal variances assumed, while for non-significant cases equal variances not assumed.

In addition to the gaps between both groups’ performance, a detailed examination focused on the advantages or disadvantages of high or low accessibility in terms of urban performance. Using a series of one-sample T-tests, for each spatial system was separately examined if the group of cities characterized by high or low accessibility level systematically deviates from the allometric trend, i.e. over- or under-performance. The results are presented in Supplementary Figure S1. The general results are less substantive than the previous part, but still, several issues are notable. The low accessibility level of cities are found to be significantly related to underperformance for all UPI, regardless to scale of accessibility, except for relatively high value in variables 10 and 13. In most cases, the level of underperformance is substantial, especially, for the relatively low incidence of crime (variables 16–19). Again, this reinforces the finding that low level of accessibility of cities in terms of UPI regarding crime indicators may have a positive effect. In contrast, for some UPI, including variables of business, motorization, and property crime (i.e. 3–4, 8–9, and 18), the high regional and national accessibility levels of cities are significantly related to relatively high incidence of cases, but with a comparatively small gap. This is also highlights the dominant role of high regional and national accessibility level rather than local accessibility. Besides, for most UPI, cities with high levels of accessibility tend to cluster around the allometric fit, or be above it, even if not significant for some cases. This is in stark contrast to the groups of cities with low accessibility level.

The results of these two analyses demonstrate several important points, most of them regarding the superiority, in terms of urban performance, of cities with a high level of accessibility at the local, regional and national scales separately. The next part of the analysis focuses on the city’s level of accessibility over all three spatial systems simultaneously, i.e. the MAP of a city (see section “Multiscale accessibility analysis”).

First, we examine if different kinds of MAP are separate in terms of urban performance. The primary result of the one-way ANOVA test shows that for most UPI, the clusters of MAP are significantly different from each other (p < 0.05), while only for indicators 6 and 20 the result is not significant.¹ In order to find the source of the differences between MAP clusters, post hoc tests were used.² The tests³ show that the main differences concerning urban performance are between the edges of the MAP, namely the HHH and the LLL clusters. Excluding variables 4 and 15, all UPI are significantly different, when the HHH cluster provides significantly higher performance for positive variables (1–3, 5, 7–9, 11, 12, and 14) and the LLL cluster provides significantly higher performance for the familial variables (10 and 13). In contrast to the positive effects of multiscale high accessibility, the HHH cluster is also characterized by a significantly higher incidence of crime (variables 16–19).

Further differences were also found between the HHH and the HLL clusters. These significant differences are consistent for variables 3, 5, 7–9, 14, and 15, noting the higher performance of the HHH clusters. Like the LLL cluster, the HLL provides significantly higher performances than the HHH cluster regarding variable 13. A third point noted on differences in terms of variables 3 and 4 between the HHH and the HHL clusters, similarly performance-level priority for the HHH cluster. The last point here regards the crime variables (16–19), when the HHH cluster was found characterized by a significantly higher level of crime than the HLL and the LLH clusters.

An examination was conducted to clarify the systematic associations among levels of accessibility across scales and urban performance. Using a series of one-sample T-tests, each MAP was examined if it systematically deviates from the allometric trend, i.e. does the urban performance of examined MAP are significantly related to over- or under-performance. The results are presented in Supplementary Figure S2. Clearly, the LLL cluster significantly suffers from underperformance, except for relatively high value in variables 10 and 13 and nonsignificant results for variable 4. For the variable at the upper figure, this can be interpreted as a negative aspect while from variable at the bottom figure on the contrary. Similarly, the HLL cluster is presented similar results of performances but for most cases, the result is closer to “0”, especially for the variables at the bottom figure. The HLL in that manner quite compatible with the HHL cluster results who characterized by less significant results.

In contrast to those, the HHH cluster reveals higher performances, mostly above the model but only a few UPI are significantly higher than expected (variables 3, 4, 8, 9, 12, and 18). For most variables, the results of HHH correspond to the HLH results, but those are less significant. Here it should be noted that in general terms for all UPI, the HHH, LLL, HLL, and the HHL clusters are more homogeneous with lower standard deviation than the LLH and the HLH clusters. Another point should be noted that for most crime variables (16, 17, and 19) the LLL and the LLH provides similar results.

For the last test, the HHH cluster was examined against all other cities with a high level of accessibility at the three spatial systems separately, to determine if they have priority over other cities with high accessibility levels. It was found that the HHH cluster has significantly higher performance than other cities with a high accessibility level at the local (variables 1–9 and 12–19) regional (variables 3, 7–9, 14, 17–19) and national (variables 5, 6, 12, 14, 16–19) systems.⁴ This may reinforces that continuity of the accessibility level is critical and contributing to the effect of accessibility on urban performance.

Discussion

This study aimed to suggest a methodological framework for describing vertical multiscale accessibility and to reveal its importance for understanding and characterizing urban performance. The conceptual framework adopted here adheres to the needs of combining two important aspects of size and scale in spatial analysis (Batty, 2020). It demonstrates the emergence of spatial scales through the road network centralities (Krenz, 2017b) related to geographical spatial units (Kaplan et al., 2020; Serra and Pinho, 2013). It also reinforces the relevance of accessibility analysis to define and characterize regional forms (Wachs and Kumagai, 1973). The results obtained from implementation of this approach in the case of Israel reveals the spatial structure of the MAP of settlements and cities based on the road network multiscale accessibility analysis.

The categorization of settlements with a high or low accessibility level for each spatial system separately showed the formation of distinct spatial systems, by increasing the scale up to the national system. This process is analogous to the units of road network accessibility, by increasing scale-up (Law and Versluis, 2015; Serra and Pinho, 2013). Further, the MAP of settlements, based on the analysis of multiscale accessibility indicates a spatial configuration pattern highlighted by the distinction between the two main MAPs—HHH and LLL clusters. Those are concomitant with settlements described with high accessibility or low accessibility at all scales, respectively. Between them, several profiles are observed by a mixed level of multiscale accessibility. For example, cities with high local accessibility in the periphery or those with low local accessibility located in the center. This highlights a much more complex and nuanced structure of accessibility than the generally acceptable division of the national space into core (center) and periphery, which was also found to be unsuitable for other countries (Bar-El and Parr, 2003; Curtis, 2011; Krugman, 1999). Hence, the vertical perspective of multiscale accessibility establishes a detailed and complex structure that should be considered further.

Our findings show that accessibility of a place is fundamentally related to its performance, as previously reported (Geurs and van Wee, 2004; Spiekermann and Neubauer, 2002), particularly, with regards to multiscale accessibility. Here, urban performance was characterized by twenty UPIs representing different aspects of urban attributes which are analyzed according to the urban allometric model (Bettencourt, 2013; Bettencourt and West, 2010). This leads to a scale-independent comparison of performance among cities (Bettencourt et al., 2010; Sobolevsky et al., 2014) with high validity accounting for the inherent nonlinear relation of variables to the city size (Bettencourt et al., 2007) in a more pertinent manner to the needs of variables normalization for the unit of analysis (Batty, 2020).

The results presented here illustrate that cities with a high accessibility level reach a significantly higher level of functional and economic performance than cities with a low accessibility level, regardless of a specific scale of accessibility. These gaps are larger regarding economic variables, such as motorization level and wages. It seems that the national high accessibility level plays an more important role than local accessibility or even regional accessibility as found in previously (Bystrov, 2008; Shefer and Antonio, 2013). This also applies to social variables. Interestingly, negative aspects of cities, such as crime, are negatively associated with high accessibility (Beavon et al., 1994; Liu et al., 2016; Setiawan et al., 2019). Conversely, accessibility shows inconsistent effects on demographics. This maybe can be extenuated by the age distribution where cities with high accessibility levels have relatively fewer children and relatively more adults, especially pensioners.

A detailed examination shows that cities with high levels of accessibility exhibit lower variance relative to the urban allometric model. Furthermore, for some variables of functional and economic performance, cities with a high level of accessibility were found to be significantly characterized by over-performance, while cities with low accessibility levels were found to significantly underperform for most UPIs (except for a demographic variables as mentioned above). Therefore, the superiority of cities is highlighted with a high level of accessibility, at least with respect to each geographic system separately.

Associations of urban performance with the combinations of the accessibility levels over the three spatial systems simultaneously were significant, namely, with the cities’ MAP. It was found that the main differences in terms of urban performance are associated with the two main MAPs—HHH and the LLL clusters. For most functional, economic, and social indicators the HHH correlates substantially with higher performance, with a positive effect for the first two, though with a negative effect for the third. For demographic variables, the results are in agreement with earlier findings from the high and low accessibility levels noted above. In contrast, the LLL, HLL, and HHL clusters suffer from significant underperformance for the functional and economic indicators, but those clusters experience a relatively low incidence of crime. In addition to those, the LLL is prominent with a large gap from its expected performance. In contrast, the HHH is closer to, or higher than, the allometric fit and cities in this cluster significantly tend to over-perform in variables for standard-of-living, such as motorization level. These points highlight, again, the predominant role of national accessibility on urban performance.

These findings concerning the relation of urban performance to accessibility show clearly that accessibility analysis should be conducted from the multiscale point-of-view considering both the horizontal (Law and Versluis, 2015; Parham et al., 2017) and also the vertical perspectives (Kaplan et al., 2019, 2020) when the full spatial context of cities should also be taken into account (Serra et al., 2015). The necessity of the multiscale view is illustrated favorably by the superiority of the HHH cluster over all other clusters with an advantage over cities with a high accessibility level within a specific spatial system. Thus, the necessity of multiscale analysis of accessibility in the study of urban performance is validated and supports the theory of pervasive centrality found in cities. Accordingly, centrality functions diffuse throughout the network at all scales (Hillier, 2009, 2012a, 2012b; Omer and Kaplan, 2019), as a much more complex structure than a simple hierarchy of location at a single scale (Hillier, 2009).

To summarize, the importance of vertical multiscale accessibility for spatial knowledge is presented when the relationships between the MAP of cities to their urban performance are revealed. The capabilities of the suggested methodology for capturing urban performance aspects may be implemented for examining related planning scenarios such as establishment of employment areas, residential places and transportation networks as well as population growth scenarios with concomitant infrastructure implementation at a national scope (Parham et al., 2017). Moreover, the use of urban scaling theory to accurately portray the effect on urban performance and to conduct a reliable benchmark for cities in a specific urban system (Bettencourt et al., 2010; Sobolevsky et al., 2014) may be powerful for other dimensions of macro-level planning.

Conclusions

This study aims to measure the multiscale accessibility level of individual cities and thereby examine its relationship to urban performance in the nation-wide urban system of Israel. The multiscale accessibility examination of the Israel road network exposes three differentiated spatial systems—local, regional, and national—and the distinctions between them. According to these three spatial systems, eight MAPs at the settlement level were determined. The current study focuses on the relation of those profiles with cities performance. Twenty UPIs representing different aspects of performance were derived from urban scaling deviation and placed cities in a scale-independent manner suitable for comparative investigation. The findings reported here show the superiority of high accessibility level not only for a specific scale but also over scales, i.e. multiscale accessibility. Cities with a high accessibility level depict significantly higher performance and even over performance in terms of functional and economic variables, but also valid for crime indicators. For the demographic variables, the results were inconsistent and may be more indicative of the complexity of the demographic aspects themselves. When the focusing on the vertical multiscale accessibility, results show a strong association between MAP and performance of cities, where the largest differences were found for the HHH cluster to all others. Moreover, some MAPs are shown to be associated with over- or under-performance of cities as measured with urban scaling methodologies. Thus, it is suggested that pervasive accessibility across spatial scales is inherently connected to the city’s performance and can shed light on how accessibility should be analyzed and interpreted. The findings and insights presented in the study may be useful for wide aspects of spatial planning at varied geographic scales. Moreover, they shed new light on the concept of accessibility and can serve as a basis for theorization of accessibility-based spatial development approaches. Particularly, this study supports the network or “semi-lattice” view of urban and regional systems (e.g. Alexander, 1965; Batty, 2018), characterized by pervasive accessibility across spatial scales (Hillier, 2009). This view stands in contrast to the classical hierarchical or “tree” view that refers to spatial systems as distinct entities. The methodological framework suggested here can assist to implement this view in research and planning.

Further work that will examine the suggested methodological approaches in other countries may contribute to a better understanding of accessibility–urban performance relationships. Further analysis of different accessibility measures may also expand the knowledge with this respect. Such an analysis in countries with a varied pattern of transportation modes and/or larger countries, in general, should take into consideration the effect of additional major networks such as rail on the accessibility level of cities. In addition, the suggested multiscale accessibility analysis and the urban scaling deviation can be adapted for intensive research at the city level as a robust benchmark indicator for various planning assessments. Other research directions may include consideration of changes in accessibility across time and include a wider range of UPIs.

Supplemental Material

sj-pdf-1-epb-10.1177_23998083211024648 - Supplemental material for Multiscale accessibility and urban performance

Supplemental material, sj-pdf-1-epb-10.1177_23998083211024648 for Multiscale accessibility and urban performance by Nir Kaplan, David Burg and Itzhak Omer in EPB: Urban Analytics and City Science

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The first author is partially supported by a scholarship from the Shlomo Shmeltzer Institute for Smart Transportation in Tel-Aviv University. This paper has been published, thanks to the support of the Ministry of Science and Technology, Israel (grant no. 3-13526).

ORCID iD

Nir Kaplan

Supplemental material

Supplemental material for this article is available online.

Notes

References

Al Sayed

Turner

Hillier

, et al. (2014) Space Syntax Methodology. 4th ed. London: Bartlett School of Architecture, UCL.

Alexander

(1965) A city is not a tree. Architectural Forum 122: 58–62.

Bank of Israel (2018) Collection of Policy Analysis and Research Issues. Jerusalem: Bank of Israel.

Bar-El

Parr

(2003) Overreliance on the core—periphery model? The case of Israel. Environment and Planning C: Government and Policy 21(3): 353–369.

Batty

(2008) The size, scale, and shape of cities. Science (New York, N.Y.) 319(5864): 769–771.

Batty

(2009) Accessibility: In search of a unified theory. Environment and Planning B: Planning and Design 36(2): 191–194.

Batty

(2018) Inventing Future Cities. Boston, MA: The MIT Press.

Batty

(2020) On scale and size. Environment and Planning B: Urban Analytics and City Science 47(3): 359–362.

Beavon

DJK

Brantingham

(1994) Inﬂuence of street networks on the patterning of property offenses. Crime Prevention Studies 2: 115–148.

10.

Bettencourt

LMA

(2013) The origins of scaling in cities. Science (New York, N.Y.) 340(6139): 1438–1441.

11.

Bettencourt

LMA

West

(2010) A unified theory of urban living. Nature 467(7318): 912–913.

12.

Bettencourt

LMA

Lobo

Helbing

, et al. (2007) Growth, innovation, scaling, and the pace of life in cities. Proceedings of the National Academy of Sciences 104(17): 7301–7306.

13.

Bettencourt

LMA

Lobo

Strumsky

, et al. (2010) Urban scaling and its deviations: Revealing the structure of wealth, innovation and crime across cities. PLoS One 5(11): e13541.

14.

Bielik

König

Schneider

, et al. (2018) Measuring the impact of street network configuration on the accessibility to people and walking attractors. Networks and Spatial Economics 18(3): 657–676.

15.

Boeing

(2019) Street network models and measures for every U.S. City, county, urbanized area, census tract, and Zillow-Defined neighborhood. Urban Science 3(1): 28.

16.

Boeing

(2020) A multi-scale analysis of 27,000 urban street networks: Every US city, town, urbanized area, and Zillow neighborhood. Environment and Planning B: Urban Analytics and City Science 47(4): 590–608.

17.

Boeing

(2017) Methods and Measures for Analyzing Complex Street Networks and Urban Form. Phd Thesis, University of California, USA.

18.

Bystrov

(2008) Spatial inequalities between the core and the periphery in Israel: A geopolitical challenge. In: European Consortium for Political Research (ECPR) joint sessions: Workshop N 4, Rennes France April, 2008, pp. 1–13.

19.

Ceccato

Persson

(2003) Differential economic performance (DEP) in the periphery: Evidence from Swedish rural areas. European Journal of Spatial Development (7): 28.

20.

Chen

Yeh AGO (2019) Accessibility inequality and income disparity in urban China: A case study of Guangzhou. Annals of the American Association of Geographers 109(1): 121–141.

21.

Curtis

(2011) Integrating land use with public transport: The use of a discursive accessibility tool to inform metropolitan spatial planning in Perth. Transport Reviews 31(2): 179–197.

22.

Geurs

van Wee

(2004) Accessibility evaluation of land-use and transport strategies: Review and research directions. Journal of Transport Geography 12(2): 127–140.

23.

Hansen

(1959) How accessibility shapes land use. Journal of the American Institute of Planners 25(2): 73–76.

24.

Hillier

(2009) Spatial sustainability in cities: Organic patterns and sustainable forms. In: Proceedings of the 7th international space syntax symposium, Stockholm: KTH June, 2009, K01: 1–20.

25.

Hillier

(2012a) Studying cities to learn about minds: Some possible implications of space syntax for spatial cognition. Environment and Planning B: Planning and Design 39(1): 12–32.

26.

Hillier

(2012b) The genetic code for cities: Is it simpler than we think? In: Portugali J, Meyer H, Stolk E, et al. (eds) Complexity Theories of Cities Have Come of Age. Berlin, Heidelberg: Springer Berlin Heidelberg, pp. 129–152.

27.

Hillier

Yang

Turner

(2012) Normalising least angle choice in Depthmap and how it opens up new perspectives on the global and local analysis of city space. The Journal of Space Syntax 3(2): 155–193.

28.

Jayasinghe

Sano

Kasemsri

, et al. (2016) Travelers’ route choice: Comparing relative importance of metric, topological and geometric distance. Procedia Engineering 142: 18–25.

29.

Jiang

(2013) Head/tail breaks: A new classification scheme for data with a heavy-tailed distribution. The Professional Geographer 65(3): 482–494.

30.

Jiang

(2015) Head/tail breaks for visualization of city structure and dynamics. Cities 43: 69–77.

31.

Jiang

(2018) A topological representation for taking cities as a coherent whole: Topological representation. Geographical Analysis 50(3): 298–313.

32.

Kaplan

Omer

Burg

(2019) How is multiscale accessibility related to population, functional and socio-economic characteristics of settlements: The case of Israel. In: Kyriakidis P, Hadjimitsis D, Skarlatos D and Mansourian A (eds) 22nd AGILE conference on geo-information science, Limassol, Cyprus, 17–20 June 2019.

33.

Kaplan

Burg

Omer

(2020) The spatial organization of accessibility and functional hierarchy: The case of Israel. Computers, Environment and Urban Systems 80: 101429.

34.

Krenz

(2017a) Employing volunteered geographic information in space syntax analysis. In: Proceedings of the 11th space syntax symposium, Lisbon July, 2017, vol. 150, pp. 1–26.

35.

Krenz

(2017b) The emergence of spatial scales in urban regions. In: Proceedings of the 11th space syntax symposium, Lisbon July, 2017, vol. 74, pp. 1–23.

36.

Krugman

(1999) The role of geography in development. International Regional Science Review 22(2): 142–161.

37.

Kwok

RCW

Yeh

AGO

(2004) The use of modal accessibility gap as an indicator for sustainable transport development. Environment and Planning A: Economy and Space 36(5): 921–936.

38.

Law

Versluis

(2015) How do UK regional commuting flows relate to spatial configuration? In: Proceedings of the 10th international space syntax symposium, London July, 2015, vol. 74, pp. 1–21.

39.

Liu

Song

Xiu

(2016) Spatial patterns of violent crimes and neighborhood characteristics in Changchun, China. Australian & New Zealand Journal of Criminology 49(1): 53–72.

40.

Manley

Orr

Cheng

(2015) A heuristic model of bounded route choice in urban areas. Transportation Research Part C: Emerging Technologies 56: 195–209.

41.

Marshall

Gil

Kropf

, et al. (2018) Street network studies: From networks to models and their representations. Networks and Spatial Economics 18(3): 735–749.

42.

Molinero

Murcio

Arcaute

(2017) The angular nature of road networks. Scientific Reports 7(1): 4312.

43.

Omer

Jiang

(2015) Can cognitive inferences be made from aggregate traffic flow data? Computers, Environment and Urban Systems 54: 219–229.

44.

Omer

Kaplan

(2019) Structural properties of the angular and metric street network’s centralities and their implications for movement flows. Environment and Planning B: Urban Analytics and City Science 46(6): 1182–1200.

45.

Parham

Law

Versluis

(2017) National scale modelling to test UK population growth and infrastructure scenarios. In: Proceedings of the 11th space syntax symposium, Lisbon July, 2017, vol. 103, pp. 1–17.

46.

Portugali

(2011) Complexity, Cognition and the City. Springer Heidelberg; New York: Springer.

47.

Serra

Hillier

(2019) Angular and metric distance in road network analysis: A nationwide correlation study. Computers, Environment and Urban Systems 74: 194–207.

48.

Serra

Pinho

(2013) Tackling the structure of very large spatial systems - Space syntax and the analysis of metropolitan form. The Journal of Space Syntax 4(2): 178–196.

49.

Serra

Hillier

Karimi

(2015) Exploring countrywide spatial systems: Spatio-structural correlates at the regional and national scales. In: Proceedings of the 10th international space syntax symposium, vol. 84, pp. 1–18.

50.

Setiawan

Dede

Sugandi

, et al. (2019) Investigating urban crime pattern and accessibility using geographic information system in Bandung city. KnE Social Sciences 2019: 535–548.

51.

Shefer

Antonio

(2013) Spatial inequality between and within urban areas: The case of Israeli cities. European Planning Studies 21(3): 373–387.

52.

Sobolevsky

Sitko

Grauwin

, et al. (2014) Mining urban performance: Scale-independent classification of cities based on individual economic transactions. arXiv:1405.4301 [physics, q-fin]. Available at: http://arxiv.org/abs/1405.4301 (accessed 21 January 2019).

53.

Spiekermann

Neubauer

(2002) European accessibility and peripherality: Concepts, models and indicators. Nordregio Working Paper 2002(9): 46.

54.

Vieira

Haddad

(2015) An accessibility index for the Metropolitan Region of São Paulo. In: Karima Kourtit, Peter Nijkamp, Roger R. Stough (eds), The Rise of the City. Cheltenham: Edward Elgar Publishing, pp. 242–258.

55.

Wachs

Kumagai

(1973) Physical accessibility as a social indicator. Socio-Economic Planning Sciences 7(5): 437–456.

56.

Yamu

Frankhauser

(2015) Spatial accessibility to amenities, natural areas and urban green spaces: Using a multiscale, multifractal simulation model for managing urban sprawl. Environment and Planning B: Planning and Design.

57.

Zhang

(2019) How do people understand convenience-of-living in cities? A multiscale geographic investigation in Beijing. ISPRS Journal of Photogrammetry and Remote Sensing 148: 87–102.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.44 MB