Quantifying and visualizing the 15-Minute walkable city concept across Europe: a multicriteria approach

ABSTRACT The disruptions associated with health and energy crisis have emphasized the need for hyperlocal cities. However, in Europe, a tool to measure the efficiency of land use and accessibility planning for localizing urban mobility is missing. In this paper, we construct a comparable 15-Minute-Walking City (15-MWC) index that assesses the walking performance of 121 European metropolitan areas and seven amenity types. The data are combined equally following the PROMETHEE II multicriteria approach to assign a final score and to present a ranking of 15-Minutes cities. The main visualization demonstrates European-level disparities and indicates that most of leading cities are in Germany, while most of the worst performers are in the UK. We also reveal a statistically significant difference in 15-MWC performance due to their GDP per capita disparities. The empirical results may serve as a referencing tool for cross-city comparisons and may support policymakers when designing transport and city-planning strategies.


Introduction
Over the past century, many cities have been places designed for cars and traffic, and many people have accepted the toxic impacts of auto-dependent environments, such as sprawl, long commutes, loss of social interaction, isolation, and health and environmental problems.However, the 15-Minute City model is a new popular planning concept, conceived in 2016 by Carlos Moreno (Moreno et al., 2021), which emphasizes enhanced temporal and spatial proximity and accessibility and presents important economic and social relevance.The model also involves significant intellectual similarities with Clarence Perry's Neighborhood Unit concept, the new urbanism (Sharifi, 2016), and the influential ideas of Jane Jacobs (1961) and Leon Krier (1984), such as mixed-use neighborhoods and cities within the city.In particular, the fundamental dimensions of closeness and access are based on the philosophy of chrono-urbanism, and they allow people (especially children, the elderly, disabled or time-poor people) to comfortably walk to essential urban amenities and opportunities in a timeframe not exceeding 15 min.To that end, car travel demand is lower and planners are allowed to create more spaces for people that foster social endeavors, as advocated by many urban theorists, such as Nikos Salingaros (2000).
The 15-Minute model, although it has received criticism regarding catastrophic socio-spatial inequalities (i.e.segregation, housing affordability) (Ferrer-Ortiz et al., 2022;Glaeser, 2021;Weng et al., 2019) has been recognized as a fundamental tool in designing healthier and more sustainable urban spaces (Allam et al., 2022).It is internationally acceptedas a critical solution for increasing active mobility and tackling some of the effects of global crises, such as the COVID-19 pandemic, physical inactivity, climate change, oil-vulnerability, etc.For instance, a recent study from China found that reduced perceived accessibility of daily necessities and activities may cause mental illness, which was exacerbated during the pandemic (Liu et al., 2022), while others indicated that 15-minutes or 1 km travel distances can promote and support short walking trips in different population groups and trip purposes (Elldér et al., 2020;Ferrer & Ruiz, 2018;Perchoux et al., 2019;Ryley, 2008).However, previous research in analyzing the 15-Minute concept have mainly focused on the local geographic variability of accessibility performance and demonstrated significant differences between global and local patterns of human mobility to neighborhood amenities (Ferrer-Ortiz et al., 2022;Graells-Garrido et al., 2021;Weng et al., 2019).Logan et al. (2022) constructed an interactive geospatial dashboard for New Zealand's urban areas that demonstrates modespecific accessibility statistics and geovisualizes the xminute concept for seven daily amenities (i.e., from 5 min.to more than 20 min.)at the neighborhood level.Furthermore, they measured the x-minute concept across the 500 largest cities in the USA and 43 metro areas of New Zealand, where they found that New York is on average a 10-minute city, while Austin or Atlanta are mainly 26-minute cities. Nicoletti et al. (2022) reported spatial inequalities in walking access to infrastructure across several socio-vulnerable urban profiles in US cities and pointed out that urban accessibility disparities are proportional to urban growth processes (Zipf's law of cities).Other researchers, indicated the need for a dynamic assessment of the 15-minute planning concept that considers people's spatiotemporal travel behavior (e.g., mobile phone signaling data) (Zhang et al., 2022), and others proposed a grammar-based method that optimizes 15-minute neighborhood layouts by minimizing street length and maximizing the street integration and the distance from all parcels to thenearest amenities (Lima, 2022).
Yet, less is known about the drivers of 15-Minute urbanism and the advantages created by the underlying economies of scale.Additionally, the discrete planning cultures and development policies in European countries have led to different accessibility dynamics (Barrington-Leigh & Millard-Ball, 2019;Bartzokas-Tsiompras & Photis, 2019;ITF, 2019;Kompil et al., 2019), different understandings of proximity measures and solutions (Gil Solá & Vilhelmson, 2019), and different levels of urban sustainability (Lafortune et al., 2019) and walking infrastructure (Bartzokas-Tsiompras, 2022).The study of Kompil et al. (2019) demonstrated a universal method to measure accessibility to generic services across Europe and estimated that the average distance to the nearest daily amenity in metropolitan areas is 3.2 km (or about 45 min' walk) and 6.7 km for regional services (large facilities for sports, culture, education, health, public administration etc.).Nevertheless, they emphasized on distance-based accessibility analytics per country and level of urbanization and they did not analyze separately walking access by destination category and city.Therefore, more detailed walking accessibility models might be useful for policy makers to discover regional disparities in planning 15-Minute places and enhancing city competition for better spatial strategies and healthier cities.
To fill this gap, we construct a new metropolitan walkability index to rank and map the walking performance of 121 heterogeneous urban areas and 30 European countries.The visual analytics summarize the complex European geography of walking accessibility to daily amenities and supports policy makers in raising awareness for healthier environments and travel.Additionally, one-way ANOVA statistical tests are used to identify potential differences in the development of 15-minute cities due to economic and population-relatedfactors (i.e., GDP per capita, population density, and population size).The findings from this research pave the way for new research questions seeking to further discover the multidimensional impacts of hyperlocal and x-minutes cities on societies, health and environmental sustainability.

Materials and methods
This research was based on the new urban accessibility framework (ITF, 2019), developed by the EU, the International Transport Forum (ITF) and the OECD, using comparable and uniform data indicators.To measure the 15-Minute concept we utilized seven walking performance indicators, available on the OECD statistics web portal (https://stats.oecd.org/Index.aspx?DataSetCode=ITF_ACCESS), which are (population-weighted) average ratios of accessible destinations within 15 min (network-based accessibility) to nearby destinations within 1 km Euclidean distance (proximity) (see Figure 1).These indicators summarize the average effect of street network connectivity as well as land use and population distribution.They are suitable for cross-city comparisons since they are independent of city size and local amenities.For instance, the calculations of these metrics were based on the harmonized Functional Urban Area (FUA) geography, defined by the EU and OECD, which includes a dense city and its surrounding commuting zone (Dijkstra et al., 2019).
To construct the 15-Minute-Walking City (15-MWC) index, we selected seven essential categories of destinations (Moreno et al., 2021), which was initially based on proprietary (i.e.TomTom) and open geospatial datasets (e.g.openstreetmap, Urban Atlas 2012, and the INSPIRE population grid by JRC): i.e. schools (primary and secondary), food markets (such as supermarkets, grocery shops and bakeries), other people (proxy for opportunities), recreation facilities (such as theaters, cinemas and museums), restaurants/bars, health care facilities, and green areas.Transit stations were not included in the analysis as the main goal was to focus only on walking accessibility and not to involve faster or motorized modes of transport.In this way, the proposed 15-MWC differs from classic walkability measureswhich analyze macro-and micro-level built environment data (Fonseca et al., 2021) such as land uses, streetscape features, pedestrian infrastructure, safety and security, accessibility, and street network connectivityand focuses on destination access, which is the most fundamental walking need when people assess walking feasibility as they choose their travel mode (Alfonzo, 2005).
After obtaining all the walking performance indicators, the next step was to aggregate them and assign a value to each FUA.This was accomplished using the PROMETHEE II method, which is an outranking multicriteria approach (Brans & Vincke, 1985) for selecting the optimum solution from a list of options based on a set of multiple criteria.All indicators (criteria) for each destination were maximized, as the aim was to make cross-city comparisons about the application of the 15-Minute planning concept.We also selected an equal weighting scheme, as the objective was to identify cities with maximized values of walking performance in all destination categories.The preference function was the Type I Usual in each criterion, which produces binary results and corresponds to the difference between pairs of cities.This outranking method has previously been used in walkability studies (Ortega et al., 2020) and was conducted using Visual Promethee 1.4 software (Academic Edition), which calculates positive φ + (a), negative φ − (a) and net outranking flows (φ(a) = φ + (a) − φ − (a)).To rank FUA's performances, we normalized φ(a) values to a scale of 0-1, using the min-max method, where 1 pertains to the FUA with the most 15-minute walking accessible destinations.
Finally, a one-way ANOVA was conducted to analyze the main effects of GDP per capita, and population density levels (based on tertiles of the OECD metropolitan statistics data at FUA level: https:// stats.oecd.org/Index.aspx?DataSetCode=CITIES#) on the creation of 15-minute walkable cities.

Data indicators
Boxplots in Figure 2 display briefly how the raw values per walking performance indicator and sub-region are spread out.Western cities have the highest medians in almost all destinations except for green spaces and hospitals.The highest median performance of all groups, though, is observed in restaurants, as 82% of them that are within a 1 km radius are accessible within a 15-minute walk.Eastern cities, however, perform better than all other subregions both in green areas and health care destinations.On the other hand, Northern European cities present significant variation and the lowest performance in all destination categories.Southern cities demonstrate high values when it comes to food markets and restaurants.
Regarding the most important outliers, we see that Belfast is significantly lagging behind in many destination categories (i.e.food markets, schools, restaurants and hospitals) both in comparison to other northern cities and at the European level.The configuration of the street network, distribution of people, and land uses in Belfast are probably not well integrated and people need to spend more travel time to reach nearby opportunities.For example, only about 50% to 63% of nearby (1 km) health care facilities, food shops, schools and restaurants are accessible by Belfast's residents on foot within 15 min.To the contrary, in Karlsruhe, about 87% of close-by food markets and schools are accessible within a 15-minute walk.Bari, a highly compact and polycentric metropolitan area, exhibits an outstanding walking efficiency and outperforms all the other Southern European cities when it comes to green spaces accessibility.The city, although it provides a small number of parks or forests, has concentrated its population nearby green areas (<1 km), thus, 80% of them are accessible in 15 min walking.Nonetheless, the performance of other Mediterranean citieswith loose grids and sprawled street patterns (in their peripheries), such as Genoa (0.57), Nice (0.59), Toulon (0.60) and Marseille (0.63)in green destinations is disappointing.These data are alarming, as a European health impact assessment study estimated high mortality burdens due to a lack of accessible urban parks or forests (Barboza et al., 2021).
Additionally, Venice, a highly touristic city, has been identified as a negative outlier in restaurants performance in southern cities.This outcome likely demonstrates a strong spatial mismatch between the local population and eatery locations, which may be present due to a geographically concentrated pattern of the night-and leisure-time economy.Stuttgart (0.84) displays significantly higher performance in schools' access than the other Western European cities, while Nice (0.76) and Leipzig (0.75) are characterized as negative outliers in the group.Of Eastern European metropolitan areas, Krakow has the most problematic accessibility performance in food shops, whereas of the northern cities, Stockholm (0.83) and Riga (0.83) outperform their peers in walking accessibility to hospitals.Finally, Nicosia (0.67), Bonn (0.73) and Leipzig (0.74) demonstrate, at the subregional level, significantly higher issues in reaching recreation facilities.

Index scores and rankings
The rankings of the 15-MWC Index per FUA are presented in Figure 3 and the scores are displayed in Figure 4.It is impressive to underline that almost six out of the top ten metros are found in Germany: namely, Frankfurt, Karlsruhe, Stuttgart, Rennes, Brno, Aachen, Mannheim, Thessaloniki, Geneva and Leipzig.Between the 10th and 20th positions are included some major European cities, such as Barcelona, Budapest, Nuremberg, Hannover and Zurich.Cities of Central Europe, including Germany, have benefited from post-war reconstruction policies that gradually focused on urban form efficiency and sustainable urban mobility (Buehler et al., 2017).In other highly walkable metros of the European periphery, for cultural reasons, rapid urbanization has contributed to the development of ultra-dense and mixed communities.On the other hand, eight out of ten cities in the bottom of the ranking are located in the UK.Specifically, the worst metros across Europe are Belfast, Liverpool, Birmingham, Manchester, Leeds, London, Cardiff, Dublin, Newcastle upon Tyne and Lisbon.Probably, the movement of garden cities, and car-based planning cultures such us modernism, have led to suburban street patterns, zoning, and car-dependence (The Prince's Foundation, 2020), which in turn have not allowed density, diversity and connectivity to thrive and make them walkable (Ewing & Cervero, 2010).
Even within the countries, the differences in developing 15-Minute-Walking cities are stark (see Figure 4).In Belgium, the scores of Antwerp and Brussels differ by more than 0.43 points compared to the smaller cities of Ghent and Liege.In the Czech Republic, Brno differs by almost 0.45 points in relation to Ostrava and Prague.In Austria, Vienna scores more than 0.34 points compared to Graz and Linz.In Poland, Wroclaw differs by 0.24 points to the second-best Polish city Wroclaw.In Italy, the leading city of Bari scores 0.19 more points than the second, Turin, and 0.62 more points than the worst Italian performer, Genoa.In France, southern and coastal cities have less walkable forms than the central and northern cities (except for Lille).For instance, the leading French city, Rennes, differs by 0.19 points in comparison to the secondplaced city, Strasbourg, and 0.64 points in relation to the bottom-ranked French City, Nice.In Germany, Augsburg, Berlin, Dusseldorf, Cologne and Bonn present more car-oriented structures and are significantly lagging behind at the national level.Particularly, the scores range between Frankfurt and Augsburg by roughly 0.64 points.In Spain there is also a large difference between the ultra-dense and mixed-use city of Barcelona (0.87) and the more dispersed island city of Las Palmas (0.41).In the Netherlands, Eindhoven (0.71) stands out, as it presents the highest national performance in all destinations except for hospitals and recreation.Similarly, in Sweden, Malmoa role model in urban sustainabilityoutperforms Stockholm and Gothenburg in all destinations except for health care facilities.

Statistical analysis
ANOVA revealed statistically significant differences in 15-MWC scores due to levels of metropolitan GDP per capita (as tertiles) (F(2,115) = 4,824, p = 0.01) but no main effects of population density (p = 0.116), and population size (p = 0.912) were found (see Figure 5).A Tukey's HSD test highlighted that there is a statistically significant difference in mean scores between FUAs of lower (<39.042USD per capita) and higher (>50.995USD per capita) economic productivity levels (p = 0.007, 95% C.I. = [−0.302,−0.039]).Nevertheless, there was no statistically significant difference in the average scores between lower-and medium-productivity FUAs (p = 0.14) or between medium and higher-productivity FUAs (p = 0.464).

Discussion and concluding remarks
This study ranked and mapped the development of 15-Minute walkable destinations using comparable metropolitan accessibility and proximity indicators, independent of city size, across Europe.The assessment method was based on PROMETHEE II multicriteria approach and helped us to summarize and visualize, for the first time, the complex European geography of walking performance in seven crucial destination categories.The new composite tool analyzes the average effects of walking accessibility indicators and can feed the ongoing discussion about the resilience of our car-dependent cities when multiple crises breaks out together.To that end, the findings can be used by decision-makers to identify, in each emergency, the most vulnerable societies.Hyperlocal and independent communities can be either an advantage (e.g.oil-vulnerability, energy crisis) as well as a public health risk (e.g. the spreading of communicable diseases).However, the scores do not necessarily associate with increased walking behavior in the city, and the index should not be used to assess mobility outcomes.Instead, the results can allow policymakers to conduct urban efficiency comparisons and revise regional development strategies in the European battle towards sustainability and carbon-neutral cities.
Visual analytics of rankings show significant regional-and national-level disparities in planning outcomes and demonstrate the European need for more integrated land-use and transportation policies to support mobility decarbonization and active travel.Infill, mixed-use and fair redevelopment plans can be beneficial solutions for providing more walkable communities where people and opportunities live closer together.Difficulties in designing the 15-Minute-Walking City are indicated mainly in the Anglo-Saxon region, in Portugal, and in many Mediterranean cities of France and Italy (except for Bari), while leading cities in walkable accessibility are chiefly found in West Germany (i.e.Frankfurt, Stuttgart and Karlsruhe).These results are in contrast to the findings of Kompil et al. (2019), where UK or Italian metropolitan regions, for example, presented higher walking or cycling accessibility as the average distances to the nearest daily service were lower than in Germany.However, our method does not depend on a single metric (e.g.average distance) and it has a composite nature which takes into account either the urban morphologies within neighborhoods (accessibility to proximity) and the diversity of destinations (separate metrics for several amenity types).Importantly, statistical tests indicated that European disparities in walking accessibility policies are due to economic inequalities between cities, and there is no effect of population density, and size in the results.It is hypothesized that larger urban economies provide a significantly higher number of urban amenities and, in turn, can serve more residents and multiple places.
As a result, people living either in more dispersed or compact but more prosperous settlements can reach a minimum number of neighborhood amenities quickly on foot.Therefore, more disadvantaged cities wishing to increase their walking accessibility performance should apply stricter planning measures that limit residential sprawl and repair the spatial mismatch between people and opportunities.Otherwise, the benefits of walkable urbanism could be an advantage only for the wealthier societies.This analysis is not without limitations.The index put emphasis on the residential environment of cities, as the data indicators are weighted by the population of the area.Hence, jobs are not included in the method, which might be a significant factor in designing hyperlocal and x-minute cities.Moreover, the approach of easy-to-communicate cumulative opportunities, which used accessibility metrics of 15-minutes and 1 km thresholds, might offend some cities where many destinations can be reached in just a few extra minutes or meters.A potential solution to this could be to use the more theoretically sound gravity measures that decrease the significance of larger distances compared to easier-to-reach locations (El-Geneidy & Levinson, 2021).Hopefully, however, researchers indicate that the correlation of these two approaches is strong and above 0.9 (Palacios & El-geneidy, 2022).Finally, the method also lacks the temporal dimension of spatial accessibility, which is important for specific social groups (e.g.night workers) and activities (e.g.shopping), while the average metropolitan values of accessibility indicators might mask significant disparities in urban morphologies of the central city and the outskirts.
Future studies can expand on our data and integrate micro-scale built environment indicators (Bartzokas-Tsiompras et al., 2021b) to construct a multiscale walkability assessment tool or an oil-vulnerability index (Wang et al., 2016) for Europe.Moreover, the development of x-minutes city indices for all modes of transport can support European-level investigations regarding the association of spatial access, travel behavior and environmental justice.

Figure 2 .
Figure 2. Box plots of walking performance metrics per destination category and subregion (Note: regions defined by the United Nations Geoscheme -Standard M49) (Data source: ITF, 2019).

Figure 4 .
Figure 4. 15-MWC Scores per FUA, country and subregion (countries in the y-axis are ordered by median values).

Figure 5 .
Figure 5. Mean plots between the 15-MWC Index and groups of average population density, population size, and GDP per capita (FUA level).