Green walking networks for climate change adaptation

Green walking networks for climate change adaptation

Transportation Research Part D xxx (2015) xxx–xxx Contents lists available at ScienceDirect Transportation Research Part D journal homepage: www.els...

2MB Sizes 4 Downloads 59 Views

Transportation Research Part D xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Transportation Research Part D journal homepage: www.elsevier.com/locate/trd

Green walking networks for climate change adaptation Salvatore Caprì, Matteo Ignaccolo, Giuseppe Inturri, Michela Le Pira ⇑ Department of Civil Engineering and Architecture (DICAR), University of Catania, Via Santa Sofia 64, 95100 Catania, Italy

a r t i c l e

i n f o

Article history: Received 17 October 2014 Revised 3 July 2015 Accepted 13 August 2015 Available online xxxx Keywords: Climate change Adaptation Mitigation Green spaces Pedestrian mobility Thermal comfort Accessibility Centrality indexes

a b s t r a c t Climate change (CC) potentially affects people travel behaviour, due to extreme weather conditions. This is particularly true for pedestrians, that are more exposed to weather conditions. Introducing the effect of this change in transport modelling allows to analyse and plan walking networks taking into consideration the climatic variable. The aim of this work is to develop a tool that can support planning and design of walking networks, by assessing the effects of actions oriented to increase resilience with respect to extreme weather conditions (CC adaptation). An integrated approach is used, thus combining transport and land-use planning concepts with elements of outdoor thermal comfort and network accessibility. Walking networks are analysed through centrality indexes, including thermal comfort aspects into a general cost function of links and weighted nodes. The method has been applied to the walking network inside the Campus of the University of Catania (Italy), which includes different functions and where pedestrian paths are barely used by people. Results confirm that this tool is sensitive to the variables representing weather conditions and it can measure the influence of CC adaptation measures (e.g. vegetation) on walking attitude and on the performance of the walking network. Ó 2015 Elsevier Ltd. All rights reserved.

Introduction Most of the current literature focuses on the relationship between climate change (CC) and transport-related greenhouse gases (GHG) emissions. Mitigation of climate change has been the mainstream research and planning practice in the last decades, mainly oriented to reduce GHG emissions. Fewer contributions can be found on adaptation to the impacts of extreme weather conditions deriving from CC. A great number of studies recommend new approaches to urban and transport planning as solutions to climate change mitigation. Examples include Smart Growth (ICMA, 2002), Transit Oriented Development (Calthorpe, 1993) and Transport Demand Management (Ignaccolo et al., 2006a). All of them advocate for development of high mixed-use densities and walkable communities in areas close to public transport nodes. These are arguably the most effective development strategies for increasing the number of trips at a walkable distance and favouring the access to transit stops and stations. Besides, promoting the shift towards more sustainable transport modes (mainly walking and cycling) is one of the best ways for limiting the increase in motorisation. Therefore it should be a priority for local authorities, especially in emerging countries. However, CC adaptation requires space for green and blue infrastructure within and around built areas, because too high densities can exacerbate the urban heat island effect and increase the likelihood of urban flooding. This conflict is an example ⇑ Corresponding author at: Via Santa Sofia 64, 95123 Catania, Italy. Tel.: +39 0957382220. E-mail addresses: [email protected] (S. Caprì), [email protected] (M. Ignaccolo), [email protected] (G. Inturri), [email protected] (M. Le Pira). http://dx.doi.org/10.1016/j.trd.2015.08.005 1361-9209/Ó 2015 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Caprì, S., et al. Green walking networks for climate change adaptation. Transport. Res. Part D (2015), http://dx.doi.org/10.1016/j.trd.2015.08.005

2

S. Caprì et al. / Transportation Research Part D xxx (2015) xxx–xxx

of how mitigation actions could compromise adaptation objectives. It suggests that climate change must be tackled by integrating mitigation measures, such as limiting the production of GHG emissions, with adaptation measures, in order to prepare our cities to its inevitable impacts. An urban green infrastructure network (Inturri and Ignaccolo, 2011) is seen as a good solution, both for CC mitigation and adaptation. It can favour people walking from residential areas across public open spaces and natural green corridors, to urban activities even under extreme weather conditions. Green infrastructure networks have both the potential to support a modal shift from road to walking, thus reducing GHG emissions, and to increase the resilience providing greater human comfort. There are guidelines to build effective green walking networks (Martincigh, 2002, 2005) and some studies show that street trees have the greatest per-unit-area potential for cooling and the greatest potential to alter the albedo of urban surfaces compared with other vegetation (NYC Regional Heat Island Initiative, 2006; Ali-Toudert et al., 2005; Ali-Toudert and Mayer, 2007). In this paper a method is presented, which establishes a functional link between climate issues and ‘walkability’ aspects. The aim of the research is to understand how climatic conditions and adaptation measures can affect the performance of a walking network, in order to assist the decision-making process in urban planning and design. Thermal comfort and travel behaviour The relationship between climate change and transport behaviour is quite an unexplored area. Many studies focus on the impact of climate change on transport and refer to increase in travel times, congestions and infrastructure disruptions (Koetse and Rietveld, 2009; Tsapakis et al., 2013). Nevertheless, there are few attempts to effectively link climate-related variables and thermal comfort issues with travel behaviour and modal shift (Saneinejad et al., 2012). Sabir et al. (2008) used a multinomial logit mode choice model and found that people switch from car to bicycle as temperature increases. However, if temperatures reach levels higher than 25 °C, people switch back from bicycle to car and public transport. Aaheim and Hauge (2005) carried out a survey in Bergen (Norway) regarding travel attitude and climatic data, showing that climate change would take to an increase in public transport, in walking and cycling. Clearly these results are significant for the specific case study: while in Bergen high temperatures lead to better walking conditions, evidence shows that, when temperature rises in hot and dry climate cities, people switch to air-conditioned cars and the attitude to walk and cycle decreases. By the way, this leads to enhanced energy requirement and further exacerbates climate change. There is a wide literature about reproducing pedestrian behaviour at the very fine scale in closed and open environments (e.g. by means of agent-based models: Bruse, 2007; Camillen et al., 2009). Nevertheless, the quantitative relationship among thermal comfort of pedestrians, microclimate variables and walking attitude at the scale of the street network is still an unexplored issue. Thermal comfort is defined by ASHRAE (1966) as ‘that condition of mind which expresses satisfaction with the thermal environment’. The substantial difference between indoor and outdoor comfort lays in solar radiation and wind chill. One of the most important parameters which governs the human energy balance is the mean radiant temperature, defined as ‘uniform temperature of an imaginary enclosure in which the radiant heat transfer from the human body equals the radiant heat transfer in the actual non-uniform enclosure’ (ASHRAE, 2001). The proposed method uses the ‘Universal Thermal _ Climate Index’ or UTCI (Jendritzky et al., 2007; Błazejczyk et al., 2010) as a comprehensive indicator of the temperature perceived by pedestrians linked with the thermal sensation, which takes into consideration solar radiation through the mean radiant temperature. It is considered a suitable index in order to incorporate the climate variable into the attributes influencing walking attitude. Methodology The methodology proposed is based on the analysis of walking networks through node centrality indexes taken from a revision of the Multiple Centrality Assessment method (Porta et al., 2008). The concepts of node accessibility, equivalent walking distance (Wibowo and Olszewski, 2005) and thermal comfort are integrated using a general cost function of links and weighted nodes. Walking-specific attributes and a thermal comfort measure are introduced into the model, since they can contribute significantly to the best path choice by pedestrians. The procedure consists of the following elements: (1) a ‘network model’ including the representation of origin and destination (OD) sites (zones), walking segments and their properties (links), intersections of segments (nodes); (2) a ‘cost function’ that links climatic attributes and walking-specific issues (slope, stairs, etc.) and the ‘walkability’ (the cost) of each segment of the network; (3) ‘centrality indexes’ used here as accessibility measures, both global (at the level of the whole network) and local (at the zone level), in order to carry out evaluations and comparisons for different situations.

Please cite this article in press as: Caprì, S., et al. Green walking networks for climate change adaptation. Transport. Res. Part D (2015), http://dx.doi.org/10.1016/j.trd.2015.08.005

S. Caprì et al. / Transportation Research Part D xxx (2015) xxx–xxx

3

The cost function is used within the network model for characterising links and finding best routes among model zones in order to build minimum cost matrices. The different centrality indexes are calculated, based on these matrices, thus representing a sort of ‘climatic accessibility’ of the single zone or the whole campus. Network model In order to represent the actual walking paths, it is needed to build a network model, consisting of the following elements:  Links, which characterise all kinds of pedestrian paths, including sidewalks, separated walking segments, greenways, ramps, stairs, pedestrian road crossings, and also links which are within the buildings; links are characterised by a set of attributes, which includes physical features (length, slope, type of path, etc.) and climate-related information (surface material, type of surrounding built environment, etc.); links should be bi-directional because attributes are generally different in the two directions.  Nodes, which are used to connect more than two links each other or, eventually, to allow link features’ variation or to introduce particular penalties.  Zones, representing buildings or, generally, all origin and destination sites within the study area; since internal paths are explicitly modelled, zones are connected to the rest of the network by means of fictitious links, whose length and cost is set equal to zero. The resulting walking network model is more detailed and complex than the corresponding road network model. A pedestrian traffic assignment model can be applied, in order to distribute the O/D demand matrix on the alternative paths, if both the procedure and the demand estimate are available. Cost function The core of the method is the cost function of the links, which should contain all the necessary elements to reflect the cost of walking as perceived by users. For planning purposes, the walking generalised cost function is usually expressed in terms of equivalent distance (Wibowo and Olszewski, 2005; Ignaccolo et al., 2006b). Here the concept of equivalent distance is extended and a further element is introduced in order to make the model sensitive to thermal comfort. The equivalent distance (ED) formulation adopted in this paper, for the generic link i, is the following one:

EDðiÞ ¼ ð1 þ a  sðiÞÞLðiÞ þ b1 XðiÞ þ b2 AðiÞ þ b3 CðiÞ

ð1Þ

where EDðiÞ is the equivalent distance for the link i; LðiÞ is the length of the link i; sðiÞ is the ascending slope of the link i; XðiÞ is the number of road crossing along the link i; AðiÞ is the number of ascending steps on the link i; CðiÞ is the number of traffic conflicts on the link i; a; b1 ; b2 ; b3 are coefficients to be determined in experimental way. This formulation determines a sort of network actual distances’ ‘distortion’ due to link attributes different from the distance that increase people’s walking effort. In order to take into account the thermal comfort, which is generally different for each link of the network, a Climatic Multiplier (CM) is introduced, depending on UTCI temperature (T UTCI ) introduced in ‘Thermal comfort and travel behaviour’. UTCI derives from the energetic model ‘UTCI-Fiala’ (Fiala et al., 2012). It simulates the thermal exchange phenomenon inside a human body and his surface by taking into account anatomic, thermal and physiologic properties: it is conceived with the aim to be universally applicable, for all type of climates, seasons and scales. The UTCI is defined, for a given combination of meteorological variables, as the air temperature (T a ) of the reference _ condition causing the same model response as actual conditions (Błazejczyk et al., 2013). It represents an equivalent temperature that can be associated with thermal stress, e.g. if UTCI is between 18° and 26° we are in the so called ‘thermal comfort zone’ (The Commission for Thermal Physiology of the International Union of Physiological Sciences, 2003). UTCI provides a measure of the equivalent temperature taking into account the following attributes:    

the the the the

effective air temperature; relative humidity; wind speed; difference between the air temperature and the mean radiant temperature.

Please cite this article in press as: Caprì, S., et al. Green walking networks for climate change adaptation. Transport. Res. Part D (2015), http://dx.doi.org/10.1016/j.trd.2015.08.005

4

S. Caprì et al. / Transportation Research Part D xxx (2015) xxx–xxx

In order to correlate the walking impedance with thermal stress, we make use of the concept of Dynamic Thermal Sensation (DTS), which is a measure of the thermal perception under rapidly changing transient conditions based on the symmetrical seven-point ASHRAE scale of thermal stress [3; +3]. UTCI is correlated to DTS (Bröde et al., 2012); we linked the walking effort with TUTCI by means of the thermal stress, as reported in Fig. 1. We assume that the walking effort – represented by the climatic factor that multiplies the actual distance – increases proportionally to thermal stress, reaching its maximum values for very low and very high temperatures and being equal to 1 when DTS ¼ 0 (neutral thermal sensation). The mathematic formulation of this correlation is:

CMi ðT UTCI ðiÞÞ ¼ 1 þ jDTSðT UTCI ðiÞÞj

ð2Þ

Combining (1) and (2), the final climatic equivalent distance (CED) formulation for the generic link i is the following one:

CEDðiÞ ¼ CM i ½ð1 þ asðiÞÞLðiÞ þ b2 AðiÞ þ b1 XðiÞ þ b3 CðiÞ

ð3Þ

where only the terms of the ED formulation (1) depending on temperature (i.e. length, slope and steps) are multiplied by CM. The coefficients a; b1 ; b2 ; b3 should be determined in experimental way. For a first test of the methodology, b1 ; b2 ; b3 can be considered those of the original formulation of Wibowo and Olszeski (2005). The additional coefficient a regarding ascending slope can be derived by correlating it with the one related to ascending steps: a  sðiÞLðiÞ ¼ b2 AðiÞ where an hypothetic step (AðiÞ ¼ 1) of 18 cm high and 1 m long (LðiÞ ¼ 1 m) corresponds to a 18% slope (sðiÞ ¼ 18%). Following this approach, (3) becomes:

CEDðiÞ ¼ CM i ½ð1 þ 0:156sðiÞÞLðiÞ þ 2:81AðiÞ þ 55:4XðiÞ þ 36:31CðiÞ

ð4Þ

It means, for example, that a 300 m slight uphill walk (2% of slope) with 2 road crossings, in ideal climatic conditions (CM ¼ 1), and without other contingent elements (ascending steps and traffic conflicts) corresponds to a walking effort of more than 500 m (CEDðiÞ ¼ 1  ½ð1 þ 0:156  2Þ  300 m þ 2:81  0 þ 55:4  2 þ 36:31  0 ¼ 504:4 m). Centrality indexes Accessibility can generally be defined as the ability to reach a destination. Accessibility measures can be used to assess to what extent the transport network is adequate to connect origins and destinations. A quite comprehensive review of accessibility measures can be found in Rubulotta et al. (2013). The Multiple Centrality Assessment (MCA) by Porta et al. (2008) is a simple and effective method to analyse network accessibility in terms of centrality indexes. MCA provides the analysis of a network in terms of primal graphs, following the rule of the ‘road-centreline-between-nodes’ and computing some indexes representative of different ways of ‘being central’, in terms of:     

number of nodes directly connected to a given node (degree centrality); proximity to the other nodes through the shortest paths (closeness centrality); position along a high number of shortest paths connecting each couple of nodes (betweenness centrality); straightness of the shortest paths from a given node with reference to the Euclidean distance (straightness centrality); resilience of the network once a given node (and relevant connected links) is suppressed (information centrality).

It can be noted that MCA networks are basically node-based, so that centrality measures are defined and computed for each node of the network, making this method more suitable for dense urban areas, where land use types are homogeneously distributed over the urban fabric. For the purpose of this paper, ‘zones’ have been introduced into the model, i.e. fictitious nodes representing a OD site, as a further element for the computation of centrality indexes. Each zone has one or more ‘weights’, which represent the dimension of the relevant site, e.g. in terms of resident population, employees, surface, etc., depending on the type of analysis which is carried out. Four centrality indexes have been adapted to this method, in order to provide consistent and different accessibility measures for the zones of the network: 1. Closeness, defined as the inverse of the mean distance of shortest paths from the zone i:

N1 CðiÞ ¼ P j2G;i–j dij

ð5Þ

2. Remoteness, defined as the mean distance of the shortest paths from the zone i (it gives the opposite indication than the closeness, but being a distance it has a useful physical meaning):

P

RðiÞ ¼

j2G;i–j dij N1

ð6Þ

Please cite this article in press as: Caprì, S., et al. Green walking networks for climate change adaptation. Transport. Res. Part D (2015), http://dx.doi.org/10.1016/j.trd.2015.08.005

5

S. Caprì et al. / Transportation Research Part D xxx (2015) xxx–xxx

Relaonship walking effort (CM) - thermal stress (DTS) equivalent temperature (TUTCI) 3

4

2 1 0

2

-1

1

CM 0 -12

DTS

CM

3

-8

-4

0

4

8

12

16

20

24

28

32

DTS 36

40

-2 -3

TUTCI [°C] Fig. 1. Correlation among DTS-UTCI-CM.

3. Betweenness, defined as the mean percentage of shortest paths between every couple of zones that pass through the zone i:

BðiÞ ¼

X njk ðiÞ 1 ðN  1ÞðN  2Þ j;k2G;j–k–i njk

ð7Þ

Also a betweenness 10 index is defined, as the mean percentage of short paths (shorter than 110% of the shortest paths) between every couple of zones that pass through the zone i. 4. Straightness, defined as the mean ratio between the Euclidian distance and the corresponding shortest path on the network from the zone i to all the other zones:

SðiÞ ¼

Eucl 1 X dij N  1 j2G;i–j dij

ð8Þ

Also three global centrality indexes have been defined in order to assess the overall performance of the network: 5. Global Straightness (or Efficiency), defined as the mean ratio between the Euclidean distance and the shortest path length for each couple of zones, that can be weighted by the selected origin and destination weights:

2 3 P dEucl X6 j2G;i–j wj dijij 7 GS ¼ P 4 P 5 i2G wi i2G j2G;i–j wj 1

ð9Þ

6. Global Closeness, defined as the inverse of the mean distance of the shortest paths connecting each couple of zones, that can be weighted by the selected origin and destination weights:

GC ¼ P

" P # X wj P j2G;i–j i2G wi i2G j2G;i–j wj dij 1

ð10Þ

7. Global Remoteness, defined as the mean distance of the shortest paths connecting each couple of zones, that can be weighted by the selected origin and destination weights:

GR ¼ P

"P # X wj dij Pj2G;i–j i2G wi i2G j2G;i–j wj 1

ð11Þ

Here follows the notation used in the given equations: G – the set of zones, N – the number of zones, dij – the actual (or equivalent) shortest distance between zones i and j along the network, Eucl

dij – the Euclidean distance between zones i and j, nij – the number of shortest paths between zones i and j, nij ðkÞ – the number of shortest paths between zones i and j passing through the zone k and, wi – the weight of the zone i. The proposed approach uses a network model, which is based on a graph where links represent roads (or walking segments) and nodes represents road junctions. This is the so-called ‘primal’ approach, in contrast to the ‘dual’ approach of the Space Syntax method (Hillier, 1996; Hillier and Hanson, 1984), where streets are turned into nodes and intersections Please cite this article in press as: Caprì, S., et al. Green walking networks for climate change adaptation. Transport. Res. Part D (2015), http://dx.doi.org/10.1016/j.trd.2015.08.005

6

S. Caprì et al. / Transportation Research Part D xxx (2015) xxx–xxx

are turned into links (also defined ‘indirect’ approach). The method described in this paper can also be applied to indirect network representation, when links (streets) are the main focus of the analysis, e.g. in a dense urban environment. Geographic information, which generally gets lost after graph transformation, should be stored in the attributes of the dual graph elements in order to allow walking cost calculations. In conclusion, as stated before, the methodology presented is able to assess the performance of a walking network (‘network model’) in terms of accessibility to zones (‘centrality indexes’) taking into account walking-specific attributes and a thermal comfort measure (‘cost function’).

Case study The methodology has been applied in order to analyse the walking network inside the campus of the University of Catania, one of the biggest open spaces in Catania, a 70 ha area experienced by students, teaching staff and employees. It contains a number of OD sites spread over the area, including lecture rooms, offices, student residences, parking lots and few utilities; car access is allowed and parking is still free in most areas. There is a general lack of green spaces, pedestrian/cycling paths and connection among different parts and the actual network of pedestrian paths is poorly experienced (see Fig. 2). The methodology should support the design of the walking network, which can include further green paths, aiming at promoting sustainable mobility and increasing the functionality and liveability of the campus. Catania, a medium-sized city (300,000 inhabitants) located in the eastern part of Sicily (Italy), appears particularly vulnerable to climate change as a consequence of the starting high base temperature as well as the peak temperature values, deriving from the location in the south of Europe and, above all, from the strong influence of the Saharan region.

Network model The network model was built by considering:  links: actual pedestrian links and potential pedestrian links, with all the related physical features (length, number of steps, slope, etc.) and climate-related information (surface material, type of surrounding built environment). They are categorised into 9 main typologies of links, which are combinations of surface type (sidewalk, paved or unpaved link) and surroundings’ features (buildings, green spaces, both of them), as shown for example in Fig. 3. In some scenarios also internal links are considered;  nodes, as explained in ‘Network model’;  zones, representing sites, accesses to the campus area and parking spaces (both parking lots and on street parking); as pointed out in ‘Network model’, zones are connected to the rest of the network by means of fictitious zero-length links. The sites are weighted according to the number of daily users (teaching staff and students). The resulting network model consists of: 342 nodes, 862 links, 20 sites (numbered from 1 to 20), 5 accesses (from n. 220 to 224), 23 parking spaces (14 parking areas and 9 on street parking – from n. 300 to 322) (see Fig. 4).

Fig. 2. Campus of the University of Catania.

Please cite this article in press as: Caprì, S., et al. Green walking networks for climate change adaptation. Transport. Res. Part D (2015), http://dx.doi.org/10.1016/j.trd.2015.08.005

S. Caprì et al. / Transportation Research Part D xxx (2015) xxx–xxx

7

Fig. 3. Typologies of sidewalks according to the surrounding environment (from left to right: buildings, green spaces, both of them).

Fig. 4. The network model of the University campus.

Climatic Multiplier evaluation Once the links are categorised according to the physical and climate-related features, it is necessary to evaluate the Climatic Multiplier (CM) related to the UTCI index (T UTCI ) to be associated with each link. The T UTCI inputs are: air temperature (T a ), relative humidity (R:H:), wind speed (v a ) and the difference between mean radiant temperature and air temperature (T r  T a ). A quite extreme summer scenario is simulated by considering typical values of a hot summer day, i.e.:

T a ¼ 33  C R:H: ¼ 80%

Please cite this article in press as: Caprì, S., et al. Green walking networks for climate change adaptation. Transport. Res. Part D (2015), http://dx.doi.org/10.1016/j.trd.2015.08.005

8

S. Caprì et al. / Transportation Research Part D xxx (2015) xxx–xxx

v a ¼ 10 km=h ¼ 2:78 m=s The mean radiant temperature T r , defined in ‘Thermal comfort and travel behaviour’, depends on the type of link. The methodology by Gill (2006), implemented in a Mathematica procedure, has been used to determine the temperature of the surface of a certain area with a small number of input data. This methodology derives from the energy exchange model developed by Whitford et al. (2001), based on the model of urban climate by Tso et al., 1991. Necessary input parameters are:  the building mass per unit area (mc (t/m2));  the percentage of evaporative surface (evaporative fraction e:f:). We use our link categorisation to derive mc and e:f: and the output temperature of the model is assumed, for our purpose, as the mean radiant temperature of the link, considering that it is ‘a measure of the combined effect of surface temperatures within a space’ (Gill, 2006). Then, T UTCI is computed1 in each case and the related CM value is derived (see Table 1): Table 1 Equivalent temperature (T UTCI ) and climatic multiplier (CM) according to the type of link and the surrounding environment.

a

Type of link

Surrounding environmenta

mc (t/m2)

e.f.

DT = Tr  Ta (°C)

TUTCI (°C)

CM

Sidewalk Sidewalk Sidewalk Paved path Paved path Paved path Unpaved path Unpaved path Unpaved path

B GS B + GS B GS B + GS B GS B + GS

2.28 0 1.14 2.7 0 1.44 2.91 0 1.455

0.001 0.76 0.38 0.001 0.96 0.48 0.031 1 0.515

15 0 3 13 0 0 10 0 0

40.6 37.8 38.3 40.2 37.8 37.8 39.6 37.8 37.8

3.76 3.35 3.44 3.75 3.35 3.35 3.67 3.35 3.35

‘B’ stands for buildings, ‘GS’ for green spaces, ‘B + GS’ stands for both buildings and green spaces.

Moreover, a link can be protected from wind or solar radiation, resulting in different values of CM. University buildings were originally planned to be crossed in order to connect some of them without interfering with research work and teaching. So, internal paths are also considered in the model, assuming ideal climatic condition for relevant internal links, i.e. CM ¼ 1. Indeed buildings are air-conditioned and, especially in summer, they provide better climatic conditions than the external paths. It is worth to point out that crossing air-conditioned buildings may have some negative consequences: from one side, airconditioning results in increasing energy consumption, and it surely has negative effects on climate change mitigation, while ‘adapting’ quite well to extreme climatic conditions; from the other side, guiding pedestrian flows through buildings could ‘disturb’ the normal activities inside them. Nevertheless, in this context, the internal layout of the buildings does not cause any type of interference between pedestrian flows and students/workers. Therefore, all the possible combinations of DT, conditions of ‘wind protection’ and ‘internal link’, and the relative T UTCI and CM values, are listed below (see Table 2): Table 2 Equivalent temperature (T UTCI ) and climatic multiplier (CM) according to the link group. Link group

1

2

3

4

5

6

7

Wind protection Shadow Internal link DT TUTCI (°C) CM

No Yes No 0 37.8 3.35

Yes No No 0 39.2 3.60

Yes Yes Yes 0 21.9 1.00

No No No 3 38.3 3.44

No No No 10 39.6 3.67

No No No 13 40.2 3.75

No No No 15 40.6 3.76

Scenarios The efficiency of the network is evaluated through local (referred to zones) and global (referred to the whole network) centrality indexes, as described in ‘Centrality indexes’. In order to test the methodology and assess the effectiveness of climate-friendly measures, three scenarios have been considered:

1

By using the online tool available at http://www.utci.org/utcineu/utcineu.php.

Please cite this article in press as: Caprì, S., et al. Green walking networks for climate change adaptation. Transport. Res. Part D (2015), http://dx.doi.org/10.1016/j.trd.2015.08.005

9

S. Caprì et al. / Transportation Research Part D xxx (2015) xxx–xxx

Fig. 5. (a) Representation of the ratio between CED and the link length L in the reference scenario (from green to red, green meaning CED=L ¼ 1). (b) Improvements in CED with the greening of the network (scenario 2): difference between CED in the reference scenario and in scenario 2 (the thickness of the line reflects the value of the difference). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

 scenario 0, or reference scenario, with the present network consisting of all the links except the internal ones;  scenario 1, where the buildings can be ‘crossed’, i.e. internal links can be used as part of the shortest path between any two zones;  scenario 2, which is a project scenario, where a sort of general ‘greening’ of the network is made in order to increase its walkability from a climatic point of view. With reference to the last scenario, it consists of adding trees to the links, given their beneficial role in cooling and their greatest potential to alter the albedo of urban surfaces compared with other vegetation. Trees spread all along the network contribute in minimising the solar radiation (DT ¼ T r  T a ffi 0) and in reaching the most favourable (but not ideal) climatic conditions for the climate setting assumed, thus obtaining T UTCI ¼ 37:8  C and CM ¼ 3:35. Introducing a climate-dependent cost function has the effect of raising the equivalent distance, which is variable according to the climatic conditions, with respect to the actual length (Fig. 5).

Results and discussion Results Results are expressed in terms of centrality indexes. Global indexes have been computed for the three scenarios, taking into account different sets of Origin/Destination couples. In detail, Global Straightness and Global Remoteness have been calculated for the following zone sets:  Internal trips: only couples of internal zones.  Outgoing trips: internal origin (site) and external destination zone (access to the campus or parking space).  Ingoing trips: external origin and internal destination zone. Table 3 shows the global indexes in the three scenarios. Improvements can be noted when passing from scenario 0 to scenario 1 and from scenario 1 to scenario 2. It is worthy of notice that the first group of improvements are bigger than the second one. The model confirms that sometimes adopting ‘soft’ measures (to favour buildings’ crossing – e.g. placing route indications) may result in a greater performance than making more important ‘green’ interventions (e.g. planting trees along paths). This result is justified by the quantitative Table 3 Global indexes in the three scenarios. Global index

Reference scenario

Scenario 1

Var% (scenario 1 VS 0)

Scenario 2

Var% (scenario 2 VS 0)

Global Global Global Global Global Global Global

0.1846 0.1529 0.1802 1.9596 2.2666 1.8518 0.5103

0.1977 0.1711 0.1982 1.8151 2.0467 1.7022 0.5509

+7 +12 +10 7 10 8 +8

0.2051 0.1774 0.2053 1.7536 1.9810 1.6473 0.5703

+11 +16 +14 11 13 11 +12

Straightness (internal trips) Straightness (ingoing trips) Straightness (outgoing trips) Remoteness (internal trips) Remoteness (ingoing trips) Remoteness (outgoing trips) Closeness (internal trips)

Please cite this article in press as: Caprì, S., et al. Green walking networks for climate change adaptation. Transport. Res. Part D (2015), http://dx.doi.org/10.1016/j.trd.2015.08.005

10

S. Caprì et al. / Transportation Research Part D xxx (2015) xxx–xxx

assumptions made on the climatic factor: internal links have ideal climatic conditions (CM ¼ 1), while green and shadowed external links are inevitably affected by the extreme climatic conditions, thus contributing partially to improving the thermal comfort of pedestrians (CM ¼ 3:35 with respect to the worst case of CM ¼ 3:76). In order to understand the contribution of the methodology and the effect of the climatic factor CM, further simulations have been performed by considering only equivalent distance, (i.e. using ED instead of CED, as explained in ‘Cost function’). When no climate effect applies, scenario 2 coincides with scenario 1 (tree planting has no effect on equivalent distance), so just two scenarios are compared in this case (Table 4). It can be noticed that the climatic factor emphasises the role of internal links (which are enabled in scenario 1/2) in reducing shortest paths and improving performances, and no clear proportionality between the two cases is evident. At the zone scale, all the indexes have been computed for the three scenarios, except for betweenness and betweenness 10, that are not considered in the reference scenario because zones cannot be ‘between’ anything, since zone crossing is inhibited. In general, there is a progressive increase in closeness and straightness indexes from the reference scenario to the ‘improved’ scenarios and, consequently, a reduction in the remoteness, i.e. the average shortest path. The effect of the climatic multiplier can be evaluated by analysing the performance increase in terms of closeness centrality of sites and parking spaces in three additive steps:  the performance increase due to allowing buildings’ crossing, without considering the climatic multiplier and relevant benefit;  the marginal performance increase due to considering the climatic benefit of building crossing;  the marginal performance increase due to scenario 2 measures (planting trees) (Fig. 6). The increase due to the measures in scenario 2 is quite plain for all the zones, while the increase due to the use of buildings’ internal links varies over the zones, even when considering the effect of the climatic multiplier on the walking cost function. The impact of the climatic multiplier is not evidently proportional to the increase in performance, if compared with the case of the ‘non-climatic’ equivalent distance. That is one of the reasons why the proposed methodology can lead to different evaluations than ordinary planning methods do.

Table 4 Global indexes in the three scenarios without considering the climatic multiplier (no CM). Global index (no CM)

Reference scenario

Scenario 1/2

Var% (scenario 1 VS 0)

Global Global Global Global Global Global Global

0.5103 0.4398 0.5133 0.6914 0.8015 0.6824 1.4463

0.5236 0.4590 0.5270 0.6687 0.7699 0.6666 1.4954

+3 +4 +3 3 4 2 +3

Straightness (internal trips) Straightness (ingoing trips) Straightness (outgoing trips) Remoteness (internal trips) Remoteness (ingoing trips) Remoteness (outgoing trips) Closeness (internal trips)

Fig. 6. Percentage increase in closeness centrality for sites and parking spaces.

Please cite this article in press as: Caprì, S., et al. Green walking networks for climate change adaptation. Transport. Res. Part D (2015), http://dx.doi.org/10.1016/j.trd.2015.08.005

S. Caprì et al. / Transportation Research Part D xxx (2015) xxx–xxx

11

Fig. 7. Betweenness centrality in scenario 1 with climate effect (BC) and without climate effect (BC  CM).

Regarding the betweenness index, there is not a great difference in comparing scenarios 1 and 2. Betweenness10 is always quite higher than betweenness and some sites that are not central by considering only the shortest paths, become central if we also consider the other ‘almost shortest’ paths (there are 29 ‘between-zones’ against 45 ‘between-10-zones’). Again, the climatic variable emphasises the role of internal links in reducing shortest paths and improving performances, as shown in Fig. 7, where betweenness centrality in scenario 1 is compared with the same index computed without considering the climatic multiplier. Model validation Generally, in order to provide results, which can be used to take decisions in planning real walking networks, the model would need to be properly validated. The general model architecture is well consolidated in transportation systems engineering and do not need a specific validation process. Actually, the new element which is introduced by the present paper within the methodology is the Climatic Multiplier; even if it is built on naturally true assumptions, related to thermal comfort, and a sound scientific basis, it should pass through an empirical validation process. In order to do that, an extensive survey campaign is needed to be implemented: a set of simple walking study areas have to be selected, each containing all ‘weather-sensitive’ elements, especially in terms of sidewalks’ types, and a continuous monitoring has to be carried out (e.g. by means of video recording), during many different weather conditions, in order to measure pedestrians’ walking choices. Ideal study areas for the survey should contain few origins and destinations and many walking path alternatives and can be found, for example, in garden squares’ (or parks’) crossings in urban environment or within the University campus. A simple network model should be set up for each study area, including all considered elements for cost function computation. Model validation can be carried out by comparing observed and simulated behaviour by means of a Maximum Likelihood procedure (Robin et al., 2009).

Conclusion The proposed methodology is a first attempt to link weather conditions to walking attitude in order to support planning and design of pedestrian networks. It allows the assessment of the impact of climate-related actions (i.e. actions affecting thermal comfort) oriented to climate change adaptation on the performance of the walking network. The model mainly consists of the definition of a walking cost function, conveniently adapted to take into account the climatic variable, and a set of centrality indexes, which can assess the performance of the network. It is based on a sound theory, thus resulting very suitable to be modified and improved, and the relevant validation process has been outlined and it is in progress. The methodology has been applied to test the walking network inside the Campus of the University of Catania (Italy), where a hot summer scenario was considered, but it can be easily extended to a winter scenario, assessing the impact of low temperatures and wind chill. Results show that the climatic factor plays an important role in defining the shortest paths connecting the origin and destination zones inside the campus, thus influencing the centrality of them and the general efficiency of the network. Please cite this article in press as: Caprì, S., et al. Green walking networks for climate change adaptation. Transport. Res. Part D (2015), http://dx.doi.org/10.1016/j.trd.2015.08.005

12

S. Caprì et al. / Transportation Research Part D xxx (2015) xxx–xxx

References Aaheim, H.A., Hauge, K.E., 2005. Impacts of climate change on travel habits: a national assessment based on individual choices. CICERO Report 2005:07, Oslo. Available from: . Ali-Toudert, F., Djenane, M., Bensalem, R., Mayer, H., 2005. Outdoor thermal comfort in the old desert city of Beni-Isguen, Algeria. Clim. Res. 28, 243–256. Ali-Toudert, F., Mayer, H., 2007. Thermal comfort in an east–west oriented street canyon in Freiburg (Germany) under hot summer conditions. Theoret. Appl. Climatol. 87, 223–237. ASHRAE, 1966. Thermal comfort conditions. ASHRAE standard 55, 66, New York. ASHRAE, 2001. ASHRAE Fundamentals Handbook 2001. (SI Edition) ISBN: 1883413885. _ K., Broede, P., Fiala, D., Havenith, G., Holmér, I., Jendritzky, G., Kampmann, B., Kunert, A., 2010. Principles of the new Universal Thermal Climate Błazejczyk, Index (UTCI) and its application to bioclimatic research in European scale. Miscellanea Geographica 14 (2010), 91–102. _ K., Jendritzky, G., Bröde, P., Fiala, D., Havenith, G., Epstein, Y., Psikuta, A., Kampmann, B., 2013. An introduction to the Universal Thermal Climate Błazejczyk, Index (UTCI). Geographia Polonica 86 (1), 5–10. _ Bröde, P., Fiala, D., Błazejczyk, K., Holmér, I., Jendritzky, G., Kampmann, B., Tinz, B., Havenith, G., 2012. Deriving the operational procedure for the Universal Thermal Climate Index (UTCI). Int. J. Biometeorol. 56, 481–494. http://dx.doi.org/10.1007/s00484-011-0454-1. Bruse, M., 2007. Simulating human thermal comfort and resulting usage patterns of urban open spaces with a Multi-Agent System. In: Wittkopf, St., Tan, B. K. (eds.), Proceedings of the 24th International Conference on Passive and Low Energy Architecture PLEA, pp. 699–706. Calthorpe, P., 1993. The next American metropolis. Princeton Architectural Press, New York. Camillen, F., Caprì, S., Garofalo, C., Ignaccolo, M., Inturri, G., Pluchino, A., Rapisarda, A., Tudisco, S., 2009. Multi agent simulations of pedestrian behavior in closed spatial environments. In: IEEE XPLORE. Proceedings to TIC-SENCS Symposium on Engineered and Natural Complex Systems-Modeling, Simulation and Analysis. Toronto, Ontario, Canada, 26–27 Settembre 2009. vol. C.N.: CFP0956G CDR, pp. 375–380, ISBN: 978-1-4244-3878-5, doi: 10. 1109/TIC-STH.2009.5444471. Fiala, D., Havenith, G., Bröde, P., Kampmann, B., Jendritzky, G., 2012. UTCI-Fiala multi-node model of human heat transfer and temperature regulation. Int. J. Biometeorol. 56, 429–441. http://dx.doi.org/10.1007/s00484-011-0424-7. Gill, S.E., 2006. Climate Change and Urban Greenspace (Ph.D. thesis). University of Manchester, UK. Available from: . Hillier, B., 1996. Space is the Machine: A Configurational Theory of Architecture. Cambridge University Press, Cambridge. Hillier, B., Hanson, J., 1984. The Social Logic of Space. Cambridge University Press, Cambridge. ICMA (International City Management Association), 2002. Getting to smart growth: 100 policies for implementation. ICMA, Washington, DC. Available from: . Ignaccolo, M., Caprì, S., Giunta, U., Inturri, G., 2006a. Discrete Choice Model for Defining a Parking-Fee Policy on Island of Ortigia, Siracusa. J. Urban Plann. Dev. 132 (3), 147–155. Ignaccolo, M., Inturri, G., Caprì, S., 2006b. Localizzazione delle fermate di una linea di trasporto pubblico urbano mediante un approccio di rete per la rappresentazione dell’accessibilità pedonale. In: Ferrari, P., Cepolina, E.M. (Eds.), Didattica e ricerca nell’ingegneria dei trasporti. pp. 187–196, MILANO: Franco Angeli, ISBN/ISSN: 88-464-7501-1. Inturri, G., Ignaccolo, M., 2011. The Role of Transport in Mitigation and Adaptation to Climate Change Impacts in Urban Areas. Springer, Resilient Cities. Jendritzky, G., Havenith, G., Weihs, P., Batchvarova, E., de Dear, R., 2007. The Universal Thermal Climate Index UTCI goal and state of COST Action 730. In: Mekjavic, I.B., Kounalakis, S.N., Taylor, N.A.S. (Eds.), Environmental Ergonomics XII. Biomed, Ljubljana, pp. 509–512. Koetse, M.J., Rietveld, P., 2009. The impact of climate change and weather on transport: an overview of empirical findings. Transp. Res. Part D 14, 205–221. Martincigh, L., 2002. Urban quality and design for pedestrians. In: Fleury, D. (ed.), A city for pedestrians: policy making and implementation. Final Report COST Action C6. Office for Official Publications of the European Communities, Luxembourg. Martincigh, L., 2005. A green-network. The integration of the green structure and the non motorized transport modes network. COST Action C11 – greenstructure and urban planning. Available from: . NYC Regional Heat Island Initiative, 2006. Mitigating New York City’s heat island with urban forestry, living roofs, and light surfaces. Final report 06–06 October 2006. Available from: . Porta, S., Crucitti, P., Latora, V., 2008. Multiple centrality assessment in Parma – a network analysis of paths and open spaces. Urban Des. Int. 13, 41–50. http://dx.doi.org/10.1057/udi.2008.1. Robin, Th., Antonini, G., Bierlaire, M., Cruz, J., 2009. Specification, estimation and validation of a pedestrian walking behavior model. Transp. Res. Part B 43, 36–56. Rubulotta, E., Ignaccolo, M., Inturri, G., Rofè, Y., 2013. Accessibility and centrality for sustainable mobility: regional planning case study. J. Urban Plann. Dev. 139 (2), 115–132. Sabir, M., Koetse, M.J., Rietveld, P., 2008. The impact of weather conditions on mode choice decisions: empirical evidence for the Netherlands. Tinbergen Institute Discussion Paper, VU University, Amsterdam. Available from: . Saneinejad, S., Roorda, M.J., Kennedy, C., 2012. Modelling the impact of weather conditions on active transportation behaviour. Transp. Res. Part D 17, 129– 137. The Commission for Thermal Physiology of the International Union of Physiological Sciences, 2003. Glossary of Terms for Thermal Physiology (2003). J. Therm. Biol. 28, 75–106. Tsapakis, I., Cheng, T., Bolbol, A., 2013. Impact of weather conditions on macroscopic urban travel times. J. Transp. Geogr. 28, 204–211. Tso, C.P., Chan, K., Hashim, A., 1991. Analytical solutions to the near-neutral atmospheric surface energy balance with and without heat storage for urban climatological studies. J. Appl. Met. 30, 413–424. Whitford, V., Ennos, A.R., Handley, J.F., 2001. City form and natural process-indicators for the ecological performance of urban areas and their application to Merseyside, UK. Landsc. Urban Plan. 57 (2), 91–103. Wibowo, S.S., Olszewski, P., 2005. Modeling walking accessibility to public transport terminals: case study of Singapore mass rapid transit. J. East. Asia Soc. Transport. Stud. 6, 147–156.

Please cite this article in press as: Caprì, S., et al. Green walking networks for climate change adaptation. Transport. Res. Part D (2015), http://dx.doi.org/10.1016/j.trd.2015.08.005