Journal of Transport Geography 30 (2013) 227–233
Contents lists available at SciVerse ScienceDirect
Journal of Transport Geography
journal homepage: www.elsevier.com/locate/jtrangeo
Mapping bicyclists’ experiences in Copenhagen ⇑ Bernhard Snizek a, , Thomas Alexander Sick Nielsen b, Hans Skov-Petersen a a Institute of Geosciences and Natural Resource Management, University of Copenhagen, 1958 Rolighedsvej 23, Denmark b Department of Transport Technical University of Denmark, Bygningstorvet 116B, 2800 Kgs. Lyngby, Denmark article info abstract
Keywords: This paper presents an approach to the collection, mapping, and analysis of cyclists’ experiences. By spa- Mapping tially relating located experiences to the availability of bicycle facilities and other aspects of the urban Experiences environment, their influence on cyclists’ experiences can be analysed. 398 cyclists responded and Cycling sketched their most recent cycle route and a total of 890 points to locations along the route where they Urban planning had had positive and negative cycling experiences. The survey was implemented as an online question- Transport planning naire built on Google Maps, and allowed up to three positive and three negative experience points to be Traffic engineering mapped and classified. By relating the characteristics of the experience points and the routes to the traversed urban area in general, the significance of the preconditions for obtaining positive or negative experiences could be eval- uated. Thereby urban spaces can be mapped according to the potential promotion of positive or negative experiences. Further, the method might be applied to assess the effect of proposed changes to the urban design in terms of cyclists’ experiences. Statistical analysis of the location attributes, traffic environments and conflicts, bicycle facilities, urban density, centrality, and environmental amenities indicates that positive experiences, or the absence of negative experiences, are clearly related to the presence of en-route cycling facilities, and attractive nat- ure environments within a short distance of large water bodies or green edges along the route. Ó 2013 Elsevier Ltd. All rights reserved.
1. Background 1.1. Technology-supported mapping of urban experiences
Encouraging motorists to stop using the car for daily urban trips Affective computing systems are a recent development within in favour of the bicycle is one of the major challenges cities cur- emotional mapping. Nold (2009) developed an electronic device, rently face in order to enhance the liveability of cities (Ewing which constantly measures a respondent’s state of arousal via a and Cervero, 2010). However, convincing motorists to choose the galvanic skin device and his/her current location while walking bicycle as the major means of transport is a significant challenge. in a city. By storing the measurements in a spatial database and Understanding the way cyclists perceive their environment as well overlaying several respondents’ tracks, unpleasant locations can as mapping and analysing these perceptions could be the key to be identified. In this way, a map can be constructed, which Leahu designing positive cycling experiences which may well encourage and Schwenk (2008) refer to as an arousal map that visualises more people to travel by bicycle, thereby contributing to sustain- the city’s psychogeography. This map could provide important able urban environments. background information for city planning and could be applied Recently, several studies have treated cycling as a special phe- during the knowledge-gathering phase of the planning process. nomenon, which differs both from motorised as well as pedestrian In Nold’s context, emotional maps are generated by means of both traffic in a number of ways (Forsyth and Krizek, 2011; de Geus stated (semantic mapping of experiences) and revealed (geospatial et al., 2008; Bonham and Suh, 2008; Raford et al., 2005). However, recording of skin resistance) preference after having completed a only a few of these studies deal with the quality of cycling in rela- particular trip. Zeile et al. (2009) developed a methodology for con- tion to the cyclists’ surroundings (Heinen et al., 2011). structing what they call emotional neighbourhood portraits by applying a similar approach to Nold’s. Rantanen and Kahila (2009) presented Internet-based methods to gather, examine, and analyse local knowledge within what they called SoftGIS. Posi-
⇑ Corresponding author. tive and negative quality spots were analysed regarding hotspots E-mail address: [email protected] (B. Snizek). and clustering, which were used as the foundation for participatory
0966-6923/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jtrangeo.2013.02.001 228 B. Snizek et al. / Journal of Transport Geography 30 (2013) 227–233 planning processes. Emotional mapping and SoftGIS approaches collection was soon replaced by online questionnaires that were are, however, still new in the context of bicycling. built around or contained map-based components (Bearman and Appleton, 2012). Today, the availability of Application Program- 1.2. Cyclists’ experiences ming Interfaces (APIs) enables software developers to build map applications based on different map technologies such as Google Cyclists’ experiences differ to some extent from those of motor- Maps or maps from OpenStreetMap.1 ists on the one hand and pedestrians on the other and are influ- A web and map-based application based on the Google Maps enced by several factors: The existence and design of cycling API was designed and implemented in connection with a compre- facilities (Sener et al., 2009; Dill and Gliebe, 2008) play a great role hensive online questionnaire survey about cyclists’ preferences in both attracting cyclists in the first place and how they subse- conducted in Copenhagen.2 In this questionnaire, respondents were quently perceive safety and appreciate the route. Furthermore, asked to, (1) draw their most recent route, (2) designate three loca- other roadway characteristics such as physical characteristics, tions where they had had positive experiences and, (3) identify three on-street parking and operational characteristics (Sener et al., locations where they had had negative experiences. In addition, the 2009) as well as land-use (Winters and Cooper, 2008) and proxim- respondents had the opportunity to classify the locations by select- ity to retail establishments (Krizek and Johnson, 2006) also con- ing from a list of given classes and finally adding comments. In total, tribute to the overall cycling experience. 554 positive and negative locations were registered by the system In general, the literature on cycling experiences is limited. For and stored for further data processing, which is described below. Re- example, the following themes have not been studied at all: spa- sponses containing fewer than three good or/and bad points were tially explicit experiences relating to land use, traffic intensity, pol- accepted. The respondents were prompted for located experiences lution and noise, the number of pedestrians and cycling indicating the focus on features of the urban environment, but all infrastructure other than bicycle paths and lanes. As these themes experiences could be recorded – and a possibility for adding text are considered important components of experience, further re- explanations was part of the survey. search is needed. This study conducts a geospatial approach to the mapping and analysis of cyclists’ experiences. The overall goal of this article is to determine whether the characteristics of elements and locations of the city are correlated to the likelihood of having a po- 2.2. Data processing sitive or negative cycling experience? A series of processing steps had to be performed so that the raw data, which had been generated by the web application, could be 2. Methodology analysed. A quick visual sweep showed that some of the lines drawn were In order to build spatial explicit models of an individual’s per- apparently the result of the respondent’s inability to operate the ception of his/her surroundings, data collection methods had to web application. These lines were either very short, did not, par- be developed. Within the literature, quite a diverse range of data tially or as a whole, follow the transportation network or their seg- collection methods can be found. Methods, where respondents ments overlapped each other to a certain extent. actually draw their routes, be it on paper or via computational rep- The routes which obviously had been erroneously entered were resentations, are quite rare: Raford et al. (2005) who asked respon- removed from the dataset as well as the points related to the dents to draw routes on paper being one example. A rather routes (see Fig. 1). uncommon, specialised and distributed system for real time data The area of analysis was defined to be composed of the munic- acquisition which makes it a tool for revealed preference studies ipalities of Copenhagen and Frederiksberg, Denmark – the latter is described in (Eisenman et al., 2009), whereby a sensing system being totally enclosed by the municipality of Copenhagen. Follow- collects data on pollution level, allergen levels, noise levels and ing the removal of erroneously entered lines those lines exceeding the roughness of terrain. Parkin et al. (2007) employ video clips the municipal boundary were cut to the boundary, those being which are shown to respondents as a means of conveying informa- completely outside of the municipality were removed from the tion on different road environments. dataset (see Fig. 2). In this section, we describe a methodology that through collect- ing data leads us to an understanding of the relations between sta- ted experiences and urban elements. An interactive, web-based questionnaire was developed. Up to three locations, a number, 2.2.1. Map matching of lines which was found a reasonable number for the respondents to As the precision of the drawn lines varied significantly due to end answering the questionnaire, of positive or negative experi- the respondents’ differing abilities, each route was projected onto ences, as well as the route on which they were experienced, could the road network, a procedure called map matching. A road net- be entered. Post processing and data cleansing were applied to the work, which had been extracted from OpenStreetMap, was used. raw data. Every point was then related to elements of the urban The resulting routes, which were identical to parts of the transport environment as well as to selected route metrics. Finally, in order network, were then ready for further analysis. Data describing cy- to establish which urban features resulted in significant positive cle-related infrastructure (bicycle paths, segregated cycle paths or or negative experiences, every point, positive or negative, was re- no bicycle infrastructure at all) were extracted from OpenStreet- lated to the locations on the routes, which did not generate posi- Map. Most routes returned by the map matcher were usable, tive or negative experiences. In the next section, we present the although some, especially those on roads with several lanes map-based questionnaire, explain the process of data cleansing (see Fig. 3), were returned erroneously. To ensure the highest and processing and justify our selection of urban elements and quality possible, each route was inspected and any apparent map route metrics. matching errors were compensated for.
2.1. The online, map-based questionnaire 1 ‘‘OpenStreetMap is a free editable map of the whole world. It is made by people like you’’ (Anon, 2012). The collection of spatial data via online media is becoming 2 This questionnaire was developed within the bikeability project, see more at more and more common within research. Simple, postcode-based http://www.bikeability.dk. B. Snizek et al. / Journal of Transport Geography 30 (2013) 227–233 229
Fig. 1. Part of the online, map-based questionnaire in Danish. Translation: ‘Positive and Negative experiences on your bicycletrip’; 1. Draw your route; 2. Pinpoint positive experiences; 3. Pinpoint negative experiences; 4. Save and exit. A respondent’s route is shown in blue, while two positive experiences are indicated by green markers and three negative experiences are indicated by red markers, one of which is described. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
2.2.2. Processing the points their daily selection of routes, bicyclists are likely to avoid negative The dots entered by the users were then projected onto their experiences. Thereby the registration of negative experiences along respective routes. Dots with a distance of over 100 m to their the chosen route only relates to experiences on routes perceived to respective projection onto the road network were deleted from be apt to travel on. Overcoming this shortcoming would require a the dataset. methodology where bicyclists experienced or rated ‘un-experi- A set of points was generated in order to compare points of enced’ places. experience – whether positive or negative – with points of non- No additional information like the time of day the trip was be- experience and thereby investigate the significance of the experi- gun at or its duration was recorded. Therefore differences in per- ences in relation to the urban elements described in Section 2.3 be- ception of the urban space between night and day were low. These 86,332 so-called non-experience points were laid out on disregarded from. the spatially corrected routes with a distance of 50 m. Values pulled from a series of relevant GIS layers, discussed be- low, were then attributed to the data layer containing the experi- ence points. In the following section, we discuss each of these layers by referring to the existing literature in the field in order to justify the existence and relevance of each factor and to develop a frame- work of how these indicators can be related to the recorded experiences. About 4700 respondents completed the questionnaire. The interactive map questionnaire presented in the current paper was only shown to about 66% of the respondents. 625 routes were drawn and 1677 dots were entered into the system. Erroneously drawn routes were removed which left 554 routes. After omitting the routes that lay outside the municipal boundaries of Copenha- gen and Frederiksberg, only 409 routes remained. After having per- formed map matching and manual correction, 398 routes were left in the dataset. The number of points decreased to 890 after remov- ing those which did not have corresponding routes and those that were located further than 100 m from their respective routes as well as those which lay outside the city boundary. In total, 65% of routes and 57% of the points were valid. The number of respon- dents that did not complete the map questionnaire was not recorded. Fig. 2. Example of an erroneously drawn route (blue). (For interpretation of the It has to be noted that the method presented here might be references to color in this figure legend, the reader is referred to the web version of slightly skewed towards an emphasis on the positive results. On this article.) 230 B. Snizek et al. / Journal of Transport Geography 30 (2013) 227–233
2.3.1.4. Distance to the nearest traffic lights. Traffic lights are a source of delay, although they also increase road safety for cyclists (Zeile et al., 2009; Rietveld and Daniel, 2004). In this study, they are considered as sources of delay and are therefore proxies for nega- tive experiences.
2.3.2. Street types Street types, as a proxy for traffic volume and speed, perceived safety, noise and pollution, have a significant impact on route choice and how cyclists value their environment and the trips themselves. Within the current context, this variable was split into the fol- lowing eight classes: primary roads, secondary roads, residential streets with detached or semi-detached housing, residential streets with multi-storey housing, cycle paths exclusively for bicycles, mixed paths for cyclists and pedestrians, and others. Data for this variable were also taken from OpenStreetMap, as this data source was the most up to date.
2.3.2.1. Distance to the closest group of bus stops. Bus stops and bus stop groups are part of the urban public transport system. In the Fig. 3. Erroneously mapmatched route (red line), respondent-drawn route (green context of this study, bus stops are regarded as sources of negative dots), OSM transport network. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) experiences as pedestrians crossing the cycle lane may pose a threat to cyclists and may also delay them.
2.3. A framework for relating cyclists’ experiences to urban phenomena 2.3.2.2. Distance to nearest intersection. The distance to the nearest intersection tells something about the position of the current expe- The objective of this section is to discuss a series of indicators rience point to a potential obstacle on a cyclist’s route. Intersections and to relate these two elements of the urban space as well as to can act as temporal obstacles in that the cyclist has to give way to route characteristics. Within the analysis, a series of thematic GIS vehicles or other cyclists who are approaching from the left or has layers were produced which delivered values for the dots. In addi- to wait for a green light. In this context, as the direction of the drawn tion, a series of calculations regarding directionality and distance routes was impossible to establish, the relation of the experience were performed. point to the nearest intersection was not given, i.e. one could not determine whether the intersection had just been passed or was 2.3.1. Cycling facilities and level of service for cyclists being approached. In this study, the distance to the nearest intersec- 2.3.1.1. Roads with cycle facilities. Sener et al. (2009) incorporated tion varied from 1 m to about 400 m. Intersections are regarded as two variables related to cycling facilities: on-road cycle lanes and contributing to negative experiences. The data for the calculation shared roadways. Based on the cycling facilities available in Copen- of intersections was extracted from OpenStreetMap. hagen, we have classified the cycling infrastructure in the follow- ing manner: street with no cycling facilities; cycle path by the 2.3.3. Urban density and centrality side of the road (separated from motor traffic by a curb); cycle lane 2.3.3.1. Distance to town hall. The distance to the town hall was ta- on the road (separated from motor traffic by markings on the ken as an indicator of the traffic environment in general. Copenha- road); off-road path exclusively for cyclists, and finally off-road gen, with its medieval urban core, exhibits different traffic path shared by cyclists and pedestrians. The road network and cy- environments. Within the city centre, and thereby close to the town cle facilities were extracted from OpenStreetMap. hall, a net of small streets displays the typical pattern of medieval cities. Here, cycle infrastructure is rare, combined with low vehicle speeds due to the narrowness of the streets. Around the city centre, 2.3.1.2. Distance to nearest cycle rack. In Copenhagen, about 4100 Copenhagen has a typical 1900s block structure with a high level of cycle racks are distributed throughout the city. The importance cycle infrastructure. At the fringe of the city, single housing is con- of cycle facilities at travel destinations to cyclists has been high- centrated around roads with cycle infrastructure at collector roads. lighted in several studies (Heinen et al., 2011). Cycle racks may The streets within the housing areas are characterised by low traffic provide a positive experience as they offer cyclists a place where speeds. In this study, the distance from the experience points to the they can safely store their often expensive cycles. town hall varied from about 50 m to 8000 m.
2.3.1.3. Number of cycle racks within 100 m. The provision of cycle 2.3.3.2. Number of companies within a distance of 100 m to the racks, as previously discussed, improves cycle infrastructure. experience point. Most companies have more than one employ- Therefore, they make a positive contribution to cycle infrastructure ee and therefore create urban life, i.e. a higher number of pedestri- and are regarded as an initiator of positive experiences. Geodata, ans in the vicinity of their location. These pedestrians might cross which described cycle racks, originated from the municipality of the cycle infrastructure to and from bus stops or parked cars, while Copenhagen and was downloaded from the municipality’s home- they may also be in search of a shop and thereby pose a potential page.3 There were between zero and 32 bicycle racks within a dis- threat to the cyclists. The data for this indicator was extracted from 4 tance of 100 m to the experience point. The Central Business Register (CVR). The number of companies ranges from 0 to 753.
3 http://www.kk.dk/Borger/ByOgTrafik/CyklernesBy/CykelFaktaOgViden/Cykeltal/ CykelData.aspx 4 See http://www.cvr.dk B. Snizek et al. / Journal of Transport Geography 30 (2013) 227–233 231
Table 1 Multinomial logistic regression model explaining the probability of positive or negative experience points against all experience spaces/potential experience points derived from the respondents’ cycle routes. B column presents regression coefficients and Exp(B) the corresponding change of odds. (0,1) after the variable name indicate a binary variable. Fully scaled variables are included as Ln transformed variables to account for non-linearity. N = 87,222 (of which 554 positive; and 336 negative experiences). Cox & Snell R- square = 0.04; Nagelkerke R-square = 0.037; P = 0.000 (Chi-square).
B Exp(B) Sig.
Positive experience Intercept À2.519 0.000 Primary road (0,1) À0.954 0.385 0.000 Secondary road (0,1) À0.510 0.600 0.000 Residential street (0,1) À0.008 0.993 0.961 Cycling facility en route (0,1) 0.436 1.546 0.001 Cycling facility is a separated cycle path (0,1) 0.400 1.492 0.004 Distance to intersection (ln) À0.034 0.966 0.444 Distance to signalled intersection (ln) À0.048 0.953 0.291 Distance to bus stop (ln) 0.179 1.197 0.004 Companies within 100 m (ln) À0.113 0.893 0.000 Distance to town hall (ln) À0.177 0.838 0.008 Distance to water body (ln) À0.162 0.851 0.000 Percentage of green edge on route segment (ln) 0.006 1.006 0.781 Distance to ‘flight of crow’ route (ln) À0.115 0.891 0.000 Negative experience Intercept À0.732 0.342 Primary road (0,1) À0.729 0.482 0.020 Secondary road (0,1) À0.278 0.757 0.083 Residential street (0,1) À0.492 0.611 0.008 Cycling facility en route (0,1) À0.398 0.671 0.006 Cycling facility is a separated cycle path (0,1) 0.199 1.221 0.459 Distance to intersection (ln) À0.166 0.847 0.004 Distance to signalled intersection (ln) À0.235 0.791 0.000 Distance to bus stop (ln) À0.128 0.880 0.058 Companies within 100 m (ln) 0.085 1.089 0.080 Distance to town hall (ln) À0.263 0.769 0.001 Distance to water body (ln) À0.047 0.954 0.317 Percentage of green edge on route segment (ln) À0.097 0.908 0.003 Distance to ‘flight of crow’ route (ln) À0.033 0.968 0.407
2.3.3.3. Number of retail units within a distance of 100 m to the cyclists’ choice of route has been documented in several studies experience point. This measure is a subset of the one above. Retail (Dill, 2009; Aultman-Hall et al., 1997). generates urban life and thereby people who will cross the cycle The distances to the direct line varied between 0 and about infrastructure and therefore it is considered to be a negative influ- 5000 m. ence on cyclists’ overall experiences. The number of retail units ranged between 0 and 61 (see Table 1). 2.3.5.2. The angle between the current segment and the line from the experience point towards town hall. The Town Hall is a proxy for the 2.3.4. Water and green areas city centre and this variable explains the direction of the road seg- 2.3.4.1. Distance to closest water or green area. In the current study, ment upon which the experience point is located. The lower the the location of water and green areas in the vicinity of cycle infra- value, the more directly the cyclist approaches or travels away structure is regarded as something, which contributes to a positive from the city centre. Higher values (up to 90 degrees) indicate that experience. This indicator is composed of the distance from the the current segment of the route leads the cyclists in a radial direc- road’s centreline to layers describing parks, cemeteries, heaths, for- tion. The role of directionality vis-a-vis the urban centre for urban ests and commons from the topographical map of Denmark, experiences and spatial behaviour have been highlighted in several TOP10DK.5 studies within urban design and geography including Kevin Lynch’s study of the city image (Lynch, 1960). 2.3.4.2. Percentage of green of route segments. In order to calculate this indicator a buffer of 50 m around each road segments was con- 3. Results structed and within the percentage of forests, parks, cemeteries, heaths and wetlands calculated. The experience data obtained from the web-based, map-en- abled survey instrument can be mapped as an independent basis 2.3.5. Route-related measures for assessment and analysis. The addition of geographical data lay- 2.3.5.1. Deviations from the direct line. By measuring the distances ers to experience points and route segments allows further statis- and the angles between route segments and the direct line, i.e. tical analysis of the environmental experiences. Maps and the line which the crow flies between the origin and the destina- environmental correlates of cyclists’ experiences are presented in tion, one gains information about the directness of the route. This the following. measurement is seen as an indicator for how complex a route is Fig. 4 presents the distribution of the positive and the negative and thereby indicates whether the cyclist went straight towards experiences reported by the survey participants. It should be noted the destination or chose to turn left or right several times, thereby that most of the dots follow roads or cycle tracks with a high load taking detours resulting in delays. The importance of directness for of cycle traffic, and reflect the spatial distribution and largely radial character of cycle travel in the area. Clusters of points are observa- 5 See http://www.geodatabiblioteket.dk/images/stories/specifikationer/doc/ ble in some areas, e.g. along a recently established green trail on spec_320.pdf derelict railroad terrain, on most bridges across the harbour or 232 B. Snizek et al. / Journal of Transport Geography 30 (2013) 227–233
Fig. 4. Geographic distribution of positive (to the left) and negative (to the right) spots; the municipalities of Copenhagen and Frederiksberg are shown in grey. the lakes separating the medieval city centre from the rest of the tive correlation displayed for the variable ‘companies within city; as well as along some circumferential routes. 100 m’ expresses density and trip destination density in the imme- The statistical analysis of environmental correlates was based diate environment of the cyclist route – and therefore congestion on a logistic multinomial regression model explaining the proba- by all modes of travel and associated conflicts. The negative effect bility of a positive experience versus no experience, and the prob- of distance to the town hall appears to be somewhat counter intu- ability of a negative experience versus no experience. The model itive to this, although the same effect applies to the probability of a was fitted based on a model search among the variables presented negative experience, while both reflect the convergence of most in the methodology section under continuous testing for multi-col- cycle routes in town centres and contribute to the probability of linearity and robustness of the result. The optimal model was cho- central locations being recorded as negative experiences by the sen based on an assessment of the level of explanation, conceptual survey participants. soundness and statistical significance. Variables with non-signifi- The correlation of distance to large water bodies such as the cant effects were excluded from the model. lakes, which surround central Copenhagen or the harbour, indi- The resulting regression model indicates significant correlations cates that attractive environments and views make a significant between the urban environment factors included in the study, but contribution to positive experiences. achieves a limited level of explanation of positive and negative ver- The distance to the direct line of the cycle route (as the crow sus no experiences (R-square). The limited ability of the model to flies) indicates that detours contribute negatively to the probability explain the recorded experiences can be explained by the many of a positive cycling experience. personal and situational factors of relevance to experiences, as well The model results for the probability of a negative experience as the limits of the approach in that it only observes environmental on a cycle route reflect the reverse probability of a positive experi- factors and amenities. The environmental correlates are, however, ence with some additions and exceptions. As for positive experi- still relevant as significant effects/correlations exist which can ences, cycling along a primary road is related to a lower form part of the conditions which influence the experiential out- probability of a negative experience, a result that may reflect the comes (see Naess, 2006). general character of these environments as linear connectors, with The model results for the probability of a positive experience large volumes of separated traffic. Cycling on residential streets is point to a significant contribution from the road environment, cy- also negatively correlated to the probability of a negative experi- cling facilities, environmental factors, factors that can be inter- ence, and in the absence of a similar correlation with the probabil- preted as annoyances and congestion and finally deviations from ity of a positive experience; this is likely to reflect lower traffic the most direct route. volumes and speeds. Similarly, the availability of en-route cycling Cycling on primary or secondary roads reduced the probability facilities contributes negatively to the probability of a negative of a positive experience, while cycling with the availability of cy- experience, but positively to a positive experience, which reflects cling facilities, especially a separate ‘Copenhagen style’ cycle path, the importance of comfort and right of way for a positive cycling increased the probability of a positive experience. A greater dis- experience. tance to a bus stop is positively correlated with the probability of The distance to intersections and signalled intersections are a positive experience, which reflects the conflict between cyclists both negatively correlated with the probability of a negative expe- and bus passengers crossing the cycle infrastructure. Most major rience. Thus, the closer to intersections or signalled intersections, roads in Copenhagen are equipped with bicycle paths adjacent to the higher the probability of a negative cycling experience. This the sidewalk and conflicts with busses only take place when very likely reflects the dangers and conflicts encountered at buspassengers crosses the bicycle paths to enter a bus. The nega- intersections. B. Snizek et al. / Journal of Transport Geography 30 (2013) 227–233 233
The ‘greenness’ of the cycling environment measured as the Computer-based cycle models could also use the study’s results percentage of green areas on the route segment is negatively cor- as an input to model experience both on a larger scale as well as on related with the probability of a negative experience, which to- an individual level. Therefore, psychogeographical cross-sections gether with the correlation between distance to water bodies and of cyclists’ routes may be very relevant for the improvement of cy- the probability of positive experiences, indicates that aquatic and cle infrastructure. nature environments both contribute to better cycling experiences. The questionnaire was distributed to cyclists within the munic- References ipalities of Copenhagen and Frederiksberg. It should therefore be noted that the analysis presented here is based on cycling routes Anon, 2012. FAQ – OpenStreepMap Wiki.