PUBLIC TRANSIT SERVICE-OPPORTUNITIES IN COMMUTER-BELT MUNICIPALITIES – A SYSTEMIC ANALYSIS OF TWO DISTRICTS IN THE VIENNA REGION

Public transport in the transition zone from cities to rural areas is increasingly becoming a focus from the financial and public service provision perspective. The (perceived) supply differences of rural and remote areas are on the agenda of policy discourse. Our survey studies the public transport supply of two districts and their municipalities in the region of Vienna, Austria, by using the parameters of service-opportunities, municipal population, acreage of settlement units and potential demand. Annual service-opportunities is a parameter recorded by the public transport authority of the Vienna region for every single station under its zone of influence. These parameters are analyzed to conclude that service-opportunities pose a viable entity for systematic public transport analysis and differences in supply of these two districts are in contrast to expectations. Finally, we address the need for future development of service-opportunities based analyses.


INTRODUCTION
Whereas most research focuses on urban public transport (PT), the changes in population structure of recent decades have led to new challenges and ideas of flexibilisation for PT in rural areas.The constant migratory flow from villages and rural towns to urban centres and their peri-urban fringes has left many regions with very diverging internal population developments and has put a pressure on the organisation and financing of PT for regions of this kind [1,2].Such strongly differing structural conditions with distinct gradients of PT service within a small distance can be found especially in transition zones from urban commuter belts to rural areas.Very often the perceived inequality between these areas governs the public demand for improvement or serves as a justification for service underutilization.Such conditions call for a quickly adaptable systemic survey of provided services.The analysis of service-opportunities serves this purpose well.The knowledge of PT supply on a regional level is of importance not only from a regional policy standpoint to balance regional inequalities but it also provides benchmarks for the improvement of services to cover the existing demand better, or even stimulate new demand.
The quality of PT performance can be appraised on different levels and from different perspectives.Such evaluations range from individual lines and corridors [3], via the behaviour of multi-line systems up to the transnational comparison of cities and their PT characteristics [4].The efficiency of public transport as a function of governance and financing structures, e.g. in the form of PT organizing bodies and fare systems has been analysed by some authors [5,6,7] as well as the efficiency of PT companies [8].As PT systems incorporate complexity, an analysis of systems instead of simple lines calls for all-encompassing quality parameters like "Transit Quality of Service" or "Level of Service" [9].Important partial aspects of PT systems like the network design impact on service regularity [10], timetables [11,12], the connectivity of PT services [13], line routing and operational efficiency [14,15] have been analysed before.Often, analysis is also based on the production perspective of PT, using input parameters like labour force, capital, vehicle and seat kilometres [3,8,16].Furthermore, from an individual perspective, the rider's valuation of services [17] and amenities at PT stations [18] have been subject of analysis as well as overall efficiency of PT rolling stock [19].
This paper is based on a recently conducted analysis of PT supply for two administrative districts close to Austria's capital Vienna: Baden (BN) and Gänserndorf (GF).These two districts have inherited different properties in terms of location, geographical extent, size and number of municipalities.But both districts are partly located in or touch the peri-urban fringe of Vienna.Both studies are aimed (a) at identifying the service quality on a systemic level and (b) at comparing the features of both districts.For this a systemic macro-scale PT supply parameter approach was adopted by utilizing the metric service-opportunity (SO) for every PT station in the region and aggregate parameters like number of PT stations in combination with population and size of settlement units on a municipal scale.
The next section focuses on data and methods used for the analysis and the results thereof are displayed in section three.Finally in the last section these results are discussed and conclusions for further systemic PT analysis are drawn.

MATERIALS AND METHODS
The input data for the analysis were provided by Verkehrsverbund Ostregion (VOR), the organising and financing body of PT in the Vienna region: service-opportunities (in German: Servicefahrten) per individual station, transport mode and day of operation for every station within the area of both districts.Serviceopportunities are defined by VOR as the number of stops of PT modes per station in a given period of time according to the timetable.The period of time under consideration is either one year in total -the regular validity of the timetable -or separated by days of operation: Monday through Friday during school time or during school holidays, Saturdays, Sundays and general holidays.Figure 1 illustrates the metric of serviceopportunities by means of three symbolic bus lines.For example, every bus line usually has stations with scheduled stops and stations that may be left out due to express service, deviant routing of singular runs or lack of passenger's needs to board or alight.So the bus run does not actually stop at that station.In the example of Figure 1 three lines cross a municipality with one bus running in every direction and having a poten-tial scheduled stop at every station.These stops may not necessarily be realized, as no passengers would show up to board or no passengers would request a stop to alight.At each station each line produces 2 service-opportunities due to their 2 buses.At stations where n lines meet 2 times n service-opportunities arise.Summing up all the municipal stations produces the service-opportunities of the municipality, in our case 18.The total number of such scheduled potential stops of runs at stations denotes service-opportunities per station.This metric is irrespective of a bus actually halting there or not, as for example is also the case for stops on demand.The service-opportunity data fielded into single stations with municipality code, days of operation and actual means of transport, so in the analysis we were able to utilize dedicated sub-sums of the dataset.VOR data include SO for all PT modes, rail and bus.The Austrian federal statistics office "Statistik Austria" measures different surface parameters of municipalities: total area, permanent residence area (PRA) and the area of settlement units (SU), which conforms to the population cluster in Regulation (EC) No. 1201/2009.While PRA includes farmland and other not directly built-up areas, SU accounts only for patches of continuously built-up land with at least 200 inhabitants.For eight out of seventy-four municipalities the SU measure was not available from "Statistik Austria", because they did not meet the threshold values.As these municipalities ought to be included in the analysis, the SU size A SU l was approximated by using the average area per inhabitant in SU for districts.Factors { and } denote the proportion of population within and outside the district's SUs respectively.Population density d is derived for SU and in permanent residence areas (PRA) for total municipal population PMUN .A denotes area and index DIST stands for district (1).(2) In the final step of analysis both districts are combined.As actual passenger demand figures were not available to check the service-opportunities we introduced a daily "potential demand (PotDem)" function for each municipality.PotDem is the sum of two functions: one considering relations between the different municipalities and Vienna and the other considering the relations between different municipalities and the capital of each district -Baden or Gänserndorf (3).
) While PotDemX denotes the potential demand of municipality X, PotDemV specifies the potential demand in relation to Vienna and PotDemDC the potential demand in relation to the district capital (Baden or Gänserndorf).PotDem is calculated for the municipalities according to a typical gravitational model that explains trip distribution between Vienna and the district capitals with their regional surroundings (4).

ANALYSIS RESULTS
Baden district's (BN) total population amounts to 138,000 inhabitants, whereas Gänserndorf (GF) has 96,000, see  1 shows extremes of parameter values for both districts.From the total number of stations via SO per district to SO per station, the span of parameters is derived from both datasets.GF values suggest a more rural situation than BN district data does.To- tal SO rates at 1.6 to 1 which is very similar to the rate of population being 1.4 to 1. On the other hand for maximum municipal SOs, Baden's value is 3.14 times higher than that of Groß Enzersdorf, where population rates behave like 2.6 to 1 in favour of Baden. Figure 2 illustrates the different inhabitant distributions by using the as-the-crow-flies distance from Vienna city centre and from the district capitals Baden and Gänserndorf and the circle size is proportional to population.The rectangles show how much the district is spread out.Baden district shows a more compact structure, as the distances to Vienna range from 22.2 (Pfaffstätten) to 40.8 (Furth/Triesting) and the furthermost municipality is placed 21.8 km (Reisenberg) from Baden.In GF the structure is more spread out.Not only are peripheral municipalities located up to 34.9 km (Hohenau) away from Gänserndorf municipality, but also the distances to Vienna vary considerably; 14.4 km (Aderklaa) to 59.5 km (Hohenau).
When the service-opportunities sums of both districts related to population size for different days of operation are compared, a difference in favour of BN district is visible on every day of operation.Also, for both districts a consecutive reduction in service quantity is easily identifiable (Figure 3).
The linear approximation of population density in SU with station density in SU of GF district shows a slightly better level of service quality than BN district (see Figure 4).In both cases the correlation coefficients show a remarkably well fit with R 2 of 0.94 and 0.69, respectively.
When plotting the relative number of inhabitants per station over population on a municipal basis (Figure 5), a convergence towards the average appears with the increment of population size for both districts.For populations up to 5,000 inhabitants the divergence of station population values shows a factor of 7, ranging from 100 to 700.GF district municipalities appear to be spread distinctively be-

Figure -Total service-opportunities per municipality over population 6
Figure 6 shows the total of service-opportunities per municipality over population and the corresponding linear approximation for GF and BN districts.In this case the district of GF lies beneath BN's performance.Both districts show remarkably high correlation coefficients ranging from 0.61 to 0.75.
As previous research shows [22,23,24,25,26], the coverage of settlements by PT networks and sta-tion locations has been an important issue in PT systems design and appraisal.In his seminal work on public transport station access, Walther [27] has derived distance-acceptance curves for two different types of trips: occasional and commuting.This is illustrated by a dashed line (commuting) and full line (occasional) in   The regression analysis of potential demand Pot-Dem with SO is shown in Figure 8, while Figure 9 shows the regression analysis of Comm with SO.Both correlations follow a square root model (5).
(5) Table 2 shows the model parameters, statistical measures and R 2 values for both models, while Pot-Dem and Comm also correlate with each other satisfactorily in linear model with an R 2 value of 0.88 (not shown here).Since P-values in the ANOVA tables of both models are less than 0.01, there is a statistically significant relationship between PotDem and SO as well as between Comm and SO at the 99% confidence level.

DISCUSSION
The PT supply of municipalities in both districts varies distinctively, but not as much as popular belief of peripheral districts would have suggested pre-study.This is indicated by the small differences in per capita service-opportunities as shown in Figure 3. Also, this diagram depicts very clearly the orientation of service provision on commuter and school needs.On school weekdays and Sundays the per capita SO supply of both districts is very close.BN with 0.120 and 0.045 has the edge over GF's 0.116 and 0.037.The base supply is therefore of similar quantity.In the case of school holidays and Saturdays BN is much better served with PT runs than GF -0.102 vs. 0.083 and 0.076 vs. 0.048.Services outside school days and base supply differ distinctively, here BN is better supplied.Albeit the GF district is more spread out, its areal supply with stations in terms of density per settlement unit meets the population density better than BN district (Figure 4).This hints at a more compact settlement structure within the municipalities of GF district.Interestingly, the station population of small municipalities scatters by a factor of about seven (Figure 5).Such a fluctuation with service-opportunities in small-sized municipalities would reflect their geographical positions: (a) small municipalities situated between larger ones, and (b) distance to larger municipalities (gravity trip distribution model).This could be explained by a location in between larger units and the more frequent services in between these.Now, as this is a similar case with station supply it remains to be studied what lies behind the fact that small municipalities close to Vienna have better station supply than more peripheral ones.Figure 5 also, suggests the increasing population size stabilizing impact on relative supply parameters.One explanation may simply be one of matters of rates and proportions: an expectable fluctuation of absolute station numbers simply plays a bigger role on station population where the population and number of stations are small on the average in comparison to larger settlements.
In contrast to station supply, BN municipalities have a better provision with service-opportunities, i.e. bus and rail services, than GF municipalities do (Figure 6).The calculation of average walking distances to access stations (Figure 7) illustrates graphically that both districts lie pretty close when it comes to the coverage.Median values are almost identical and the capital and minimum values (Großhofen and Furth/Triesting) lie close to each other.Only maximum values (Oberwaltersdorf and Ringelsdorf) are separated by a calculatory access distance of 123 m.The acceptance diagram shows the potential of public transit to reach its customers according to chosen function -commuter or occasional trips.It is worth noting that some rural municipalities reach values as well or even better than in dedicatedly urban areas [26,27], where PT station coverage area radii of 300 m or more are used for the calculation of passenger potential [28].So average maximum walking distances to stops of 90 and 102 m pose impressively good values in theory.What of course is very different to urban areas is the much lower frequency of services in combination with high motorization rates, leaving most of the stations' passenger potential unexploited.It remains to be studied which district exploits this passenger potential better and if being better served with stations or with runs plays the more important role.Setting commuters in relation to service-opportunities shows that daily work commuters can be explained at a statistically significant level by service-opportunities.The adoption of the PotDem concept in absence of actual passenger data and the use of model parameters from Valencia and Barcelona regional mobility surveys thus prove to be transferrable to the Vienna region.

CONCLUSION
In this study we have studied public transport supply parameters of municipalities in two districts in the Vienna region by assessing parameters like station density, service-opportunities, settlement unit density, commuters and by introducing a proxy named potential demand.
A strong opinion is persistent in suburban to rural regions that these may be less well served than others, and often the expectations of urban PT service quality and frequency to suburban or rural areas are applied.The systemic parameter analysis of serviceopportunities, population, settlement unit area and potential demand puts an equality discussion on solid, evidence-based ground.Popular perception, even among transport experts is prone to misperceptions of demand comparisons over different territories.In both of our cases the commonly accepted view is that the more rural district of Gänserndorf has generally worse PT services to offer than the more suburban district of Baden.Our analysis proves this general notion to be wrong, and we have proven the picture to be more complex than expected.
The advantage of such a systemic supply analysis is that it does not only help to clarify a blurred picture in daily transport policy discourse on a regional scale but also provides a regional development planning tool for the analysis and comparison of public transport supply.The comparison of diverse areas is necessary especially under two perspectives: fair funding of PT base services of differing regions, and the regional planning equality/disparity debate.As real passenger data were not available for this research, we derived the proxy parameter potential demand from proven transport models and model parameters and used it for the verification which led to satisfying results.But future service-opportunity studies shall extend to include actual passenger data (per stop) instead, to gain more precise insight and correlate these for example within a multi-criteria model, e.g. with the average distance to access stations as another model parameter.
One limitation of this study is the lacking distinction between service-opportunities of road and railbound services, even though the data provided by VOR differentiates between different modes.
Although the models of potential demand and commuters already establish service-opportunities as a useful method, real ridership figures need to be added to increase the proficiency and explanatory value.Fur-thermore, additional supply parameters like the average number of transfers to destinations, location of stations (e.g.within major settlement areas) or minor municipalities between bigger municipalities are necessary to be incorporated in the development of this analysis method.We suggest applying a method with additional and improved data to potentially verify and eventually explain the convergence nature of station supply from Figure 5.
Figure -The service-opportunity concept explained 1 in an exemplary scheme

withT.
Brezina et al.: Public Transit Service-Opportunities in Commuter-Belt Municipalities -A Systemic Analysis of Two Districts in the Vienna Region P being the population of a municipality, dij being the as-the-crow-flies distance between municipalities i and j.Parameters a , b and c define the gravitational model and n was chosen as a position variable to produce coefficients within reasonable values, i.e. to avoid obtaining too small or too big numbers that may be difficult to use.We set the values to. the experience obtained with the surveys of Valencia and Alicante strategic urban mobility plans[20,21].PotDem is then correlated with SO and commuters (Comm) and the fits are examined using the Statgraphics software.Comm are daily work commuters departing from the municipality to work in another one.Tadat from micro census 2009 have been provided by "Statistik Austria".

Figure -
Figure -Scatter plot of geographical structure of municipalities in districts Baden (BN, dashed dark grey) 2 and Gänserndorf (GF, dotted light grey) with circle area proportional to population.The diagram axes show as-the-crow-flies distances from Vienna city centre (x-axis) and from the district capital (y-axis) respectively.

Figure 7 .Figure -
Figure -Average access walking distance to stations calculated from average SU area per station, 7 shown per district: minimum, maximum, district capital and district median.Dashed line: commuter trips; Full line: occasional trips; Light grey columns: GF district; Dashed dark grey columns: BN district.Modified after [27].

Figure - :
Figure -: Fitted model for Comm = f (SO); 9 top bottom: observed values vs. values predicted by the model

Figure 7
depicts the calculated average access distances of municipalities of both districts with Walther's commuting and regular trips curves.The average access distance is calculated from SU area per station based on a circular area.The shortest average access distances to stations are 90 m in Großhofen and 102 m in Furth/Triesting.Maximum values are Oberwaltersdorf with 378 m and Ringelsdorf-Niederabsdorf with 501 m.The medians of Baden and Gänserndorf districts are situated very close together: 261 and 264 m, respectively.

Table 1 -
Extreme parameters of both case study districts; Names in brackets denote the name of the station or municipality.

Table 2 -
Estimated parameters, statistical measures and R 2 for both models T. Brezina et al.: Public Transit Service-Opportunities in Commuter-Belt Municipalities -A Systemic Analysis of Two Districts in the Vienna Region STICHWORTE ÖV-Angebot; ländlicher Bezirk; Haltestellendichte; Servicefahrten; potentielle Nachfrage;