A GIS and multi-criteria-based analysis and ranking of transportation zones of Vilnius city/Susisiekimo sistemos analize ir rangavimas Vilniaus miesto transporto rajonuose naudojant GIS.

Jakimavicius, Marius ; Burinskiene, Marija

1. Introduction

The motivation for this research arose from an effort to assess transportation system performance in Vilnius city. The approach taken in that research (Casello 2003) was to preselect a series of origin destination pairs for which public transportation might compete well with private automobile, and test the sensitivity of modal split, and overall system performance, to changes in transit service provided and the cost of auto travel. A review of the literature suggests that transit is most competitive in high-density commercial areas and to a lesser extent in residential ones (Pushkarev and Zupan 1982). To preselect the origin and destination pairs, it was necessary to have a quantitative definition of "high-density" areas.

The urban studies contain definitions of activity centres, typically defined as areas with higher than adjacent concentrations of employment at the traffic analysis zone (TAZ) level. This definition has proven satisfactory in the analysis of polycentric areas' employment patterns, residential location theory, and overall economic analysis.

The accessibility concept can be applied to many spatial problems: e.g. service centre location, hospital-sitting, school closure and many others. The analysis based on the concept of accessibility is therefore ideally suited to be integrated within geographic information systems (GIS). This paper expands the work in modelling accessibility fields taken by Donnay and Ledent (Donnay and Ledent 1995) for the urban region of Liege (Belgium) and Juliao (Juliao 1999) for Tagus Valley Region (Portugal), as well as one-stage model for Slovene municipalities (Drobne 2003; Black et al. 2002). In this paper, travel time (by car) and territorial allocation to the Lithuanian administrative regions have been modelled using the road network and GIS approach.

Accessibility matrix was implemented with origin-destination (OD) matrix computation used in travel demand analysis in transportation geography. In both cases, GIS is used in determination of user-defined arbitrary analysis zone or area of interest (AOI), corresponding to TAZ (Miller and Shaw 2001).

The research presented here proposes an extension to a commonly used activity centre definition to improve that definition's applicability to transportation research. This extension involves identifying activity centres based on the trip-attracting strength of disaggregate employment types within TAZs. This approach identifies areas that are responsible for a disproportionate number of regional trips. The proposed methodology has 2 positive characteristics. First, the approach computes attraction strengths using standard socio-economic data available at the municipality planning organization level. Second, employment is still the fundamental unit of the activity centre definition, and the pedagogical approach of identifying sub areas that exceed certain thresholds remains unchanged.

The efficiency of urban transportation is getting more and more important because of the increasing rate of mobility demand. To plan, control and organize urban transportation in the most efficient way, we also need to consider the aspects of land use (Tanczos and Torok 2007).

Accessibility from the centre of traffic analysis zone to the central part of Vilnius was taken as the main factor for transport system analysis in Vilnius city. Also, other factors have been included, like population density in TAZ, number of working places in TAZ, street network density in traffic zone, public transport density, average number of daily trips in each analysis zone. By comparing the georeferenced data like street network, the territorial allocation and statistical data for each traffic analysis zone in Vilnius can argue about the equity of investments distribution for each TAZ. Also the created GIS application could be used for transport analysis zones ranking by various aspects and problematic zones identification.

2. Case in Vilnius city

Growing Lithuanian economy and increasing quality of the living conditions prompts population's mobility, the motorization level and increasingly high transport flow on the countries streets and roads (Burinskiene and Paliulis 2003).

Average percentage of Vilnius city automobiles quantity is increasing per year about 3%. Number of personal cars in Vilnius city rose from 265 automobiles for 1000 inhabitants in 1999 till 450 in 2005. Sharp bounce of motorization level invokes a lot of transportation problems. Many researchers analyze transportation system from the point of system sustainability, which influences economical, social and environmental implications (Black et al. 2002; Camagni et al. 2002; Grigonis and Burinskiene 2002). Other scientists also indicate political and institutional aspects (Ciegis and Gineitiene 2008).

Number of public transport passengers rose from 229.5 mln. in 1999 year till 277.1 till 2004 year. This indicator increases by about 3.7% each year. The main Vilnius city transport system indicators are in Table 1.

Vilnius city is divided into 51 traffic analysis zones. TAZ and population density (inhabitants in hectare) in each zone are shown in Fig. 1.

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

Analysis of modal split of Vilnius city transportation system showed that trips by public transport decreases (Fig. 2) and trips by private transport are increasing.

The most concentration of working places is in the central part of Vilnius city (Fig. 3). Largest density of working places in the central part of Vilnius involves parking and traffic flow problems.

[FIGURE 3 OMITTED]

3. Methodology

For this research, several changes to the Bogart and Ferry model are implemented. First, three "levels" of activity centres are defined (Bogart and Ferry 1999): major urban centres of large cities, secondary urban centres of smaller cities, and suburban centres. Decreasing employment and employment density thresholds are utilized in each case. Establishing differing thresholds for inclusion ensures that the method will identify those TAZ's with higher than adjacent employment characteristics, the essence of an activity centre. The second set of modifications involves formation of activity centre clusters. Recall that in the Bogart and Ferry method, those zones which by themselves do not meet the activity centre employment thresholds may be clustered with adjacent zones and so meet the criteria to form larger areas. Bogart and Ferry added zones until the whole cluster density fell below the threshold. The authors have adopted this method, but only for suburban activities centres, to avoid the case where a single ultra-high density zone in an urban centre dominates that all adjacent zones would be included to form a "superzone." Further, we require that individual zones being added to meet a minimum employment density threshold. This requirement avoids the case where an open space adjacent to a high density employment centre is considered a part of a suburban activity centre. Finally, we relax the adjacency requirement such that any two zones are considered adjacent if they share a common border of any length. The most significant change we propose is motivated by the following observation. A hypothetical TAZ with a 100 mining jobs attracts far fewer trips than a TAZ with sufficient retail development to employ 100 persons. Furthermore, Targa has shown that different employment types tend to respond to agglomerative location forces more readily than others, with retail among the most responsive (Targa 1990). Transportation models specifically for retail activity have been developed by Hamed and Easa. Generally, retail activities produce more trips, are more likely to agglomerate, and therefore are likely to have stronger impact on regional transportation patterns (Hamed and Easa 1998). For transportation analysis, then the method to identify transportation activity centres TACs should not be based solely on employment density, but rather on the trip-attracting strength of the disaggregate employment types present in a TAZ. To incorporate trip attraction strength into the TAC definition, one could compute the product of employment and trip attraction rate per job for each disaggregate employment type. Those zones that exceeded a threshold value of trips and trip density trips per unit of area would be then considered part of a TAC. The decision statistic, however, would then no longer be the well-established gross employment and employment density thresholds frequently used in the literature. The approach advanced here is to define a hypothetical "mean trip-attracting" MTA job. Suppose that there is a TAZ with exactly one job in each of the 11 standard disaggregate employment types: agriculture, mining, construction, manufacturing, transportation, whole sale, retail, fire, service, government, and military employment. In this case, a total number of daily trips would be attracted to this zone, and an average number of trips per job could be computed. The relative strength of each employment type can be calculated as the ratio of each employment type's attraction rate to the mean attraction rate. This ratio can be used to express each actual job in terms of equivalent MTA jobs. A zone that exceeds the gross employment and employment density levels in terms of MTA jobs would then be considered for inclusion in a TAC.

Trip attraction to TAZs in their metropolitan region equals:

TA = 1.4Ag + 1.2Mi + 3.0Re + 2.4Se, (1)

where:

TA--number of trips attracted; Ag--number of agricultural jobs; Mi--number of mining jobs; Re--number of retail jobs; Se--number of service jobs.

If a TAZ had only 4 jobs, one of the above categories, the zone would attract 8 trips, or 2 trips per job. Thus, an MTA job would attract 2 trips. Retail, in contrast, attracts 3 trips per job; thus, a retail job can be considered 3/2 or 1.5 MTA jobs. Similarly, an agricultural job attracts only 1.4 trips per job, and therefore can be considered 1.4/2 0.7 MTA jobs. The example is generalized as follows. If [[alpha].sub.k] is defined as the trip attraction rate for employment type k, then:

[[chi].sub.k] = [[alpha].sub.k] n/[n.summation over (k=1)] [[alpha].sub.k] [for all]k, (2)

where [[chi].sub.k]--MTA factor for each employment type, k; and n--total number of employment types. A TAZ would be considered as a TAC if:

[summation over (k)] [E.sub.k] [[chi].sub.k] [greater than or equal to] [greater than or equal to] [xi], (3)

and

[summation over (k)] [E.sub.k] [[chi].sub.k]/A [greater than or equal to] [phi], (4)

where [E.sub.k]--actual employment of type k; [xi]--gross employment threshold (MTA jobs); A--area of the TAZ (hectares) and [phi]--employment density threshold (MTA jobs per hectare). Thus, TAZs that meet or exceed the employment and employment density thresholds using MTA jobs are considered TACs. The creation of TAC clusters is done by adding adjacent candidate zones (those with MTA employment density greater than 3.0 MTA jobs per acre), such that the total cluster remains above the threshold level. For our research, we utilized MTA employment and MTA employment density thresholds equal to gross employment thresholds typically used in the literature.

[FIGURE 4 OMITTED]

The following sections demonstrate the analysis of the Vilnius city area using standard activity centre definitions and the TAC method presented here.

The map of traffic analysis zones of Vilnius city (Fig. 4) presents the areas where traffic analysis zones could be considered like transport activity centres (these zones are presented in black colour).

This analysis showed that TAZ could not considered like TAC that in the central part and old town of Vilnius, also in areas of Vilnius city which are in a distant of central part of Vilnius city. The main reason is that in the central part of Vilnius there is a big concentration of working places and in areas around Vilnius city residential houses are dominating, with less working places.

The second stage is to perform an estimated traffic analysis zones ranking using various transportation indicators. For TAZ ranking 2 methods of decision support system were used--Topsis and SAW. GIS-based application computes the ranks of transport analysis zones.

3.1. SAW (Simple Additive Weighting) method in GIS application

For a fragment of input from Vilnius traffic analysis zones socio-economic data for GIS application (Fig. 5).

[FIGURE 5 OMITTED]

Input data for calculation is the criteria and their values of importance; criteria matrix is normalized according to these conditions (Shevchenko et al. 2008):

If the criterion is maximized:

[X.sub.ij] = [X.sub.ij]/[X.sup.max.sub.j], (5)

If the criterion is minimized:

[X.sub.ij] = [X.sup.min.sub.ij]/[X.sub.ij]. (6)

A normalized matrix for each criterion of concrete municipality is multiplied with its importance. Multiplied criteria are summed for each row (for each TAZ). The biggest value means the best transport situation in certain traffic analysis zone.

3.2. TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) method in GIS application

Criteria matrix is normalized by formula

[X.sub.ij] = [X.sub.ij]/[square root of [n.summation over (i=1)] [X.sup.2.sub.ij]]. (7)

It is multiplied by matrix of importance values (Ustinovichius et al. 2007):

P*=[X]x[q], (8)

where: q--matrix of creations importance values.

Normalized matrix is used for calculating ideal positive ([f.sup.+.sub.j]) and negative ([f.sup.-.sub.j]) variants. Calculation of variant's deviation to ideal positive variant is based on:

[L.sup.+.sub.i] = [n.summation over (j=1)] [([f.sub.ij] - [f.sup.+.sub.j]).sup.2]. (9)

Calculation of variant's deviation to negative variant is based on:

[L.sup.-.sub.i] = [n.summation over (j=1)] [([f.sub.ij] - [f.sup.-.sub.j]).sup.2]. (10)

Calculation of proportional variant's deviation to ideal variant [K.sub.BIT] is based on:

[K.sub.BIT] = [L.sup.-.sub.i]/[L.sup.+.sub.i] + [L.sup.-.sub.i]. (11)

The best variant of transport system situation in TAZ is the one with the highest [K.sub.BIT] value. Indicators of Vilnius city transport system analysis for each traffic analysis zone are in Table 2.

Importance for each indicator was estimated by a transport specialists' questionnaire. The results of analysis (Fig. 6) showed that the best transport situation is in Santariskes and Zemieji Paneriai transport activities centres. There are no major disproportion of working places and inhabitants in these zones, there is enough street network density.

4. Conclusions

Research of traffic analysis zones in Vilnius city showed that not all traffic analysis zones could be possible to consider like transport activity centres. Such kind of problematic situation is in the central part of Vilnius and in the TAZ, which are in a distant area of the central part of Vilnius. The main reason is a large disproportion of population and working places density in these areas.

[FIGURE 6 OMITTED]

The second stage of this research represents a GIS based methodology for Vilnius city traffic analysis zones ranking. The created GIS application with 2 calculation methods of decision-support system Topsis and Saw performs TAZ ranking. The analysis of Vilnius city TAZ showed that the best transport situation is in Santariskes and Zemieji Paneriai transport activities centres.

The investigation of TAZ identified major car parking and traffic problems in the following traffic zones: Centras I, Centras II, Lazdynai, Karoliniskes, Antakalnis, Senamiestis, Snipiskes and Naujamiestis. Public transport problems were also identified in these Vilnius TAZ: Verkiai, Dvarcionys, Valakupiai, A. Paneriai and Tarande.

The created methodology is flexible and could be successfully adopted for TAZ analysis and ranking in other cities. It is necessary to have TAZ GIS and socio-economic statistical data.

Received 2 October 2008; accepted 23 January 2009

References

Black, J. A.; Paez, A.; Suthanaya, P. A. 2002. Sustainable urban transportation: Performance indicators and some analytical approaches, Journal of Urban Planning and Development 128(4): 184-209.

Bogart, W.; Ferry, W. 1999. Employment centres in greater Cleveland: Evidence of evolution in a formerly monocentric city, Urban Studies, 2099-2110.

Burinskiene, M.; Paliulis, G. 2003. Consistents of cars parking in Lithuanian towns, Transport 18(4): 174-181.

Camagni, R.; Gibelli, C.; Rigamonti, P. 2002. Urban mobility and urban form: the social and environmental costs of different patterns of urban expansion, Ecological Economics 40(2): 199-216.

Casello, J. 2003. Improving regional transportation system performance through increased suburban intermodalism: A user cost approach. PhD. dissertation, Univ. of Pennsylvania, Philadelphia.

Ciegis, R.; Gineitiene, D. 2008. A concept of sustainable development for regional land use planning: Lithuanian experience, Technological and Economic Development of Economy 14(2): 107-117.

Donnay, J. P.; Ledent, Ph. 1995. Modelling of accessibility fields, in Proc. of JEC-GI '95, I: 489-494.

Drobne, S. 2003. Modelling accessibility fields in Slovene municipalities, in Proc. of the 7th Symposium on Operation Research in Slovenia (SOR'03), 89-96.

Grigonis, V.; Burinskiene, M. 2002. Information technologies in energy planning of cities and towns, Journal of Civil Engineering and Management 8(3): 197-205.

Hamed, M.; Easa, S. 1998. Integrated modeling of urban shopping activities, J. Urban Plann. Dev., 115-131.

Juliao, R. P. 1999. Measuring accessibility using GIS, in Geo Computation Proceedings. Available from Internet: <http://www.geovista.psu.edu/sites/ geocomp99/Gc99/010/gc_010.htm>. [Accessed 20-04-2005].

Miller, H. J. and Shaw, S. L. 2001. Geographic information systems for transportation (GIS-T): Principles and application. Oxford University Press.

Pushkarev, B.; Zupan, J. 1982. Where transit works: Urban densities for public transportation, in H. S. Levinson and R. A. Weant (eds.). Urban Transportation: Perspectives and Prospects. Eno Foundation, Westport, Conn.

Shevchenko, G.; Ustinovichius, L.; Andruskevicius, A. 2008. Multiattribute analysis of investments risk alternatives in construction, Technological and Economic Development of Economy 14(3): 428-443.

Tanczos, K.; Torok, A. 2007. Linear optimization model of urban areas operating efficiency, Transport 22(4): 225-228.

Targa, F. 1990. Traffic congestion and suburban activity centers. Transportation Research Circular #359, Transportation Research Board, Washington, D.C.

Ustinovichius, L.; Zavadskas, E. K.; Podvezko, V. 2007. Application of a quantitative multiple criteria decision-making (MCDM-1) approach to the analysis of investments in construction, Control and Cybernetics 36(1): 256-268.

doi: 10.3846/1392-8619.2009.15.39-48

Marius JAKIMAVICIUS. PhD student of Vilnius Gediminas Technical University, Faculty of Environmental Engineering, Dept of Urban Engineering, LT. Member of Association of Lithuanian Surveyors. Research interests include GIS and GPS systems, GIS-based solutions for transport analysis tasks, optimization of transport system according to urban areas.

Marija BURINSKIENE. Professor, Dr, Head of Urban Engineering Department and Director of Territorial Planning Research Institute of Vilnius Gediminas Technical University. She was project manager of more than 40 national projects from 1983, participated in more than 30 intern conferences and was involved in 8 Framework 5 and 6 program projects. The main area of research interest is regularities and specificity of urban and regional sustainable development, development of urban transport system as well as creation of decision-support system for implementing engineering solutions.

Marius Jakimavicius (1), Marija Burinskiene (2)

Dept of Urban Engineering, Vilnius Gediminas Technical University, Sauletekio al.11, LT-10223 Vilnius, Lithuania E-mail: (1) mjakimavicius@hnit-baltic.lt; (2) marbur@ap.vgtu.lt

Table 1. Transport system indicators in Vilnius city, 1999, 2005

Indicator 1999 2005

Street network density (km/[km.sup.2]) 1.9 2.4
Public transport network density (km/[km.sup.2]) 0.55 0.62
Bicycle paths network density (km/[km.sup.2]) 0.10 0.16
Average traffic flow in peak hours (aut./h) 1275 1521
Percentage of trucks in average flow 3.4 2.4
Average speed in peak traffic flow (km/h) 37.5 29.3
Modal split
--pedestrian trips % 31.3 34.8
--trips by bicycles % 0.3 0.3
--trips by public transport % 45.4 34.2
--trips by car % 23.0 30.7
Maximum number of public transport passengers 5300 3600
 in peak hours
Transit of trucks in peak hours % 21.3 13.2
Number of traffic accidents for 1000 inhabitants 1.07 1.77

Table 2. Transport system indicators for Vilnius TAZ analysis

 Importance
Indicator description Function (%)

Street network density (km/[km.sup.2]) in maximize 19
 each TAZ
Public transport network density maximize 15
 (km/[km.sup.2]) in each TAZ
Length of streets for 1000 inhabitants in maximize 16
 each TAZ
Disproportion for population and employees minimize 22
 densities
Density of parking places (parking places/ maximize 10
 hectare)
Accessibility from the central part from each maximize 9
 transport activities centre to Vilnius city
 central part
Average number of daily trips in each maximize 9
 analysis zone