You are on page 1of 16

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/23538716

Accessibility, Network Efficiency, and


Transport Infrastructure Planning

Article in Environment and Planning A August 1998


DOI: 10.1068/a301337 Source: RePEc

CITATIONS READS

77 149

3 authors, including:

Javier Gutirrez Andrs Monzn


Complutense University of Madrid Universidad Politcnica de Madrid
89 PUBLICATIONS 1,360 CITATIONS 128 PUBLICATIONS 536 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

SocialBigdata-CM | Using Big Data for social change monitoring and analysis View project

INSIGHT | Innovative Policy Modelling and Governance Tools for Sustainable Post-Crisis
Urban Development View project

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Javier Gutirrez
letting you access and read them immediately. Retrieved on: 27 September 2016
NT SUPPLY SERVICES

Your order details Your shipping address


Our Order Ref: 00992895-001
Your Ref: 441MP0127349
Despatched on: 29/9/2014

Your item details


UIN: BLl01007853364
Title: Environment & planning. A
Publisher:
ISSN:
Year: 1997 Volume: 30
Part: 8 Comments
Pages: 13371350
Author name(s):
Article title
words:

Thank you for using Document Supply Services!

2
5124
Copyright Statement
Unle.. out of eopyrlllhl. the contents dlhe docu ....nt(.).n.,;:hed 10 or 9COI)ITIj)anying !hi, P'Qe are ~ecte<l by copyright. They lire aupplied on condition that. exceptio

_ble. aingl. paper copy 10 be primed out by or torlhe individ ..... whoorig"llIllly <eq ....1ed tile ~m(.). you may no:A copy (even lor internal puI'l)OSes), store or relain

any ....::t:onie me6um, retransmit, .<!HI, 01 hire " - eontent, (indt.ldng the single paper copy referred 10 iIbove). However !heM ruin do not appty where:

1. you h _ written pe<mission c11toe copyright 0WI'\8f 10 do otherwise;

2. you h_ the perrrission 01 The Copyright Licensing Agency Ltd, or similar licensing body:
3. tIMt document benefit s from 8 free and open licence iMUed with the consent 01 the COpyright owMr:

4. th e intended u$8ge is covered by statute.

Breach at the terms of this notice is enfor<:eable against you by the copyright owner or their representati ve,

This document ha$ been supplied under our Llbr.ry Privil.g. $eIViCe, You are thererore '!Ireeing to \he terms 01 supply lor our Ubl'ary Privile-ge service. avai lable at:
http :I'-w.bI,uklreshelpl.tyourdeskidoc.upply/helpltennlllndex.html

The British library Document Supply Service, Boston Spa , Wetherby, United Kingdom, LS23 7BQ
bldss.bl.uk
Environment and Planning A 1998, volume 30, pages 1337 - 1350

Accessibility, network efficiency, and transport infrastructure


planning

J Gutierrez
Departamento de Geografia Humana, Universidad Complutense de Madrid, 28040 Madrid,
Spain; e-mail: ghhumOI @sis.ucrn.es
A Monzon, J M Pinero
Departamento de Ingenieria Civil y Transportes, Universidad Politecnica de Madrid,
28040 Madrid, Spain; e-mail: tr05 @dumbo.caminos.upm.es
Received 9 November 1996; in revised form 17 April 1997

Abstract. Most accessibility measures reported in the literature do not meet certain requirements of
transport planners, as the results offered by these indicators are heavily influenced by the geographic
location of the places. PeripheraJ locations invariably figure as being of low accessibility and conse-
quently as possible receptors for new investments, even though they may already have a very good
provision of transport infrastructure. In this paper an accessibility indicator is formulated that
neutralizes the effect of geographic location. To this end, the customary notion of distance (length,
time, or costs) is substituted by another which expresses the ease of access in terms of network
efficiency. This indicator can be useful to assist decisions to invest in transport networks, as it is
more sensitive than the potential model to the transport infrastructure needs of each individual
region. It has been tested by application to the current Spanish Infrastructure Master Plan.

1 Introduction
Accessibility is currently an important issue in transport and regional planning policies
in Europe. Successive enlargements of the European Union, proposals for trans-
European transport networks, and the introduction of the Cohesion Fund as part of
the Maastricht Treaty package have over the past few years given rise to a number of
studies on accessibility at the European scale (for example, Bruinsma and Rietveld,
1993; CEDRE, 1993; Gutierrez and Urbano, 1996; Keeble et aI, 1988; Lutter et aI, 1992;
Simmonds and Jenkinson, 1995). The accessibility impact of the Channel Tunnel has
also been investigated (for example, Dundon-Smith and Gibb, 1994; Smith and Gibb,
1993) and research is being conducted at the national scale by using different indicators
(Geertman and Ritsema van Eck, 1995; Linneker and Spence, 1992; Spence and
Linneker, 1994). In this context, economic potential measures seem to have been
recently rediscovered in the research literature (Frost and Spence, 1995) and new
accessibility indicators have been developed (Vickerman, 1996).
New transport infrastructure improves accessibility conditions. By reducing travel
times, the new transport infrastructure produces what Janelle (1969) called a "time - space
convergence". A shrinking space arises (Spiekermann and Wegener, 1994) in which the
relative locations of places are constantly being modified. However, this contraction of
space is not uniform, but benefits some places more than others, according to their
situation on transport networks. Hence a space is configured that is more and
more discontinuous, polarized, and hierarchized (Plassard, 1992), in which the chief
beneficiaries would seem to be the great cities because they are rapidly and efficiently
connected to each other. Counter to the classical view of core - periphery imbalances,
a new perspective opens up that underlines the imbalances between the chief cities and
their surrounding hinterlands (Vickerman, 1995).
Most of the accessibility indicators reported in the relevant literature (Izquierdo
and Monzon, 1992; Keeble et aI, 1988; Linneker and Spence, 1992; Pooler, 1995;
1338 J Gutierrez, A Monzon, J M Pinero

Vickerman, 1974) are based on absolute measures of spatial separation among places.
In this paper an accessibility indicator that expresses ease of access in relative terms
is formulated and empirically tested for the Spanish Infrastructure Master Plan. A
geographic information system (GIS) is used in which transport networks are modelled,
accessibility values are calculated and mapped, and in which the benefits of the plan
are evaluated from a twofold perspective, both social and spatial. GIS technology is
eminently su ited to this type of analysis: one of the fundamental contributions of GIS
has been to automate and to improve visualization of spatial analysis methods which
have long been carried out without GIS (Ding and Fotheringham, 1992), as in network
analysis.

2 Formulation of an accessibility indicator for transport infrastructure planning


Accessibility has generally been defined as some measure of spatial separation of
human activities. Essentially, it denotes the ease with which activities may be reached
from a given location by means of a particular transport system (Morris et ai, 1979).
Typically, accessibility measures such as those based on economic potential (for example,
see Geertman and Ritsema van Eck, 1995; Harris, 1954; Keeble et aI, 1988; Linneker
and Spence, 1992; Spence and Linneker, 1994), potential transport costs (Botham, 1980;
Harris, 1954; Linneker and Spence, 1992), or weighted mean distances (Gutierrez and
Urbano, 1996; Gutierrez et ai, 1996), combine the distance to, and utility of, different
destinations in a single index (see the appendix). Some indicators (accessibility measures
based on unweighted mean distances) take only distances into account, without
considering the utility of the ditTerent destinations (Allen et ai, 1993; Ingram, 1971;
Lutter et ai, 1992). But in each type, distance (length, time, costs) is part of the
calculations, so that the results of applying each type display a typical concentric
core - periphery pattern (Vickerman, 1996), where peripherality is interpreted as indic-
ative of remoteness or inaccessibility (Ball, 1996). In this sense, these models behave
as indicators of peripherality and are particularly useful for studies of economic
development (Clark et aI, 1969; Harris, 1954; Keeble et ai, 1988).
The fact that the results offered by these indicators are heavily influenced by the
geographic location of the nodes makes these measures unsuited to determining the
transport infrastructure needs of each region. They are undoubtedly useful in the
planning of transport infrastructure when changes in accessibility between two tem-
poral situations are analyzed: before and after a specific action such as the building of
the M25 London orbital motorway in Britain (Linneker and Spence, 1992), or a whole
transport plan such as the master plan for the trans-European road network (Gutierrez
and Urbano, 1996). However, they do not offer an accessibility measure in terms of
relative ease of access (network efficiency) in each of these temporal situations. Remote
locations invariably figure as being of low accessibility and consequently as possible
receptors of new investments, even though they may already have a very good provi-
sion of transport infrastructure. For transport planners the issue is to what degree are
spatial variations in accessibility a result of the infrastructure they provide rather than
a result of the geographic location of the places. It is therefore necessary to formulate
an accessibility indicator that will neutralize the effect of geographic location, in order
to meet this requirement of transport planners. To this end, the customary notion of
distance (length, time, or costs) must be substituted by another which expresses ease of
access in relative terms, that is in terms of network efficiency, as follows:

(I)
Accessibility and transport infrastructure planning 1339

where
Ai is the accessibility of node i,
N,} is the network impedance through the network between node i and destination
activity centre j,
Eij is the Euclidean impedance between the two (assuming optimal infrastructure
efficiency), and
M} is the mass (income) of the destination economic centre.
Thus, for each node a weighted average of the ratio N Ij / Elj is calculated, taking as
weights the incomes of the economic centres in the destination, ~. The ratio N,; / E;j
expresses the relative ease of access in each relationship: the closer network impedance
comes to the Euclidean impedance, the lower this figure will be. Network impedance
from any node on the network to each economic centre [NIj in equation (I)] is that
obtained by the minimal route, adding the impedances of the arcs crossed and the turn
impedances on several nodes (see section 3). Euclidean impedance (;j) should be
specifically defined for each transport mode. For example, for road transport the
Euclidean impedance between any two nodes is the estimated travel time through a
hypothetical straightline noncongested motorway link. The importance of each relation-
ship is valued according to the economic weight of the destination activity centre (Mj ).
The greater or lesser proximity of activity centres (as in the economic potential model)
is not taken into account in the weight, for a national plan should guarantee efficient
connections to all chief economic activity centres, with particular regard for long-
distance interregional relationships rather than short-distance relationships, the
analysis of which pertains to regional transport plans. Consequently, the approach
adopted does not take into account the real transport demand but rather the network
efficiency for these multiple connections. However, in future work other similar acces-
sibility indicators can be used that do in fact include the interaction in the weights.
The proposed indicator gives a value in meaningful units: how much network
impedance surpasses Euclidean impedance. The results are in an inverse scale: the
lower the value attained the more accessible is the place in relative terms. Thus,
when the infrastructure connecting the nodes to the various economic activity centres
is very efficient (that is, relatively straight and high speed), nodes receive an accessibility
value which tends towards 1. A value equal to 2 means that the network impedance is
double the Euclidean one and so forth.(I)
It could thus occur that a region which is peripheral according to the potential
model is highly accessible in terms of network efficiency, according to the measure
introduced in this paper. In such a case, if the target is to improve the economic
potential of this region, the volume of its activity should be increased, as an improve-
ment in the provision of transport infrastructure will produce only minor effects. The
less accessible a region is in terms of network efficiency the more an improvement in
the provision of transport infrastructure will produce effects on the economic potential.
[n consequence this accessibility indicator can be useful to assist decisions to invest in
transport networks, as it is more sensitive than the potential model to the transport
II)A weak point in this measure of accessibility is (hat, as it is formulated here, it can be applied
only to compact spaces. If the form is not compact (as is the case, for example, of tbe whole of
Europe), there is no sense in using the Euclidean impedance (in a straight iine along a hypothetical
high-quality infrastructure) as we have done in this paper, because geographical conditionings
would be incorporated into the indicator. Thus, in the case of road transport in Europe, there is no
point in using a straightline road connection between Valencia and Naples over the Mediterranean
when calculating Euclidean impedance. Yet in these cases there is the alternative of calculating the
shortest land path (outside the transport network) between both nodes and considering high-
quality infrastructure over this minimal path as the best hypothetical connection between the
two points. This calculation is quite easily approached with a raster GIS.
1340 J Gutierrez, A Monzon, J M Pinero

infrastructure needs of each individual region.(2) This is the main contribution of this
measure of accessibility. Certainly many indices would indicate more or less similar
changes in accessibility between two temporal situations (depending on the distance
decay function if any), because the effect of the geographic location is neutralized by
the calculation of accessibility changes, but results would be very different in each of
these temporal situations if the proposed indicator is used.
This accessibility indicator is inspired by the measure known as the route factor.
which is used to measure the sinuosity of individual links. This is the ratio of the actual
link-distance between any two points to the straightline distance between them-the
larger the ratio, the less direct the route (Chapman, 1979; Chorley and Haggett, 1969).
In order to obtain a single value for each node Timbers (1967) calculated the unweighted
average route factor. Basically we start from this index but: (1) As a measure of the
separation between places we use impedances instead of mere distances. We do not
measure sinuosity itself but accessibility in terms of relative ease of access (network
efficiency). (2) The importance of each relationship is valued according to the economic
weight of the destination activity centre.

3 Empirical testing of the accessibility indicator


In order to test the utility of this indicator for the planning of transport infrastructure,
we have analyzed accessibility conditions by road, rail, air, and sea transport in the
situations 'before' and 'after' the adoption of the Spanish Infrastructure Master Plan
1993 - 2007, but in this paper only results for road and rail are included. This Master
Plan (MOPTMA, 1994) is the fundamental planning document drafted by the Ministry
of Public Works, Transport and Environment, as a frame of reference for developing
the totality of the infrastructure for which the national government is responsible.
For the road network this includes the building of 5300 km of motorways and dual
carriageways and 1400 km of connecting roads; for the rail network it foresees the
construction of four new high-speed lines (Madrid - Barcelona, Saragossa - Bilbao,
Madrid - Valencia, and Madrid - Lisbon), and the upgrading of several lines for a
speed of 200 km h- '.
This analysis is mainly focused on the effect of infrastructure on accessibility,
without directly taking into account the supply of services that cannot be modified
by this plan. Accessibility was calculated between 450 nodes and 30 economic centres
in Spain in the preplan and postplan situations. The 30 economic centres are urban
agglomerations with populations of at least 150000 inhabitants. These centres account
for 51 % of the whole of Spain's population. The economic measure (income) of the
centres in the indicator was fixed, so that every economic centre has the same value in
both situations (1993 and 2007) in order to demonstrate the pure infrastructure effect.
The 450 populated places were chosen according to demographic criteria and also in a
manner that provided regular coverage across the country, to facilitate isoaccessibility
mapping. Based on the accessibility values of these 450 nodes, contour maps were
generated by interpolation in order to obtain a continuous view of the space.
A GIS was used in which transport networks were modelled , accessibility values
were calculated and mapped, and in which the benefits of the plan were evaluated from
a twofold perspective, both social and spatial. In order to evaluate these benefits,
population and area were calculated according to the different accessibility levels in
the preplan and postpian situations.

( 1) It is assumed that improvement in network efficiency not only reduces interaction costs. but also
increases the competitiveness of regions and allows for more specialisation such that economies of
scale and specialisation benefits may be utilised (Forslund and Johansson. 1995).
Accessibility and transport infrastructure planning 1341

The calculation of the area affected by each level of accessibility is a very simple
operation in a GIS, for the system automatically calculates the area of each accessibility
polygon. In order to calculate the population at each level of accessibility, we used a
dense network of points in which the spatial distribution of the Spanish population was
reflected. By combining both layers (population and accessibility) by means of a point
in polygon overlay, we were able to calculate in the GIS the population (according to
the 1991 Census) that lives at each level of accessibility.
Comparisons with other indices are not made, because spatial patterns resulting from
classical accessibility indicators are well known (see the above cited literature): a typical
concent ric core - periphery pattern in the fo rm of a succession of accessibility rings more
or less distorted by the effect of infrastructure and the location of economic centres.
3.1 Road network
3.LI Accessibility calculations
All national roads have been taken into account. Likewise, several Portuguese roads
have been considered, to analyze ease of access to economic activity centres in
Portugal and three fictitious nodes were defined on the French frontier to simulate
the effect of activity centres situated beyond the Pyrenees.
The network impedance (N;j in equation (I)] by road from any node on the network
to each economic centre is that obtained by taking the minimal route, adding the
impedances of the arcs crossed (I.) and the impedances (penalties) on several nodes
(I.) as follows:

(2)

Time is the basic variable taken into account in calculating the impedances of the
arcs (time measurements along each arc were provided by the Ministry of Public Works
and Transportation). However, the choice of routes is not made solely in terms of
minimum travel time. Comfort and safety in driving are also influential, and to a great
extent depend on the type of road and the traffic conditions. In order to simulate this
effect, two coefficients are added as follows:
fa = Ta(l +c~ +c~ ), (3)
where
T. is the time required to traverse the arc,
c~ is the coefficient of infrastructure quality,
c~ is the coefficient of traffic.
The coefficient of infrastructure takes the following values: - 0.2 for motorways, 0.0 for
main roads, and 0.2 for other roads. The coefficient of traffic (only applied for roads)
was calculated as fo llows:
c~ = expa + expb-2, (4)
where
a is the average daily traffic (ADT)/500000.
b is the percentage of heavy vehicles/200.
Thus, a road with very bad traffic conditions (an ADT of 125000 vehicles, 50% of
which are classified as heavy), would have a c~ penalty of 57%.
Node impedances were used only to simulate delays linked to crossing international
frontiers (10 - 20 minutes, depending on the intensity of crossing traffic). The delays
caused by crossing an urban area were handled with surrogate arc impedances.
The Euclidean impedance from any node to each economic activity centre [Eij in
equation (1)1 is calculated as simply a straight line between node and centre, assuming
1342 J G utierrez, A Monzon. J M Pinero

the best quality road (motorways) and without applying the coefficient of traffic
described above (for it represents an ideal situation with no congestion). That is, the
time to be spent in a straight line at a speed of 120 km h- I was calculated (the speed
limit allowed in Spain), with an infrastructure coeffi cient of - 0.2.
3. 1.2 Results
Accessibility contours and accessibility values for selected cities in the preplan situation
are shown in figu re I (a) and table I (the lowest values correspond to the highest
accessibility levels). There are outstanding high-accessibility corridors along the molof-
ways: most of them are radial and some transversal. Within these corridors, there is
extremely high accessibility in the nodes at which motorways converge, some of them
in central positions-such as Madrid (1.23) - and others in peripheral positions- such as
Barcelona (1.18). Highly accessible nodes correspond not only to big cities, but also to
small ones if they are located at important road intersections, such as Benavente (15000
inhabitants) in the northwest of the country. Less accessible areas are located outside
these corridors, particularly in mountainous regions. For example, Somport, in the
Pyrenees, has the worst value (2.26) among the selected places. The far northwest of the
peninsula (Galicia) is an inaccessible region too, because it is not linked by motorway
with other Spanish regions: La Coruna shows a n accessibili ty value of 1.90 and Vigo a
value of 2.10. This means that the network impedance is more or less double the
Euclidean one in both cities.
In the situation after the plan has been implemented [figure I(b) and table I],
generalized im provements are to be observed. The new motorways improve accessibil-
ity above all in the west and northwest of the country. The values for Vigo decrease
from 2.10 to 1.59, for La Coruna from 1.90 to 1.48, for LeOn from 1.72 to 1.42, and for
Salamanca from 1.78 to 1.52. Minor changes are observed in Madrid and Barcelona, as
they were a lready very accessible cities in the preplan situation.
Table 2 shows the percentage of area and population within each of the accessi-
bility levels in the preplan and post plan situations. The proportion of population in the
first accessibility levels is higher than the proportion of area, for logically most
Table I. Accessibility values in the preplan a nd post pla n situations for selected ci(ies.

Cities Accessibility by road Accessibility by railway

Val ue - Difference Value Difference

preplan postplan absolute % preplan postplan absolute %

Madrid 1.23 1.18 0.05 4.01 2.32 1.36 0.96 4 1.38


Barcelona 1.18 1.16 0.02 1.69 2.52 1.30 1.22 48.41
Va lencia 1.52 1.36 0.16 10.53 3.30 2.26 1.04 31.52
Seville 1. 59 1.47 0.12 7.55 2.36 1.78 0.5 8 24.58
Bilbao 1. 5 1 1.42 0.09 5.96 4.20 2.08 2. 12 50.48
Malaga 1. 63 1. 54 0.09 5.52 3.27 2.49 0.78 23 .85
La Coruna 1.90 1.48 0.42 22.11 4.23 3.04 1.19 0.28
Valladolid 1.59 1.49 0.10 6.29 3.23 1.82 1.41 0 .44
Sa ragossa 1.56 1.46 0.10 6.41 3.3 1 1.68 1.63 0.49
A lmeria 1.17 1.55 0.22 12.43 4.50 3.30 1.20 0.27
Salamanca 1.78 1.52 0.26 14.6 1 3.50 2.51 0.99 28.29
Leon 1.72 1.42 0.30 17.44 3.30 2.09 1. 2 1 36.67
Vigo 2.10 1.59 0.51 24.29 4.34 2.81 1.53 35.25
Somport 2.26 1.86 0.40 17.70 5.99 3.54 2.45 40 .90
3 The lower the value a ttained the more accessible is the place in relative te rms.
Accessibility and transport infrastructure planning 1343

Accessibility value
< 1.6 (level I)
1.6 - 1.8 (level 2)
1.8 - 2.0 (level 3)
2.0 - 2.2 (level 4)
2.2 - 2.4 (level 5)
> 2.4 (level 6)
Road type
roads
mOloTways
km
(a) ..... o 100 200

Figure 1. Relative accessibility by road (network efficiency): (a) preplan, 1993; (b) post plan. 2007.
1344 J Gutierrez, A Monzon, J M Pinero

Table 2. Changes in road accessibility in terms of area and population in the preplan and
post plan situations.
Accessibility value" Percentage of area Percentage of population
preplan postplan preplan postplan
< 1.6 2.3 18.9 32.5 56.8
1.6 - 1.8 44 .3 58.3 48.2 38.7
1.8 - 2.0 35.7 18.5 11.4 4.0
2.0 - 2.2 15.0 3.1 7.8 0.5
> 2.2 2.7 1.1 0.2 0.0
Total 100 100 100 100
A A low value indicates high accessibility.

populated areas tend to have good accessibility. Changes between preplan and postplan
situations are remarkable. The proportion of area in the first accessibility level 1.6)
increases from 2.3% to 18.9% and the proportion of population from 32.5% to 56.8%.
On the other hand, the proportion of area in the two worst accessibility levels (> 2.0)
decreases from 17.7% to 4.2%. and the proportion of population from 8.0% to 0.5%.
3.2 Rail network
3.2.1 Accessibility calculations
All Spanish interregional railways were taken into account, those stations at which long-
distance trains stop being marked as nodes. Portuguese railways that allow for access to
Lisbon and Oporto were also included, as were certain fictitious nodes along the French
frontier to reflect the connections towards northern countries. In view of the fact that the
rail network does not densely cover the whole territory, road transport, the modelling of
which was described previously, was used as a complementary mode of access to stations.
Within the rail network, resistance to movement can occur both on the arcs and on
the nodes. The impedance of each arc is given by a combination of the variables travel
time and congestion, and it is the result of applying the following formula;

la = 0.4 (S)'
T+7 C T , (5)
where
la is the impedance of the are,
n is the number of tracks,
S is the number of services (a day),
C is the capacity (a day),
T is the minimum travel time on the arc by the fastest passenger train.
The above formula implies recording the minimum travel time on the arc together with
an additional quantity that is a function of this time and of two variables: the number
of tracks on the line and the ratio SI C which represents congestion on the line. For a
congestion that one could consider normal in which SIC = 0.7 the ratio
(S/C)2 = 0.5, which means that travel time is increasing by 5% on a double track
and by 20% on a single track. For serious congestion in which S I C = 1, the time
increases considered would be 10% on a double track and 40010 on a single track. It is
quite clear that such penalties do not respond to an increase in real time (for the means
of regulating rail transport would facilitate circulation without delays or with mini-
mum delays), but that they reflect the greater or lesser resistance to movement on a
given arc. In this way, not only improvements in the travel times foreseen in the plan
could be reflected in the indicator, but also actions foreseen in order to increase the
capacity of the network.
Accessibility and transpof[ infrastructure planning 1345

The following impedance has been formulated for nodes on the rail network:
In = PI + Pw + Pd , (6)
where
In is the impedance of the node,
Pt is the penalization through changing trains,
P.., is the penalization through changes in track width,
Pd is the penalization through change of direction.
The formulation thus established consists of a series of added items whose value is
expressive not only of the time spent but also of the discomfort certain movements
may cause passengers. These penalties are directional and therefore depend on both
arcs amongst all those that converge on the node taken into account:
Changing trains: There are some lines, mostly in the north of Spain, that have a different
track width to the rest of the country. Switching from one to the other means changing
trains, the penalty of which has been estimated at 60 minutes to reflect not only the time
taken but also the discomfort and uncertainties caused to passengers by such changes.
Changing track width: The new Spanish high-speed lines were built with the European
track width. This means that there are breakage points (because of the change from
Spanish to international width) not only as until recently incertain stations along the .
French frontier, but also in certain stations within Spain. However, systems have been
developed that enable trains to switch from one track width to another relatively easily
and quickly without passengers having to change trains, in about 30 minutes.
Change of direction: A penalty of 20 minutes was established for the change of direction
that trains had to carry out at certain stations. This reflects not only the time spent in
the operation, but also the increase in congestion this means for the station and the
psychological discomfort caused to passengers.
Thus, the impedance between the node of origin and the node of destination on the
rail network (Jrw) is the one obtained by the minimal route, bearing in mind not only
the travel time of the sections crossed and the degree of congestion (arc impedances),
but also a series of delays linked to switching trains, changing widths, and changing
directions (node impedances):
1~~LJ.+ LI", (7)
in which the first term on the right-hand side is the sum of the arc impedances crossed
and the second is the sum of the corresponding directional node impedances.
When the point of origin of a trip does not coincide with one of the nodes on the
rail network, the roads play the part of a complementary mode. In such cases, road
impedance is considered from the node of origin to the nearest station by the minimal
route, plus an impedance for changing the mode associated with this node. The method
for calculating road impedance was described above. On the other hand, impedance
through change of mode, which simulates the time taken and the discomfort caused by
the change of mode from road to rail was estimated at 80 minutes.
Consequently, in rail transport the network impedance [Ny in equation (1)] between
the node of origin and the destination economic centre is finally given as:
Ny = It + 1m + Irw , (8)
where
NIj is the network impedance between node i and destination activity centrej,
Ir is the impedance by road from node of origin to the nearest station,
1m is the impedance through change of mode,
Irw is the impedance by rail from the nearest station to the destination economic
activity centre.
1346 J Gutierrez, A Monzon, J M Pinero

Accessibility value
<2.5 (level 1)
2.5 - 3.0 (level 2)
3.0 - 3.5 (level 3)
3.5 - 4.0 (level 4)
4.0- 5.0 (level 5)
> 5,0 (le~'el 6)
Line type
- Other lines
- High speed
,
(aJ ... o

Figure 2. Relative accessibility by rail (network efficiency): Ca) preplan, 1993; (b) postpla n. 2007.
Accessibility and transport infrastructure planning 1347

The Euclidean impedance (E;"in eq~ation (1)] b~tween the node of origin and the
destination economic activity centre is that obtained in a straight line by a high-speed
train at a speed of 250 km h- l .
3.2.2 Results
In the preplan situation [figure 2(a) and table I], the high level of accessibility
attained along the new high-speed line between Madrid and Seville is outstanding.
The stations along this line mark the points of maximum accessibility in the whole
geographical area of Spain (for example, Madrid shows a value of 2.32 and Seville
2.36). The effects of this high-speed line go beyond Seville and even reach certain cities
on the southwestern coast. There are also good accessibility conditions in certain
'islands' that correspond to important nodes on the network connecting relatively
fast sections: some of these nodes correspond to large cities (such as Barcelona), but
others are relatively small places in not very populated regions (such as Albacete or
Alcazar de San Juan, in the region of Castilla - La Mancha). The worst accessibility
conditions are to be found in certain areas that are served by lines with low-quality
infrastructure (La Coruna, Vigo, Almeria, and Bilbao show values higher than 4),
particularly those which have poor access conditions to the nearest station by road
(a value of 6 in Somport).
Most Spanish territory, approximately 79%, is situated between levels 5 and 6
(table 3). Level 5 is also the one which affects the highest percentage of the Spanish
population, almost 33%.
In the postplan situation [figure 2(b), there is a radical variation in the values of
accessibility, because of the building of new high-speed lines and improvements in
many of the existing lines. The future high-speed lines Madrid-Barcelona-French-
frontier and Saragossa-Basque-country will produce important effects on accessibility
values. Values for Bilbao decrease from 4.20 to 2.08, for Saragossa from 3.31 to 1.68,
and for Barcelona from 2.52 to 1.30. The zones with worst accessibility are the same as
those in the preplan situation, although they improve by about two levels (I unit) with
respect to the preplan situation.
The predominant levels of accessibility become I and 2, which between them
cover 63% of the territory (table 3). The difference is even more marked in popula-
tion, for 82% of the population is situated in regions affected by the first two levels
of accessibility.
Table 3. Changes in rail accessibility in terms of area and population in the preplan and post plan
situations.
Accessibility value<l Percentage of area Percentage of population
preplan postplan preplan post plan

<2.5 0.0 27.7 0.0 57.5


2.5 - 3.0 1.9 35.2 23.8 24.6
3.0 - 3.5 3.7 19.0 9.5 11.8
3.5 - 4.0 15.0 10.7 24.4 4.4
4.0 -5 .0 50.5 6.0 33.0 1.3
> 5.0 28.9 1.3 9.2 0.3
Total 100 100 100 100
a A low value indicates high accessibility .
1348 J Gutierrez, A Monzon, J M Pinero

4 Final remarks
An accessibility indicator has been formulated, which synthesizes in a single value
the access conditions of each node in terms of network efficiency, emphasizing the
infrastructure effects. This indicator offers advantages for certain purposes, but logically
cannot claim to replace other accessibility measures. In this paper we have demon-
strated the practicality of the idea. The indicator allows us to identify not only corridors
(linear regions lying on networks) and shadow regions (those apart from the networks),
but also nodal and interstitial places in a corridor (for a fuller explanation about these
concepts see Vickerman, 1996). In fact nodes situated along the main transport corri-
dors, and particularly those located on the intersections of these corridors, are those
which show better accessibility conditions in terms of network efficiency, independently
of their geographic location.
Railway and road transport offer different accessibility patterns: in the first place,
because access to the railway network occurs only at specific points through stations
(in our model, those at which long-distance trains stop); second, because there are
great differences in speeds within the railway network, from 250 km h~l on the high-
speed Madrid - Seville line to approximately 50 km h- l on certain obsolete sections.
Although corridors and shadow regions can be identified in both modes, corridors are
more evident in road transport (shaping the space in a relatively continuous way),
whereas rail transport highlights the better accessibility of nodal places, especially
along the high-speed line (creating a more discontinuous space).
Results offered by this indicator differ from those obtained by the potential model.
Certainly many of the most accessible nodes correspond to big cities, with large
demands for transport, but others correspond to relatively small cities in not very
populated regions. Less developed regions in peripheral locations, such as Andalucia,
can have very accessible nodes in terms of network efficiency, and it is possible to find
poor accessibility conditions in several nodes located in central regions, such as some
areas near Madrid, so that there is not a strong relationship between accessibility in
terms of network efficiency and accessibility in terms of economic potential. On the
other hand, the results offered by this indicator are easily interpretable, for the network
impedance is compared with the Euclidean impedance, there being a minimum refer-
ential value of 1 (theoretical) for the nodes with the best accessibility. This is an
important advantage with respect to other measures such as the market-potential
measure: as Geertman and Ritsema van Eck (1995) point out, a potential value of
say 12554 for location x is not easy to interpret.
The indicator we use here highlights in particular the imbalances between the chief
nodes in the network and their surrounding territory, a matter to which more and more
attention is being paid (Vickerman, 1994; 1996). Measures of access to networks, such
as the ICON indicator (CEDRE, 1993; see the appendix), do emphasize these imbal-
ances too. But differences between the two approaches are substantial: the indicator of
network efficiency measures accessibility to the economic centres and the CEDRE
indicator measures access to the networks, so that the same nodes may have very
different accessibility values according to the measure adopted. For example,
La Coruna and Vigo, in west Galicia, would be very accessible cities in road transport
terms according to the CEDRE indicator, because there is a motorway along this area,
but these cities have very poor accessibility conditions according to the indicator of
network efficiency, because the existing motorway in the area is not well connected
with the other motorways in the country, so that access from these cities to most
Spanish economic centres is difficult.
Although in this paper the indicator has only been empiricaJly tested for road and
rail transport, there is no a priori difficulty in using it in the case of air transport as well.
Accessibility and transport infrastructure planning 1349

T he main traffic centres, with direct connections to most airports, should give
much lower values on this indicator than small airports which have only few direct
connections. Road (and railway) transport may be considered as a complementary
mode (access to airports) in a similar way to that in the previous analysis of rail
accessibility (access to stations).
Acknowledgements. Support from Ministerio de Obeas Publicas, Transporte y Media Ambiente,
and rrom Comision Interministerial de Ciencia y Tecnologia, data from Dr J L Calvo, and referees'
comments are gratefully acknowledged.
References
Allen W B, Liu D, Singer S,I993, "Accessibility measures of US metropolitan areas" Transportation
Research A 27439-449
Ball R, 1996, "Local sensitivities and the representation of peripherality" Journal 0/ Transport
Geography 4 27 -36
Botham R W, 1980, "The regional development effects of road investment" Transportation Planning
and Technology 6 97 - 108
Bruinsma F R, Rietveld P, 1993, "Urban agglomerations in European infrastructure networks"
Urban Studies 30 919 -934
CEDRE, 1993 Etude prospective des regions at/antiques report to European Commission, Centre
Europeen du Developpement Regional, DG XVI, Strasbourg
Chapman K, 1979 People, Pallerns and Process (Edward Arnold, Sevenoaks, Kent)
Chorley R J, Haggett P, 1969 Network Analysis in Geography (Edward Arnold, Sevenoaks, Kent)
Clark C, Wilson F, Bradley J, 1969. "Industrial location and economic potential in Western Europe"
Regional Studies 3 197 - 212
Ding Y, Fotheringham A S, 1992, "The integration of spatial analysis and GIS" Computers,
Environment and Urban Systems 16 3-14
Dundon-Smith D M, Gibb R A, 1994, "The Channel Tunnel and regional economic development"
Journal oj Transport Geography 2 178 - 189
Forslund U, Johansson B, 1995, ''Assessing road investments: accessibility changes. cost benefit
and production effects" The Annals 0/ Regional Science 29 ISS - 174
Frost M E, Spence N A, 1995, "The rediscovery of accessibility and economic potential: the critical
issue of self-potential" Environment and Planning A 27 1833 -1848
Geertman SCM, Ritsema van Eck J R. 1995, "GIS and models of accessibility potential: an
application in planning" International Journal oj Geographical In/ormation Systems 9 (I) 67 - 80
Gutierrez J, Urbano P, 1996, ''Accessibility in the European Union: the impact of the Transeuropean
road network" Journal oj Transport Geography 4 15 - 25
G utierrez J, Gonzalez R, Gomez G, 1996, "The European high-speed train network: predicted
effects on accessibility patterns" Journal o/Transport Geography 4 227 -238
Harris C 0, 1954, "The market as a factor in the localisation of industry in the United States"
Annals of the Association of American Geographers 44 315 - 348
Ingram 0 R, 1971, "The concept of accessibility: a search for an operational form" Regional Studies
5101 -107
Izquierdo R, Monzon A, 1992, "La accesibilidad a las redes de transporte como instrumento de
evaluacion de la cohesion economica y social", paper presented at 12th International
Symposium of European Conference of Transport Ministers, Lisbon; copy available from
Departmento de Ingenieria Civil y Transportes, Universidad Politecnica de Madrid, 28040
Madrid
Janelle 0 G, 1969, "Spatial reorganization: a model and concept" Annals o/the Association 0/
American Geographers S9 348 - 364
Keeble 0, Offord J, Walker S, 1988 Peripheral Regions in a Community of Twelve (Office for Official
Publications of the European Communities, Luxembourg)
Linneker B, Spence N A, 1992, "Accessibility measures compared in an analysis of the impact of
the M2S London Orbital Motorway on Britain" Environment alld Planning A 24 1137 - IIS4
Lutter H, Piitz T, Spangenberg M, 1992 Accessibility alld Peripheralizy of Community Regions: Tile
Role of Road, Long-distallce Railways and Airport Networks report to European Commission,
Bundesforschungsanstalt fUr Landeskunde und Raumordnung, DO XVI, Bonn
MOPTMA, 1994 Injrastructure Master Plan 1993 - 2007 Ministry of Public Works, Transportation
and Environment, Madrid
Morris J M. Dumble P L, Wigan M R, 1979, ''Accessibility indicators in transport planning"
Transportation Research A 1391 -109
1350 J Gutierrez, A Monzon, J M Pinero

Plassard F, 1992, "L'impact territorial des transports a grande vitesse", in Espace et dinamiques
terri/oriales Ed. P-H Derycke (Economica, Paris) pp 243 - 261
Pooler J A, 1995, "The use of spatial separation in the measurement of transportation accessibility"
Tran.~portatjon Research A 29 421 - 427
Simmonds 0, Jenkinson N, 1995. "The impact of changing transport services in Europe", in
Proceedings of PTRC Summer Annual Meeting, Seminar A, Pan-European Transport Issues,
Warwick (PTRC. London) pp 121 - 134
Smith D M. Gibb R A, 1993, "The regional impact oftbe Channel Tunnel: a return to potential
analysis" Geoforum 24 183 - 192
Spence N, Linneker B, 1994, "Evolution of the motorway network and changing levelsofaccessibility
in Great Britain" Journal o/Transport Geography 2 247 - 264
Spiekermann K, Wegener M, 1994, "The shrinking continent new time- space maps of Europe"
Environment and Planning B 21 653 - 673
Timbers J A, 1967, "Route factors in road networks" Traffic Engineering and Control 9 392 - 401
Vickerma n R W, 1974, "Accessibility, attraction and potential: a review of some concepts and their
use in determining mobility" Environment and Planning 6 675-691
Vickerman R W, 1994, "Transport infrastructure and region building in the European Community"
Journal of Common Market Studies 32 1 - 24
Vickerman R W, 1995, "The regional impacts of trans-European networks" The Annals of Regional
Science 29 237 - 254
Vickerman R W, 1996, "Location, accessibility and regional development: the appraisal of trans-
European networks" Transport Policy 2 225 - 234

APPENDIX
Some accessibility measures
(I) The economic potential model is well known. Its mathematical expression is:
" M.
Pi = LD ~ ' (AI)
j = J if
where
Pi is the economic potential of node i,
M j is a measure of the economic activity of node j,
Dij is a measure of the transport costs between nodes i and J,
a is a parameter reflecting the rate of increase of the friction of distance.
(2) Potential transport costs (7;) are given by
"
7; = LMjDij. (A2)
j .. l

(3) Weighted mean distances are calculated as follows:

A; ~ 1;(D;j MJ /1; Mj . (A3)

(4) The ICON index, J, has the following formulation:


I; ~ ( I +P,)(Tm +T, +P'1;), (A4)
where
is the sum of times of connection to each network,
is a penalty for the absence of a particular network, in terms of extra time,
is a penalty for any discontinuities or gaps in the network,
are weights which reflect the importance of the mode and the penalties for
having to substitute an alternative.

p e 1998 a Pion publication printed in Great Britain

You might also like