Role of GIS in Traffic Engineering

Introduction:
GIS stands for Geographic Information System. It consists of hardware, software and data that helps to
collect,store and manipulate any information based on its geographic location. To put it in perspective,
GIS helps to visualise data by spatially distributing them across a map, to make it more comprehensive
than when it is trapped in tables and charts and sheets. Thus, seemingly unrelated information once
placed on a map may help to unravel hidden links and patterns that might not have been so obvious
otherwise. Hence, GIS has widespread applications on almost all fields of science and research. Here,
we concentrate on the application of GIS in traffic engineering.
Traffic Engineering or Transportation Engineering aims to create smooth and safe flow of traffic.
Traffic is something that changes temporally as well as spatially. Therefore a realtime analysis of traffic
taking into account the spatial variation is something that would help in traffic management. That is
where GIS comes in handy.

Applications of GIS in Traffic Engineering:

GIS can be used in the following ways in traffic engineering:
 Road Network and Public Transport Route Planning during preliminary design
 Realtime analysis for fastest route,journey time,stoppage time, not only for regular traffic but
during emergency situations
 Accident Hotspot Analysis
 Traffic Rule Violation Reduction by realtime monitoring
 Tracking of Public Transport and Availability
 Parking Demand and Solution
 Traffic Induced Pollution Study, and finally,
 Using the output of all the above analysis for a better management of traffic congestion,delays
and accidents.

The above applications are discussed in detail in the following sections with regard to the methods
involved and a few case studies.

1. Route Planning
The traditional method of determining road networks and public transport routes through Origin and
Destination studies can be performed much more easily with the help of a GIS. A GIS based software
can prepare maps with the landuse and demographic information overlayed on it. This, together with
the data on public demands of destination and origin and factors like time, and cost can create a more
comprehensive understanding. Furthermore, more recent criteria like environmental and greenhouse
considerations, energy consumption etc can be taken into account in such softwares. There are specific
Route Analysis Tools inbuilt in such software that take into account the following info:
1. Route Specific Data: Desired number of trips in a timeperiod, the distance involved, the net
cost, the no of public transport vehicles required to accommodate the demand, as well as the
projected ridership in future using statistical analysis based on available data.
2. Demographic Data: Total population, households, private vehicle owmership data, income
category, age of population, percentage of population with disability
3. Employment: Number of employees, current transit riders, people interested in transit.

After planning of the route by considering the above criteria, the implementation of the route involves
deciding the terminal points and stoppages for the public transport. The boarding points should be
distributed according to the demand area densities shown in the map. The effect of the boarding points
on the existing traffic flow should also be taken into account by using realtime modelling and
simulations.

The next step after implementation of boarding points is monitoring them. Every data regarding the
boarding point like the location id,the number of bays, the passenger facilities provided are fed into the
GIS for further monitoring and analysis of performance.

2. Fastest Route Planning

A map of a city will show us the possible routes to reach a destination from a source. A database can
also be made available of the details of these routes, like the road widths, the variation of traffic along
these routes at different times of the day, the length of the routes. If we know the speed of the vehicle
and speed of the traffic stream, an estimate can be made of the journey time through available
formulae. Clubbing the database of the routes with the map through geocoding is what GIS does.

The visual and geographical aid provided by GIS can be further exploited to find out the fastest routes.
There are several algorithms based on graph theory for finding out the shortest path. The principle lies
in the fact that the shortest path is the path between 2 vertices or nodes in a graph such that the sum of
the weights of its constituent edges is minimized. We can imagine the road network as a graph with
positive weigths, and the nodes represent road junctions and each edge of the graph is associated with a
road segment between 2 junctions. The traffic equivalent of this problem is finding the shortest path
between 2 intersections on a road map each weighted by the length of its road segment. The weight of
an edge may also correspond to the time needed to traverse the segment, or the cost of traversing the
segment. Using directed edges it is also possible to model one-way streets. Such graphs are unique as
certain edges are more important than others for long distance travel (e.g. highways). This property has
been formalised using the notion of highway dimension.

The most primary shortest path algorthm is the Dijkstra's algorithm which solves the single source
shortest path problem. Other notable ones are Bellman-Ford Algorithm, Viterbi Algorithm etc. All of
the specialised road network algorithms work in 2 phases. In the first phase, the graph is preprocessed
without knowing the the source or target node. The second phase is the query phase. In this phase,
source and target are known. The idea is that the road network is static, so the proprocessing phase can
be done once and used for a large number of queries on the same road network.

But the challenge in case of shortest path problem for a road network is the traffic which is varying at
every point spatially and temporally. A traveller traversing a link daily may experience different travel
times on that link due not only to the fluctuations in travel demand ( origin-destination matrix) but also
due to such external factors like work zones, bad weather, accidents, and vehicle breakdowns. As a
result a stochastic time dependant modelling and solution is more suited for optimisation of the shortest
path problem in traffic. Therefore the shortest route might not be the fastest. A GIS system with
continous shared dataflow system that will collect the changing traffic data at every point of time from
the traffic signals and detectors is required to update the constraints for the algorithm. The shortest path
therefore needs to be modified with time and change in traffic.

While navigation apps like Google Maps and cab-rental apps like Uber provide their own GIS based
systems for shortest path and expected journey time with updates of realtime traffic, the most critical
application of this facet of GIS might be in routing ambulances and fire engines to sites of fatality and
guiding the ambulances to nearest hospitals. It also has great use for delivery businesses where
optimisation through algorithms like The Travelling Salesman problem can help to minimise wastage
of fuel and time and reduce congestion.

Case Study
A simple application of the shortest path problem is shown here using Djikstra's algorithm, for its
simplicity. The problem in hand is finding out the shortest route from home to school. Some routes
have greater congestion than others which has been taken into account as weight. The diagram is given
below.

From observation, the shortest route is along Home-B-D-F-School.

Dijkstra’s algorithm finds a shortest path tree from a single source node, by building a set of nodes that
have minimum distance from the source.

The graph has the following:

 vertices, or nodes, denoted in the algorithm by u or v;

 weighted edges that connect two nodes: (u,v) denotes an edge, and w(u,v) denotes its weight. In
the diagram on the right, the weight for each edge is written in gray.
This is done by initializing three values:

4. dist, an array of distances from the source node s to each node in the graph, initialized the
following way: dist(s) = 0; and for all other nodes v, dist(v) = ∞. This is done at the beginning
because as the algorithm proceeds, the dist from the source to each node v in the graph will be
recalculated and finalized when the shortest distance to v is found
5. Q, a queue of all nodes in the graph. At the end of the algorithm's progress,Q will be empty.
6. S, an empty set, to indicate which nodes the algorithm has visited. At the end of the algorithm's
run, S will contain all the nodes of the graph.
The algorithm proceeds as follows:

1. While Q is not empty, pop the node v, that is not already in S, from Q with the smallest dist(v). In the
first run, source node s will be chosen because dist(s) was initialized to 0. In the next run, the next node
with the smallest dist value is chosen.

2. Add node v to S, to indicate that v has been visited

3. Update dist values of adjacent nodes of the current node v as follows: for each new adjacent
node u,
 if dist(v) + weight(u,v)<dist(u), there is a new minimal distance found for u, so update dist(u) to
the new minimal distance value;
 otherwise, no updates are made to dist(u).
The algorithm has visited all nodes in the graph and found the smallest distance to each node. dist now
contains the shortest path tree from source s.

Note: The weight of an edge (u,v) is taken from the value associated with (u,v) on the graph.

3. Accident Data Validation and Hot Spot Analysis

A major part of traffic management deals with minimisation of accidents. Accidents can occur due to a
variety of causes, starting from vehicle defects, road user negligence, faulty traffic planning to factors
beyond control like climate and weather and natural disasters. Hence a totally accident free system is
not feasible, only minimisation is what can be aimed for. GIS tools helps in analysing accident data and
records to find spatio-temporal correlations which can be then be rectified with due consideration.

The first part of accident analysis has to deal with validating the accident records already available with
the respective departments like the police, the insurance companies and hospitals and rectifying the
mistakes,if any, in the record. This can be done by creating a map of the road network where all the
roads and intersections are geocoded. Then, the available accident records of a particular accident are
checked in terms of spatial attributes, surrounding features etc to get a match on the geocoded map. If a
match is available, the concerned accident data is validated and geocoded. If not, the data has to be
manually checked for typographical errors and then entered in to the system.

Case Study
A workflow diagram for the above process is presented from “Becky P. Y. Loo,2006:Validating Crash
Locations For Quantitative Spatial Analysis : A GIS-based Approach” :

This validation of accident records is absolutely essential as the correctness of the spatial analysis
depends on this data. Based on study done in the above paper for Hong Kong in 2004, it was found that
the police crash database had 12.7% mistakes for road names and 9.7% mistakes for district boards.
This gives a rough idea of the human error in non-geocoded accident records.

The next step involves spatial analysis of the accident location to find out relations. An accident hotspot
may be loosely defined as a high density accident zone. There are in built functions in GIS softwares to
find out the spatial autocorrelation between accident site locations. The Marvin's I method is used to
find out whether the accident locations in an area, in terms of distance, forms a cluster pattern or a
dispersed pattern or a random pattern. It uses both feature locations and feature values simultaneously.
To quantize the spatial locations into clusters, K-means clustering is a standard methods that is
used. It aims to partition n observations into k clusters in which each observation belongs to the cluster
with the nearest mean, serving as a prototype of the cluster.
There is also the Kernel Density Estimation technique. It involves placing a symmetrical surface
over each point and then evaluating the distance from the point to a reference location based on a
mathematical function and then summing the value for all the surfaces for that reference location. This
method is repeated for successive points. This therefore allows us to place a kernel over each
observation and summing these individual kernels gives us the density estimate for the distribution of
accident points. The main advantage of this method lies in the determination of the spread of risk of an
accident. The spread of risk is defined as defined as the area around a cluster where there is an
increased likelihood of occurrence of accident based on spatial dependency.
After determination of the clusters of hotspots, further analysis can be made on causes of accident
and measures taken based to reduce the risk.
Case Study
A case study for the accident analysis is presented from Anitha SD Selvasofia,Prince G
Arulraj(2016):Accident and Traffic Analysis using GIS :

The study area selected is Coimbatore district, concentrating on NH 47 Gandhipuram to Avinashi, NH

209 from Gandhipuram to Annur and NH 67 Gandhipuram to Mettupalayam.The methodology is
described as follows:
1. Data Collection : The identification of accident prone and congested zones is done by collecting
data, both spatial and non-spatial. Spatial data involves exact geometric information like location,
boundary extent, road network, which is done by collecting maps from concerned authorities. Non-
spatial data includes accident details obtained from police,insurance and hospital authorities.

2. Data Processing : The Ground Control Points were collected with the help of handheld GPS in
WGS1984 stratum. The accident spots are converted into shape files in the ArcGIS software and the
accident information added as attribute. The road network and taluk maps are georeferenced to extract
vector data from base rastor data. The extracted layers like district boundary layer, NH , SH, Major
District Roads are saved as shape files. The extracted and generated layers are added to geo-database.

3. Buffer Analysis : The buffer analysis was done to create a buffer polygon to find the area in and
around accident location.

4. Accident Spot : The accident spots are visualised by overlaying the accident location with the NH
road and accident hotspots are found out from the available statistics like type of vehicles involved,
frequency of accident, type of damage.

5. Analysis : Kernel Density Analysis is done to find the accident probability zone by using the hot spot
information. The probability zones are classified as Low, Medium and High.

6. Medical information like nearby hospitals around the accident site and travel time to reach those,
police jurisdiction information and local governing body information are also mapped into the GIS
system followed by a weighted overlay analysis.

7. Suggestions to reduce accident are then presented as a conclusion of the analysis.

4. Traffic Rule Violation Reduction by Realtime Monitoring: