Review of Past Urban Transportation Studies and ...

Journal of the Eastern Asia Society for Transportation Studies, Vol.13, 2019

733

Review of Past Urban Transportation Studies and Implication for

Improvement on Travel Surveys and Demand Forecasting Methods in

Developing Countries

Sadayuki YAGI a, Deo NOBEL b, Hirohisa KAWAGUCH c

a,b ALMEC Corporation, Tokyo, 160-0022, Japan

c Oriental Consultants Global Company Limited, Tokyo, 163-1409, Japan

a E-mail: [email protected]

b E-mail: [email protected]

c E-mail: [email protected]

Abstract: Japan International Cooperation Agency has provided assistance in urban transport

planning for over 60 metropolitan cities of developing countries by conducting urban

transportation studies that included large-scale household travel surveys (HTS) as a main

means of understanding current situation of travel movements. In the past HTS in developing

countries, it is common to set sampling ratios of 1% to 3% in order to secure statistically

effective samples involving a great number of respondents and spending considerable time

and cost. As for a travel demand forecasting method based on the HTS, a trip-based approach

represented by the four-step method has been widely in use; however, this approach has

several drawbacks. With these issues as a background, this paper aims at extracting issues of

travel surveys and demand forecasting methods and studying directions for improvement with

a view to reorganizing the contents for future cooperation. Based on the result of the reviews

and interviews of the 12 urban transportation studies, this paper identified and organized

common issues and problems in terms of travel surveys with a main focus on HTS, travel

demand forecasting methods, and cooperation needs in the urban transportation sector.

Directions for improvement on travel surveys and demand forecasting methods in developing

countries that should be further pursued in this research are 1) improvement of accuracy in

travel demand forecast, 2) travel surveys and demand forecasting methods tailored for study

objectives and cooperation needs, 3) reduction of required time for travel survey and demand

forecast, and 4) reduction of cost for travel survey and demand forecast.

Keywords: Urban Transportation Studies; Travel Surveys; Travel Demand Forecast;

Developing Countries

1. INTRODUCTION

In metropolitan regions of developing countries, various urban transportation problems are

being tackled such as rapid urban development, extraordinary growth of population and the

number of vehicles, traffic congestion on roads and low service standards of public transport,

and endangered traffic safety. Under these circumstances, Japan International Cooperation

Agency (JICA) has provided assistance in urban transport planning for over 60 metropolitan

cities of developing countries by conducting urban transportation master plan (M/P) studies or

feasibility studies (F/S). However, JICA has been faced with the following three main

issues: travel surveys with a main focus on household travel surveys, travel demand forecast

methods, and cooperation needs in urban transportation sector.


734

Travel Surveys with a Main Focus on Household travel Surveys

JICA has conducted many large-scale household travel surveys (HTS) as a main means of

understanding necessary current situation of travel movements. While HTS have been

conducted in Japan as well, various issues and problems in terms of accuracy have been raised

such as: too coarse traffic analysis zone (TAZ) system; lack of information on the payment of

cost such as drivers’ parking choice, transit users’ route choice, and use of monthly passes; no

attention paid to the seasonal fluctuation of transport demand; complicated survey forms

leading to non-response or invalid sample biases; and difficulty in collecting samples from

single-person households living apart from their families (1).

In the past HTS in developing countries, it is common to set sampling ratios of 1% to 3%

in order to secure statistically effective samples involving a great number of respondents (e.g.,

necessary samples of minimum 50,000 persons or about 10,000 households in the case of a

metropolitan region with a population of 5 million). Thus, it is a large-scale survey which

takes a period of 6 to 8 months and needs a large cost including surveyors’ training and

preliminary survey periods. Furthermore, based on the authors’ experiences, the following

serious problems have been pointed out in the case of developing countries:

• Access to the high-income households who tend to use autos is often difficult and

cooperation from them is least expected;

• Complex survey system and too many survey items make it difficult for surveyors

and respondents to understand the method, leading to incorrect responses or even

misconduct of surveys;

• Surveys on a project basis make it difficult to train many surveyors in a short time;

and

• Sometimes problems occur in the accuracy of residents’ registration system or

electoral roll.

Meanwhile, as for travel surveys with a main focus on HTS, the data need updating after

completion of the surveys; however, in most cases, governments of developing countries

cannot secure the budget and human resources, and more realistic method of data updating is

required. Furthermore, it has widely been acknowledged in the US and Europe that direct

estimation of OD tables by aggregating and expanding the survey data is not realistic; hence,

development and use of disaggregate transportation demand forecast models are given more

priority, so smaller-scale sampling is more common (2, 3).

Travel Demand Forecast Methods

As a travel demand forecasting method based on the HTS, a trip-based approach represented

by the four-step method has been widely in use; however, this approach has several

drawbacks as below (4, 5):

• It does not consider a series of activities made by individuals or intra-household

interactions (e.g., car sharing and picking up/sending off household members);

• Daily traffic demand forecast which does not consider hourly fluctuations is often

applied for a trip-based approach, making it difficult to reflect changes in travel

behavior in the peak hours caused by transportation control measures (TCMs) such

as pricing policies;

In order to overcome these drawbacks, new travel demand forecast methods such as

activity-based travel demand modeling (ABM) have been researched and developed, and

some have already been in practical use in the U.S. or in Europe (6). Some travel behaviors

typical of developing countries, such as walking over a long distance and sending off children

to school before going to work by chauffeur-driven car, could also be reflected, realizing

forecast based on the actual situations. Furthermore, devices which can keep track of the


735

users’ travel behaviors such as cellular phones and smartphones with a built-in GPS feature

have rapidly been diffusing in the world. Information obtained from these devices have also

been utilized as a new transportation data collection method by various researchers and

practitioners. Thus, they have a potential to be utilized by JICA for efficiency of HTS.

Cooperation Needs in Urban Transportation Sector

While JICA has provided assistance in urban transport planning by conducting urban

transportation M/P studies or F/S, there are increasing needs of technical cooperation such as

analysis of policies like TCMs in a specific area. Therefore, travel demand forecast for such

policy analysis needs to be refined as well.

In the context of those issues, this paper aims at extracting issues of travel surveys and

travel demand forecasting methods and studying directions for improvement with a view to

reorganizing the contents for future cooperation in the urban transportation sector in

developing countries. This paper presents the result of the first step of the grand research,

that is, review of the previous urban transportation M/P studies and its implication for further

improvements on the travel surveys and travel demand forecasting methods.

2. REVIEW OF PREVIOUS STUDIES ON URBAN TRANSPORTATION MASTER

PLAN

Methodology

In this research, as listed in Table 1, a total of 12 major urban transportation M/P studies in

developing countries were first reviewed through literature review of the final study reports

and major fact data were developed for comparison. Since those final study reports

presented only the results of the travel survey and travel demand forecast that would lead to

formulation of the master plan, issues and difficulties that were faced in the process of the

studies are hard to be revealed from the literature review. Therefore, except for NIUPLAN

in Nairobi, it was followed by interviews with the ex-JICA experts and/or the counterpart in

charge of travel surveys and travel demand forecasting at the times in order to find out issues

and tendencies in travel surveys with a focus on the HTS and in travel demand forecasting

methods, as well as in cooperation needs of the urban transportation sector.

It is true that selection of these 12 M/P studies were determined based on availability of

the information and persons in charge; however, it can be assumed that these 12 M/P studies

are enough to capture the overall tendencies in the urban transportation M/P studies

conducted by JICA. For each of the above-listed urban transportation M/P studies, the

following major items were reviewed and asked:

• Objective,

• Detail of comprehensive large-scale travel survey (such as HTS),

• Supplemental transport surveys,

• Detail of activity diary survey (ADS) (if any),

• Detail of socioeconomic survey (if not included in the comprehensive travel survey),

• Detail of demand forecast modeling, and

• Development needs that are expected of the study.


736

Table 1. List of urban transportation M/P studies for literature review and interviews

No. Project Name Country Study Period

Major Travel surveys

Remarks

1

Transport Improvement Master Plan Project for Santa Cruz Metropolitan Area (SCZMP)

Bolivia 2016- 2017

HTS

Commuter Survey (using tablets)

Activity Diary Survey

Source: (7)

2

The Project for Capacity Development on Transportation Planning and Database Management in Manila (MUCEP)

The Philippines

2011- 2015

HTS (for updating OD matrices)

HTS conducted by the counterpart

Source: (8)

3 Project on the Revision and Updating of Strategic Transport Plan for Dhaka (RSTP)

Bangladesh 2014- 2016

HTS

Source: (9)

4 Data Collection Survey on Railways in Major Cities (METROS)

Vietnam 2013- 2016

HTS Source: (10)

5 The Roadmap Study for Sustainable Urban Development in Metro Cebu

The Philippines

2013- 2015

HTS

Source: (11)

6

The Project for the Development of Urban Master Plan in Greater Abidjan (SDUGA)

Cote d’Ivoire

2013- 2015

HTS (first time)

Activity Diary Survey

Source: (12)

7 Project for Comprehensive Urban Transport Plan of Greater Yangon (YUTRA)

Myanmar 2012- 2015

HTS Source: (13)

8

Urban Transport System Development Project for Colombo Metropolitan Region and Suburbs (CoMTrans)

Sri Lanka 2012- 2014

HTS (first time) Including F/S

Source: (14)

9

JABODETABEK Urban Transport Policy Integration Project (JUTPI)

Indonesia 2009- 2011

Commuter Survey (with updating OD matrices)

Technical cooperation project

Source: (15)

10

The Study on Integrated Urban Transportation Master Plan for Istanbul Metropolitan Area (IUAP)

Turkey 2007- 2009

HTS (without updating OD matrices)

Survey and demand modeling conducted by the counterpart

Source: (16)

11 Dar es Salaam Transport Policy and System Development Master Plan (DSM)

Tanzania 2007- 2008

HTS Including pre-F/S

Source: (17)

12

The Project on Integrated Urban Development Master Plan for the City of Nairobi (NIUPLAN)

Kenya 2013- 2014

HTS Updating previous M/P

Source: (18)

Comparison of Fact Data of the Previous Studies

Fact data of the 12 urban transportation M/P studies for literature review and interviews are

summarized in Table 2. It is understood that scope of the travel surveys and travel demand

modeling in each M/P study was designed in accordance with its own background. Overall,

in the metropolitan areas where the HTS was conducted for the first time, namely, in Santa

Cruz, Abidjan, Yangon, and Colombo metropolitan areas, comprehensive travel surveys

including a variety of supplemental surveys were conducted for development of the travel


737

demand forecast models. As a result, input of experts dispatched by JICA who were in

charge of travel surveys and travel demand modeling in each of these metropolitan areas tends

to have more man-months (MMs) than other metropolitan areas, leading to relatively higher

price of HTS per sample.

On the other hand, trip rates per person-day including walk trips in these metropolitan

areas tend to be relatively lower than those in other metropolitan areas. Compared to other

metropolitan areas where HTS have been conducted twice or more, development of mobility,

which may also have been in an early stage, may be related to this tendency.

As for the sampling rate in the HTS, formerly it was around 3% of the population in the

past surveys in Hanoi, Ho Chi Minh, Manila, and Jakarta though it is not shown in the table.

Meanwhile, as a recent trend in JICA’s urban transportation M/P studies, sampling rates are

observed as around 1%, whether the HTS was conducted for the first time or not.

Table 2. Fact data of the urban transportation M/P studies for literature review and interviews City Santa Cruz Manila Dhaka Hanoi Ho Chi Minh Cebu Abidjan

Country Bolivia Philippines Bangladesh Philippines Cote d’Ivoire

Abbreviation for Study Name SCZMP MUCEP RSTP Metro Cebu SDUGA

Population (million) 1.80 18.05 9.83 7.60 10.93 2.91 4.90

No. of Households (million) 0.40 5.10 2.40 2.06 2.60 0.65 1.17

Implementation Year 2016 2014 2014 2014 2014 2014 2013

First Time? Yes No No No No No Yes

Cost for Local Consultant *1

(million JPY) 40.81 20.31 27.19 17.76 39.64

(other) 356,920 USD - - - 416,262 USD

Input of JICA Experts (MM) 5 9.0 2 2.5 9.30

Total Cost (million JPY) 53.99 44.03 32.46 24.35 64.16

Collected Samples

(households) 8,500 51,330 15,897 27,151 20,000 6,527 20,000

(persons) 34,000 177,489 66,246 100,168 60,551 29,675 74,309

Sampling Rate (%) 2.00 1.01 0.67 1.00 1.00 1.16 2.00

Price per Sample (JPY)

(per household) 4,407 858 2,042 3,731 3,208

(per person) 1,102 248 490 821 863

Trip Rate per person-day 1.74 1.97 2.26 4.00 3.25 2.97 1.60

Activity Diary Survey Yes No No No No No Yes

Cordon Line Survey Yes Yes Yes Yes Yes Yes Yes

Screenline Survey Yes Yes Yes Yes Yes Yes Yes

Intersection Traffic Volume No No No No No No Yes

Public Transport OD Interview No No Yes No No Yes Yes

Parking Survey Yes No No No No No Yes

Stated Preference Survey Yes No No No No No Yes

Travel Speed Survey Yes No No Yes Yes Yes Yes

Freight OD Survey Yes No No No No No Yes

Road Inventory Survey Yes No Yes No No No Yes

Modeling Platform CUBE STRADA/CUBE CUBE STRADA STRADA STRADA CUBE

Target Time Period one day one day one day one day one day one day one day

Traffic Analysis Zones 433 432 190 444 275 389 173

Household Classes 3 2 3 3 3 4 4

Purposes 5 5 8 5 5 5 4

Modes 4 5 5 5 4 5 4

PT Survey

Other

Surveys

Modeling

METROS

Vietnam

43.00

3

50.91

-

1,080

317

Note: Unit cost of JICA expert (approx. 2.5 million yen per man-month) is evenly applied to all studies. *1 For calculation of the cost of HTS in Santa Cruz, since it also included cost of a commuter survey (CS)

targeting 7,500 households, ratio of unit cost between HTS and CS is assumed to be 2 to 1, which is the same as the ratio of the number of home visits.

Source: (7, 8, 9, 10, 11, 12)


738

Table 2 Fact data of the urban transportation M/P Studies for literature review and interviews

(cont’d) No. 7 8 9 10 11 12

City Yangon Colombo Jakarta Istanbul Dar es Salaam Nairobi

Country Myanmar Sri Lanka Indonesia Turkey Tanzania Kenya

Abbreviation for Study Name YUTRA CoMTrans JUTPI IUAP DSM NIUPLAN

Population (million) 5.69 5.80 28.00 11.60 3.03 3.66

No. of Households (million) 1.24 1.50 7.30 - 0.71 1.13

Implementation Year 2013 2012 2010 2007 2007 2013

First Time? Yes Yes No No No No

Cost for Local Consultant (commuter survey)

(million JPY) 28.3 58.89 57.41 - 41.3 (unknown)

(other) - 574,858 USD 6,212,512,500 IDR - - -

Input of JICA Experts (MM) 6 15 12 - 7.20 (unknown)

Total Cost (million JPY) 44.12 98.43 89.05 - 60.28 (unknown)

Collected Samples

(households) 11,330 35,850 179,000 72,280 7,694 10,000

(persons) 44,980 125,000 657,000 263,768 26,687 -

Sampling Rate (%) 1.00 2.30 2.60 2.20 1.70 1.02


(per household) 3,894 2,746 497 - 7,835 (unknown)

(per person) 981 787 136 - 2,259 (unknown)

Trip Rate per person-day 2.04 1.87 1.04 - 1.51 2.19

Activity Diary Survey No No Yes*2 No No No

Cordon Line Survey Yes Yes No Yes Yes Yes

Screenline Survey Yes Yes Yes Yes Yes Yes

Intersection Traffic Volume Yes Yes No No Yes No

Public Transport OD Interview Yes Yes No No Yes Yes

Parking Survey Yes No No No No No

Stated Preference Survey No Yes No No Yes Yes

Travel Speed Survey Yes Yes Yes Yes Yes No

Freight OD Survey Yes Yes No No Yes No

Road Inventory Survey No Yes No No No No

Modeling Platform CUBE CUBE CUBE TransCAD STRADA STRADA

Target Time Period one day one day one day one day one day one day

Traffic Analysis Zones 187 475 343 460 170 106

Household Classes 2 3 3 1 2 3

Purposes 5 4 5 4 5 4

Modes 4 5 4 4 3 5

PT Survey

Other

Surveys

Modeling

Note: Unit cost of JICA expert (approx. 2.5 million yen per man-month) is evenly applied to all studies. *2 ADS in Jakarta metropolitan area applied utilization of a mobile device (i.e., GPS logger) and was

renamed as Person Tracking Survey. Source: (13, 14, 15, 16, 17, 18)

Moreover, ADS was focused on among the 12 urban transportation M/P studies, and three

M/P studies where ADS had been conducted were highlighted with a comparison of the scope

and the cost presented in Table 3. It should be noted that ADS in Jakarta metropolitan area

applied utilization of a mobile device (i.e., GPS logger) and was renamed as Person Tracking

Survey. ADS in these three M/P studies mainly aimed at confirming or adjusting the trip

rates obtained from the HTS, and it is quite natural that the trip rate analyzed from the ADS

was higher than that from the HTS.

ADS usually has a smaller sample size and is also smaller in scale; whereas, the price of

ADS per sample is estimated at around 1,000 yen (or approximately 9 US dollars) per

person-day. Compared to the HTS, the unit price of ADS is several-fold higher (in Jakarta)


739

or around the same range (in Santa Cruz and Abidjan). ADS can capture a series of trips

over several days ideally without omission as an advantage. However, the same level of

detailed trip attributes need to be collected as in the HTS for developing travel demand

forecast models. If collected ADS data do not contain as detailed trip attributes as in Jakarta,

it will be difficult to develop travel demand forecast models based on the ADS data.

Table 3. Fact data of ADS conducted in the urban transportation M/P studies No. 1 6 9

City Santa Cruz Abidjan Jakarta

Country Bolivia Cote d’Ivoire Indonesia

Abbreviation for Study Name SCZMP SDUGA JUTPI

Population (million) 1.80 4.90 28.00

No. of Households (million) 0.40 1.17 7.30

Implementation Year 2016-17 2013 2010

First Time? Yes Yes No

Use of Any Mobile Device None None GPS Logger

Cost for Local Consultant

(million JPY) 1.60 6.20 8.24

(other) 14,400 USD 69,700 USD 891,650,000 IDR

Collected Samples

(households) 900 1,010 600

(persons) 1,800 3,088 2,461


(per household) 1,777 6,142 13,734

(per person) 889 2,009 3,348

(per person-day) 889 1,004 1,116

Survey Targets

Intervals (minutes) 15 15 -

Duration (days) 1 2 3

Survey Zones 433 101 95

Trip Rate per person-day 2.13 3.05 2.37

Household Attributes Yes

Individual Attributes Yes

Origin/Destination Yes Yes Yes

O/D Facility Type No No 20 types

Departure/Arrival Time every 15 min. every 15 min. Yes

Purpose 12 purposes 11 purposes 7 purposes

Cost No No Yes

Transfer Points None None 6 unlinked trips

Travel Modes 20 modes 19 modes 27 modes

Driver/Passenger No Yes Yes

Number of Occupants No No Yes

Access/Egress Cost and Time No No Yes

Transit Wait Time No No Yes

Parking Place No No No

Parking Cost No No Yes

Act

ivit

y D

iary

Su

rvey

Tri

p A

ttri

bu

tes

Gen

era

l In

fo.

Linked with PT

Survey

Linked with

Commuter Survey

Note: ADS in Jakarta metropolitan area applied utilization of a mobile device (i.e., GPS logger) and was

renamed as Person Tracking Survey. Source: (12, 15)

Review and Interview Results

Subsequent tables summarize review and interview result of each urban transportation M/P

study in terms of: (a) travel surveys with a main focus on HTS, (b) travel demand forecast

methods, and (c) cooperation needs in the urban transportation sector. Based on the result of

the reviews and interviews for each of the 12 urban transportation M/P studies, common

issues and problems are identified and organized in the following section.


740

Table 4. Review of previous M/P studies: SCZMP, MUCEP, RSTP Summary

1.

(a)

(b)

(c)

SCZMP (7)

The simplified census was conducted in urban areas by the institution of statistics of the Santa Cruz Departmental Government. This data was then utilized for sampling in rural areas. HTS of 1.7% sampling rate or equivalent to 7,500 households was the first time ever conducted in Santa Cruz, Bolivia. Tablet with specific software/application was used in exchange for traditional paper-based survey form carried out by surveyors. In reality, application or the software did not run smoothly and often encountered error messages. Those facts discouraged and demotivated some of the surveyors.

ADS was conducted for 1,800 persons and 900 households at intervals of 15 minutes for one day. In Latin American countries, culture of having 2-3 hours lunch break is quite common and many workers take this time to go back home. This fact made the result of ADS was quite unique to be modeled.

Demand forecast model was developed based on four-step method considering: household income, four trip purposes, and four travel modes by household income. This model might be useful for estimating yearly economic benefit of projects and can be utilized for series of transport projects including transit system and road network development.

So far, Bolivian central government is planning to construct urban railway system with funding from private sector. However, none of these plans are actually making significant progress just yet.

2.

(a)

(b)

(c)

MUCEP (8)

Household Interview Survey (HIS) was not the first time ever conducted in Metro Manila. MUCEP was a study to update existing OD as well as the database. Sampling rate was 1.01% or equivalent to 51,330 households or 177,489 persons. Trip rate from the survey is 2.26 trips per person-day. Available demographic data was population and household census 2010 only, therefore, population and number of households as of 2014 were estimated by framework of the study area and HIS was performed to build present OD matrices.

Demand forecast model was developed based on four-step method targeted 365 internal and 67 external TAZs considering: two household classes by the number of vehicle ownership, five trip purposes, and five travel modes. This study was based on M/P; therefore, daily demand forecast was required to calculate yearly economic/finance analysis. Socioeconomic frame was a trend frame only in this study. Therefore, fratar method was basically adopted as trip distribution model in this study.

Governmental agencies as counterpart of this project are more interested in development of road sector: road traffic management, ring roads, highways, and also transit-oriented development (TOD) system in urban areas development. Also, the government is interested in tools for self-evaluation and self-development. They see themselves in the future as independent for updating database and conducting large-scale survey. In this sense, they need more capacity-building-related programs so that they can practice, do small pilot projects, and get lectern session.

3.

(a)

(b)

(c)

RSTP (9)

HIS was conducted in this study to update OD from the previous study. Sampling rate for this survey was 0.67% or equivalent to 15,897 households or 66,246 persons. There is a significant different of trip rate due to the gender influenced by religion in Dhaka. Male’s trip rate was 2.26 trips/person-day and female’s trip rate was 1.18 trips/person-day.

During the study, the University of Tokyo developed OD update with a different kind of method, that is, CDR (Call Detail Record) method. Collected data was provided by one of the cell phone carriers in Dhaka, which took 40% of the total market share. Such sizeable data would have been more beneficial if demographic data attached to each mobile phone user was available. For the privacy purpose, demographic data was kept hidden; thus, it was hard to complement the HIS.

Demand forecast model was developed based on four-step method targeting 141 internal TAZs and 49 external TAZs and considering: three household income classes, eight trip purposes, and five travel modes. Model from this study was then utilized by follow-up pre-F/S studies for MRT Lines 1 and 5. Additional traffic count survey data was utilized to improve the model during the pre-F/S and the model platform was changed.

Development needs for Dhaka are mainly for all sectors of inland transportation as the city/country is growing: road development, flyover/underpass, BRT, TOD, TCM, road/area pricing, traffic management, and capacity development for official personnel.


741

Table 5. Review of previous M/P studies: METROS, Metro Cebu, SDUGA, YUTRA Summary 4.

(a)

(b)

(c)

METROS (10)

Hanoi and Ho Chi Minh are the two major cities in Vietnam which were surveyed in this study. The only available census data was from 2009. Population and number of household in 2014 were estimated based on the 2009 census. Local consultant was faced with difficulties during the HIS due to lack of experience. Also, the survey form was designed inefficiently and made impartial result as questions were just partially answered. Thus, the quality was only moderate.

Demand forecast was not the first time ever developed for both cities. However, lack of previous models database management made it impossible to update the existing models, so they had to start it over from scratch. Models were developed based on four-step method and targeted for: 436 internal TAZs and 8 external TAZs in Hanoi; and 265 internal TAZs and 10 external TAZs in Ho Chi Minh. Models are developed based on: three household income classes, five trip purposes, and five travel modes for Hanoi and four travel modes for Ho Chi Minh.

Direction of recent development needs in both cities is toward the development of urban railways conjugated with the development of TOD. Capacity building and technology transfer, however, are necessary considering limitation from the Vietnamese side.

5.

(a)

(b)

(c)

Metro Cebu (11)

HIS was not the first time ever conducted in Metro Cebu. Sampling rate of 1.16% or equivalent to 6,527 households or 29,675 individuals was surveyed in this HIS. Not only for the OD update purpose, this survey also collected citizens’ opinion regarding the development direction in Metro Cebu. From the latest census data of 2010, 2014 data was estimated.

Demand forecast model was not the first time ever conducted in Metro Cebu. However, previous model was out of date, thus, no longer accurate. Such situation left no other option than to build it from scratch. The time consuming process within limited period covered the depth of analysis of, among others: shift of private vehicle users to railway, increase of unemployment rate, and two-wheeler users. One-month extension was given at the consultant’s cost to finalize the model.

Future development needs for Metro Cebu are towards the development of road and public transport (bus- and rail-based). Intelligent Transport System (ITS) and comprehensive traffic management system are also needed to complement such development. On top of those, capacity building is the most necessary to enable the government to keep up with the growth of the city.

6.

(a)

(b)

(c)

SDUGA (12)

For the first time in Greater Abidjan, HIS was conducted simultaneously with the planned census survey in 2013. For demographic and population data, this study initially intended to utilize the 2013 census data. However, the census survey was never realized due to budget problem and debatable survey method utilizing mobile device. Eventually, 1998 census data (before the civil war) was utilized but with comprehensive adjustment. ADS was also conducted during the study period for the purpose of capturing all activities and trips on two consecutive weekdays to include underreported trips adjust the trip rates estimated from the HIS.

Demand forecast model was the first time ever conducted in the study area. Lack of socioeconomic data made it difficult to forecast the socioeconomic values for transport and spatial sectors. Even findings from Cairo were utilized to measure the truck trips in study area since data was so scarce.

The Government made a plan, namely: “Grand Projects of Emergence”. This project listed up, among others: four road projects and six public transport projects to be implemented through PPP scheme. Thus, private sector participation is the key of urban transport project implementation.

7.

(a)

(b)

(c)

YUTRA (13)

HTS was the first time ever conducted in Yangon and in parallel with HIS data collection process. HTS also triggered positive curiosity among the first timers: respondents and surveyors. Not like many other cases, the survey received very low number of rejections even from what usually is caused by the privacy-related issue. The uniformly “low” household income level made the study team consider 1% sample rate as an adequate number to represent the whole population of Greater Yangon. However, in line with JICA’s consideration of numerous potential investors in the city center, the sample rate in the city center was increased to 2%.

Demand forecast model was developed by updating the existing model, especially to reflect the motorcycle banning policy in the model. Motorcycle ban made huge difference of transport pattern between the city and the rural areas. In exchange for the low share of motorcycle in the city, taxi was highly utilized. This situation made the model quite unique. The study team wanted to develop a disaggregate model for modal split. It was impossible because: there were insufficient samples to further calibrate the model and no reasonable variation of value of time since stated preference (SP) survey was not conducted. Also, the fact that public transportation service was still poor in quality made people stick with their private car.

Direction of recent development needs is toward the development of urban railways conjugated with the development of TOD.


742

Table 6. Review of previous M/P studies: CoMTrans, JUTPI, IUAP Summary 8.

(a)

(b)

(c)

CoMTrans (14)

HIS that was conducted for the first time brought significant challenges to the study team. Many high-income households and Moslem community were reluctant to participate for some reasons. The survey company also failed to bring up the best quality of work since they cut off significant amount of budget for surveyor to gain more profit. The half-hearted surveyors were found unable/did not even try to convey the essence of the survey. Cases of surveyors’ cheating – filling up the survey form by themselves – were also found.

Above-mentioned conditions impaired the demand forecast model input and furthermore the demand forecast model quality. It was found that there was huge discrepancy of traffic volume resulted from the network assignment and the screenline survey. Many adjustments had to be done to get the modeling work done, such as: reconsideration of household’s expansion factor, reconsideration of do-nothing scenarios and impedances, and more. After this study, the study of “Western Region Megapolis Plan 2030” was conducted in 2016 and it utilized the demand forecast model from CoMTrans. Major modification for the population framework was done following the new urban development concepts.

Development direction is mainly towards all sectors of transportation, namely: rail sector, bus sector, road sector, TOD, TCM, and pricing system. The capacity development in terms of engineering and good governance is also necessary. As numerous projects in transport sector are under way, Sri Lankan government thinks it essential to maintain transport data and model under one agency/institution although they are still clueless on how to do that.

9.

(a)

(b)

(c)

JUTPI (15)

Significant number of high-income household was reluctant to be surveyed for the privacy reason during the commuter survey. Some of them asked for official governmental letters before the survey started. Such situation implicated the gap of vehicle ownership between the survey result and the number from vehicle registration data. Expansion factors needed to be adjusted following such constraints.

Commuter travel survey was also conducted within the scope of the study. Trip was recorded with paper-based interview, GPS data, and telephone interview. Both GPS data and the telephone interview were essentially calibrating the paper-based interview.

Demand forecast model was not the first time ever developed for the study area. It involved updating the existing model. The variable of travel time surprisingly became insignificant in the modal split model mainly due to high fluctuation in travel time. It indicated unpredictable travel time regardless of the transport modes. There was also big difference between the traffic volume from the screenline survey and from the network assignment. After further analysis, it was found that the cause was because there were many on-site traffic management system applied during the screenline survey, among others: two-way became one-way street, U-turn prohibition, and detouring route.

Development direction for the study area is related to the development of rail-based transport and the policies of TOD, parking pricing, road pricing, and TCM. In 2016, the first MRT (mass rapid transit) line was constructed and the 3-in-1 policy was omitted in exchange for odd and even number plate regulation.

10.

(a)

(b)

(c)

IUAP (16)

HTS was conducted by the Turkish counterpart a year prior to this study. However, result of the HTS was not utilized for expansion or adjustment. Even data cleaning was not conducted. This is because no aggregation models were applied by the Turkish counterpart. Existing OD matrices were then developed in IUAP for 4 trip purposes by 4 modes (walk, car, service, and transit) for several target years. However, since the study’s target years were different from Turkish counterpart’s projection, interpolation program was made to create OD matrices that fit IUAP’s target year. The present OD matrices for 2006, which were provided to the study team, were the output of the disaggregated models; thus, OD matrices are subject to change once the models have been modified.

During the process of trip production/attraction model, the court sentenced that the base land use was invalid. However, the study team had to use the socioeconomic framework because that was the only source available. Frequent changes in trip distribution due to the inconsistency between observed and synthesized screenline traffic volumes caused a delay in finalization of the distribution model. There was not enough time to finalize the modal split but to continue the process to the next step, that is, network assignment.

Development direction in Istanbul was more related to the road, bus, and rail sectors. For policy development, Istanbul emphasized more on the development of pricing system and TCM. Since there are many areas in Istanbul that are categorized as the world heritage by UNESCO, applying changes may be challenging.


743

Table 7. Review of previous M/P studies: DSM, NIUPLAN Summary 11.

(a)

(b)

(c)

DSM (17)

The wealthy household made HIS difficult to conduct because some households refused to be surveyed. Accuracy of the data, especially the high-income-related data such as car ownership and income was affected. Not only that the surveyors received obstacles collecting data in high-income household area, but also security was also problem in low-income area, especially for female surveyors.

The objective of the traffic demand for the M/P was planning and evaluation of integrated transport network and service of the whole study area. For the traffic control and management of CBD area, microscopic simulation model and vehicular OD in morning/evening peak which was built based on HIS and traffic count at the boundary of CBD was used.

Trip attraction model, especially home-based work trip, was difficult to build. Therefore, dummy variable was used. Coefficient of the trip distribution model was not high enough. Therefore, two household types (car-owning and no-car-owning household) and five trip purposes were made. As of 2007, number of motorcycle and bajaj (three-wheeler taxi) in the study area was not so many. Therefore, motorcycle and bajaj were not included in the modal split model. However, motorcycle and bajaj are becoming popular these days. For car-owning household, two models (NMT or motorized and car or bus) by five trip purposes were estimated. For no-car-owning household, one model (NMT or bus) by five trip purposes were estimated.

Since population growth, motorization, and urban development have proceeded much more rapidly than the initial projections by DSM, Dar es Salaam still needs major development on their bus-based transport development and infrastructure that supports the plan such as road widening, road improvement (flyover or underpass), and capacity development. In terms of capacity development, Dar es Salaam needs establishment of agency/authority that handles database update and planning.

12.

(a)

(b)

(c)

NIUPLAN (18)

HTS was conducted not for the first time but involved no update of the existing OD data. Sampling was based on the 2009 population and housing census and was carried out for 1.02% or equivalent to 10,000 households. In addition to the screenline survey, roadside traffic count survey for 30 stations was conducted for the purpose of calibration of current traffic movement estimated by demand forecast model.

Modal split model consists of three steps: walk split model, private-public split model, and bus-rail split model. Walk split model incudes inter-zonal walk split models by four trip purpose and intra-zonal walk split models by four trip purpose and car ownership. Private-public split models for inter-zonal trips are built by 4 trip purpose, and for intra-zonal trips are built by four trip purpose and car ownership. Bus-rail split model was decided by access and egress distance as well as distance between origin and destination. For network assignment model, in general, traffic assignment of statistic model often computed very high volume capacity ratio of road links. In this study, very highly congested road links were estimated.

Road projects in Nairobi are mainly funded by the World Bank, EU, AfDB, and China. Other than the road-based development, bus- and rail- based transportation developments are also necessary as well as the capacity building development in planning, engineering, and governance sectors.

3. ISSUES ON HOUSEHOLD TRAVEL SURVEYS AND TRAVEL DEMAND

FORECAST

Based on the result of the reviews and interviews for each of the 12 urban transportation M/P

studies in the previous section, common issues and problems were identified and organized in

terms of travel surveys with a main focus on HTS, travel demand forecast methods, and

cooperation needs in the urban transportation sector. Implications for improvement are also

added in the discussion.

Travel Surveys with a Main Focus on HTS

Problems of HTS Quality

• In recent years, sampling rate of a HTS, whether it is conducted for the first time or

not, has been fixed to a low level of around 1%. Such a low sampling rate impairs

accuracy, making it difficult to analyze comprehensive travel patterns as well as to

compare them with the previous survey result, if any.


744

• Meanwhile, in some HTS where sampling rates of 2% to 3% were applied, local

consultants were required to complete a HTS of an unprecedentedly large scale in

such a short time of 2 to 4 months. As a result, problems occurred in many aspects

such as employing incapable surveyors, delay in payment to the surveyors, and

misconduct of surveys in the field. These may cause further delay in survey

implementation, and in reality it sometimes takes nearly one year to be ready for

analysis of the survey result.

• Random sampling based on population census data or residents’ registrations usually

takes time and increases burden to the surveyors, leading to larger cost. There may

be also a problem of outdated data. Meanwhile, sampling by the surveyors in the

field tend to target easily accessible households, thus causing a bias. Solutions to this

may be sampling using satellite images, strict application of the rules in the field like

counting a certain number of households to select the target household, etc.; thus,

sampling method needs to be well prepared taking the balance between survey

efficiency and randomness into consideration.

• HTS conducted only by local consultants with little involvement of JICA experts in

charge often have problems in the survey data quality such as no data cleaning

conducted. Sometimes the problems become fatal with unusable data. Problems related to HTS quality may be in large due to technical issues or problems. For

developing countries particularly where available transportation data are scarce, HTS data may be appreciated and thus long utilized as precious data; hence, it is important to secure the quality of HTS considering existing other transportation surveys as well as the country’s mobility development stage. Meanwhile, it may also be necessary to confirm that increased sampling rate will lead to assured quality of HTS.

Issues on the Survey Purpose and Contents

• While it usually takes nearly one year to implement, complete, and analyze all the

necessary transportation surveys including HTS, the counterpart agencies can hardly

wait for the results, especially when it becomes necessary to move forward part of

the M/P study schedule for political reasons or for quicker implementation of

subsequent F/S projects. Some government officials who do not well understand the

whole process of urban transportation M/P study may even develop a sense of

distrust.

• In cities where mass transit system is under-developed and citizens cannot even

picture a new mode of public transport, it is difficult to implement the stated

preference (SP) survey and to develop a mode choice out of the SP survey. In that

sense, attention has to be paid before implementing SP surveys.

• For analysis of panel data or for time-series comparison of HTS, it may be better not

to drastically change the survey form design. However, if ADS is to be conducted in

place of the HTS, the survey form design will be drastically changed to a more

complex one, causing a burden to the respondents as well. As such, the issue is

which to select, existing HTS plus supplemental ADS or comprehensive ADS from

the beginning.

• In the case of Japan, HTS have long been implemented because of the data reliability

and relative stability of the situation surrounding the transportation. On the other

hand, in developing countries where the surrounding situation is drastically

changing, necessary type of travel surveys as well as travel demand forecast methods

should be discussed in light of the difference of the situation.

Needless to say, it is important to obtain an understanding or an agreement about the


745

process and approach of the M/P study from the counterpart agencies beforehand. Some of

the interviews with the counterpart as to the time-consuming large-scale HTS followed by

model development showed various responses such as: it is generally understandable

(Indonesia); it is impossible to wait for two years or longer (Vietnam); and it is

understandable if important and useful outputs are to be provided (Tanzania).

Furthermore, based on the cooperation needs in urban transportation sector, it is important

to understand which kind of forecast the counterpart truly needs, transportation demand on the

planned infrastructure(s) (i.e., trips) or reaction in the travel behavior to some transportation

policies such as TCM (i.e., activities); then, most appropriate travel survey contents should be

proposed accordingly. It should also be noted that a travel diary survey using mobile

devices such as smartphones that can automatically detect trips is one way to reduce

respondents’ burden of filling complicated survey forms.

Issues on Socioeconomic Data Collection

• Socioeconomic zonal aggregation data should be collected from the latest population

census. However, in reality, that implementation of the population census is often

delayed and the data is not ready for the study. In such a case, the only option is to

utilize and modify old population census with some adjustments often by the experts.

If the latest population census result becomes available, redoing the analysis will

become another burden. Delivery of the latest population census result on time is

most important.

• For HTS, access to the high-income households who tend to use autos is often

difficult and cooperation from them is least expected. As a result, weight factors

often need to be adjusted to avoid the bias. While the survey method should be

improved to have more responses from the high-income households, it is also

necessary to obtain accurate socioeconomic data such as vehicle registration data and

income statistics.

• As for large-scale development, if official development plans of satellite cities are

already given, future socioeconomic values should be modified based on the degree

of maturity after reviewing the plans. However, even if development plans have been

politically proposed and the realization seems questionable, it is usually difficult to

turn down the plans in the formulation of the master plan.

Comprehensive socioeconomic data with accurate detail obtained from regularly

conducted population census and other surveys are as important as HTS data. Therefore, it

is vital to confirm with relevant agencies about the contents, accuracy, and schedule of

delivery of the census and survey data to make sure to smoothly analyze the survey data. As

for future socioeconomic settings in the M/P with many urban developments, it should be

noted that they are usually based on the ideal situation in the target year.

Issues on New Methods of Travel surveys

• There has already been new effort of utilizing electronic tablet devices for large-scale

surveys as in Santa Cruz. While this new survey method has merits such as

simultaneous data input with interviewing, prevention of inputting errors and

cheating, and inputting precise locations, some issues to be solved have been pointed

out such as fixing unexpected system errors and bugs in the application.

• According to the comparison of total trips and trip patterns analyzed from the CDR

(Call Detail Record) and from the HTS in Dhaka, while no major difference in the

total trips was observed in the whole metropolitan area, comparison of trips of

certain TAZ OD pairs showed a low level of correlation. However, considering that


746

collection of CDR is possible at a low cost throughout the metropolitan area

including districts that are not accessible to the public, CDR may be worth analyzing

and hence latest information on other cases of analyzing CDR needs to be gathered

continuously.

It seems necessary to keep gathering information on practical applications of new method

of travel surveys such as utilization of mobile devices for travel survey and analysis of CDR.

In Dar es Salaam, Tanzania, a travel diary survey was conducted by the World Bank as a pilot

survey utilizing smartphones carried by the respondents. This pilot survey has revealed

various problems in the field as well as issues for the next trial. It is expected to proceed

with improvement on such a new method of travel surveys by overcoming the problems and

issues. In addition, analysis of CDR was also conducted as a research funded by the World

Bank, and OD matrices by trip purpose were developed. Thus, it is necessary to explore the

possibility of analyzing CDR at least to update the existing OD matrices.

Travel Demand Forecast Methods

Issues on Improvement of Demand Forecast Models

• In most cases, R2, coefficient of determination, of trip distribution models where

gravity models are applied is not so close to 1, implying an issue in the model

validation. Thus, trip distribution models are considered as the hardest in the

four-step method, and disaggregated approach including ABM shall be discussed as

an alternative.

• Large-scale travel surveys such as HTS are conducted in urban transportation M/P

studies for the purpose of developing as accurate OD matrices and demand forecast

models as possible; whereas, as for freight vehicle trip distribution, though existing

OD matrices could be developed from a freight OD survey, it is difficult to develop

freight demand forecast models and simple future estimation is often made based on

future socioeconomic indicators such as zonal GDP and population of workers.

Likewise, future forecast of external OD matrices that are estimated from cordon line

surveys often follows a simple methodology.

• As for the choice of daily trips or peak-hour trips for travel demand forecast, daily

trips are usually modeled in M/P studies unless there is a request of analyzing

peak-hour traffic from the counterpart agencies, because calculation of benefit of the

proposed projects or programs is necessary on a daily basis for conversion to annual

benefit in the benefit-cost analysis. If peak-hour traffic needs to be forecasted in an

F/S, directional peak-hour ratios are calculated from the survey data and utilized for

estimation of the peak-hour traffic. It is true that there is an issue in terms of

forecasting accuracy, but it usually takes enormous amount of time and cost to

redevelop the demand forecast models.

• Total man-months of the experts required for the task of travel demand forecasting

tend to be reduced; whereas, the same or more detailed output of forecast is required.

In short, there is not enough time for elaborating the models in reality.

About issues on freight trip distribution models and demand forecast of external ODs, it

may be worth analyzing how much these problems contribute to the gap between the

forecasted values and actually observed data. Meanwhile, practical application of ABM as a

disaggregated approach is also worth trying, since final output from ABM is usually OD trips

for each time of day. However, according to interviews with experts of ABM, reduction of

necessary input of man-months as well as time may not be expected. Application of ABM

will just contribute to reduction of cost and time that are necessary to collect a smaller number


747

of samples for the ADS.

Issues on Socioeconomic Indicators and External Factors

• For setting future zonal socioeconomic indicators, there should be an expert in

charge of socioeconomic framework or land use planning, who is capable of

considering the existing upper-level national or regional development plans along

with planned macro-values of socioeconomic indicators in the metropolitan area and

distributing the total values of the metropolitan area to each administrative district

and then to each zone. However, it is often the case that the M/P study lacks such an

expert, and some other experts in charge of travel surveys or demand forecasting are

taking this important and difficult task.

• Generally, in developing countries, there is a rapid change in urban structure,

economy, and travel behavior in metropolitan areas, for example, rapid motorization

or sudden economic recession. Developing countries are subject to these external

factors, which are difficult to assume. Since travel demand forecast is a kind of

future mathematical simulation based on the assumed conditions, forecasted figures

should be treated as values with a range. As such, there is a trade-off between the

degree of accuracy and cost and time.

Out of zonal socioeconomic settings, unexpected external factors, travel surveys, and

demand forecast models, in order to find out which of these caused a gap between the initial

forecasts and the actually observed values, it is necessary to investigate the cases of the

previous M/P studies. If demand forecasts are just figures with a range, there may be some

comments that disaggregate models such as ABM based on ADS would be better rather than

to conduct a large-scale HTS. Hence, practical application of demand forecasting methods

based on disaggregate models using a smaller number of samples may be worth studying.

Cooperation Needs in Urban Transportation Sector

Issues on the Development Directions

• Overall, developing countries’ interests in urban transportation sector are toward

public transport development rather than road development. Road network

development is implemented to some extent, they are faced with a necessity for road

maintenance and management. As for development of relatively inexpensive public

transport development such as bus rapid transit (BRT), it is often combined with the

road development as the road sector is a key infrastructure. Metropolitan areas of

which infrastructure development relies on public-private partnership (PPP) are

increasing in number as well.

• As the concept of transit-oriented development (TOD) is becoming widely prevalent,

its inclusion in the transportation demand forecast is requested in some cases. The

issue to be considered is in what detail TOD should be reflected in the forecast.

While applying additional population to the zones around the target public transport

may involve limited increase of tasks, evaluating the effect of shorter walking

distance may involve complicated and time-consuming work. It should also be noted

that, in developing countries, TOD is a commercial strategy that attracts not only

investors but also politicians.

Interviews were made with the counterpart agencies as to: who have been utilizing the

M/P as well as the HTS data; and whether the M/P fits stakeholders’ needs. Their answers

were: adhering to the M/P though the progress is slow (Vietnam); adopting only development

directions and policies (Tanzania); incorporating some of the proposed projects from the M/P


748

without any reference (Tanzania); and basically following the M/P along with some top-down

“ad hoc” projects implemented in parallel (Indonesia). Thus, they have various ways of

utilizing the M/P.

As for TOD, it is foreseen that TOD will draw attention sooner or later while it depends

on the stage of maturity of each metropolitan area. For evaluation of TOD, disaggregate

models including ABM where accessibility to the station can be directly reflected may be

suitable demand forecasting methods. Therefore, even at this present stage, it may be

desirable to study the possibility to apply these disaggregate models.

Issues on TCMs and Short-term Measures and Policies

• In some cases there is a political request to share the results in a short time and/or to

propose more short-term projects. While quite a few TCMs may comprise the

short-term projects, required accuracy of the travel surveys and demand forecast for

these short-term projects will be different from what is required for large-scale

infrastructure development projects. Therefore, it is necessary to obtain consent from

the counterpart agencies about the policies to be prioritized at the initial stage of the

M/P study.

• TCM policies will become alternatives to be discussed in the M/P studies only in

cities where the road and public transport developments have been completed to

some extent. Furthermore, effect and necessity of traffic demand management

policies including intelligent transport system (ITS) will highly depend on the local

custom and situation of each city such as driving manners; rather, education and

enlightment on transportation may be a key “input” into the demand forecast models.

Introduction of TCM policies that particularly involve new technology system will

need capacity development and technology transfer first of all.

While there are some comments that M/P should play a part just as a tool by means of

which decision makers can explain to the public, interviews were made with the counterpart

as to whether they understand large-scale HTS and subsequent demand forecasting model

development that will require quite long time to reach the output. Their answers were quite

varied such as: they generally agree (Indonesia); they cannot wait for the result for over a year

(Vietnam); and they can agree as long as the result is really important and useful (Tanzania).

As for evaluation of TCM policies, since it is essentially difficult to forecast the change in

travel behavior through the conventional four-step method, disaggregate models including

ABM should be used for this evaluation but should be applied only to cities where the

infrastructure developments have been completed to some extent and a need for TCMs has

been arising.

4. CONCLUSION: IMPLICATIONS FOR IMPROVEMENT

With the three main broad issues as a background, namely, travel surveys with a main focus

on household travel surveys, travel demand forecast methods, and cooperation needs in urban

transportation sector, this paper identified and organized common issues and problems in the

urban transportation studies in developing countries by conducting reviews and interviews of

the past 12 urban transportation M/P studies. Through the discussions in the previous

section, directions and implications for improvement or “new” issues to be tackled further in

this research can be reorganized and summarized in the following four main directions: (1)

improvement of accuracy in travel demand forecast, (2) travel surveys and travel demand

forecasting methods tailored for study objectives and cooperation needs, (3) reduction of


749

required time for travel survey and travel demand forecast and (4) reduction of cost for travel

survey and travel demand forecast.

Improvement of Accuracy in Travel Demand Forecast

Four major causes of gaps between previous projections and actual observations: namely, 1)

socioeconomic framework, 2) travel survey method, 3) travel demand forecast model and 4)

other external factors, are described below. In any case, continuous update of database and

models are essential in the long run.

Socioeconomic Framework

Gaps of socioeconomic framework between previous projections and actual observations are

observed in some M/P studies that were reviewed in this paper. The impact of these

differences and a travel demand model on accuracy can be analyzed by inputting actually

observed socioeconomic and network variables to the previous demand forecast models with

parameters unchanged. If the model outputs are similar to the actual observations, it can be

concluded that the model is valid and socioeconomic framework caused the gap.

Although it is important to forecast population distribution considering concentration to

city center and migration, future socioeconomic framework is often determined by agreement

with counterpart agencies. It is not easy to maintain both accuracy and accordance with policy

of the counterpart agencies. Therefore, further detail analysis on the socioeconomic

framework will not be conducted.

Travel Survey Method

With regard to travel survey method, the following method will be analyzed utilizing ICT and

state-of-the-art survey method.

• Utilization of tablet computers can improve accuracy of HTS. By examining an

example of HTS in Santa Cruz, Bolivia and a number of application of the “Survey

Solutions” by the World Bank (19) for interview survey, improvement in accuracy

and efficiency can be analyzed. Inputs from the ongoing example of application of

tablet computers for a commuter survey and a paper-based ADS as a part of “Project

for Urban Transport Master Plan in Kinshasa City” will also be taken into account.

• For the area with previous HTS data, a smaller-scale survey such as ADS and the

survey with mobile device and CDR can be utilized for analysis. As there is no

example of application in developing countries, feedback from ongoing project, the

“JABODETABEK Urban Transport Policy Integration Project Phase 2” (JUTPI2)

can be included in the analysis.

• Impact of reduction of sampling rate will be examined by comparing coefficients of

determination (R2), t-values and parameters of models estimated with full sample

(e.g., 2% sampling rate) and with partial sample (e.g., 1% sampling rate). Factors

affecting precision of the model and the countermeasures will be analyzed.

Travel Demand Forecast Model

According to the analyses on previous studies, errors were observed in trip distribution

models and modal split models of some projects. Further detail analysis will be conducted

to identify the cause of errors by inputting current socioeconomic and network data to the

previously estimated model with the same parameters as shown in 1) Socioeconomic

Framework. If the cause is the travel demand model, fundamental causes should be

examined and the countermeasures will be proposed.


750

For modal split models, it is understood that disaggregate model is widely applied due to

accuracy, flexibility in analysis and smaller sample size. Disaggregate approach can also be

applied for trip distribution models. Furthermore, joint mode and destination choice models

that could improve accuracy will also be included in analysis. In addition, disaggregation of

first to third steps of the four-step method: namely, trip generation/attraction, trip distribution

and modal split, that is, development of ABM will also be discussed.

Other External Factors

As mentioned in this paper, other external factors will be listed and analyzed based on the

previous M/P studies as a reference for future studies.

Travel Survey and Travel Demand Forecasting Methods Tailored for Study Objectives

and Cooperation Needs

As study objectives and cooperation needs are different by type of projects such as M/P, F/S

and others (e.g., TCM policies), directions are discussed by type of projects.

Urban Transportation Master Plan (M/P)

Accuracy is considered as prerequisite for urban transportation M/P study for proposal and

evaluation of short, intermediate and long term projects. It should be noted that future

demand sometimes depends on grand vision of urban development according to the review of

previous M/P studies.

Review of the previous M/P studies also showed that quality of travel surveys and travel

demand forecast has never been compromised. In case expeditious work is expected

especially for a specific project, F/S can be commenced earlier while M/P can be worked on

in parallel. It also should be noted that accurate transportation database of M/P are utilized

by other F/S including the ones conducted by other development partners.

In terms of sample size, 500 – 1,000 samples are considered as large enough for

disaggregate model (20). Assuming that high-income or auto-owing household is 10%,

5,000 – 10,000 households are expected for model development of households’ choice model.

On the other hand, if the sample size is too large and survey period is too short, management

of field survey can be significantly complex. Empirically, 10,000 household samples can be

conducted by 40-50 surveyors, which can be managed effectively, can be completed in 4

months. Considering these aspects as well as examples in other cities, 10,000 households

are expected for disaggregate modeling for several categories.

If there are constraints such as capacity of local consulting firm for survey

implementation, theoretically the sample size can be reduced further while there is a risk that

sample size for model development for a specific group might be too small. Impact of

reduction of sample size of HTS and ADS will be examined with case studies.

Feasibility Study (F/S)

For the F/S, accurate forecasts for the relatively short and intermediate terms are required.

Since F/S is usually based on M/P forecasts and data, it is generally easy to maintain accuracy

of the demand forecast in the F/S as long as M/P data is utilized. Therefore, accuracy of

M/P studies will be focused on and examined. Review of previous F/S will be further

continued.

Others (TCM Policies etc.)

For the evaluation and travel demand forecast of TCM policies, conventional method with


751

daily passenger analysis is not accurate enough for policy evaluation. While SP surveys can

be conducted for analysis of TCM policies, it is not easy to apply in some countries and areas.

This is because it is quite difficult for some respondents to image new policies. Therefore,

ABM is considered as appropriate method for TCM policy evaluation.

Reduction of Required Time for Travel Surveys and Demand Forecast

According to the interview survey results in the U.S., reduction of time is not expected with

utilization of mobile devices; in fact, minimum 3 to 5 months are required for survey

implementation. In terms of travel demand modeling, development of ABM also requires

almost the same amount of human resources (man-months) with the conventional aggregate

four-step method. Therefore, reduction of time required for travel surveys and travel

demand forecast may not be expected.

Reduction of Cost for Travel Surveys and Demand Forecast

As total man-months for applying ABM and those for the conventional four-step method are

nearly the same, reduction of cost is also not expected. If it is possible to reduce the number

of samples, it will lead to reduction of cost of travel surveys. However, it should be noted

that availability of population census data is a prerequisite.

ACKNOWLEDGEMENTS

This study was conducted as a part of the “Project Research on Travel Survey and Demand

Forecast in Developing Countries” by the Japan International Cooperation Agency (JICA)

with advice from the academic committee members.

REFERENCES

1. Kitamura, R. (2002). Modeling Travel Behavior, p.p. 63-64, Tokyo, Gihoudou.

2. Smith, M. E. (1979). “Design of Small-Sample Household-Interview Travel Surveys.”

Transportation Research Record, 701, p.p. 29 35.

3. THE ON-LINE TRAVEL SURVEY MANUAL: A Dynamic Document for Transportation

Professionals, Provided by the Members and Friends of the Transportation Research

Board’s Travel Survey Methods Committee (ABJ40),

http://www.travelsurveymanual.org/HomePage.html. Accessed August 1, 2017.

4. Stopher, P.R. (1993). “Deficiencies of Travel-Forecasting Methods Relative to Mobile

Emissions,” Journal of Transportation Engineering, 119(5).

5. Deakin, E., and G. Harvey (1993). A Manual of Regional Transportation Modeling Practice

for Air Quality Analysis, Washington, D.C.: National Association of Regional Councils.

6. Vovsha, P., M. Bradley, and J.L. Bowman (2004). “Activity-Based Travel Forecasting

Models in the United States: Progress since 1995 and Prospects for Future.” Paper

presented at the EIRASS Conference on Progress in Activity-Based Analysis, Maastricht,

The Netherlands

7. Plurinational State of Bolivia Secretariat of Public Works and Land Use of the Department

of Santa Cruz, and Japan International Cooperation Agency (JICA) (2017). Transport

Improvement Master Plan Project for Santa Cruz Metropolitan Area, Final Report,


752

prepared by Nippon Koei Co., Ltd., Tamano Consultants Co., Ltd., Oriental Consultants

Global Co., Ltd., Nippon Koei Latin America-Caribbean Co., Ltd.

8. Department of Transportation and Communication (DOTC), Republic of the Philippines

and Japan International Cooperation Agency (JICA) (2015). The Project for Capacity

Development on Transportation Planning and Database Management in the Republic of the

Philippines: MMUTIS Update and Enhancement Project, prepared by ALMEC

Corporation and Oriental Consultants Global Co., Ltd.

9. Dhaka Transport Coordination Authority (DTCA), People’s Republic of Bangladesh and

Japan International Cooperation Agency (JICA) (2016). The Project on the Revision and

Updating of the Strategic Transport Plan for Dhaka, Final Report, prepared by ALMEC

Corporation, Oriental Consultants Co., Ltd, Katahira & Engineers International.

10. Hanoi and Ho Chi Minh People’s Committee, Socialist Republic of Vietnam and Japan

International Cooperation Agency (JICA) (2016). Data Collection Survey on Railways in

Major Cities in Vietnam (METROS), Final Report, prepared by ALMEC Corporation,

Oriental Consultants Co., Ltd, Nippon Koei Co., Ltd, Japan Transportation Consultants,

Inc.

11. Metro Cebu Development and Coordination Board (MCDCB), Republic of the

Philippines and Japan International Cooperation Agency (JICA) (2014). The Roadmap

Study for Sustainable Urban Development in Metro Cebu, prepared by ALMEC

Corporation and Oriental Consultants Global Co., Ltd.

12. Ministry of Construction, Housing, Sanitation and Urban Development (MCLAU),

Republic of Côte d'Ivoire and Japan International Cooperation Agency (JICA) (2015).

The Project for The Development of the Urban Master Plan in Greater Abidjan in the

Republic of Côte d'Ivoire (SDUGA), Final Report, prepared by Oriental Consultants Co.,

Ltd, Japan Development Institute, International Development Center of Japan, Asia Air

Survey Co., Ltd.

13. Yangon Region Government, Republic of the Union of Myanmar and Japan International

Cooperation Agency (JICA) (2014). Project for Comprehensive Urban Transport Plan of

the Greater Yangon (YUTRA), Final Report, prepared by ALMEC Corporation, Oriental

Consultants Co., Ltd, Nippon Koei Co., Ltd.

14. Ministry of Transport (MOT), Democratic Socialist Republic of Sri Lanka and Japan

International Cooperation Agency (JICA) (2015). Urban Transport System Development

Project for Colombo Metropolitan Region and Suburbs (CoMTrans), prepared by Oriental

Consultants Global Co., Ltd.

15. Coordinating Ministry of Economic Affairs (CMEA), Republic of Indonesia and Japan

International Cooperation Agency (JICA) (2012). Jabodetabek Urban Transportation

Policy Integration Project in the Republic of Indonesia, Final Report, prepared by Oriental

Consultants Co., Ltd and ALMEC Corporation.

16. Istanbul Metropolitan Municipality (IMM), Republic of Turkey and Japan International

Cooperation Agency (JICA) (2009). The Study on Integrated Urban Transportation

Master Plan for Istanbul Metropolitan Area in the Republic of Turkey, Final Report,

prepared by ALMEC Corporation and Nippon Koei Co., Ltd.

17. Dar es Salaam City Council (DCC), the United Republic of Tanzania and Japan

International Cooperation Agency (JICA) (2008). Dar es Salaam Transport Policy and

System Development Master Plan, Final Report, prepared by Pacific Consultants

International and Construction Project Consultants.

18. Nairobi City County (NCC), Republic of Kenya and Japan International Cooperation

Agency (JICA) (2014). The Project on Integrated Urban Development Master Plan for


753

the City of Nairobi in the Republic of Kenya, Final Report, prepared by Nippon Koei Co.,

Ltd., IDCJ Inc.

19. Loksin, M. Survey Solutions: Computer-assisted Personal Interviewing,

https://undataforum.org/WorldDataForum/wp-content/uploads/2017/01/ta2.05_Michael-Lo

kshin-Survey-Solutions.pdf. Accessed August 1, 2017.

20. Ort´uzar, J. D. and Willumsen, L. G. “Modelling Transport.” 4th Edition. 2011. Aptara

Inc., New Delhi, India.

Date post:	28-Feb-2022
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times

Review of Past Urban Transportation Studies and ...

Documents