Microsoft Word - CR2009-321 ADB Inception Report Final.doc

Confidential 2009

GNS Science Consultancy Report 2009/321

EXECUTIVE SUMMARY

The Asian Development Bank (ADB) are funding Technical Assistance, TA 6496-REG
Regional Partnerships for Climate Change Adaptation and Disaster Preparedness to develop
exposure databases - information on the built environment and its relationship to hazards -
that support greater resilience to climate impacts and natural disasters through facilitating
decision-making on hazard exposure and risk minimization. The databases will also support
an assessment of the feasibility of a regional pooled catastrophe insurance scheme and its
subsequent development, an interrelated World Bank project, which requires data form this
project to carry out loss modelling.

GNS Science International Ltd in association with the Pacific Disaster Center, and the Pacific
Islands Applied Geoscience Commission (SOPAC) have been contracted to carry out the
TA. Eight Pacific island countries: Cook Islands, Fiji, Papua New Guinea, Samoa, Solomon
Islands, Tonga, Tuvalu and Vanuatu are to be involved in this initial project. Critical to the
success of the project is the partnership with the SOPAC, who have an understanding of
existing data, and have built relationships with these countries over many years.

As part of the inception mission the following activities have taken place
• Preparation of a précis of the project concept
• Engagement with SOPAC in September 2009, where the composition of the project

team and timetable were discussed, along with a review of SOPAC’s existing data
holdings for the eight PICs;

• Presentation of the Project to the SOPAC Annual Session, Vanuatu;
• Attendance at the FEMM meeting, Cook Islands, October, 2009 and engagement with

Cook Island government officials;

•  Development of Project introductory letters to country counterpart agencies;
•  A data capture trial in Nadi, and
•  An inception meeting held at SOPAC 2-4 December 2009 to discuss the inception

activities and finalise work programme, in conjunction with personnel from a related
World Bank project.

SOPAC and the countries involved have some infrastructure data. Five capital cities (Apia,
Nuku’alofa, Honiara, Suva and Port Vila) have building data collected as part of the Pacific
Cities project captured by SOPAC 1998 to 2001. Other infrastructure and physiographic data
are held within the countries but appear to be inconsistent in coverage and often poorly
attributed. This TA will collect existing infrastructure and physiographic data, and where data
are lacking or absent, the location and attributes of buildings will be collected by field survey
and extrapolation from census data. The field data collection will be reliant on SOPAC and
in-country staff and will utilise integrated handheld data capture devices. Building foot prints
and major, above-ground assets will be captured by digitizing from aerial or satellite imagery
carried out by SOPAC and AIR Worldwide, prior to field data collection.

All data collected will have a spatial reference but will be in a form that is GIS platform
independent, i.e. it can be utilized in any GIS. This will ensure that the data have greater
utility in risk modelling. Experience has shown that when a database is dependent on third-
party proprietary software it introduces additional cost in implementation and problems
following new releases of the proprietary software. We envisage using an open source
database and GIS management software that can be accessed by existing proprietary GIS
software or open source GIS.

Confidential 2009

GNS Science Consultancy Report 2009/321

CONTENTS

EXECUTIVE SUMMARY ......................................................................................................... II

INTRODUCTION .......................................................................................................... 1

INCEPTION ACTIVITIES ............................................................................................. 3

2.1

Initiation Report.................................................................................................3

2.2

Engagement with SOPAC/Fiji, September 2009 ..............................................3

2.3

Presentation of Project − SOPAC Annual Session, Vanuatu ...........................4

2.3.1

STAR Seminar ...........................................................................4

2.3.2

Special meeting .........................................................................4

2.4

FEMM meeting, Cook Islands, October 2009...................................................5

2.5

Project introductory letters to country counterpart agencies.............................6

2.6

Data capture trial, Nadi .....................................................................................6

2.7

Inception Meeting, 2 - 4 December 2009..........................................................6

INCEPTION FINDINGS................................................................................................ 7

3.1

Hazard data ......................................................................................................7

3.1.1

Tropical Cyclone ........................................................................7

3.1.2

Earthquake.................................................................................8

3.2

Infrastructure Data ..........................................................................................10

3.2.1

Buildings ..................................................................................10

3.2.2

Building Data Analysis – Pacific Cities, RiskScape .................12

3.2.3

Building data analysis - Nadi trial.............................................13

3.2.4

Transportation..........................................................................15

3.2.5

Underground Pipe Networks....................................................18

3.2.6

Power Distribution Networks....................................................19

3.3

Physiography ..................................................................................................20

3.3.1

Imagery ....................................................................................20

3.3.2

Topography..............................................................................20

3.3.3

Bathymetry...............................................................................21

3.3.4

Soils/Geology...........................................................................23

PROJECT IMPLEMENTATION PLAN ...................................................................... 25

4.1

Activity 1: Inception, engagement and liaison.................................................25

4.2

Activity 2: Database design ............................................................................25

4.3

Activity 3: Data Collection ...............................................................................26

4.3.1

Data collection Phase 1 ...........................................................26

4.3.2

Data collection Phase 2 ...........................................................26

4.3.3

Methodology for data capture in the field.................................26

4.3.4

Hazard data collection .............................................................28

4.3.5

Physiographic data ..................................................................28

4.4

Activity 4: National and Regional Systems .....................................................28

4.5

Activity 5: Training ..........................................................................................30

ACKNOWLEDGEMENTS .......................................................................................... 33

REFERENCES ........................................................................................................... 33

Confidential 2009

GNS Science Consultancy Report 2009/321

Appendices

Appendix 1:

Country Contacts – ADB Pacific Exposure Database Project. TA- 6496-REG

Appendix 2:

Participants of the special session held at SOPAC Annual Session in Vanuatu, October
2009

Appendix 3:

Example letters sent to country Government officials

Appendix 4:

Satellite imagery data and coverage maps held by SOPAC

Appendix 5:

Attribute forms and reference material for field trial − Nadi

Appendix 6:

Minutes of the Inception meeting and Attribute Table

Appendix 7:

Status of SOPAC Map servers for each country

Appendix 8:

Detailed project plan

Appendix 9:

Detailed data collection plan

Figures

Figure 1:

The eight Pacific Island Countries involved in the project are Cook Islands, Fiji, Papua New
Guinea, Samoa, Solomon Islands, Tonga, Tuvalu and Vanuatu........................................................2

Figure 2:

The average number of tropical cyclones per year passing within 555 km (a circle of radius
equal to 5° of latitude) of the main island groups of the Southwest Pacific over the full
cyclone season (November through May) (source Glassey

et al.

2004). ...........................................8

Figure 3:

Epicentres of earthquakes of M≥5.5 from the Pacific Catastrophe Risk Financing Initiative
simulated 10 000 year catalogue (source Pacific Catastrophe Risk Financing Initiative)...................9

Figure 4:

Field checking of building footprints captured from high resolution satellite imagery .......................14

Figure 5:

SW Pacific map showing bathymetry survey locations carried out under SOPAC/EU funding
2003-2008........................................................................................................................................23

Figure 6:

GIS Options for the inventory database ...........................................................................................29

Tables

Table 1:

Mean return period (in years) of tropical cyclones for the 8 PICs (source PCRFI 2008) ....................8

Table 2:

Mean return period (in years) of earthquake magnitudes occurring close (<200 km) to the
capital cites of the eight selected countries (source Pacific Catastrophe Risk Financing
Initiative). The distance is the site-to-rupture distance not site-to-epicentre.....................................10

Table 3:

RiskScape earthquake building classes parameters and possible values. ......................................12

Table 4:

PCRFI building classes parameters and possible values.................................................................12

Table 5:

PC database building class parameters and possible values. .........................................................13

Table 6:

Accuracy of building attributes data collected in the Nadi field trial..................................................15

Table 7:

Existing Topographic data................................................................................................................21

Table 8:

Existing Geology and Soils data ......................................................................................................24

Table 9:

Proposed workplan ..........................................................................................................................32

Confidential 2009

GNS Science Consultancy Report 2009/321

1 INTRODUCTION

South Pacific nations are exposed to a range of natural hazards such as earthquakes, floods,
tsunami, volcanic eruption, cyclones and severe storms. There is a need to compare the risk
posed by each hazard in a consistent way using potential impacts such as cost and
casualties. Such a comparison of risk can then support decision-making by:

• Determining which hazard represents the greatest risk to various communities;
• Enabling mitigation investments to be prioritised (e.g. earthquake strengthening of

buildings versus cyclone strengthening);

• Avoiding inappropriate land development through planning; and
• Contributing to effective emergency management plans.

• Support the development of up to eight national infrastructure inventories for

countries most exposed to the risk of earthquakes and cyclone; and

• Develop a consolidated, regional database encompassing hazard to create an

exposure database and, if available, hazard-specific fragility functions.

The data will enhance the preliminary assessment of the feasibility of a Pacific Catastrophe
Risk Financing Initiative (PCRFI), which is being carried out as part of a related World Bank
project (Phase 2). The information gathered will enhance the World Bank project in their
development of country specific risk models. In addition it will assist informed decision-
making in Disaster Risk Reduction, minimizing the negative social and environmental
impacts of catastrophic events.

The exposure databases will build upon existing information. The key tasks for this TA are to:

•  Assess the current in-country use of hazards and risk information;
•  Collate and analyse existing hazard models and risk profiles;
•  Collect national georeferenced building and infrastructure (inventory) data from the

Pacific Islands Applied Geosciences Commission (SOPAC) and local government
agencies to develop national and regional exposure databases;

• Incorporate historical natural disasters and climate change models into risk profiling

and disaster exposure;

• Involve policy makers in stakeholder consultations, meetings, and relevant training to

develop an awareness of the type of information available and its range of uses;

• Identify and implement an appropriate GIS platform for acquiring, managing,

analysing, and displaying hazard and exposure data and risk information; and

• Provide in-country training, user manuals, and reports.

Key to the success of the project is SOPAC who have an excellent understanding of
available data as well as relationships with Pacific Island Countries (PICs) developed over
many years. The eight Pacific Island Countries (PICs) that are involved in the project are
Cook Islands, Fiji, Papua New Guinea, Samoa, Solomon Islands, Tonga, Tuvalu and
Vanuatu (Figure 1). These PICs have agreed with the ADB to support the initiative,
predominantly via the ministries of finance and/or planning (Appendix 1).

Confidential 2009

GNS Science Consultancy Report 2009/321

2 INCEPTION

ACTIVITIES

It was in initially planned to visit all eight Pacific Island Countries involved in the project to
engage with government officials as part of this inception activity. However, on discussion
with SOPAC, it became obvious that this approach would be unrealistic in achieving effective
engagement in such a short timeframe, and very costly. It was decided to take advantage of
a number of existing meetings, such as the SOPAC Annual Session in Vanuatu and the
Forum Economic Ministers Meeting in the Cook Islands, to present the project concept to the
countries. The project was also presented at the Pacific GIS and RS users Conference to be
held in Suva from the 1-4 December 2009. Full engagement will be done prior to data
collection in each country. Inception activities are described below.

2.1 Initiation

Report

A précis (Glassey, 2009) of the project concept was prepared and supplied to ADB as
requested.

2.2

Engagement with SOPAC/Fiji, September 2009

The international project team engaged with SOPAC counterparts in Suva Fiji from the 28
September until the 2 October 2009 and activities included:

• Discussing the project team composition (Michael Bonte-Grapetin, Litea Biukoto);
• Presenting the project concept to SOPAC staff and University of South Pacific (USP)

Staff (Conway Pene and Eberhard Weber);

• Determining SOPAC data holdings for each country, (Joy Papao, Wolf Forstreuter –

SOPAC);

• Engaging with the SOPAC Ocean and Islands team (Arthur Webb, Robert Smith and

Herve Damlamian);

• Engaging with the SOPAC Economic Assessment Unit (Paula Holland) and Pacific

Disaster Network coordinator (Jutta May);

• Discussing and planning the data capture process, including running a data capture

trial in Nadi, based on Pacific Cities attributes and utilising 60 USP students to assist
with data collection; and

• Revision of the Project Implementation Plan to consolidate the work plans and

maximise synergies with the related World Bank-led Pacific Catastrophe Risk
Financing Initiative (PCRFI) Phase 2.

In addition we took the opportunity to meet with the following people based in Suva

• John Prasad, Acting Permanent Secretary, Finance, Fiji;
• Rasmih Rita (Fiji Land Information System), regarding the content and availability of

data on Fiji, including census data;

• Terrence Erasito, Structural Engineer, Suva, Fiji to discuss building construction in

the Pacific, and;

• Alain Goffeau of the ADB, South Pacific Regional Office, Suva.

Confidential 2009

GNS Science Consultancy Report 2009/321

2.3

Presentation of Project − SOPAC Annual Session, Vanuatu

2.3.1 STAR

Seminar

The opportunity was taken to present the project at the Science, Technology and Research
(STAR) seminar of the SOPAC Annual Session in Vanuatu in October 2009. A presentation
entitled “Hazard and Risk in the Pacific: Towards an understanding of exposure with
RiskScape

and the development of a Pacific exposure database” was given by David Heron

of the Project Team, and outlined the previous work done by the World Bank and ADB, as
well as the concept of this current project. The asset types (buildings and other structures,
roads and other infrastructure) that would be captured under the ADB project were described
together with the hazards (earthquake, tsunami, and cyclone), and RiskScape used to
illustrate how these spatial asset data could be overlain on spatial hazard maps to determine
the exposure (that is the number of buildings, kilometres or roads, etc. that are exposed to
the hazards), vulnerability and loss.

2.3.2 Special

meeting

During the STAR session in Vanuatu a side meeting was held and facilitated by David Heron
and Michael Bonte-Grapetin (SOPAC) to further brief the country counterparts that attended
from the Cook Islands, Papua New Guinea, Samoa, Tonga and Vanuatu on the project
(Appendix 2). Fiji, Solomon Islands and Tuvalu were not represented at this meeting.

The country delegates were briefed on the purpose of the project and potential applications
of the national exposure databases. Details on the types of hazard and asset data to be
collected were discussed and the organisations in each country most likely to hold relevant
data were identified. Also discussed were the points of contact within each country and the
best approach to take. It was clear that further multi-stakeholder discussions in-country will
be needed to ascertain the best focal point to host and sustain the national exposure
database.

ADB has made initial contact with the officials in each country as per Appendix 1, but it was
considered at this meeting, that a follow up letter with copies to Internal Affairs, the National
Disaster Management Office and other departments and ministries to whom the project
would be of interest and/or that could hold relevant data, would be appropriate. Further, it
was suggested SOPAC should equally inform in writing the SOPAC National
Representatives and copy these letters to the other relevant ministries. The following
ministries were considered relevant in the named countries:

Cook Islands

ο  Ministry of Finance (initial contact)
ο  Ministry of Foreign Affairs
ο  Ministry of Infrastructure and Planning
ο  Cook Islands Statistics Office
ο  National Environment Services (NES)

Kingdom of Tonga

ο Ministry of Finance (initial contact)

RiskScape is a multi-hazard risk assessment system being developed in NZ

Confidential 2009

GNS Science Consultancy Report 2009/321

ο Ministry of Lands, Survey, Natural Resources & Environment
ο Ministry of Works
ο Department

Statistics

ο Met

Office

Vanuatu

ο  Office of the Prime Minister (initial contact)
ο  Ministry of Finance
ο  Ministry of Lands and Natural Resources
ο  Department of Geology, Mines and Water Resources
ο  Vanuatu Statistics Office

Samoa

ο  Ministry of Finance (initial contact)
ο  Department of Foreign Affairs
ο  Department of Lands, Survey & Environment
ο  Public Works Department
ο Department

Statistics

ο Electricity Power Corporation
ο Samoa Water Authority

Papua New Guinea

ο  Department of National Planning and Monitoring (initial contact)
ο  Ministry for Finance and Planning
ο  Ministry for Works and Supply
ο  Department of Minerals Policy and Geohazards Management
ο  National Statistics Office

2.4

FEMM meeting, Cook Islands, October 2009

Phil Glassey, TA project team leader, along with Edy Brotowisoro of the ADB, and Olivier
Mahul and Francis Ghesquiere of the World Bank, visited Rarotonga from the 27–30 October
to attend the Forum Economic Ministers Meeting (FEMM) and engage with Cook Island
government officials. The project was discussed with Dr Gerald Haberkom, Manager of the
Statistics and Demography programme at South Pacific Commission (SPC), who indicated
that the census data that they held will be available for the project.

This team also met with Cook Island government officials from the Ministry of Infrastructure
and Planning (MOIP), National Environment Service (NES), Ministry of Agriculture,
Emergency Management Cook Islands (EMCI), the Prime Ministers Office, Central Policy
and Planning and the Police. We visited Timote Tangiruaine, the MOIP GIS specialist, and
determined that the Cook Islands hold much data that can be utilised for the project. The
Secretary of the Ministry of Infrastructure and Planning expressed interest in supporting the
project and along with the National Environment Service agreed to make staff available for
the field survey in the Cook Islands planned for February 2010.

Confidential 2009

GNS Science Consultancy Report 2009/321

2.5

Project introductory letters to country counterpart
agencies

Letters of introduction to the project will be sent by the International Team to each of the
primary contacts in each country (Appendix 1) and copied to other government officials well
in advance of data collection. An example of such a letter sent to the Cook Islands
Government is attached as Appendix 3. These letters will be followed up with further
correspondence to confirm logistics.

SOPAC are also sending letters to the SOPAC National representatives for each country as
per Appendix 3. Another request to the national representative will be made immediately
prior to going on mission into the country, which is a SOPAC operating requirement.

2.6

Data capture trial, Nadi

A trial in capturing building attributes was carried out in Nadi from 27−30 October to
capitalize on the offer of 60 students from the University of the South Pacific and to test field
methodologies and data templates. Building attributes, based on the Pacific Cities data, were
collected utilizing different methods to determine the most efficient and reliable means of
data collection. Two methods were trialled, a paper-based system and hand held devices. In
both cases photographs were taken of the buildings and the image number recorded.
Reference materials in the form of field guides and various maps were provided as part of
the exercise.

2.7

Inception Meeting, 2 - 4 December 2009

The draft inception report was discussed at the inception meeting held at SOPAC in Suva,
Fiji, 2-4 December 2009. Representatives of the World Bank and AIR Worldwide, who will be
carrying out the risk modelling as part of a related World Bank project, attended these
meetings, and the attributes to be captured have been discussed and prioritised. AIR
Worldwide will be capturing major assets utilising satellite imagery, and this imagery will be
utilised to capture building footprints to which attributes will be assigned as part of this ADB
project. The minutes of this meeting along with a table of data attributes to be captured are
included as Appendix 6. At the same time the project was presented at the Pacific GIS /RS
conference held at University of South Pacific (USP), Suva Fiji.

Confidential 2009

GNS Science Consultancy Report 2009/321

3 INCEPTION

FINDINGS

3.1 Hazard

data

Numerous online catalogues of hazard events are available for tropical cyclone, earthquake,
tsunami and volcanic activity. The Pacific Catastrophe Risk Financing Initiative recognised
that because of the short time over which historic records have been kept, these databases
do not represent all possible future events. Consequently they created synthetic earthquake
and cyclone catalogues representing 10 000 years of events that are statistically consistent
with, but not identical to, the historical catalogue. These catalogues contain 150 000 cyclone
and 2.2 million earthquake events and provide a better estimate the hazard. These data, or
at least links to the data, should be included in the hazard database created by this project.

Regional Natural Hazards Potential Maps of the Circum Pacific Region have been compiled
by Johnson

et al.

(1995) at a scale of 1:10 million. These maps include earthquakes,

cyclones, landslides, tsunami, floods, bushfires, volcanoes and droughts. Multi-hazard maps
for major urban areas in Fiji (Suva, Nausori, Labasa, Nadi, and Ba) were reported by Blong
(1994). PDC has developed and maintains an Asia Pacific Hazards and Vulnerabilities Atlas
(

http://atlas.pdc.org

) which has now been extended to become the Global Hazards

Information Network (GHIN). This resource, plus the Pacific Disaster Net hosted by SOPAC,
contain information on hazards, historical damage from hazard events, as well as hazard
models and risk surfaces (probability surfaces) for storms and earthquakes.

3.1.1 Tropical

Cyclone

Figure 2 shows tropical cyclone occurrences in the South West Pacific for all years for the
period 1970/71 to 2003/04, during the modern period of satellite records. Tropical cyclones
develop in the South Pacific over the wet season, usually from November through April. Peak
cyclone occurrence is usually during January, February and March. On average nine tropical
cyclones occur during the November to April season, but this can range from as few as four
in 1994/95 and 2003/04, to as many as 17 in 1997/98, during the strong El Niño episode
(Glassey

et al.

2004).

Tracks of tropical depressions (<34 knots wind speed), tropical cyclones (34–64 knots) and
severe tropical cyclones (> 64 knots) are held in the US Navy Global Tropical/Extra-tropical
Cyclone Climatic Atlas database (GTECCA)

and the Joint Typhoon Warning Center’s

(JTWC) Southern hemisphere Best Track data. These data include date, time, position,
storm-stage and, where known, maximum wind speed, minimum central pressure and storm
name.

The Tropical Cyclone Hazard Model developed by AIR Worldwide as part of the World Bank,
Pacific Catastrophe Risk Financing Initiative (PCRFI) project, uses this historic catalogue as
well as other data from Bureau of Meteorology (BoM, Australia) and the Fiji Meteorology
Service, to generate a synthetic catalogue of 10 000 events to calculate wind speeds and
precipitation for a 2.5º latitude/longitude grid, and estimate storm surge. Return periods for
various category storms for the eight different countries are report by the PCRFI as in Table
1.

http://navy.ncdc.noaa.gov/products/gtcca/gtccamain.html

Confidential 2009

GNS Science Consultancy Report 2009/321

Figure 2:

The average number of tropical cyclones per year passing within 555 km (a circle of

radius equal to 5° of latitude) of the main island groups of the Southwest Pacific over the full cyclone
season (November through May) (source Glassey

et al.

2004).

Table 1:

Mean return period (in years) of tropical cyclones for the 8 PICs (source PCRFI

2008)

Saffir-
Simpson
Category

Cook Islands

Fiji

PNG Samoa Solomon

Islands

Tonga Tuvalu

Vanuatu

≥1 5

103

20 4

≥2 8

200

40 5

≥3 21

500

88 160

182

≥4 294 169

2500

185

270

313

769

400

5 10000

10000

3333

10000

Note: Estimated mean return periods higher than 1000 years should be interpreted with caution as they are
calculated from 10 or less simulated events out of 10000.

A Tropical Cyclone hazard model, determining return periods for wind speeds using a 5 000
event synthetic catalogue, has been developed for Port Vila, Vanuatu (Shorten

et al

. 2003)

and losses estimated.

3.1.2 Earthquake

The South Pacific is one of the most active seismic regions in the world (Figure 3).
Earthquakes result in multiple threats to island nations. Ground shaking can be amplified by
soft sediments, resulting in significant damage to buildings and infrastructure at some
distance from the epicentre. Water-saturated sediments around wharves can settle,
disrupting these facilities. Ground shaking can also trigger landslides in areas of unstable
steep slopes. However it is tsunami that can be the most widespread and damaging
consequence of an earthquake. It is intended that the hazard database include the
epicentres of earthquakes that generated tsunami affecting the island nations together with
any locations where tsunami impacts have been recorded

Confidential 2009

GNS Science Consultancy Report 2009/321

The SOPAC Pacific Cities Database includes an assessment of the ground conditions in the
capitals of five of the eight island nations that are the focus of this project. Similar data will be
captured for other heavily populated centres, where available, to assist in earthquake loss
modelling.

The Pacific Catastrophe Risk Financing Initiative showed a short return period for large
earthquakes occurring within 50 km of the capital cities of the nations included in this project
(Table 2). The data indicate that the Solomon Islands, Tonga and Vanuatu have a high
earthquake hazard whereas Cook Islands and Tuvalu have a low earthquake hazard.

A probabilistic earthquake hazard model was developed for Vanuatu by Suckale

et al.

(2005)

and recently published (Suckale and Grünthal, 2009). The earthquake risk for Suva was
reported by the South Pacific Disaster Reduction Programme (SPDRP 2002) and a
probabilistic earthquake hazard assessment for Fiji was carried out by Jones (1998).
Denham and Smith (1993) gave a review of earthquake risk in the Southwest Pacific region.
Ripper and Letz (1993) determined return periods and probabilities of occurrence of
earthquakes greater than M 7.0 normalized 10 000 km

areas of Papua New Guinea.

Figure 3:

Epicentres of earthquakes of M≥5.5 from the Pacific Catastrophe Risk Financing

Initiative simulated 10 000 year catalogue (source Pacific Catastrophe Risk Financing Initiative).

Confidential 2009

GNS Science Consultancy Report 2009/321

Table 2:

Mean return period (in years) of earthquake magnitudes occurring close (<200 km)

to the capital cites of the eight selected countries (source Pacific Catastrophe Risk Financing
Initiative). The distance is the site-to-rupture distance not site-to-epicentre.

Distance

from

Capital

Earthquake

magnitude

Cook

Islands

Fiji

Papua

New

Guinea

Samoa

Solomon

Islands

Tonga

Tuvalu

Vanuatu

<50 km

≤

M<6

- 23 182 270 2 3 -

<100 km

≤

M<7

- 40 124 135 3 3 2000 1

<150 km

≤

M<8

- 244 98 128

8 12 -

<200 km

≥

- - - 1667 94 106 - 833

Tsunami hazard models have been attempted for a number of the nations under
consideration. Modelling of the wave as it approaches the coast can be achieved using one
of several different open-source software models available (e.g. ANUGA, MOST, COMCOT,
etc.) and existing bathymetric data. Inundation beyond the coastline is more problematic
because of the coarseness of the elevation data currently available. At best, elevation data
are based on interpolated 2 m contours but good modelling requires elevations down to cm
accuracy. Because of the uncertainty created by this lack of key data, no inundation
modelling will be included in the hazard database.

3.2 Infrastructure

Data

As part of the first project inception mission at SOPAC (28 Sep–2 Oct), the International
Team was provided with digital GIS data by SOPAC analysts. These data were catalogued
by PDC analysts and assessed against preliminary criteria. SOPAC has commissioned
inventories of geospatial data related to tsunami inundation and risk modelling regionally and
for some individual countries (SOPAC 2008, 2008a-d). They also collected building data for 5
Pacific capital cities − Apia, Nuku’alofa, Suva, Port Vila and Honiara between 1998 and 2001
(Biukoto

et al.

2001). In addition other data such as roads and land use for example were

supplied by the governments or other sources. Analysis of the Pacific Cities database and its
utility in loss estimation is discussed in Section 3.2.2 below. Other infrastructure data may be
available in the countries, but is not held by SOPAC.

3.2.1 Buildings

A key focus of this activity is the development of a building database for each country and
the region. At a minimum, buildings will be represented as point features. Building “footprint”
datasets will be retained as polygons and converted to points for consistency. Attributes
which characterize buildings such that they can be assigned a fragility class to enable loss
computations will be collected. Replacement costs are required to compute direct
replacement costs and hence financial loss to hazards. For the hazards under consideration
(seismic and wind), fragility can be estimated using attributes including: use type, age,
structural type, construction materials, roof configuration, number of stories, area and floor
level. The attributes to be captured have been discussed and prioritised with AIR Worldwide
who will be carrying out the loss modelling as part of the related World Bank project and
included in Appendix 6. The following assessment of the building data, either held by
SOPAC, or known to exist in each country, was based on these information requirements.

Confidential 2009

GNS Science Consultancy Report 2009/321

Cook Islands:
None held by SOPAC. However, from the inception mission to the Cook Islands in October
2009, it was discovered that the Cook Islands do have building information including location
(point), use and type of structure for buildings constructed since 1996. These data are
collected as part of the building permitting and control system. Hence some building data is
available for Rarotonga. In addition building locations and use have been captured for
Aitutaki, Atiu, Mauke, Mitiaro, and Mangaia as part of a survey conducted under an NZODA
Outer Islands Water & Power Reticulation Feasibility Study (

http://www.maps.gov.ck

Fiji Islands:

Suva only – Pacific Cities data.

Papua New Guinea:
There are five separate building footprint datasets for different areas of the City of Port
Moresby. However, the information (values) within the attribute tables is indiscernible.
Bougainville, Chimbu, East New Britain, Enga, East Sepik, Madang, Manus, Milne Bay,
Morobe, New Ireland, Oro, Port Moresby, Southern Highlands, Western, Western Highlands,
West New Britain and West Sepik have health building layers that include the following
attributes: Name, Alternate_Name, CU_Name, Census2000_Key, Feature, Province, District,
LLG, Ward, RMU, Access and Health_Type. These areas also have a school buildings layers
that includes the following attributes: Name, Alternate_Name, CU_Name, Literacy_Name,
Village_Key, Province, District, LLG, LLG NSO Map, Access and School.

Samoa:
There is one dataset (AassetsGPS) of GPS collected buildings for the Apia area only
captured as part of Pacific Cities. This dataset has very detailed attributes that include the
type of structure (e.g., house, commercial, church, etc.), wall materials, windows, roof
materials, roof shape, roof pitch, number of stories, maximum height, base floor area, and
date of survey.

Solomon Islands:
Honiara Only – Pacific Cities data.

Tonga:
Building locations for the whole of Tonga were captured from satellite imagery as part of the
CERM project 2004-2007. Large buildings (>1000 m

) were captured as polygons and

smaller ones as points. There are few attributes assigned to these buildings. There is a
points layer (Nasset) for Nuku’alofa captured as part of the Pacific Cities Project that includes
attributes such as building use (e.g., house, church, commercial, etc.), base floor area,
height above sea level, foundation type, wall materials, number of stories, roof material, roof
shape, roof pitch, % of windows, and burglar bars.

Tuvalu:
The following islands have buildings footprints, but no attributes: Vaitupu, Nukulaelae,
Nukufetau, Nui, Nuitao, Niulakita, Nanumea, and Nanumaga. The island of Funafuti has a
detailed buildings polygon dataset (funafuti_Govbuild) with attributes including building type
(household, government, etc.), water storage tank capacity, presence of gutters, gutter
material, roof type (coded) and roof material.

Confidential 2009

GNS Science Consultancy Report 2009/321

Vanuatu:
There is a points dataset (Vasset) for Port Vila, collected as part of the Pacific Cities project,
with attributes that include building use (e.g., house, church, commercial, etc.), base floor
area, height above sea level, foundation type, wall materials, number of stories, roof material,
roof shape, roof pitch, % of windows, and burglar bars. Also for Port Vila is a point dataset
(Vsites) with places such as hotels, public places, government buildings, etc.

3.2.2

Building Data Analysis – Pacific Cities, RiskScape

Prior to undertaking the data capture trial in Nadi, analysis was undertaken on the Pacific
Cities (PC) data to determine how readily they might be used for loss estimation. The data
were compared to those used as input to RiskScape and the Pacific Catastrophe Risk
Financing Initiative (PCRFI). The data were further analysed in an attempt to provide several
general building categories to speed up field data capture. In general, the Pacific Cities (PC)
data can be used for loss modelling but requires some assumptions to be made. For
earthquake hazard, building data in RiskScape are classified into 72 construction classes
using 5 parameters (Table 3). PCRFI lists 45 building construction classes of in Annex 4
(Table 4) based on 4 parameters.

The PC data contain a total of 20 781 points representing buildings in Apia, Honiara, Port
Vila, Suva, and Nuku’alofa. Attributes are listed in Table 5. The database holds a
classification of use, building name, minimum elevation of floor above ground, roof material,
slope and shape, building size and the proportion of walls in windows plus the presence of
concrete cantilever balconies.

It was determined that with the use of a suitable lookup table the Pacific Cities data can be
translated to fit within the proposed building classification and therefore can be used to form
the basis of the building dataset for this project.

Table 3:

RiskScape earthquake building classes parameters and possible values.

General Description

Load System

Height

Age Band

Quality

Reinforced Concrete

Shear wall

Low (1-3 storeys)

pre-1940

Sound

Steel Moment-resisting

frame

Medium

(4-7 storeys)

pre-1970

Deficient

Timber stud, steel stud

Braced frame

High (8+ stories)

1960-1979

Reinforced masonry

Tilt-up panel

post-1970

Reinforced concrete

Portal frame

Light Industrial

Energy absorption

Advanced Design

Unreinforced

Brick Masonry

Table 4:

PCRFI building classes parameters and possible values.

Building Type

Construction Class

Height

Year of
Construction

Traditional

Thatched / Palm leaf

Low (1-3 storeys)

pre-1990

Shacks

Other

Medium (4-7 storeys)

pre-1990-2005

Detached

Unreinforced masonry

High (8+ stories)

post 2005

Detached / Flat

Confined masonry walls

Confidential 2009

GNS Science Consultancy Report 2009/321

Flat / Commercial / Industrial

Reinforced confined masonry walls

Commercial / Industrial

Reinforced concrete

Light

metal

Other

metal

Table 5:

PC database building class parameters and possible values.

Building use

UD material

UD structure

Wall material

Storeys

House Slab Soft

Concrete

Flats wooden

poles

Stiffened

Timber

Shed concrete

columns

Metal 3

Commercial

steel columns

fibre-cement sheet

public services

load bearing walls

Brick

health services

Accommodation

etc

Etc

3.2.3

Building data analysis - Nadi trial

Building attributes were collected for over 1 200 buildings in both residential and commercial
areas in Nadi. Some buildings were visited by two survey groups and so only 714 buildings
were surveyed. Based on these trials a daily building attribute collection rate of between 50
and 100 buildings (depending on building density and visibility) per survey group is estimated
and can be used to evaluate time requirements for future survey work.

About 100 (10%) of the buildings surveyed required the roof polygon captured from satellite
imagery to be split (that is to say a single roof polygon represented 2 or more individual
buildings) or were missing from the roof polygon dataset (Figure 4).

A preliminary analysis of the data collected shows that 53% of the sampled buildings were
residential, 31% commercial, 10% public or communal facilities, and the remainder were
critical facilities, hazardous facilities, other facilities or unknown.

For residential buildings most were either single or double storeys. The predominant roof
pitch was low (56%), the predominant shape was gable (49%), and the predominant material
was sheet metal (83%). The foundations were predominantly slab on ground (91%). Of those
buildings with wood or concrete piles, wood poles or concrete column foundations very few
had bracing. 83% had concrete walls - they were assumed to be plastered concrete block.
For 38% of these, concrete columns were visible. It is assumed that where the building is
above 1 storey those buildings without visible concrete columns had internal concrete
columns and where it is 1 storey it is likely they consist of a mix of internal concrete columns
and thin concrete columns (the width of a concrete block) between the wall blocks.

Confidential 2009

GNS Science Consultancy Report 2009/321

Figure 4:

Field checking of building footprints captured from high resolution satellite imagery

For commercial buildings most were double storeys (49%) with 13% being 3 storeys and 1%
being 4 storeys. The predominant roof pitch was a flat (55%), and where the roof was not flat
the predominant shape was gable (45%). The predominant material was sheet metal (50%)
with concrete a close second (40%). The foundations were predominantly slab on ground
(91%). Of those buildings with concrete column foundations most had some form of bracing.
87% had concrete walls. There were assumed to be plastered concrete block. For 66% of
these, concrete columns were visible. It is assumed that where the building is above 1 storey
those buildings without visible concrete columns had internal concrete columns and where it
is 1 storey it is likely they have internal concrete columns.

Confidential 2009

GNS Science Consultancy Report 2009/321

Only a cursory analysis of survey accuracy has been possible. Five buildings surveyed by a
trainer were compared with the surveys of others who had undergone ½ day of training. In no
instance was any building described completely consistently. An accuracy assessment of the
survey is provided in Table 6.

Table 6:

Accuracy of building attributes data collected in the Nadi field trial

Building Attribute

Percent Correct

Within 1 Class

Note

Main Use

Number of Storeys

100

Roof Pitch

Roof Shape

Roof Material

Foundation 100

Foundation Bracing

Wall Material

100

Wall Structure

Windows 40

Shutters 80

Defects 60

Parapets/Towers 100 4
Concrete Cantilever

Plan Shape

Base Floor Area

Minimum Floor Height

Notes:
1: In most instances it is not possible to see the roof clearly, especially on commercial buildings due to observer’s position.
Consequently the information in these fields is often inferred rather than observed.
2: All foundations in the sample were slab. Analysis of random photos indicated some surveyors were confused whether to
record concrete columns below the living levels as foundation or wall structure.
3: The data capture form was ambiguous and as a result some surveyors understood ‘building with windows forming less than
70% of the wall’ as ‘building with less than 70% wall ‘.
4: None of these occurred in the sample.
5: Concrete cantilevers are any unsupported concrete structure extending from a building wall a significant distance (>1 m). On
commercial buildings concrete cantilevers are used to provide shelter to the foot path. In a number of instances the concrete
had been covered with timber and so was difficult to recognise.
6: Base Floor Area can be taken from the roof polygons but in the sample it was compared to the estimate made by the trainer.
7: In the Nadi survey Minimum Floor Height was estimated as in m. Only in one instance was the sample estimate the same as
the trainer. However future surveys will use classes and had they been used the accuracy would have been 60%.

3.2.4 Transportation

Transportation infrastructure, including roads, bridges, airports, and seaports, constitutes a
significant category of assets exposed to hazards. As with building datasets, attributes that
characterize the degree to which a hazard will damage or destroy the asset are required
along with information that can be used to estimate replacement costs.

Key attributes for roads include number of lanes, traffic flow (one or two way), road surface
(sealed, unsealed, etc), and highway number/name. Additional attributes for bridges, often
depicted as point features, include span (length), construction type and materials.

Airports, either represented as lines or polygons, should include basic information on runway

Confidential 2009

GNS Science Consultancy Report 2009/321

length, width and height above sea level. Additional information on associated structures
(e.g., terminals, hangers, communications equipment, etc.) would be required to fully
estimate replacement costs.

Cook Islands:
There is a major roads dataset for Aitutaki Island, but it has no attributes. There is also a
polygon layer for the airstrip data on Aitutaki. There are several roads datasets for Rarotonga
with varying accuracy and completeness. The most detailed dataset (RaroRoads_WGS84)
has attributes including name, type (minor or major) and surface type. However, not all
features have completed attribute data. Roads have been captured for Aitutaki, Atiu, Mauke,
Mitiaro, and Mangaia as part of a survey conducted under an NZODA Outer Islands Water &
Power Reticulation Feasibility Study (

http://www.maps.gov.ck

Fiji Islands:
There are two national level roads datasets (Major roads and Roads). Major roads has a
name attribute, however it is not complete. Neither dataset has any other attributes other
than coded ones that appear to be symbology related. These codes relate to attributes such
as pavement type or number of lanes.

There are various roads datasets for very small areas of Fiji which appear to be based on
cities or small level administrative units. Some examples are Suva (SUVroad), Nausori
(nAUSORI) and Lau (LAUroad). These datasets do not have attributes.

Papua New Guinea:
There are national level transportation datasets for:

• The nation’s airstrips including attributes for airstrip name, longitude and latitude,

category (e.g., national, provincial, private, etc.), capacity and owner.

• The nation’s roads including attributes for road name, surface type, condition and

type (e.g., major, minor, etc.).

There are also several other transportation datasets organized by province/region:

• Bougainville: An airstrips layer with type, lanes and approximate span (m); apparent

airport buildings data; a bridges layer with standard attributes

, and a roads layer with

standard attributes

. The column for road names is incomplete.

• Chimbu: An airstrips layer with standard attributes

and a roads layer with standard

attributes

• East New Britain: An airstrips layer with airstrip name, longitude and latitude,

elevation, category, owner and capacity; a bridges layer with standard attributes

; a

roads layer with standard attributes

; plus a very incomplete road number column.

• Enga: An airstrips layer with standard attributes

; a bridges layer with standard

attributes

and a roads layer with standard attributes

• East

Sepik: An airstrips layer with standard attributes

; a bridges layer with standard

attributes

; and a roads layer with standard attributes

. The column for road names is

incomplete.

• Madang: An airstrips layer with airstrip name, longitude and latitude, elevation,

category, owner and capacity; a roads layer with standard attributes

; plus a very

incomplete road number column.

• Manus: An airstrips layer with standard attributes

; a bridges layer with standard

attributes

; and a roads layer with standard attributes

• Milne Bay: A roads layer with standard attributes

. The road name column is very

Confidential 2009

GNS Science Consultancy Report 2009/321

incomplete.

• Morobe: An airstrips layer with standard attributes

; a bridges layer with standard

attributes

; and a roads layer with standard attributes

. The road name column is

incomplete.

• New Ireland: An airstrips layer with standard attributes

and a bridges layer with

standard attributes

• Oro: An airstrips layer with standard attributes

; a bridges layer with standard

attributes

; and a roads layer with standard

but very incomplete attributes.

• Port

Moresby: A roads layer with no discernable attributes.

• Southern

Highlands: An airstrips layer with standard attributes

and a roads layer with

standard attributes

• Western: An airstrips layer with standard attributes

and a roads layer with standard

attributes

plus a very incomplete road number column.

• Western Highlands: An airstrips layer with standard attributes

and a roads layer with

no discernable attributes.

• West New Britain: An airstrips layer with standard attributes

and a roads layer with

no attributes.

• West Sepik: An airstrips layer with standard attributes

and a roads layer with

standard attributes

. The road name column is very incomplete.

Standard attributes for an airstrips layer are airstrip name, longitude and latitude, category (national, provincial, private, etc.),

capacity and owner.

Standard attributes for a bridges layer are with type, lanes and approximate. span (m).

Standard attributes for a roads layer are road name, surface type, condition and type (major, minor, etc.).

Samoa:
There is a roads layer for each of the main islands - Savaii (S_ROAD) and Upolu (U_ROAD).
However, this dataset has no attributes. There are three transportation layers for the town of
Apia (i.e., bridge, roadsealed and roadunsealed). These datasets only have an elevation
attribute.

Solomon Islands:
There is one roads dataset of Honiara (H_roads) but it has no attributes. Also, there is a
polygon layer of the Honiara Airport (sb_airport) but it does not have attributes either.

Tonga:
There is a national level road dataset for Tonga collected as part of the CERM Project
between 2005 and 2007 and held by Ministry of Lands, Survey, Natural Resources and
Environment (MLSNRE). Attribute fields include Road-id, Name, Status, Surface, Width,
Number_of_lanes, Description. There are also layers for Tracks (Track_id, Track_name,
Status, Surface), Bridges (Bridge-id, Name, Status, Type, Construction, Surface, Use),
Airport runway (Airstrip_id, Airstrip_name, Type, Surface, Status), large and small wharves
(Wharf_id, Name, Use, Owner) and boatramps. However, it should be noted that many of the
attribute fields have not been populated. The airport terminals are captured in the building
database.

Tuvalu:
There are airport runway datasets for Nukufetau and Funafuti. There are also roads datasets
for the following islands: Vaitupu, Nukulaelae, Nui, Niutao, Niulakita, Nanumea, and Funafuti.
None of the roads datasets have any attributes.

Confidential 2009

GNS Science Consultancy Report 2009/321

Vanuatu:
There is an airports point dataset (Airports) with attributes including name, type, runway
length, surface type and direction. There are roads data for Malampa, Penama, Sanma,
Shefa, Tafea and Torba with attributes including length, classification, number of lanes,
surface type, and width. Some attribute data are not populated. There is also a national level
wharfs dataset including attributes for name and island.

3.2.5

Underground Pipe Networks

Utilities represent another significant category of infrastructure exposure to hazards. They
have been broken into underground pipes (3.2.5) and power distribution (3.2.6) since they
each have unique attributes and hazard exposure modes.

Underground pipe networks include features such as pipes, reservoirs, valves and hydrants,
some of which exist above ground, or partially above ground. Key attributes for risk
assessment include physical properties like construction materials, age, quality and
diameter. For this project, up to three separate networks are anticipated: 1) potable water
reticulation, 2) sewage reticulation, and 3) storm-water reticulation. The data need to be
attributed with replacement costs (e.g. $ replacement cost per meter of storm water pipe in a
given area) and elevation, or level above the surrounding ground as well.

The components of the networks will need to be modeled as lines, points or polygons. The
linear components will comprise primarily the water pipelines (either principal or lateral),
while the nodal components will include features associated with the movement of water
(pumps, valves, hydrants, meters, etc.). Polygons will be used to represent water storage
tanks and reservoirs.

Cook Islands:
No data is held by SOPAC. However as part of the survey conducted under an NZODA
Outer Islands Water & Power Reticulation Feasibility Study (

http://www.maps.gov.ck

) a water

reticulation network has been captured for Aitutaki, Atiu, Mauke, Mitiaro, and Mangaia.
These data include attributes on water intakes and pipe diameter at least. Similar data are
available for the water network on Rarotonga.

There is no reticulated sewage in the Cook Islands.

Fiji Islands:
None held by SOPAC. However, there might be some data held by PWD Water and Sewage
Dept.

Papua New Guinea:
Port Moresby has the following datasets:

• A reservoir dataset with name attribute;
• Water (waterPS) and sewage (sewagePS) datasets. What exact features they

represent is unknown as attributes are coded;

•  A water tank dataset with identification (TAG) and name attributes;
•  A water trunk dataset with diameter and description attributes;
•  A borehole dataset that includes well-code and altitude attributes.

Port Moresby, Oro and Western PNG have drainage datasets. However, it is not clear
what their attributed values represent.

Confidential 2009

GNS Science Consultancy Report 2009/321

Samoa:
None held by SOPAC. GIS data is held by the Samoa Water Authority

Solomon Islands:
The city of Honiara appears to have a water pipeline dataset (Hpipes) with detailed attributes
including system network name, pipeline size, and date. Most of the attributes are coded
and, therefore, need a data dictionary to fully interpret, while some are not complete. There is
also a valves dataset (H_av) for Honiara with attributes of size, year installed and location.
Most attribute values for most features are not present. There are three other datasets that
appear to be related to the water network in Honiara (H_bf, H_fh and Hdrain). More
information is needed from SOPAC about the exact nature of these datasets.

Tonga:
There are data layers for water towers (owner, construction, capacity (not complete) and
comments), as well as Water tanks (Tank_id, Owner, Construction, Capacity), and Water
bores (Wellbore_id, Name, Type, Depth, Owner, Pump_type, Motor, Volume, PH). Not all
fields have been populated.

Tuvalu:
There is a water tanks dataset (funafeti_Tanks2_84) for Funafuti Island with attributes
including construction material, year installed, condition, size and capacity. Also, well
locations are available for Nukufetau, Nui and Niutao, but with no attributes.

Vanuatu:
Detailed water reticulation data are apparently held by a French company (UNELCO).

3.2.6 Power

Distribution Networks

The Power Distribution Network is anticipated to include 1) information on distribution circuits
operated either via underground cables or overhead lines, 2) substations, and 3) service
area. Relevant attributes for the risk assessment include the size and the material of the
cables and substations, the voltages, the year of installation, the relationship with the other
components of the grid or the typology. Information on replacement costs is also required.

Cook Islands:
Power generation stations, sub-stations, pillar boxes and power poles have been collected
for Aitutaki, Atiu, Mauke, Mitiaro, and Mangaia as part of a survey conducted under an
NZODA Outer Islands Water & Power Reticulation Feasibility Study
(

http://www.maps.gov.ck

), and are also available for Rarotonga.

Solomon Islands:
There is a dataset held by Solomon Islands Electricity Authority (SIEA) which contains power
poles, power lines, transformers, meters, switches and street lights for Honiara.

No data have been sighted for any other PIC’s although it exists at least for Tonga and parts
of PNG.

Confidential 2009

GNS Science Consultancy Report 2009/321

3.3 Physiography

3.3.1 Imagery

All high and moderate resolution image data available at SOPAC are listed in the SOPAC
Image Data Catalogue, including relevant meta-data, such as acquisition date, resolution,
sensor, and image type. These data are partly available online at www.sopac.org/map or
geonetwork.sopac.org. These data sets include:

Cook Islands:
There are high resolution satellite images available for the islands of Aitutaki, Atiu, Mauke,
Mitiaro, Mangaia, Manihiki, Pukapuka and Rarotonga. The latest imagery for Rarotonga
appears to be May 2009.

Fiji Islands:
There are various images of Viti Levu and Rotuma islands. Coverage of Rotuma is complete
but scattered for Viti Levu as most imagery is focused on the towns of Suva and Nadi.

Papua New Guinea:
There is high resolution imagery for the towns of Lae, Manam, Port Moresby, Sissano
lagoon, Vanimo and Wewak.

Samoa:
There are several high resolution images of Apia, one moderate resolution (3.5m) image of
the island of Savaii and various images (at different resolutions) of Upolu. New imagery will
most likely be available, taken shortly after the September 2009 tsunami.

Solomon Islands:
High resolution images of Gizo, Ranongga, Simbo, Vellalavella and a small part of
Kolombangara. Moderate resolution images of Savo and Honiara.

Tonga:
There are multiple sets of high resolution imagery for Tongatapu as well as high resolution
imagery for Niuas, and Vava’u. Quickbird imagery, 2003–2006 was purchased for the whole
of the Kingdom of Tonga part of the World Bank CERM Project.

Tuvalu:
There are high resolution images of Nanumanga, Nanumea, Niulakita, Nui, Nuitao and
Nukulaelae. There are also moderate resolution imagery of Nukufetau, Funafuti and Vaitupu.

Vanuatu: There are high resolution images of Port Vila. There is also moderate resolution
imagery of the entire island of Efate.

Maps for each country showing imagery extent plus tables showing metadata etc are given in
Appendix 4.

3.3.2 Topography

Table 7 shows available topographic data for each of the countries based on maps and
information available at SOPAC and would need to be verified in-country. In general, there
are National Topographic Map series available in all countries, with the exception of Tuvalu.

Confidential 2009

GNS Science Consultancy Report 2009/321

Contour intervals are rarely less than 20 m and the existing relief information available for the
low-lying coastal areas is inadequate to allow meaningful inundation modelling of tsunami,
storm-surge or river flooding hazard.

Map scales are usually 1:50 000 or larger, with the exception of Papua New Guinea, where
the national mapping scale is 1:100 000. In some places, (e.g. Honiara) larger scale maps
are available for urban centres providing more topographic detail. For the five capital cities
Apia, Honiara, Nuku’alofa, Suva and Vila large-scale digital terrain models (DTMs) with a
resolution of 5 m are available as part of the Pacific Cities Project. The only DTMs currently
available nationwide for all countries are either based on the topographic maps or SRTM
data with ground resolutions of 70 m and less. Vanuatu, Solomon Islands and Papua New
Guinea have recently been mapped using airborne radar, which might allow production of
higher resolution DTMs. Some very high resolution (dm) data might be available for
Fongafale, the main population area of Funafuti, Tuvalu (the data have not been reviewed by
SOPAC) and a small area around Avatiu Harbour, Cook Islands, which had been used for
storm-surge modelling (ADB 2005). Fiji has just commissioned a high-resolution aerial
survey with the aim to produce large-scale topographic maps for selected coastal areas with
1 m contours.

Table 7:

Existing Topographic data

Topography Maps

Country

Scale Contour

interval

Year Scanned

Raster

Contour
Layer

Cook Islands

1:25 000

10 -15 m

1989

√

Fiji 1:50

000

m 1989 √

√

Papua New Guinea

1:100 000

m 1965

√

Samoa 1:50

000

m 2000 √

√

Solomon Islands

1:50 000

1:10 000

20 m
1.5 m

1976
1971/6

√

Tonga 1:25

000

1.5

m 1975

Tuvalu 1:10

000

n/a

1971

√

Vanuatu 1:50

000

40 m

1968

Rarotonga Only

Partly unclear reference and data source, interpolated(?) 2m contour lines for coastal areas and 10 –15 m contour

For various areas around Fiji including Lautoka and Suva

National topographic map series

Various maps of Upolu and Savaii

Guadalcanal

Honiara Town West/East

Tongatapu group, otherwise 5 m contour interval

Efate

3.3.3 Bathymetry

Traditionally, nautical charts have been the source of depth information, although relatively
detailed in some areas, they are dedicated to safety in navigation and are therefore not fully
suited as fundamental background information for multiple applications in science,
engineering, geophysical and environmental studies, as well as coastal resource and

Confidential 2009

GNS Science Consultancy Report 2009/321

ecosystem management. In particular, the assessment of hazards due to storm tides or
tsunamis relies on a high-resolution, precise, and seamless bathymetric terrain model that
extends from the littoral zone (shallow water depths from about 10-20 m) to the shoreline and
topography for at least a distance of 20 m run-up inland.

While there is a growing list of near shore sites that have been surveyed by high-resolution
multibeam bathymetry systems (see Figure 5 below for example), there is an almost
complete lack of data for the shallow waters of the intertidal zone including reef flat and reef
crest areas. These shallow water areas present the primary focus of hazard studies, but are
the most poorly resolved and understood. The reef provides many ecosystem goods and
services such as the production of sediments and providing a protective barrier during
extreme storm wave conditions. Detailed shallow water bathymetry is also a vital component
in modelling the contribution of wave-induced radiation stresses on lagoon circulation,
sediment transport and water quality. It is therefore essential that the morphology and
shallow water bathymetry of the reefs and atoll rim is accurately determined. The nature of
shallow reef environments does not permit the use of traditional hydrographic survey
methods, and an airborne LiDAR surveys present the best option to fill this data gap.

Cook Islands:
There is a bathymetry line dataset for Penrhyn Island. However, there are no available
attributes. Data is available for Aitutaki.

Fiji Islands:
Only a small area of the southern coast of Viti Levu.

Papua New Guinea:
There are bathymetry line datasets for the Bougainville, East New Britain, East Sepik,
Madang, Manus, Milne Bay, Morobe, NIP, Oro, Western, West New Britain and West Sepik
areas.

Samoa:
Upolo and Savaii.

Solomon Islands:
Some bathymetry around Honiara, Marovo and Gizo.

Confidential 2009

GNS Science Consultancy Report 2009/321

Figure 5:

SW Pacific map showing bathymetry survey locations carried out under SOPAC/EU

funding 2003-2008

Tonga:
Tongatapu.

Tuvalu:
Bathymetry exists for Funafuti, Nanumanga, Nanumea, Niulakita, Nui, Nuitao, Nukulaelae,
Nukufetau, and Vaitupu.

Vanuatu:
Efate.

3.3.4 Soils/Geology

Physiographic data of the countries needs to be available including geology (to determine
soil class for earthquake shaking, amplification and liquefaction hazard). Available geological
soil data are summarised in Table 8. All larger, high-volcanic Pacific Island Countries have
Regional Geological Map series and mapping programmes, which cover major parts of the
country in scales of 1:50 000 to 1:250 000. Some of these data are available in vector format
(PNG and Fiji) and include stratigraphic information and major fault lines.

Digital soils data and borehole information suitable for assessing ground shaking
amplification exists for the capital cities of Fiji, Samoa, Solomon Islands, Tonga and Port Vila
through the SOPAC Pacific Cities database. Where possible, similar data will be created for
the heavily populated areas not covered by the Pacific Cities Database in the Cook Islands
(Avarua), Fiji (Nadi and Lautoka) and Papua New Guinea (Port Moresby, Mt Hagen and
Lae), Tuvalu (Funafuti), and Vanuatu (Luganville) from the available geological maps.

AIR Worldwide have indicated that in the absence of soils and geology data, topography can

Confidential 2009

GNS Science Consultancy Report 2009/321

be used to assign a soil class based on slope, as a surrogate.

Cook Islands:
Geology for Rarotonga based on geomorphic units (two units). Soils for Rarotonga mapped
by Landcare Research Ltd

Fiji Islands:
Regional geological and soil maps (Twyford and Wright 1965), including soil texture and FAO
classes for major islands, Viti Levu, Vanua Levu and Taveuni

Papua New Guinea:
Regional geological maps available in vector format.

Samoa:
There are soils and land use layers for the island of Savaii, however these have coded
attributes which need interpretation by SOPAC. There is also a geological map at 1:100 000
but it is not digital.

Solomon Islands:
Geological maps of some areas available at 1:100 000 and 1:50 000 and have been
scanned.

Tonga:
There is a soils dataset with soil code attribute for approximately half of the islands

Tuvalu:
None

Vanuatu:
A regional geological map at 1:100 000 – not digital

Table 8:

Existing Geology and Soils data

Geology Maps

Soil Maps

Country

Scale

Year

Scanned
Raster

Vector
Data

Scale

Year

Scanned
Raster

Vector
Data

Cook Islands

√

Fiji 1:50

000

1960/70s √

√

Papua New
Guinea

1:250 000

1970s

√

Samoa 1:100

000

1958

Solomon
Islands

1:100 000

1:50 000

1970/80s
1969

√

Tonga

Tuvalu

Vanuatu 1:100

000

√

Regional Geological Map series 1:50 000 for most parts of Viti Levu and Vanua Levu

Generalised vector data (1:250 000) available as national data set, more detailed geology (1:50 000) vectorised, but individual

sheets are not yet edited to form one comprehensive data set

Regional Geological Map series 1:250 000, e.g. Tingey and Grainger (1976): Markham <Lae>, PNG Sheet SB 55/10

Kear, D., Wood, B.L. (1959). The Geology and Hydrology of Western Samoa. New Zealand Geological Survey Bulletin, No.63.

Regional Geological Map series 1:100 000, e.g. BGS (1986): Geology of the New Georgia Group, Solomon Islands, report

MP/86/6, 7 sheets

Regional geological map series 1:100 000, e.g. Ash et al. (1974): Geology of Efate and offshore islands

Confidential 2009

GNS Science Consultancy Report 2009/321

PROJECT IMPLEMENTATION PLAN

4.1

Activity 1: Inception, engagement and liaison

In consultation with SOPAC, and taking into account existing work programme commitments,
it was decided that engagement will be done prior to data collection in each country. There
will be regular liaison with the ADB and the World Bank and its consultants during the
project, and particularly at progress meetings. An inception report meeting was held in Suva,
Fiji 2-4 December in Suva and included representatives from the World Bank and Air
Worldwide. From this meeting we have determined the attributes needed for loss modelling,
who will be responsible for collecting the information and have finalized a work plan
summarized in Table 9, with a detailed project plan given as Appendix 8. In addition, during
this time we presented the project to the Pacific GIS RS conference held at the University of
South Pacific, liaised with representatives form countries, and staff from the South Pacific
Commission.

Output: From this activity include this inception report and finalised work programme (see
Table 9 and Appendix 8 and 9)

4.2

Activity 2: Database design

Based on the review of data held by SOPAC, and missions to Fiji and the Cook Islands, and
discussions with World Bank project consultants, we have analysed the existing
infrastructure data and determined the data gaps. There is some building data in each
country and detailed data albeit dated for 5 major cities as part of the Pacific Cities database.
There are generally country-wide coverage of roads but limited attributes. Utilities data
appears to be limited but may well sit with the organisation responsible for the utilities in each
country. In some cases the lack of the metadata to fully describe dataset means that
interpretation of the features is required. We anticipate that many of these gaps will be filled
when we data is collected in each country.

As part of the inception meeting, the attributes required by AIR Worldwide were defined, and
it was determined which project would collect the information. The attributes to be collected
are given in table 1 of Appendix 6. Clearly there is a trade-off between the level of detail
collected and the number of buildings that can be surveyed in a given time. Discussion on
the level of detail required by AIR Worldwide has determined that some attributes have
higher priority than others.

A database structure will be designed prior to field data surveys, and the Nadi data trial
menus (Appendix 5) will be modified to accommodate the essential attributes. The database
design will be re-iterated during and after the data capture. It will be imperative to ensure that
any new data can be easily incorporated into to existing in country databases.

Output: From this activity a draft database design to store collected data, menus for data
collection, will be developed and field data collection devices purchased and programmed.
The database design will be refined following all field data collection.

Confidential 2009

GNS Science Consultancy Report 2009/321

4.3

Activity 3: Data Collection

Details of the data collection phase of the work are given in Appendix 9. Detailed building
data will be captured by this project. Larger infrastructure data such as dams, airports and
large industrial facilities will be captured by the World Bank project using high resolution
imagery. Existing transport, network and physiographic data will be collected from the
countries where it exists by this project. SOPAC will be the central repository for all data, but
data for each country will be held by the nominated country host organisation.

Output: Infrastructure data collected for each country. Regional database held by SOPAC

4.3.1

Data collection Phase 1

Data collection will start in the Cook Islands in February 2010 and will be carried out by staff
experienced in the use of handheld devices. The Cook Islands is a relatively small country
with existing, mature GIS datasets. The Cook Islands data capture exercise will be reviewed
and lessons learnt incorporated into training for the remainder of the team in Fiji in early
March before continuing with Phase 1 data collection in the Solomon Islands and Vanuatu.
Data will be collected by SOPAC staff in some parts of Fiji in this period.

Papua New Guinea has potential exposure greater than six of the other countries combined,
and during the inception meeting, it became apparent that Papua New Guinea data collection
should be a priority. However as PNG is large and highly-rural, data collection should be
concentrated where hazard was greatest (i.e. Lae, Madang, and Raubal). Data collection in
PNG has been bought forward into Phase 1 which extends this data collection period. Given
this, it is strongly recommended that in-country data capture involve additional SOPAC staff
and International Team personnel that can be exchanged with the listed SOPAC personnel
and consultants, due to the concentrated and prolonged Phase 1 field data capture period.

4.3.2

Data collection Phase 2

Following a progress report, review of phase 1 data collection and a short break, Phase 2 of
the data collection will start in Fiji in July, and simultaneously be carried out in Tuvalu. Once
Fiji is complete data collection will move to Samoa and Tonga, finishing towards the end of
September 2010.

4.3.3

Methodology for data capture in the field

From the field trial it became apparent that the preferred method of data capture is portable
handheld devices such as a small laptop or PDA. Preconfigured portable devices allow for
maps, building ID’s, GPS, images and reference information to be stored and allow for the
immediate upload of information into GIS files. Hence, field data will be captured using
integrated handheld devices with pre-programmed attribute menus. The field survey
equipment will be specified and purchased based on the requirements of the database
design but devices like the Trimble Juno SB (

http://www.trimble.com/junosb.shtml

SurveyLab IkeGPS (

http://www.ikegps.com/

) or TopCon GMS-2

(

http://www.topconpositioning.com/products/mapping-and-gis/hand-held-devices/field-

controllers/gms-2.html

), will be used for the field data collection, although the latter two may

be to expensive or not meet the field data capture requirements.

The field data capture trial in Nadi proved useful to get an estimate of the time to collect
building attribute data using handheld devices, the training effort required, and also exposed

Confidential 2009

GNS Science Consultancy Report 2009/321

other issues to be considered such as, the need for a structural engineer(s) to review
building classifications, and be involved in the training, and that there is insufficient time to
capture all asset data. From the data trial it is likely that more than ½ a day of training is
required to get accurate building assessments from previously unskilled people. With
additional training, an updated field guide, and the use of class values rather free entry fields,
it is expected that a significantly higher level of accuracy could be achieved for individual
buildings.

Data will be collected in partnership between the International Team, SOPAC and country
counterparts. Local counterparts will be integrated into field survey activities whenever
possible to utilize local knowledge and act as liaison in communities when necessary or
appropriate. The collection of infrastructure data will build capacity of local staff to capture
their own data which can be extended to less populated areas as required.

Laminated field guides with photos and descriptions of terms were beneficial for field workers
to gain familiarity with the building attribute terms and concepts, and to reference when
particularly challenging buildings were surveyed (Appendix 5). It is recommended that prior
to field surveys, time is allowed to interview local architects and construction companies, to
gain a better understanding of local building practices. Training from structural engineers and
construction companies based in Fiji is planned for the project team along with training in the
use of field equipment and has been incorporated into the workplan.

Digitizing of building plans (rooftop areas) from satellite imagery before conducting a survey,
as is currently being done by SOPAC as part of a World Bank project, allows for ground-
truthing during field data collection. Because of resolution, visibility and changes after
acquisition imagery is obtained, there will be some errors and assumptions when performing
initial rooftop digitizing. Reference maps of the areas surveyed were useful for navigation
and to provide building identification numbers for survey forms. An index map of the entire
area surveyed should be produced with appropriate labelling showing adjacent
corresponding maps should be placed on the reference maps to assist in navigation during
surveys. Hardcopy maps become less important when using portable devices with GPS and
integrated digital maps or imagery.

Access to buildings may be limited by gates, foliage, or other buildings. Such cases were few
during the trial, but there were some buildings that posed particular challenges for recording
attributes because of access or visibility.

Field survey areas can vary greatly in climate, cultural and social settings, and physical
terrain. Before a field campaign, field surveyors should consider appropriate precautions to
protect themselves, equipment, and field documents against adverse weather and potential
environmental hazards such as aggressive animals (dogs, insects, etc.), confrontational
residents, or unsafe areas (construction sites, poisonous plants, unsafe neighbourhoods).

Informational materials and/or press releases could be used to notify residents of the survey
work and to minimize confusion about the purpose of surveys and potentially disarm any
hostility from residents towards field workers.

Where there are no field surveys, attributes will be inferred based on similar styles of
buildings in areas where attributes are available. Census data from the countries or supplied
by SPC, which contains information on dwellings and households, will be used to populate

Confidential 2009

GNS Science Consultancy Report 2009/321

the building database in rural areas.

It is our understanding that digital Census data are available for most of the eight countries,
held by SPC or the statistics departments of each country. Vanuatu and the Solomon Islands
are to hold censes within the next six months or so. These data contain some household and
dwelling information that can be used to extrapolate building attributes, particularly in rural
areas, using basic assumptions about the dwelling types.

4.3.4

Hazard data collection

The International Team will also determine which earthquake and cyclone models and
climate change models are available and which ones are commonly used or most
appropriate for risk assessment. The models will be catalogued in a data repository and
reported. For this we will be reliant on the existing PDC Global Hazards Information Network
(GHIN) and the Pacific Disaster Net hosted by SOPAC. These contain both information on
hazard models and historical hazard event damage reports.

Output: Hazard model data collected for region and countries held by SOPAC

4.3.5 Physiographic

data

The International team will collect any imagery, topographic, bathymetric and soils/geology
data as available when in country.

Output: Catalogue of data stored at SOPAC

4.4

Activity 4: National and Regional Systems

Once data are collected they will be analysed to produce recommended “Pacific” vulnerability
classes based on construction type and materials, for example. Replacement costs will be
reported for each country which is likely to vary from country to country depending on the
availability of timber, sand for concrete, etc.

National databases, developed to hold hazard models and asset data for each country, will
be consolidated to form a regional database, most likely in the form of a Pacific “RiskScape”
or some equivalent and held by SOPAC and updated by them and the countries. However,
Risk Models are not likely to be included as this will be done by World Bank.

Currently the 8 countries run predominantly MapInfo GIS software, although ESRI ArcGIS is
also used. Some open source software is also used but it is not commonplace. All of the
countries have MapServers which serve the data via the internet. However, the status of
these servers varies (Appendix 7) – some have not been maintained, some have been
disconnected due to the internet costs, and some have failed. The hardware requirements of
each country to store and maintain there databases will be assessed during the data
collection phase.

The Project Team will design and deploy a GIS-based information system to store, manage
and serve the regional and national databases that are developed under this project. The
system’s proposed architecture is shown in Figure 6. It is anticipated that SOPAC will host
the system. Some of the scenarios for such a system include:

•

Act as a central repository for the regional and national databases;

Confidential 2009

GNS Science Consultancy Report 2009/321

•

Serve as a data store for running RiskScape models;

•

Publish a browser-based map viewer to view the GIS data online;

• Disseminate the GIS using various open standards for GIS data services (e.g.

WMS, WFS, KML) − these data services can be consumed by desktop GIS
clients or as general web services;

•

Support periodic data updates from national GIS departments; and

•

Support exporting GIS data for extraction to other systems.

Concurrent with the data collection effort, the International Team will:

•

Develop a Concept of Operations (CONOPS) for the system. The CONOPS will
formalize use cases, information flow, and high-level functional requirements;

•

Develop a system architecture and hardware specification;

•

Procure any necessary hardware and software;

•

Perform the system installation and confirmation at SOPAC;

•

Develop Standard Operating Procedures (SOPs) addressing data management,
updates and dissemination as well as system maintenance; and

• Create user guides and training materials for system installation and

maintenance.

Some specific technologies are suggested in Figure 6, but the software components
ultimately selected will balance SOPAC’s operating environment, the project budget,
SOPAC’s existing infrastructure and skills, and the overall project requirements.

Figure 6:

GIS Options for the inventory database

The architecture will focus on standards for GIS system interoperability, such as the OGC
standards WMS, WFS, and KML (

http://www.opengeospatial.org

), regardless of the software

chosen. Using these standards in place of proprietary data structures, communications, and
application programming interfaces (APIs) will prevent a “vendor lock-in” situation, where a
customer becomes so dependent on a given vendor’s products that it is too costly or difficult
to switch to other solutions. These customers often find themselves buying additional
software or extensions just because it interacts with the proprietary software they already
own. It is anticipated, however, that some of the components, especially those used to edit

Confidential 2009

GNS Science Consultancy Report 2009/321

and update GIS database and metadata files, may be commercial-off-the-shelf (COTS)
products, such as ArcGIS or MapInfo, which must be licensed from a vendor. These products
may offer functionality that doesn’t exist elsewhere, or enable an enhanced workflow that
saves time. Where possible, a free or low-cost alternative will be identified that can fulfil a
given need.

Each nation will have the option to create a similar GIS infrastructure to host their national
database. The installation instructions and any source code generated for the SOPAC-
hosted solution will be made available to the nations. Each nation can then create a solution
that is sized for their capacity and needs. At the most basic level, a GIS department could
have shape files available on a walk-up computer for users to view/edit the data with free
desktop GIS software. For nations with more resources and better Internet connectivity, they
can replicate the online system delivered to SOPAC for their own nation. The nations may
also have some existing infrastructure, into which they integrate the new data generated by
this project. The budgets available to replicate these solutions at the national level could vary
greatly from one nation to another, further underscoring the need for free alternatives to any
COTS software chosen.

Standard Operating Procedures (SOP) will be developed to facilitate two-way updating of the
GIS data. The following considerations will be accounted for:

•

The updates may be generated by SOPAC or the individual nations;

• The national level GIS databases may have more detail than is needed at the

regional level, or may include attributes that are not relevant to the regional risk
assessments. Procedures will ensure that the attribute fields and values map
between the originating source and the project database and vice versa;

• SOP’s will pay special attention to use file/data formats that are easily

interoperable with various systems;

• The data must be re-projected into the coordinate system supported by the

recipient; and

• The “database update” SOP must account for bulk updates and for situations

where just a few edits are made.

The RiskScape tool aids disaster management in that scenarios can be generated quickly.
The hazard and exposure data collected as part of this TA would have this end-use in mind,
and ultimately a Pacific Risk modelling tool could provide the vehicle for regional and country
risk analyses. In this way SOPAC could interrogate any asset data and analyse impacts from
hazard events. If suitable, such a system could be based in each country such that they can
analyse their own data, run simplistic analyses, receive upgraded hazard and fragility
functions as they are developed, and easily incorporate population casualties and indirect
losses.

Outputs:

National and regional infrastructure databases, connected to existing GIS
systems. Standard Operating Procedures (SOP) developed to facilitate
updating of the GIS data.

4.5 Activity

Training

We have met with the president of the Fiji Institute of Engineers (FIE). In conjunction with the
Institute of Professional Engineers New Zealand Inc. (IPENZ), the FIE are launching a South
Pacific Engineers Association (SPEA) in March 2010. We have tentatively agreed to
contribute to this launch and will arrange seminars with Fiji structural engineers for the

Confidential 2009

GNS Science Consultancy Report 2009/321

TA/SOPAC team to gain a better understanding of Pacific building construction. This will
precede our field surveys in seven of the 8 countries. Training of in-country staff in field data
capture will occur at the beginning of the mission into each country and with ongoing
supervision during data capture. This training has been incorporated into the revised work
programme.

The FIE have also indicated that they will provide structural engineering contacts for each of
the countries. They will also provide replacement cost and construction cost, as well as
building codes/guidelines for each country where available to the project and AIR Worldwide.

Training in the use of the asset database and regional risk system will occur in March 2011.
In SOPAC’s experience of providing capacity building to PICs is that it is more sustainable
and efficient to target training in GIS applications and DRR at a national level, in terms of
people trained/training costs and impact created. SOPAC and the TA team will undertake in–
country training once the databases have been completed. This would be subject to ADB
approval to use the provisional funds allocated for training to cover travel costs of SOPAC
and International Team trainers. Otherwise regional training will be held in Fiji

It is proposed that a presentation of the project and its planned interventions will be given at
next Pacific DRM Platform meeting in 2010. Further, project results and achievements will be
disseminated at a side-meeting of the Pacific DRM Platform meeting in 2011 to senior
government officials (CEOs for Finance/Planning and disaster management) and
international Partners.

Output: In-country staff trained in collection of field data and database system.

Confidential 2009

GNS Science Consultancy Report 2009/321

Table 9:

Proposed workplan

Activity

Name

Description

Who

Duration

Output

Project Start 21 Sep 09

Initiation report for presentation at WB meeting Istanbul

Inception Mission
(28 Sep 09–24 Nov 09)

Engage with SOPAC and plan country engagement. Introduce

project at SOPAC Annual session and FEMM meeting in Cook
Islands. From this the detailed project plan will be formulated.
Nadi Data trial

Int’l TA team
SOPAC

10 weeks

Country government engagement

Analyse of existing data.

Inception report 24 Nov 09

Inception report by 24 Nov 2009
Inception Meeting Fiji, Dec 2009
Inception report finalised and approved by ADB Dec 2009

Database Design
(26 Oct 09–28 Jan 10)

Define level of infrastructure data that can be obtained and feasibly

collected and finalise attributes. Develop database design and
structure. Specify and purchase field survey equipment.
Document and advise countries. Revised and finalised in Oct 10
once data collection complete.

Int’l TA team
SOPAC
Countries

12 weeks

Draft database design document

Guidelines for capturing field data

Specification, tender and purchase field equipment

Collect Data Phase 1 –

see Appendix 8

(11 Feb 10–21 May 10)

TA team Training Fiji
(4 Mar 10–8 Mar 10)

Collect data in Cook Is, Vanuatu, Solomon Is and Papua New

Guinea. SOPAC and in-country staff to gather data using
integrated PDAs etc. Utilize foot prints captured from
aerial/satellite imagery. Involve policy makers in stakeholder
consultations and meetings.

Training in building construction and field capture devices for the TA

team in Suva in association with launch of South Pacific
Engineers Association.

SOPAC
Countries
Int’l TA team
Geoscientist

Int’l TA team
SOPAC
Structural engineer

subcontract

13 weeks

Data captured in 4 countries

Revised database design

Database hardware requirements determined

All TA team trained in use of field equipment, and in building

construction nomenclature. Reference material updated

First progress report 31 May 10

Progress report and meeting

Collect data (Phase 2)
See Appendix 8
(12 Jul 10–24 Sep 10)

Capture data in Tuvalu, Fiji, Tonga and Samoa

SOPAC
Countries
Int’l TA team
GIS Specialist

11 weeks

All field data collected

Database hardware requirements determined

Mid term report 29 Oct 10

Mid-term report and meeting

National and Regional

database
development

(1 Nov 11–28 Feb 11)

Specify in Country database hardware needs and purchase

Develop national databases, populate and consolidate

encompassing hazard and vulnerability data. Recommend
classes. Report replacement costs.

Develop consolidated database system and/or Pacific Risk Develop

web-based system for loading national updates to the regional
database

Build database of existing hazard models

Int’l TA team
SOPAC

16 weeks

Database structure developed and documented for each

country.

Regional system developed and documented

Regional System and

in-country training

(1 Mar 11–13 May 11)

Train Fiji and SOPAC in the use of the software design user

interface

Install Regional system

Install country systems and train

Int’l TA team
SOPAC

10 weeks

Regional system installed in SOPAC and staff trained to use.

Training modules and documentation for in-country training
prepared

Hardware and database installed in each country. Country

counterparts trained in use and update of database.

Training report.

Final DRAFT report - 30 Jun 11

Final Draft report due. Meeting 2 Aug 11 to discuss.

Final report - 16 Aug 11

Project Closed - 30 Sep 11

Confidential 2009

GNS Science Consultancy Report 2009/321

5 ACKNOWLEDGEMENTS

This report has been prepared by Phil Glassey and Dave Heron (GNS Science), Chris
Chiesa (PDC) and Steven Clegg, Joy Papao, Atu Kaloumaira of SOPAC. Michael Bonte-
Grapetin, Litea Biukoto and Susan Vocea (SOPAC) have also contributed significantly. The
authors also acknowledge the input of Edy Brotowisoro (ADB), Olivier Mahul and Francis
Ghesquerie and Stuart Gill (World Bank), SOPAC, Phil Bright (SPC), Paolo Bazzurro and
Bishwa Pandey (AIR Worldwide), Pratarp Singh (president FIE), the representatives of the
various government agencies that the team met with, and USP and its students. This report
has been reviewed by Andrew King and Dr. Terry Webb of GNS Science. Delia Strong
assisted with some of the diagrams.

6 REFERENCES

ADB, 2005:

Climate Proofing – A risk based approach to adaptation

, Pacific Studies Series,

Publication Stock No. 030905.

Blong, R., 1994: Natural Perils and Integrated Hazard Assessment in Fiji; Final Report for
Queensland Insurance (Fiji) Limited, National Insurance Company of New Zealand and Fiji
Reinsurance cooperation.

Biukoto, L., Swamy, M., Shorten, G. G., Schmall, S., and Teakle, G., 2001a: Pacific Cities
CD, Nuku’alofa. GIS Hazards Dataset, Version 1.1. SOPAC Data Release Report, 3.

Denham, David and Warwick Smith, 1993:

Earthquake hazard assessment in the Australian

and Southwest Pacific Region,

Annali di Geofisica, 36, 3-4, 27-39.

Glassey P.J., 2009: Pacific Exposure Database – Project Initiation report. ADB TA 6496-
REG: Regional Partnerships for Climate Change and Disaster Preparedness. GNS Science
Client report CR 2009/256.

Glassey, P., Heron, D., Ramsey, D., and Salinger, J., 2005: Identifying Natural Hazards and
the Risks they pose to Tonga. Study Report 4: GeoSource Tonga. Cyclone Emergency
Recovery and Management Project: Component B2: Land Hazards and Information
Management.

Johnson, W., Blong, R., and Ryan, C. (compilers), 1995: Natural Hazards Potential Map of
the Circum Pacific Region (1995) 1:10 M.

Jones T., 1998:

Probabilistic earthquake hazard assessment for Fiji

. Australian Geological

Survey Organisation AGSO Record 1997/46, 1998

Ripper, I.D. and Letz, H., 1993

: Return periods and probabilities of occurrence of large

earthquakes in Papua New Guinea

. Papua New Guinea Geological Survey Report

93/1.

Shorten, G., et al. (2003): Catastrophe Insurance Pilot Project, Port Vila, Vanuatu: SOPAC

Confidential 2009

GNS Science Consultancy Report 2009/321

Joint Contribution Report 147.

Shorten G., Shapira A., Regnier M., Teakle G., Biukoto L., Swamy M., and Vuetibau L.
(Compilers), 2001: Site-specific earthquake hazard determinations in capital cities in the
South Pacific. Second Edition.

SOPAC Technical Report

. 300.

SOPAC, 2008:

Inventory of Geospatial Data and Options for Tsunami Inundation & Risk

Modelling – Pacific Island Country Summary

, SOPAC Miscellaneous Report 657.

SOPAC, 2008a:

Inventory of Geospatial Data and Options for Tsunami Inundation & Risk

Modelling – Tonga

, SOPAC Miscellaneous Report 651.

SOPAC, 2008b:

Inventory of Geospatial Data and Options for Tsunami Inundation & Risk

Modelling – Solomon Islands

, SOPAC Miscellaneous Report 654.

SOPAC, 2008c:

Inventory of Geospatial Data and Options for Tsunami Inundation & Risk

Modelling – Fiji Islands

, SOPAC Miscellaneous Report 655.

SOPAC, 2008d:

Inventory of Geospatial Data and Options for Tsunami Inundation & Risk

Modelling – Tuvalu,

SOPAC Miscellaneous Report 656.

South Pacific Disaster Reduction Programme (SPDRP), 2002: Suva Earthquake Risk
Management Scenario Pilot Project (SERMP), Summary Report Part 1.

SOPAC

Joint Contribution

139.

Suckale, J. and Grünthal, G., 2009: A probabilistic seismic hazard model for Vanuatu,
Bulletin of the Seismological Society of America, 99, 4, 2108-2126.

Suckale, J., Grünthal, G., Regnier, M., and Bosse, C. (2005): Probabilistic Seismic Hazard
Assessment for Vanuatu, Geo Forschungs Zentrum Potsdam, Scientific Technical report
STR 05/16, ISSN 1610-0956.

Twyford, I.T. and Wright, A.C.S., 1965: The Soil Resources of the Fiji islands, 2 Volumes,
Government Printer, Suva, Fiji.

Appendix 1: Country Contacts – ADB Pacific Exposure Database Project. TA- 6496-REG

Country

First
Name

Last
Name

Position Ministry

Location Address

City

Cook Islands Garth

Henderson Aid Manager

Ministry of Financial and
Economic Development

PO Box 120

Avarua,
Rarotonga

Fiji John

Prasad

Acting Permanent
Secretary for Finance

Ministry of Finance

Government Buildings

PO Box 2212,

Suva

Papua New
Guinea

Joseph Lelang The

Secretary

Department of National
Planning and Monitoring

Level 3, Vulupindi Haus

PO Box 631

Waigani,
N.C.D.

Samoa

Hinauri

Petana

Chief Executive Officer

Ministry of Finance

PO Box 30017 Apia

Solomon
Islands

Sadrach Fanega

Permanent Secretary of
Finance and Treasury

Ministry of Finance

PO Box 26

Honiara

Tonga Aisake

Eke

Secretary of Finance
and Economic Planning

Ministry of Finance and
National Planning

PO Box 87

Nuku’alofa

Tuvalu Temate

Melitiana

Assistant

Secretary

Ministry of Finance and
Economic Planning

Headquarter Division

Vaiaku,

Funafuti

Vanuatu George

Maniuri

Director

General

Ministry of Finance and
Economic Management

Private Mail
Bag 9058

Port Vila

Appendix 3: Example letters sent to country Government officials

10 November 2009

Garth Henderson
Aid Management Division
Ministry of Financial and Economic Management
PO Box 120
Rarotonga
Cook Islands

Dear Sir

ADB RETA 6496: REGIONAL PARTNERSHIP FOR CLIMATE CHANGE ADAPTATION AND
DISASTER PREPAREDNESS – PACIFIC EXPOSURE DATABASE

GNS Science International Ltd, a subsidiary of GNS Science, has been awarded the above Asian
Development Bank (ADB) contract to develop a national and regional exposure databases for the
Pacific. I understand from the ADB that you have been nominated as the focal point for the Cook
Islands in relation to the project.

Along with representatives of the ADB and World Bank, I visited Rarotonga from the 27–30
October and discussed the project with Cook Island government officials including representatives
from the Ministry of Infrastructure and Planning (MOIP), National Environment Service (NES),
Ministry of Agriculture, Emergency Management Cook Islands (EMCI), the Prime Ministers Office,
Central Policy and Planning and the Police. I noted that you were out of the country during our
mission to Cook Islands.

From this visit we determined that the Cook Islands already has significant amounts of relevant
data and we want to extend the existing data for use in the project and for the benefit of the Cook
Islands. The development of the database involves the collection of data on building construction
and infrastructure in each of the countries. My team is planning to be in the Cook Islands for about
two weeks in late February 2010 to undertake field surveys. To expedite the data collection we will
require local assistance and have discussed this with the MOIP and Environment staff which you
have made a commitment to providing. The Secretary of the Ministry of Infrastructure and Planning
expressed interest in supporting the project and along with the National Environment Service
agreed to make staff available for the field survey.

We intend that our visit in February will firstly involve a briefing of all interested parties on the
project. We will then train the Cook Island counterparts and undertake the field survey with them.
The data collected will be left with the Cook Islands government and we recommend it be
incorporated into the GIS held by MOIP. Follow up training is planned for later in the project

We will be in contact with you once our planning has advanced. Please advise if you are happy for
us to liaise with MOIP regarding the logistics. If you require any further information regarding the
project please contact me via e-mail at

p.glassey@gns.cri.nz

, phone +64 3 4799684 or fax at +64

3 4775232.

Yours sincerely

Phil Glassey
Project Leader

Cc

Edy Brotoisworo, Senior Safeguards Specialist, Pacific Department, Asian Development Bank, 6
ADB Avenue, Mandaluyong City, 1550 Metro Manila, Philippines

Michael Bonte-Grapetine, SOPAC Secretariat, Private Mail Bag, GPO, Suva, Fiji.

Mike Mitchell National Representative to SOPAC and Secretary, Ministry of Foreign Affairs &
Immigration, PO Box 105, Rarotonga Cook Islands

Charles Carson Emergency Management Cook Islands, Office of the Prime Minister Private Bag,
Avarua, Rarotonga, Cook Islands

Taukea Raui, Secretary, Ministry of Infrastructure & Planning PO Box 383, Te Atukura, Avarua,
Rarotonga, Cook Islands.

TECHSEC:

November

2009

National Representative to SOPAC <CK, FJ, PG, SB, TO, TV, WS>

Dear

WB/ADB/SOPAC Initiative on Risk Exposure Databases for Disaster Risk Reduction and
Risk Financing

The first phase of the World Bank (WB) Risk Financing Initiative for the Pacific developed preliminary risk profiles for
8 Pacific Island countries. These profiles once fully developed and endorsed by the subject Pacific countries will help to
inform catastrophe risk financing options for future consideration. Under Phase 2 the risk models will be refined and the
initiative is being extended to all 14 Pacific ACP countries and Timor Leste.

Further, the Asian Development Bank (ADB) is contributing to the collection of hazard, vulnerability and risk
information to facilitate the development of risk databases in the initial 8 Pacific Island countries. The information and
data gathered will help to inform sound decision-making for disaster risk reduction and climate change adaptation
initiatives.

To facilitate the work under these initiatives a series of in-country missions are being planned with GNS Science and
SOPAC for data collection and training. In this connection a mission to <country> from to <dates> has been planned.
During the mission we would like to meet with you and other relevant stakeholders to discuss how best to move
forward.

We also aim to collect data/information relevant to the development of the national risk database and humbly request
the use of existing national datasets for these initiatives. We are aware of the concerns around data ownership and
would like to assure you that the World Bank and ADB have acknowledged that these would remain the property of all
the countries concerned.  In turn, we have also had agreement that any data or information products from these
initiatives will be released to countries.

We look forward to your favourable consideration of these initiatives and your continued support.

Yours sincerely,

Cristelle Pratt
Director

CC
ADB/WB focal point in-country
NDMO
Relevant Technical Agencies

Appendix 4: Satellite imagery data and coverage maps held by SOPAC

Overview of available imagery, resolution, population covered and other available GIS data: P=Places
(Towns, Villages, Settlements), R=Roads, C=Critical facilities, U=Utilities, A=Agriculture, B=Buildings

Countries

Islands/Areas

Imagery available at
SOPAC and resolution

Percent of
population

Asset vector
data available

Cook Islands

Rarotonga
Aitutaki
Atiu
Mangaia
Manihiki
Mauke
Mitiaro
Pukapuka

1m
2.4m
0.6m
1m
4m
0.6m
0.6m
0.6m

72%
10%
3%
3%
2%
2%
1%
3%

P, R, C, U

Fiji

Greater Suva
Nadi
Nausori
Labasa
Ba
Lautoka
Lami
Sigatoka
Coral Coast
Rotuma

0.6m
0.6m
0.6m
0.6
0.6m
0.6m
0.6m
0.6m
4m
0.6m

21%
5%
6%
4%
3%
6%
2.5%
1%
n/a
0.2%

P, R, C, U ,A,
B(Suva)

Papua New
Guinea

Port Moresby
Lae
Kavieng
Vanimo
Wewak
Manam

0.6m
0.6m
0.6m
0.6m
0.6m
0.6m

5%
2%
0.2%
0.2%
0.4%
>0.1%

P, R

Samoa

Apia
Upolo

0.6m
1 m

35%
65% incl. Apia

P, R, C, B

Solomon
Islands

Honiara
Gizo
Rannonga
Simbo
Savo

0.6m
1m
0.6m
0.6m
2.4m

12%
1%
<1%
<1%
<1%

P, R, B

Tonga

Tongatapu Group
Vavau Group
Ha’apai Group
Nuia’s

1m, 2.5m
2.5m
2.5m
2.5m

70%
15%
10%
5%

P, R, C, B, A

Tuvalu

Funafuti
Vaitupu
Nanumanga
Nanumea
Niulakita
Niutao
Nui
Nukufetau
Nukulaelae

1m
0.6m
0.6m
0.6m
0.6m
0.6m
0.6m
0.6m
0.6m

47%
17%
6%
7%
0.4%
7%
6%
6%
4%

P, R, U

Vanuatu

Port Vila
Luganville

0.6m
0.6m

15%
5%

P, R, C, B (Vila)

Appendix 5: Attribute forms and reference material for field trial − Nadi

Name

Date

Camera

Photo No

Building ID

Location

GPS No

GPS Waypoint

Use Main

use

Subsidary use

residential

house

(single household in a single building)

flats

(multiple households in single
building)

fale

(open house))

shed

(building that is not used for sleeping)

commercial/industrial

commercial

(shops, offices, restuarant)

industry

(factory, workshop, warehouse)

accommodation

(hotel, motel, resort)

public/communal

government

(government department buildings)

public services

(bus station, boardcasting, libraries,
etc)

church, temple etc.

community facilities

(meeting halls)

education

(high school, university, kindergarten)

critical

facilities

police station

fire station

defence

(army or navy building)

health services

(health clinics, hospital, chemist)

communication

(telephone exchange, cellphone)

utility

(electricity, water, gas)

hazardous

facility

petrol station

chemical store

other

(specify ……….……………………….…)

unknown

Age

No of storeys

Roof pitch

Flat

Low (1-25)

Roof shape

hip

Moderate (25-40)

gable

Steep (>40)

arch

other

Roof material

concrete

sheet metal

sheet

fibre-cement
sheet

metal tile

tile

heavy tile

wooden shakes

traditional material

other

(specify ………………………………………….…………..)

unknown

Foundation

slab

Bracing

timber wall

wooden/conc piles

concrete wall

wooden poles

masonry wall

concrete columns

other (specify)

steel columns

load bearing wall

Wall materials

concrete

Wall structure

timber frame

fibre-cement sheet

wooden poles

fibre-cement board

concrete columns

timber

steel columns

masonary

load bearing wall

metal

unknown

traditional material

other

(specify ……………………………………….)

unknown

Windows

<75% wall

Shutters

present

>75% wall

brackets

open space

none

unknown

Parapet/tower

Defects

none

minor

cracks

Concrete cantilever

major cracks

minor

rot

Plan shape

regular

major rot

irregular

poor construction

unknown

< 50 sqm

Min floor
height (m)

Base floor area
(sqm)

50-100 sqm

100-200 sqm

200-400 sqm

> 400 sqm

(sqm)

50-100 sqm

100-200 sqm

200-400 sqm

> 400 sqm

Use Main

use

Subsidary use

residential

house

(single household in a single building)

flats

(multiple households in single
building)

fale

(open house))

shed

(building that is not used for sleeping)

commercial/industrial

commercial

(shops, offices, restuarant)

industry

(factory, workshop, warehouse)

accommodation

(hotel, motel, resort)

public/communal

government

(government department buildings)

public services

(bus station, boardcasting, libraries,
etc)

church, temple etc.

community facilities

(meeting halls)

education

(high school, university, kindergarten)

critical

facilities

police station

fire station

defence

(army or navy building)

health services

(health clinics, hospital, chemist)

communication

(telephone exchange, cellphone)

utility

(electricity, water, gas)

hazardous

facility

petrol station

chemical store

other

(specify
…………….……………………….…)

unknown

INTRODUCTION

This document is to be used as a field guide for the classification of building and roof types for the
ADB RETA 6496 Regional Partnerships for Climate Change Adaptation and Disaster Preparedness
and the World Bank Pacific Catastrophe Risk Financing Initiative Phase 2 projects. This guide
contains a glossary of the technical terms that will be used during the data collection phase of the
projects. This guide is to be used in conjunction with the building data entry sheets and/or the pre-
programmed data dictionaries for mobile digital devices, such as PDAs. Also contained within this
document is a summary of the overall observations made during the Nadi building survey which
was carried out in October this year.

MAPPING BUILDINGS WITH REMOTE SENSING

High resolution satellite imagery is utilized to identify buildings and to create digitized polygons of
estimated roof areas. Georeferenced images are loaded into a GIS and buildings are identified
individually by a GIS user. Polygons representing building roof areas are constructed based on the
best estimate taken from the imagery. Satellite imagery may not reflect recent changes such as new
construction or damages to existing buildings. Therefore, digitized rooftop polygons are verified
during field survey work and corrections are made accordingly.

Buildings are assigned a unique identification number (Building ID#) which becomes the primary
reference for cataloguing and for attaching attributes collected during building and roof
classification surveys.

GLOSSARY

ROOF TYPE
All existing roofs over occupied areas are described by the pitch, shape and material of the roof. In
the case of more than one roof type is used to construct the roof all pitches, shapes and materials
will be listed. This excludes different roof types for verandas or entries and/or tower like features as
in the picture below of a flat roof, which will be listed as parapet/tower.

Roof Pitch
The roof pitch refers to the angle of the roof.

Term Description

Example

Flat

A roof that is horizontal (0°)

Low

A roof that is nearly
horizontal or at a low angle
(1°- 25°)

Term Description

Example

Sheet Metal

Roof made from
corrugated iron or metal
sheets. It does not break
but often rusts and bends.

Fibre-cement
Sheets

Similar to sheet metal
except that it does not rust
or bend. Hard to
differentiate from sheet
metal from a distance.

Metal Tiles

Thin tiles, usually with
stones glued on. Typically
stones come off, may rust

Heavy Tiles

Tiles are made of concrete
or clay and are thicker
than metal tiles. Look at
ridge line for concrete
plaster

Wooden
Shakes

Thin wood tiles, generally
grey

Traditional
materials

Roof material consists of
thatching or other
traditional materials

Other

Any other materials.
Please specify

FOUNDATION
The foundation is a part of a building, usually below the ground, that transfers and distributes the
weight of the building onto the ground.

Term Description

Example

Concrete Slab

Concrete slab sits on
ground and building
walls sit on slab

Wooden/Concrete
Pile

Timber or concrete post
less than 1m tall that
supports the floor

BUILDING TYPE
More than one wall material and construction type can be present in one building and will be noted.
In the case this can be used to distinguish between entire buildings, these will be assessed separately
and more than one form will be filled.

Wall Material
This refers to the material used to construct the walls of the building’s occupied levels.

Term Description

Example

Concrete

Concrete used to
construct walls. The
concrete may or may not
be reinforced

Fibre Cement
sheets

Sheets of fibre-cement
used in walls, normally
has thin timber over
vertical joints

Fibre Cement
Board

Looks like timber but
often has strong visible
grain pattern and joiners
at end of boards

Masonry

Brick or block used in
walls.

Timber

Timber planks or sheet
used for walls

Metal Sheet

Sheet metal used for walls

Traditional
Material

Wall constructed from
thatching or other
traditional materials

Term Description

Example

Other

Any other material not
mentioned above

Building Structure
This refers to the building structure above the foundation.

Term Description

Example

Timber
Frame

This is the assumed structure
for wall materials comprising
of metal, timber or fibre-
cement

Wooden
Pole

Wooden poles hold up the
roof

Concrete
Column

Building has concrete
columns to support the roof

Steel
Column

Usually used for factories
and warehouses without
internal walls

Load
Bearing
Wall

Concrete walls bear the
weight of the roof without
concrete columns

Unknown -

Windows
This section looks at the percentage of wall space with windows along a wall with the most
windows.

Term Description

Example

<75%
wall

Less than 75% of the wall
with the most windows is
covered in windows

>75%
wall

More than 75% of the wall
with the most windows is
covered in windows