Geographic Information

System for Coastal Hazards -

Application to a pilot site

in

Final Report

BRGM/RP-55553-FR June, 2007

Geographic Information System for Coastal Hazards - Application to a pilot site in Sri Lanka

Final Report

BRGM/RP-55553-FR June, 2007

M. Garcin, B. Prame, N. Attanayake, U. De Silva, J.F. Desprats, S. Fernando, M. Fontaine, D. Idier, N. Lenotre, R. Pedreros, C.H.E.R. Siriwardana

Checked by: Approved by: Name: Oliveros C. Name: Modaressi H. Date: Date: Signature: Signature:

BRGM's quality management system is certified ISO 9001:2000 by AFAQ

Keywords: Sri Lanka, Coastal risks, Coastal hazards, Geographic Information System, Remote sensing, Tsunami, Storm surge, Coastline erosion.

In bibliography, this report should be cited as follows:

Garcin M., Prame B., Attanayake N., De Silva U., Desprats J.F., Fernando S., Fontaine M., Idier D., Lenotre N., Pedreros R., C.H.E.R. Siriwardana (2007) – A Geographic Information System for Coastal Hazards - Application to a pilot site in Sri Lanka (Final Report). BRGM Open file BRGM/RP-55553-FR, 124 p., 94 figs, 1 DVD.

© BRGM, 2007. No part of this document may be reproduced without the prior permission of BRGM.

GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Synopsis

he project « Geographic Information System for Coastal Hazards - Application to a T pilot site in Sri Lanka » was funded by the French Government, Ministère des Affaires Etrangères (MAE) and the BRGM. The contract between the MAE and the BRGM has been signed on November 15th, 2005, under the reference SME/PAF/SUB EJ/2005/2148.

This project aims to design and develop an effective tool (coastal GIS) in order to prevent and mitigate the impact of natural hazards, as well as to optimize preparedness to a potential crisis. The coastal Geographic Information System (coastal GIS) has been built using homogeneous data on the land/sea interface. Relevant parameters can be cross-referenced so that the exposure of coastal populations to natural hazards can be studied. The GIS is a decision-support tool for policy-makers and political and economic decision-makers in the field of risk management and protection.

The project's main objectives are: - to characterise the different coastal risks in the pilot site; - to contribute to the prevention and reduction of coastal risks, as part of a coastal defence plan; - to identify the probability of their occurrence and define the most exposed areas of the pilot site; - to provide managers with useful data and information on coastal areas, which can be directly used for the safety of people and goods as well as for including environmental constraints in territorial development planning at an early stage; - to build capacity in Sri Lanka’s technical departments (by transferring technology and know-how) to enable them to take over data input and use of the coastal risks GIS (cross-referencing of parameters); - to provide managers of the land/sea interface with guiding principles for preventive policies, as recommended by international bodies (information, training, regulations, etc.).

The project has been divided in two successive phases, with ongoing training throughout these two phases. Phase 1 of the project focused on the configuration and architecture of the coastal GIS as well as on the definition of a pilot site and of the working program. Phase 2 focused on the creation of the GIS including field works.

This project has been conducted in collaboration with the Geological Survey and Mines Bureau of Sri Lanka (GSMB, co-leader). Other organizations have also contributed in their own areas of expertise: - Coastal Conservation Department (CCD);

BRGM/RP-55553-FR – Final Report 3 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

- National Science Foundation (University of Moratuwa – Dept of civil engineering); - National Aquatic Ressources Research and Development Agency (NARA); - Meteorology Department.

This report describes the various tasks carried out during the project and illustrates the GIS developed within the framework of this project. The DVD provided with this report includes the Coastal GIS and all the data acquired during the project. The GIS has been created with the ArcGIS 9.1 software from ESRI.

4 BRGM/RP-55553-FR – Final Report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Contents

1. The project ...... 11

1.1. GENERAL OBJECTIVES ...... 11

1.2. PROJECT STRUCTURATION ...... 12 1.2.1. Phase 1: Choice of the Pilot site and definition of the GIS ...... 13 1.2.2. Phase 2: GIS development and integration of the data ...... 13

1.3. PROJECT ORGANIZATION ...... 14 1.3.1. Project partners...... 14 1.3.2. Project organization ...... 14

1.4. WORKING PROGRAM ...... 16 1.4.1. First mission...... 16 1.4.2. 2nd to 4th mission ...... 16 1.4.3. Training in France ...... 17

2. Pilot site...... 19

2.1. PRESENTATION...... 19

2.2. COASTAL PROCESSES, HAZARDS AND RISKS...... 24 2.2.1. Definitions ...... 24 2.2.2. Coastal processes characterization ...... 25

2.3. COASTAL MULTI-HAZARDS MAPPING: THE PROBLEM ...... 30 2.3.1. Proposal of the creation of a composite marine inundation hazard .....30 2.3.2. Classification of coastal hazard ...... 30 2.3.3. Two possible approaches of the cartography of marine inundation hazard...... 30 2.3.4. Evaluation of the elements at risk ...... 31 2.3.5. The specific case of the coastline ...... 32

3. GIS...... 33

BRGM/RP-55553-FR – Final Report 5 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

3.1. BUILDING STEPS ...... 33

3.2. LIFE CYCLE...... 33

3.3. MAIN TASKS ...... 34

3.4. LAYERS OF THE COASTAL GIS...... 37 3.4.1. Background folder ...... 37 3.4.2. Land use folder...... 42 3.4.3. Coastal folder ...... 45 3.4.4. Assets folder...... 58 3.4.5. Elements at risk sub-folder...... 58

3.5. THE 100 M LIMIT...... 63

3.6. DIFFICULTIES ENCOUNTERED DURING GIS CONSTRUCTION...... 68 3.6.1. Location of data...... 68 3.6.2. Data format...... 68 3.6.3. Structure of numerical data ...... 68 3.6.4. Coordinate system and accuracy in projection system ...... 68

4. The Risk Scenario ...... 71

4.1. VULNERABILITY FUNCTIONS ...... 71 4.1.1. Field work ...... 71 4.1.2. Damage scale...... 73 4.1.3. Building typologies...... 76 4.1.4. Vulnerability functions...... 77

4.2. MODEL AND SCENARIO...... 82 4.2.1. The software: ARMAGEDOM...... 82 4.2.2. Testing on Unawatuna...... 84 4.2.3. ...... 91

4.3. RESULTS OF THE SIMULATION APPLIED TO GALLE...... 96

4.4. PROSPECTS...... 96

5. Numerical modelling ...... 99

6 BRGM/RP-55553-FR – Final Report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

5.1. MONSOON WAVE MODELLING...... 99 5.1.1. Data inputs...... 99 5.1.2. Numerical simulation of SW monsoon waves...... 104

5.2. TSUNAMI MODELLING ...... 109 5.2.1. Data input...... 109 5.2.2. Numerical simulation...... 110

5.3. MODELLING CONCLUSION ...... 114

6. Recommendations and conclusion ...... 115

7. Aknowledgments ...... 117

8. Bibliography...... 117

9. Appendices ...... 119

9.1. LOCATION AND TYPE OF DATA...... 119 9.1.1. Data at the Survey Department ...... 119 9.1.2. Data at UDA...... 119 9.1.3. Data at the NARA ...... 120 9.1.4. Data at the Meteorology Department...... 120 9.1.5. Data at the Coastal Conservation Department (CCD) ...... 121 9.1.6. Data at the Civil engineering department of Moratuwa University .....121

9.2. ACRONYMS/ABBREVIATIONS...... 121

List of Illustrations

Illustration 1 - Project organization, thematic coordinators...... 15 Illustration 2 - Scientific items included in the coastal risks GIS...... 15 Illustration 3 - Pilot site location (Bentota to Weligama Bay, )...... 19 Illustration 4 - The districts of the Pilot site...... 20 Illustration 5 - Large flat beach at the south of Bentota (location 5)...... 21 Illustration 6 - Coastline morphology and groynes at Ambaladuda (location 10)...... 21 Illustration 7 - Steep slope beach at Ratgama (location 13)...... 21 Illustration 8 - A large spit at Pitiwela (location 15 & 16)...... 21

BRGM/RP-55553-FR – Final Report 7 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 9 - Medium sized bay at Uwanatuna(location 17)...... 21 Illustration 10 - Small bay with narrow beach and discontinuous barrier and patch reefs at Habraduwa (location 18)...... 21 Illustration 11 - Impact of tsunami at Habraduwa (house destruction and beach erosion; location 18)...... 22 Illustration 12 - Breackwater and tombolo morphology at Danuwala. Steep slope and beach erosion (location 21)...... 22 Illustration 13 - Large flat bay with ripples (Welligama bay; location 22)...... 22 Illustration 14 - Small soft cliff at the East of Welligama bay (location 23)...... 22 Illustration 15 - Small barrier reef, lagoon and sandy beach (location 19)...... 22 Illustration 16 - Location of the illustration 4 to 14...... 23 Illustration 17 - Links between processes, hazards and risk. Risk is a combination of hazards and the vulnerability of exposed elements. The vulnerability of exposed elements is specific to each kind of hazard...... 24 Illustration 18 - Graph showing the processes, parameters and hazards and their relationships (dotted lines are feedbacks). Nota: Hazard include the notion of the probability of occurrence of a given phenomenom with a given intensity at a given place...... 25 Illustration 19 - Sea level rise from different models as an example of a continuous process and associated uncertainties (IPCC 2001)...... 26 Illustration 20 - Typology of coastal processes...... 27 Illustration 21 - Example of interaction between Sea Level Rise and Storm surges and the potential impact on extreme surge events...... 28 Illustration 22 - The different steps of coastal hazards GIS development...... 33 Illustration 23 - Coastal GIS life cycle...... 34 Illustration 24 - Tasks of the first step of GIS development...... 35 Illustration 25 - Tasks of the second step of GIS construction...... 36 Illustration 26 - Tasks of the third step of GIS construction...... 36 Illustration 27 - Ikonos image in the background folder (Galle bay)...... 39 Illustration 28 - The Digital Elevation Model created during the project (20 m resolution; original data from Survey Department)...... 40 Illustration 29 - The topographic folder (DEM, contour line and altitude points)...... 41 Illustration 30 - Building and communications layers...... 43 Illustration 31 - Map of the land use layer (CCD data)...... 44 Illustration 32 - Coastal folder (defense work and coastal geology)...... 46 Illustration 33 - Limit of the 2004 Tsunami after the retreat (Galle to Welligama road)...... 47 Illustration 34 - Within the destruction limit (A: total destruction, only foundations remain in the foreground, church remains intact at the background, the destruction limit is located at the road; B: Highly and definitively damaged houses in the destruction area)...... 49 Illustration 35 - Schematic representation of the tsunami empirical model...... 50 Illustration 36 - The empirical model results (tsunami hazard) compared to the 2004 inundation limit in the Galle area, GSMB limit of inundation and destruction...... 51 Illustration 37 - Map of the 2100 Sea Level Rise hazard...... 52

8 BRGM/RP-55553-FR – Final Report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 38 - Overview of critical erosion areas and erosion rates along the west and southwest coasts of Sri Lanka (CCD 2004)...... 53 Illustration 39 - Evolution of the permanent vegetation line from 1956 (green line) to 2007(red line) on the 1956 aerial photo, South of ...... 55 Illustration 40 - Example of coastline progression during the last 50 years (PVL 1956 in green, PVL 2007 in red; South of Hikkaduwa) - 1956 aerial photo...... 56 Illustration 41 - Example of coastline erosion during the last 50 years (PVL 1956 in green, PVL 2007 in red; South of Hikkaduwa) - 1956 aerial photo...... 56 Illustration 42 - 57 m of coastline retreat from 1956 to 2007 in the Danuwala area (in green: 1956 permanent vegetation line; in red: 2007 permanent vegetation line)...... 57 Illustration 43 - Example of a composite hazard map (Sea Level Rise and Tsunami) on the Galle city...... 59 Illustration 44 - Assets layer (Galle)...... 60 Illustration 45 - Elements exposed to the tsunami hazard (Galle)...... 61 Illustration 46 - Elements exposed to Sea Level Rise risk (Galle)...... 62 Illustration 47 - Quantification of elements at risk in the pilot site for Tsunami and Sea Level Rise...... 63 Illustration 48 - The “buffer zone” (continuous red line), the 2004 tsunami inundation limit and tsunami hazard around Hikkaduwa...... 64 Illustration 49 - The “buffer zone” (continuous red line), the 2004 tsunami inundation limit and tsunami hazard around Galle...... 65 Illustration 50 - The “buffer zone” (continuous red line), the 2004 tsunami inundation limit and tsunami hazard around Koggala...... 66 Illustration 51 - The “buffer zone” (continuous red line), the 2004 tsunami inundation limit and tsunami hazard around Welligama bay...... 67 Illustration 52 - Post-tsunami form...... 72 Illustration 53 - Form used in Unawatuna, specific to building damages...... 72 Illustration 54 - Damage scale for masonry structures...... 73 Illustration 55 - Illustration of structural damage category D1...... 74 Illustration 56 - Illustration of structural damage category D2...... 74 Illustration 57 - Illustration of structural damage category D3...... 75 Illustration 58 - Illustration of structural damage category D4...... 75 Illustration 59 - Buildings distribution observed on the field...... 77 Illustration 60 - Buildings distribution used during simulations...... 77 Illustration 61 - Values of the vulnerability function for masonry constructions...... 78 Illustration 62 - Vulnerability functions for masonry constructions...... 79 Illustration 63 - Vulnerablity functions for light constructions...... 81 Illustration 64 - Vulnerability functions for reinforced concrete constructions...... 82 Illustration 65 - Distribution of the different types of buildings (Unawatuna test area)...... 85 Illustration 66 - Damage repartition: comparison of calculated results and observations...... 85 Illustration 67 - Results of the first “test simulation” (Unawatuna test area)...... 86 Illustration 68 - Random repartition of buildings typology...... 86

BRGM/RP-55553-FR – Final Report 9 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 69 - Random distribution of the different types of buildings (Unawatuna test area)...... 87 Illustration 70 - Damage repartition: comparison of calculated results and observations...... 87 Illustration 71 - Results of the second “test simulation” (Unawatuna test area)...... 88 Illustration 72 - Unawatuna Bay: assets and aggression (height of water from 6 to 0 m)...... 89 Illustration 73 - Random vulnerability functions distribution (Unawatuna bay)...... 90 Illustration 74 - Aggression converted in height of water (m)...... 90 Illustration 75 - Unawatuna: Impacts of the 2004 tsunami...... 91 Illustration 76 - Galle: assets and aggression (height of water from 6 to 0m)...... 92 Illustration 77 - Vulnerability functions distribution (random) - Galle...... 93 Illustration 78 - Aggression applied to buildings (height of water)...... 94 Illustration 79 - Galle: Impacts of the 2004 tsunami...... 95 Illustration 80 - Comparison of the number of buildings affected...... 96 Illustration 81 - Example of bathymetric sounding realized by the NHO in the Galle offshore...... 100 Illustration 82 - 3D view of the bay of Galle (20 m x 20 m mesh size)...... 101 Illustration 83 - Measurements of the buoy deployed off Galle harbour at a depth of 70 m. A: during the 1989-1995 period, B: details of the 1991 year...... 102 Illustration 84 - Extracted data from WaveWatch3 for the Southwern Sri Lanka...... 103 Illustration 85 - Nested grids used used for modelling...... 105 Illustration 86 - Wind field June 06/02/1991 00:00 (constructed with ERA40 database)...... 106 Illustration 87 - Hs computed with the medium resolution grid, values in m...... 106 Illustration 88 - Hs computed with the high resolution grid, values in m...... 107 Illustration 89 - Set-up computed with the high resolution grid, values in m...... 108 Illustration 90 - ETOPO2’s bathymetry...... 109 Illustration 91 - Characteristics of the 5 segments of seismic source after Grilli et al., 2007...... 110 Illustration 92 - The total coseismic seafloor vertical displacement (in m) obtained for the 5 combined tsunami sources...... 111 Illustration 93 - Snapshot at 25, 110, 230 and 250 minutes of the tsunami propagation...... 112 Illustration 94 - Maximum simulated tsunami elevations above sea level...... 113

10 BRGM/RP-55553-FR – Final Report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

1. The project

1.1. GENERAL OBJECTIVES

The impact of the tsunami of 26 December 2004 clearly showed the importance of a thorough knowledge of coastal risks and of taking them into account in order to reduce their effects. This project aims at implementing and designing an effective tool in order to limit the impact of natural hazards, to anticipate and to optimize preparedness to a potential crisis.

1 - The coastal Geographic Information System (coastal GIS) has been created using homogeneous data on the land/sea interface. This allowed relevant parameters to be cross-referenced, and then to conduct studies on the exposure of coastal populations to numerous natural hazards. The GIS was designed to serve as a decision-support tool for policymakers and political and economic decision-makers in the field of risk management and protection.

2 - This approach is supported by the creation of one or more risk scenarios in order to determine whether the cross-referenced data provide an adequate basis for the management of these coastal areas or whether there is added value in the scenario approach to coastal risks. Coastal risks result from a combination of the hazard itself (aggression), vulnerability (response of exposed elements) and value (stakes involved). A coastal risk scenario involves the assessment of the impact and consequences of an event (tsunami, storm, etc.), particularly on assets. Simulating risk scenarios should allow to (i) identify “weak points” in human settlements; (ii) test the exposure of future development.

The results obtained form a basis for anticipating coastal risks as part of a risk prevention and crisis management policy.

Implementing the GIS therefore requires the following: - Characterisation of the aggression resulting from the event. This may be deduced from an historical or hypothetic reference event or from a regional statistical and/or probabilistic assessment of the hazard. The aggression would need to be modulated in accordance with local hazard characteristics (e.g. angle of incidence of ocean swell in relation to the coastline). - An inventory of the physical elements exposed, which could be supplemented by functional analysis. - A valuation of the exposed elements, so that they can be ranked according to the issues at stake. - An assessment of their vulnerability (in terms of physical vulnerability such as damage transfer function, functional vulnerability, etc.).

BRGM/RP-55553-FR – Final Report 11 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

- Application at the pilot site of the aggression factor(s) formely determined, in order to evaluate their consequences in terms of potential losses and disruptions.

This GIS supplements the international tsunami alert system now being established in the Indian Ocean and is fully consistent with natural risk management policies.

Policy-makers need concrete assessment of the impacts of coastal hazards. This requires an integrated multi-hazard approach which is linked to potential conflicts between land projects. This enables identification of risk-prone areas and subsequent implementation of an integrated strategy including harmonious development and risk prevention and management in the relevant zones.

The project has been conducted on a pilot site extending some 84 km along the coastline from Bentota to the Welligama bay.

The main project's objectives are: - to characterise the different coastal risks within the pilot site; - to contribute to the prevention and reduction of coastal risks as part of a coastal defence plan; - to identify the probability of their occurrence and to define the most exposed areas within the pilot site; - to supply useful data products to managers of coastal areas, which can be directly used for the protection of people and goods and for the purpose of including environmental constraints in territorial development planning at an early stage; - to build capacity in Sri Lanka’s technical departments (by transferring technology and know-how) to enable them to take over data input and use of the coastal risks GIS.

This is a pilot project which can be used for demonstration purposes and may subsequently be applied to other risk-prone coastal sectors, before extending it to the entire Sri Lankan coastline.

Ultimately, this coastal risks GIS could be adapted for use by civil security departments by creating reference scenarios for full-scale safety exercise and crisis management operations.

The project aims at transferring know-how in the areas of coastal risks and hazards, geographic information systems and cross-referencing of parameters.

1.2. PROJECT STRUCTURATION

In order to optimize its management, this project has been divided in two successive phases, with ongoing training throughout these two phases.

12 BRGM/RP-55553-FR – Final Report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

1.2.1. Phase 1: Choice of the Pilot site and definition of the GIS

Phase 1 of the project has focused on the configuration and architecture of the coastal GIS. It has been subdivided in the following tasks:

1) The choice of a representative pilot site made together with the project partners and managers responsible for the development of the coastal zone. The site needs to be representative of the various coastal risks to which the Sri Lankan coastline and is exposed to many coastal hazards (tsunami, cyclonic ocean swell, erosion, storm and sea-level rise).

2) The different layers of data required (existing or yet to be built up) have been specified according to needs expressed and methodologies.

3) An inventory of existing and available data has been made with regards to the project objectives and requirements, specifying their coverage in space and time, their accuracy and the formats in which they are available (print or digital). Meetings have been organized with the data providers in order to complete the inventory and to draw up agreements for the use of the data. Once this task has been completed, the project team has defined an optimum strategy for acquiring missing data, the objective being to ensure that they are homogeneous in terms of density, accuracy and quality so that the project can guarantee the relevance and reliability of its scenarios for every location in the pilot site.

4) Development of the architecture of the dynamic coastal risks GIS.

1.2.2. Phase 2: GIS development and integration of the data

Phase 2 of the project has focused on the realization of the Coastal GIS. This has included: - data acquisition (geographical, environmental, land use, communication networks…); - a critical analysis of geographical spacing and accuracy; - homogenization and processing of data before integration in the GIS. This includes extraction, restructuration of information and georeferencing; - acquisition of supplementary data to obtain greater information density, to fill in data gaps or to correct existing data.

Satellite images have been used as a data source, to check existing georeferenced data and as background map. In some cases, satellite images have provided a multi- temporal perspective that helped to identify underlying trends for coastal erosion study for example. During this phase we created the GIS structure (layers) and we have integrated data processed before.

BRGM/RP-55553-FR – Final Report 13 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Field investigations were conducted on some particular spots (Pitiwela, Galle, Unawatuna) in order to integrate new data (coastal data, damages generated by the 2004 tsunami…) or to validate existing data (tsunami inundation limit for example). During this phase, the training of the Sri Lankan team both in Sri Lanka has been organised during missions and in France during two weeks in December 2006 (cf. § Working Program).

1.3. PROJECT ORGANIZATION

1.3.1. Project partners

Co leaders of this project are: - Geological Survey and Mines Bureau of Sri Lanka (GSMB); - BRGM.

The following institutions are partners of this project: - National Aquatic Resources Research and Development Agency; - (NARA) ; - Meteorology Department ; - Coastal Conservation Department (CCD) ; - National Science Foundation (University of Moratuwa – Department of Civil Engineering).

1.3.2. Project organization

The structure of the team project is as follows (Illustration 1): - Two project coordinators: Dr B. Prame (GSMB) and Dr M. Garcin (BRGM); - Two coordinators per thematic working group (one Sri Lankan and one French): · Database, GIS: Dr. N. Attanajake, Eng. J.F. Desprats, · Remote sensing and GIS: Dr. U. De Silva , Eng. J.F.Desprats, · Geomorphology and coastal risk : Dr. S. Fernando, · Dr R. Pedreros, · Coastal processes : Prof S. Hettiarachchi/Eng. S. Samarawickrama, · (Civil Eng. Dept), Dr R. Pedreros, · Civil engineering : Eng M. Fontaine, · Coastal geology : Dr. C.H.E.R. Siriwardana/Dr. S. Fernando, Dr M. Garcin.

14 BRGM/RP-55553-FR – Final Report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Project coordinators

Geomorphology, coastal risks coordinators

Coastal Databases, GIS processes coordinators coordinators

Coastal geology Remote sensing coordinators coordinators

Civil engineering coordinators

Illustration 1 - Project organization, thematic coordinators.

The coastal risks GIS includes 13 scientific items. These items are presented in the Illustration 2.

GIS

Historical Coastal Topography Geology coastline Land use Defence works morphology Bathymetry evolution

Coastal Tsunami Storm surge Sea level rise evolution

Risk Element at risk Vulnerability evaluation

Illustration 2 - Scientific items included in the coastal risks GIS.

BRGM/RP-55553-FR – Final Report 15 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

1.4. WORKING PROGRAM

Four missions at and on the field of the pilot site have been organised. At the end of each mission, a brief meeting between project participants took place in order to define the working program until the next BRGM mission.

1.4.1. First mission

A first phase mission in Sri Lanka has been carried out from January 29th to February 10th 2006. The main objective of this mission was to meet all partners and institutions involved in the coastal management and hazards and to kick off the project.

The French team met the following institutions: - GSMB (Geological Survey and Mines Bureau); - NARA (National Aquatic Resources Research and Development Agency); - NHO (National Hydrographic Office); - Survey Department; - Department of Meteorology; - UNDP (Tsunami Disaster recovery and Humanitary Information Center-UN); - UDA (Urban Development Agency); - DMC (Disaster Management Center); - Disaster Management Minister; - CCD (Coastal Conservation Department); - University of Moratuwa (Department of Civil Engineering); - LHI (Lanka Hydraulic Institute Ltd).

A 2 day field trip has been organized in order to study the coastal morphology, coastal defences and effects and impact of the tsunami on the coast from the south of Colombo to Welligama bay. These observations help to identify the variability of the coast in terms of beach morphology, defence works and exposure to coastal hazards. This led to select the area from Bentota to Welligama as the Pilot site (see §4).

1.4.2. 2nd to 4th mission

During the 2nd (from 8 to 26 May 2006) 3rd (from 5 to 16 February 2007 and 4th mission (from 28 May to 8 June 2007), work was focused on the integration of data coming from Sri Lankan institutions and field survey in order to acquire and check data on the field. The third mission, initially planned from 23 October to 3 November 2006, was cancelled due to security problems in the Galle area. As a consequence, the 3rd mission has been postponed to the beginning of 2007 (February 5 – 16) and the end of the project postponed to June 2007.

16 BRGM/RP-55553-FR – Final Report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

During the last mission we organized the final meeting at Colombo with presentations of the project, methodologies, GIS contents, results and recommendations. A short live presentation of the GIS has been given. The following institutions were represented at this meeting: GSMB, CCD, NARA, DMC, Ministry of Disaster Management, Survey Department, NHO, French Embassy.

All these missions have been realized in collaboration with the Sri Lankan team and more particularly with the help of the GSMB team.

1.4.3. Training in France

Training in France has been organised for 4 participants from December 11th to 22nd 2006. The four participants from GSMB were Nishantha Attanayake, C.H.E.R. Siriwardana, Udaya De Silva and Starin Fernando.

The training was structured in two parts: - GIS tools training during one week (Dr. Garcin & Eng. Desprats). This training was conducted mainly with ArcGIS 9.1 software. Basic functions to more elaborate processing have been presented during practical work by the trainees on the Sri Lankan data set coming from the project. An overview of existing free GIS software has been presented (QGIS, JvSIG, OpenJump…). - During the second week, specific presentations have been given by experts of the BRGM: risk scenario and the simulation software ARMAGEDON (Dr. Sedan), Satellite Synthetic Aperture Radar interferometry : uses and limits (Dr. Carnec), Tsunami numerical modelling (Dr. Pedreros), Coastal process, coastal hazards and coastal monitoring (Dr. Pedreros), cliff monitoring using the high resolution laser (Dr. Dewez). Between these presentations, the trainees have worked on the GIS software in order to improve their skills.

BRGM/RP-55553-FR – Final Report 17

GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

2. Pilot site

2.1. PRESENTATION After discussions between partners, field investigations and data mining, the pilot site has been defined as the 84 Km-long and 2 Km-wide strip of coastline running from Bentota to Welligama bay (Illustration 3) The pilot site includes 13 districts (Illustration 4).

This site has been chosen as this area presents various contexts (illustrations 4 to 14) in terms of coastal morphology (sandy beaches, sand spit, bays from large to small (Welligama, Galle, Unawatuna…)), general orientation of the coast (from E-W to N-S…), coastal processes (erosion, accretion …) and coastal defence works. Moreover, various hazards have to be taken into account (tsunami, coastal erosion, storm surge, sea level rise). Numerous important assets characterize this coastal area: towns and villages, fishermen settlements, transport networks (roads, railway), tourist resorts. This area is within a unique administrative district (Galle) which will make easier further data acquisition.

Thus the area defined as the pilot site is representative of an important part of the coastline in Sri Lanka.

Illustration 3 - Pilot site location (Bentota to Weligama Bay, Galle district).

BRGM/RP-55553-FR – Final report 19 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 4 - The districts of the Pilot site. - The districts Illustration 4

Illustration 4 - The districts of the Pilot site.

20 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 5 - Large flat beach at the south of Bentota (location 5).

Illustration 8 - A large spit at Pitiwela (location 15 & 16).

Illustration 6 - Coastline morphology and groynes at Ambaladuda (location 10).

Illustration 9 - Medium sized bay at Uwanatuna(location 17).

Illustration 7 - Steep slope beach at Ratgama (location 13).

Illustration 10 - Small bay with narrow beach and discontinuous barrier and patch reefs at Habraduwa (location 18).

BRGM/RP-55553-FR – Final report 21 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 13 - Large flat bay with ripples (Welligama bay; location 22).

Illustration 11 - Impact of tsunami at Habraduwa (house destruction and beach erosion; location 18).

Illustration 14 - Small soft cliff at the East of Welligama bay (location 23).

Illustration 12 - Breackwater and tombolo morphology at Danuwala. Steep slope and beach erosion (location 21).

Illustration 15 - Small barrier reef, lagoon and sandy beach (location 19).

22 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka Illustration 16 - Location of the illustration 4 to 14.

Illustration 16 - Location of the illustration 4 to 14.

BRGM/RP-55553-FR – Final report 23 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

2.2. COASTAL PROCESSES, HAZARDS AND RISKS

2.2.1. Definitions

In order to better estimate the coastal risk in a specific area, it is critical to have a good understanding of the chain that leads from the process to the risk (Illustration 17). Firstly, the existing and potential processes which are the cause of the hazards must be clearly identified. These initial processes or phenomena can be purely natural or partially modified by human action at local, regional or global scale. The hazard originates from these processes and the interactions between these processes can lead to minimize or to increase the hazard. Each process can be the cause of one or several hazards (Illustration 18). When human settlements and activities are located in the coastal zone, they are considered as being exposed to the relevant hazards (exposure to the coastal hazards). Identifying the exposed element is not obvious due to the variety of processes and their interactions. The assessment of the vulnerability of the element exposed (building, communications network, infrastructure…) to a given hazard leads to the risk assessment. The specificities of the evaluation of processes, hazards and risks in the coastal strip are further explained in this chapter.

Element Process Hazard exposed

Process Vulnerability Risk

Illustration 17 - Links between processes, hazards and risk. Risk is a combination of hazards and the vulnerability of exposed elements. The vulnerability of exposed elements is specific to each kind of hazard.

24 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Legend

Process Coastal parameter

Hazard

Marine Storm submersion surge Hazard

Sea Level Rise Coastal morphology

Coastal erosion Erosion Hazard

Coastline (retreat) Tsunami

Tsunami hazard

Illustration 18 - Graph showing the processes, parameters and hazards and their relationships (dotted lines are feedbacks). Nota: Hazard include the notion of the probability of occurrence of a given phenomenom with a given intensity at a given place.

2.2.2. Coastal processes characterization a) Processes typology

Coastal processes can be sorted into 2 classes: - Continuous processes (with various rates). The Sea Level Rise (SLR) induced by the climate change is an example of this type of hazard (Illustration 19). - Discontinuous processes or resulting from a crisis period like the storm surges linked to storms or cyclones. Tsunamis are part of this category.

BRGM/RP-55553-FR – Final report 25 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Some processes can be attributed to both previous types depending on the temporal resolution of the analysis. For example, coastal erosion will be perceived as a continuous process at a pluriannual temporal scale whereas it will be perceived as a succession of crisis and stability periods at a high resolution temporal scale (days). Thus coastal erosion can be characterized by an average rate on a pluri-annual basis, whereas this process cannot be quantified by a rate at a monthly timescale.

Illustration 19 - Sea level rise from different models as an example of a continuous process and associated uncertainties (IPCC 2001). b) Temporal dimension

Another specific aspect of the coastal multi hazards approach is the variability of the return period associated to each hazard. In the Sri Lankan case, it is clear that return periods of processes are highly uneven: - infra annual to pluri annual timescale : erosive crisis; - annual to pluri annual timescale : storm surge triggered by cyclone or tropical storm; - pluri centennial to millennial timescale: major tsunami.

The Sea Level Rise linked to climate change is specific as it cannot be characterized by a return period but more adequately by a characteristic time of several centuries. This process will be perceptible in few decades and is classically evaluated for 2100.

26 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

c) Typology of the effects

3 types of effects induced by coastal hazards affect directly the safety of people and properties: - Coastline changes and more precisely the coastal retreat and erosion. This effect is induced by coastal erosion, land slide and mass wasting of the coastal cliffs. During particular events like major tsunamis, coastline retreat can occur (Sri Lanka, Indonesia…). - Instantaneous reversible marine inundation of short duration (from a few hours to a few days) of some area of the coast. This violent submersion is considered as reversible because duration is few hours to some tens hours. It is generated either by a storm surge or by a tsunami. - Progressive and irreversible marine inundation. This soft process is a consequence of the progressive sea level rise linked to climate change, isostasy … These processes are considered as irreversible at the human time scale, i.e. from a few years to centuries.

One of the characteristics of the multi-hazard approach in coastal fields for durations of several decades is the fact that there is a combination of all these processes. The cumulated effects of these processes increase the hazard. A typology of processes including the type, the return period, typology of effects and reversibility is presented in the Illustration 20.

Thus, progressive and irreversible process, such as the sea level rise induced by the climate change (CC), will be superimpose with very short, reversible and discontinuous events (storm surge, tsunami…).

Process Type Time/return Typology of Reversibility period effects Tsunami Discontinuous Centennial to Submersion & Reversible (submersion), millennial Coastline irreversible (coastline retreat retreat) Storm surge Discontinuous Supra annual Submersion Reversible Coastal Discontinuous / Infra annual Coastline Irreversible erosion Continuous retreat Sea Level Continuous Century Submersion & Irreversible Rise coastline retreat

Illustration 20 - Typology of coastal processes.

An example of the cumulative effects is presented in Illustration 21. On this figure the purple curve corresponds to the sea level rise (under the pessimistic assumption of IPCC 2001), the blue curve shows a simulated height of the storm surge (random

BRGM/RP-55553-FR – Final report 27 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

variable of maximum value of 2 m for a return period of one year) and the red curve is the sum of the two effects (SLR and the storm surge height). This figure makes it possible to visualize the incidence of the combination of these two processes on the height of inundation that could possibly occur between 2000 and 2100. Not taking into account the combination of these two processes would lead to underestimating the hazard.

In the same way, the changes of the coastline due to erosion will modify the surfaces potentially affected by an inundation hazard (tsunami or storm surge). These last two examples justify and show the advantages of a multi hazards approach in the coastal field.

Legend Simulated storm surge Sea level rise 3 Submersion height (SLR +SS)

2 ges height (m) ges height r 1 m su r Sto

0

2000 2020 2040 2060 2080 2100 Years

Illustration 21 - Example of interaction between Sea Level Rise and Storm surges and the potential impact on extreme surge events. d) Interactions

We have considered that the coastal processes and events account for 3 main hazards in the Sri Lankan pilot site: marine submersion, erosion and tsunami.

Although one process alone can generate one hazard (the erosion process can generate the erosion hazard for istance), some of the hazards are controlled by 2 or more processes. For example, as previously presented, the marine submersion hazard is controlled by 2 processes (sea level rise and storm surge). The previous paragraphs express only a direct causality relation at a given time.

In a complex system like the coastal area, some parameters of the system are modified by physical processes during events. These state parameters partially affect the

28 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

amplitude of a hazard started/controlled by another process. As an example, the occurrence of a tsunami on the coastal areas can lead to changes in close bathymetry, of the coastline itself (c.f. Bandah Aceh, Sri Lanka…), and of the topography (erosion, deposition). These parameters are fed back into the coastal system, thus affecting the hazard level if a new tsunami occurs but more especially they modify completely the parameters that control coastal erosion and storm surges. In an opposite way, the storm surges can be responsible of geomorphological processes which partially affect erosion processes. e) Uncertainties

Uncertainties on the forcing factors - Whereas the climate change (CC) and the associated sea level rise (SLR) seem to be admitted by the whole scientific community, uncertainties remain regarding the rate and the extent of the phenomenon (c.f. results of the various models, Illustration 19). Uncertainties relate to the rate of the SLR as well as to climatic parameters (precipitation: quantity, distribution; temperature; frequency and amplitude of the storms). Moreover, the impact of the CC and the SLR on the oceanic circulation and coastal currents is absolutely unknown. In addition it is considered that the CC will modify the storms climate by a probable increase of their intensity and frequencies. Over the 1980-1988 period the average annual number of tropical storms in the Indian Ocean was 4,3 (Vitart et al., 1997). Singh et al., (2001) in addition showed that between 1877 and 1997, the frequency and the intensity of the tropical storms and the cyclones have increased.

Uncertainties of the response to the climate change – Another source of uncertainties resides in the response of coastal processes to the CC. The behavior of the coastline is linked to regional and local conditions, to the area’s coastal geomorphology, to defense works and others human actions. The problem is to find past analogs with characteristics sufficiently close to those of the awaited CC, such as, for example, high rates of sea level rise occurred during the beginning of the Holocene. This sea level rise occurred during a transition between a glacial and an interglacial period and began at a very low eustatic level, whereas the present climatic change occurs during an interglacial period with high eustatic level and high average temperatures. Therefore, although the past analog approach is necessary, it must be completed by a modelling approach, that is aimed at assessing the response of coastal area to climatic change including sea level rise, temperature, rainfall, storm frequencies and intensity changes.

As a conclusion, uncertainties concern both forcing factors and coastal system behaviour. This implies that a particular attention must be brought for the analysis of the results.

BRGM/RP-55553-FR – Final report 29 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

2.3. COASTAL MULTI-HAZARDS MAPPING: THE PROBLEM

2.3.1. Proposal of the creation of a composite marine inundation hazard

The preceding paragraph illustrates that it is thus imperative to integrate the SLR as a long-term component of the marine submersion hazard. It appears more appropriate to consider the marine submersion hazard as a single one that includes both sea level rise linked to climate change (SLR) and Storm Surge. The first one is slow and presents an irreversible evolution and the second one which is a fast process results from a crisis but with reversible effect.

2.3.2. Classification of coastal hazard

The studies of the coastal hazards aim at establishing useful data bases for the synthetic mapping of the hazards. This cartography must be sufficiently clear and explicit to be usable and useful for the decision makers in charge of the coastal area. The issue is to integrate many hazards within a synthetic and homogeneous document, according to a single classification. It is indeed necessary to define a classification able to integrate hazards characterized by slow and irreversible evolution as well as those characterized by reversible crises. Return periods are highly variable: from one year, to decades, hundreds or even thousands of years. In the same way the risks affecting people and goods caused by these hazards can be of various amplitudes. How to integrate frequent hazards with relatively moderate impacts with those far from frequent but whose impact is major and destroying?

2.3.3. Two possible approaches of the cartography of marine inundation hazard

Two cartographic approaches of the inundation hazard are possible. The hazard in 2100 can be evaluated with a static approach. The hazard value will be dependent on the height of inundation of the coastal zone (SLR + Storm Surge)

The classification of the level of hazard can be done according to two parameters: - the height of inundation at a given place: as example the hazard will be strong if the height of water is higher than 2 m, medium between 2 and 1 m of inundation and low if the inundation is less than 1 meter; - the frequency and reversibility-irreversibility of inundation: in the 2100’s, all zones only submerged by the SLR (+0.8 m) will be regarded as zones with strong hazard as this immersion is irreversible. The zones submerged by a storm surge of 1 m (frequent event, 1m + 0.8 m) will be considered as zones with medium hazard. The zones which will be flooded only at the time of strong storm surge (rare and with a value of 2 m) superimposing itself with the SLR (2 m+ 0.8 m) will be classified as zones with low hazard.

30 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Two limitations have to be noted: the first relates to the height of water due to storm surges. The present day value will not necessarily be representative of what will be the storm surge in 2100. The second limit is that the hazard level is overestimated for the near future (next years and decades)

The second approach would consist in estimating the inundation hazard (SLR + storm surge) by taking into account the temporal dimension of the SLR. Thus, for some key dates which remain to be defined (every 25 years for example), an evaluation of the hazard would be associated. In this case, one would consider that any continental surface easily flooded a time “t” would be in strong risk if it is flooded by the SRL of the time to which one adds the value of storm surge. It would be then necessary to constitute diachronic maps of the inundation hazard. These maps would provide indications of the hazard in 2006, 2025, 2050, 2075 and 2100. They would allow users and decision makers to prioritize and adjust adaptation strategies according to the emergency of the situation. An additional possibility would be to provide the plausible coastlines at the same dates (2025, 2050, 2075 and 2100). The cartographic representation of these coastlines would make it possible to raise awareness of the decision makers and the populations on the impact of this phenomenon on the territory. It would be of course a rough estimation of these future shorelines because the adaptation and the impact strength of the coast are not taken into account.

2.3.4. Evaluation of the elements at risk

Taking into account the previous remarks, the evaluation of the elements exposed at risk is not obvious. Indeed, the same element (dwelling, infrastructure, Communities’ building…) is confronted with hazard at more or less long term. According to the amplitude of the hazard, its reversibility or irreversibility and its return period, it will be possible to evaluate if the element concerned will be maintainable or, on the contrary, will have to be moved (retreat strategy).

From a sustainable development point of view, any new installation of high lifespan and paramount importance for the society (hospital, help center, major road …) should be built nor in zones subject to a strong hazard with a short return period nor to any irreversible hazard. Taking into account, the temporal dimension and the reversibility character of the hazard must be integrated in land planning policies. Due to the diversity of the hazards to take into account, it appears judicious to us to assign to each exposed element the degree of exposure to each hazard, its return period or its characteristic time. It will be thus possible to carry out a request within the GIS in order to select: - elements subject to a particular hazard, with a given intensity and over one duration; - elements subject to a hazard at a given time independently of the hazard’s nature; - elements subject to a high hazard, independently of the hazard’s nature.

BRGM/RP-55553-FR – Final report 31 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

2.3.5. The specific case of the coastline

The erosion of the coastline appears more complex to deal with. It could be planned to apply an annual erosion rate over the 100 years duration (to be compatible with that of the SLR) but keeping in mind that this annual average rate has been evaluated on a period when climatic and eustatic conditions were different. It is thus completely believable that this annual average rate will not be anymore valid in the future decades due to the climate change (changes in the storm climate, currents, swell…). This average rate of erosion by segment is however the only variable available for the evaluation of the plausible coastline at a given date. From there, it will be possible to assign to each stake the date from which it will be put at the risk.

32 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

3. GIS

3.1. BUILDING STEPS

The GIS development was undertaken in 3 steps, presented in Illustration 22: - the first step was to elaborate the elementary layers of the GIS. This task was a necessary input for the next step. During this first step, the following layers have been brought together: geology, coastal morphology, topography, hydrography, land uses, defense works, roads, railways, …; - the second step was to assemble added value layers, i.e. more precisely, the different coastal hazards layers (tsunami, coastal evolution, storm surge, sea level rise); - the third step is dedicated more specifically to the risks assessment (element at risk, vulnerability, risks evaluation).

Illustration 22 - The different steps of coastal hazards GIS development.

Each task of these steps is detailed in the following chapter 3.3.

3.2. LIFE CYCLE

In order to be useful, the GIS must evolve and integrate new or updated data taking into account the evolution of the environment or changes in land uses or defence works. Some of the parameters integrated in the GIS could become out of date (as for example the near shore bathymetry which has been drastically modified by the tsunami).

BRGM/RP-55553-FR – Final report 33 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

As a consequence, the GIS reflects the state of the coast system at a given time. Thus, it is necessary to consider the coastal hazard GIS as an evolving system needing updates as presented in the Illustration 23.

Land use, Coastal urbanization changes changes

Acquisition or New expertise update of (update) data

Hazards Risk Databases GIS evaluation evaluation

Illustration 23 - Coastal GIS life cycle.

3.3. MAIN TASKS

Each task of each step is detailed in the following illustrations (Illustration 25, Illustration 26). For each task, Sri Lankan and French experts are identified.

34 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Tasks GIS STEP 1 Sri Lanka team French team

GIS Dr. Attanajake Dr. Garcin GSMB 1) Definition of structure (format, projection, …) Eng. Desprats 2) Integration of layers 3) Multicriteria analysis

GEOLOGY Dr. Siriwardana Dr. Garcin 1) Extraction of geology (lithology/stratisgraphy/faults) on Dr. Fernando coastal strip (geology 1/100.000) GSMB 2) Integration into GIS

COASTAL MORPHOLOGY Dr. Siriwardana Dr. Garcin & Dr. Pedreros Characterisation with following items : “sandbar-reef – beach Dr. Fernando rock – berm – beach cusp – sand dune – mangrove – lagoon – inlet – sandspit – outlet – soft and hard cliff” GSMB

DEFENCE WORKS Eng. Dr. Pedreros Wickramarachchi Photo interpretation of rock revetment, breakwater, groyne, harbour … CCD

TOPOGRAPHY/BATHYMETRY To determine Dr. Pedreros 1) Contourlines and altitude points integration (Survey Dpt) NHO 2) Sounding points (Bathymetry) and cross shore profile (NARA) 3) Interpolation to build DEM DR. Arunalanthan 4) Integration into GIS NARA

LANDUSE/LANDCOVER MAPPING Dr. De Silva Eng. Desprats 1) Remote sensing data selection GSMB a. Images processing (correction + classification) : High resolution (SPOT 4/ Landsat) b. Very high resolution (Ikonos/Quickbird/ SPOT 5)

2) Classifications (based on Survey Dpt/UDA’s land use maps)

3) Integration into GIS

HISTORICAL COASTLINE EVOLUTION Dr. De Silva Dr. Pedreros (1) Integration of aerial photo/Ikonos into GIS GSMB Dr. Garcin (scan/georeferencement) (2) coastline extraction (3) comparison

Illustration 24 - Tasks of the first step of GIS development.

BRGM/RP-55553-FR – Final report 35 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Tasks HAZARDS STEP 2 Sri Lanka team French team TSUNAMI Dr. Siriwardana Dr. Pedreros 1) Inundation extension Dr. Fernando Dr. Garcin 2) Integration of observed damages limit GSMB 3) wave height / Run up COASTAL EVOLUTION (erosion, aggradations, Eng. Wickramarachcchi Dr. Pedreros stability) CCD 1) Definition of sedimentary cells 2) Analysis of hydrodynamic 3) Evaluation of sedimentary budget STORM SURGE Chandrapala Dr. Pedreros 1) Wave set up METEO 2) Meteorological forcing (wind, atmospheric Prof. S. Hettiarachchi pressure) Eng. Samarawickraam 3) Tide Dept Civil Eng. SEA LEVEL RISE (global change) Dr. Wijeratna Dr. Garcin NARA

Illustration 25 - Tasks of the second step of GIS construction.

Tasks RISKS STEP 3 Sri Lanka team French team ELEMENTS AT RISK Dr. Fernando Eng. Fontaine Integration of UDA building types ….

VULNERABILITY Dr. Fernando Eng. Fontaine Set-up of an enquiry form …. RISK EVALUATION Dr. Fernando Eng. Fontaine Dr O. Sedan

Illustration 26 - Tasks of the third step of GIS construction.

36 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

3.4. LAYERS OF THE COASTAL GIS

The GIS integrates presently over 1 500 files for a total amount of data of about 5 Go of disk memory. All data used in the GIS have been georeferenced in the Sri Lankan Kandawala coordinates system. ArcGIS and others GIS softwares are able to convert “on flight” the coordinates from one to another system, but for a better accuracy, it was more appropriate to convert initial data files in the Kandawala system. The GIS is structured in folders and sub-folders each one containing several layers of information. These folders and sub-folders are structured to reflect the themes treated by the GIS. Naturally, some layers could be perceived as background layers or as assets layers (roads & buildings for example) depending on the final objective of the GIS. However, depending on the needs, layers can be moved from one folder to another without changing the data.

The main folders are as follows: - Background: base map layers ; - Land use: general land use information including buildings and communication networks; - Coastal: specific descriptive coastal information and data about hazards and processes; - Assets: this folder contains presently an assets layer only on the Galle area;

The sub-folders and layers are detailed below.

3.4.1. Background folder

This folder corresponds to the base map and contains the sub-folders: - Limit of the pilot site sketched as a polygon; - Satellites images, mainly Spot and Ikonos images (Illustration 27). SPOT Images were acquired as part of the scientific ISIS program. The ISIS program can involve any scientist working in a French research laboratory, on a scientific or demonstration project within the scope of GMES (Global Monitoring for Environment and Security). Data is strictly limited to scientific use or to activities aiming at demonstrating the benefits of remote sensing data for the implementation of GMES, in particular for the control of International Conventions (Kyoto Protocol), the monitoring of environmental stress, food security and the prevention of major hazards; - Geological vector digital map provided by GSMB. This layer provides information on the geological background. Surface coastal geology is provided in the Coastal layer; - Topography. This folder contains 2 Digital Elevation Models and vector data: · A new DEM created by the team project at a 20 m resolution (grid size) by interpolation (triangulation with linear interpolation) of the Survey Department

BRGM/RP-55553-FR – Final report 37 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

topographic data (altitude point and contour lines). This DEM covers the entire pilot site from Bentota to Welligama (Illustration 28). The accuracy is generally of good quality, but we identified some errors on some areas, mainly in the transition from flat area to small hill. These errors during the interpolation are due to relatively low density of measurement point in the transition zone between flat lands and hills, · A DEM provided by the UNU data set (computed with survey data and data acquired with GPS on the field). This high quality DEM with a 5 m grid size covers only the urban area of Galle, · Topographic contour lines extracted from Survey Department data (Microstation file). These contour lines are discontinuous, with a 5m spacing between lines, and some of them missing, · Topographic contour lines extracted from the general DEM of the pilot site. These contour lines are continuous and cover the entire site. Spacing of the lines isn’t customized: 0.5 m, 1, 2, 4 , 6, 8, 10, … The choice of this particular spacing is justified by the need for accurate topographic information for the low altitude areas prone to coastal hazards, · Altitude points extracted from Survey Department files. These points give an accurate and local value of measured altitude, · A view of the topographic folder content is presented in the Illustration 29; - Administrative districts including district and sub-district limits (Illustration 4). It will be possible to attach some non geo-referenced data such as population etc. directly to each district polygon in order to realize thematic analysis; - Hydrogaphy folder containing both polylines layer for rivers and polygons layer for lagoons and lakes extension extracted from Survey Department data; - Coastline layer corresponds to the coast line extracted from Survey Department data.

38 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 27 - Ikonos image in the background folder (Galle bay). bay). (Galle folder in the background image - Ikonos Illustration 27

Illustration 27 - Ikonos image in the background folder (Galle bay).

BRGM/RP-55553-FR – Final report 39 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka created during the project (20 m resolution; resolution; (20 m the project created during original data from Survey Department). Department). data from Survey original Illustration 28 - The Digital Elevation Model Elevation Model - The Digital Illustration 28

Illustration 28 - The Digital Elevation Model created during the project (20 m resolution; original data from Survey Department).

40 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

(DEM, contour line and altitude points). line and altitude points). (DEM, contour Illustration 29 - The topographic folder - The topographic Illustration 29

Illustration 29 - The topographic folder (DEM, contour line and altitude points).

BRGM/RP-55553-FR – Final report 41 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

3.4.2. Land use folder

This folder is made of different sub-folders corresponding to general land use information. a) Constructions sub-folder - Buildings layer with both polylines (contours of buildings) and polygons (thematic analysis in function of the type of building: private, public). This layer is a high resolution file extracted from SD (Survey Department) This dataset was created before the 2004 Tsunami so that it could be a good indicator of the state before the tsunami. However, superposition of these data with an Ikonos image acquired before the tsunami showed that the building file was not updated: in some area, a lot of small private buildings are not visible on these data. Moreover, in some areas very close small houses have been grouped in the same polygon. Thus, using this data for the realization of high resolution processing cannot be reliable. However this building layer provides a broad overview of the type and density of buildings in a district; - Bridges have been especially extracted from SD data files. Bridges are prone to be destructed or altered by coastal and others natural hazards. In addition, they are strategic infrastructures. Bridges integrated in the GIS are both road and railways bridges; - Communication networks includes five types represented with different symbols: · Type A roads (Major road), · Type B roads (major roads), · Type C roads (minor roads), · Tracks, · Railways, These 3 layers are presented in the Illustration 30. b) Land use - Land use layer has been included from CCD (Coastal Conservation Department) data on the Galle area. Thematic analysis can be done using the CCD typology. The CCD typology includes information about human activities (trade, services, tourism etc…) but also information on the natural context (Submerged rocks, water bodies, cliffs, sandy beaches..); - Toponimy layer has been included in the Land Use. This layer has been extracted from the SD data.

42 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 30 - Building and communications layers. layers. communications - Building and Illustration 30

Illustration 30 - Building and communications layers.

BRGM/RP-55553-FR – Final report 43 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka Illustration 31 - Map of the land use layer (CCD data). (CCD data). layer - Map of the land use Illustration 31

Illustration 31 - Map of the land use layer (CCD data).

44 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

3.4.3. Coastal folder

This folder contains 2 main sub-folders: one about coastal data and the second one about coastal hazards (Illustration 32). a) Coastal data sub-folder

The coastal data sub-folder includes: - Coastal lithology types with a typology of the different rock types and/or sediment located near to the coastline. The areas concerned are both the shore face the foreshore and the backshore. These layers have been created by extraction of data provided by the CCD and by digitalization during the project. Presently, this layer is uncomplete and doesn’t cover the whole pilot site area. Moreover, existing data has to be checked. The different types are: Rocky area, Rock, Rock submerged, Rock washed, Rock exposed, Coral reef and Sandy beach; - Defence layers include the following types: groyne, dykes, break water, boulders and walls. This layer has been created by extracting data from a 2000 CCD file and has been partially updated by digitalizing satellite images. This layer is not complete and needs to be updated in some sectors of the pilot site; - Boulder layer is a specific polygons layer drawing on extensions of some previous defence layers when the extension is significant. Thus, some defence works are present as polylines in the Defence Layer and as polygons in the Boulder Layer; - Bathymetry folder: . Low resolution bathymetry which is made up by low resolution bathymetric contour lines of the 5, 10, 15 m depth.

BRGM/RP-55553-FR – Final report 45 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka Illustration 32 - Coastal folder (defense work and coastal geology). geology). coastal work and folder (defense - Coastal Illustration 32

Illustration 32 - Coastal folder (defense work and coastal geology).

46 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

b) Coastal hazards sub-folder The coastal hazards sub-folder includes the different hazards identified on the pilot site. Tsunami sub-folder - 2004 tsunami limits of inundation layer. Two limits of tsunami inundation are available in the GIS. The first one has been realized just after the tsunami by the GSMB during a field survey. This low resolution limit was obtained by drawing lines between each point of observation. After checking this limit both with the GIS and on the field, the accuracy appears to be insufficient for our purpose due to the fact i) the number of point of observation are insufficient, ii) the topography wasn’t taken into account in the drawing iii) coordinates system used were not accurate. We have thus created a new inundation limit on the Pitiwela-Galle-Unawatuna area. This new limit has been obtained from numerous GPS field controls including interviews of population. The topography and the results of an empirical model of tsunami inundation were also used (see below). Our field missions confirmed that the inundation limit given by the empirical model was consistent with the 2004 tsunami inundation. Using these different sources of information we created a new limit of inundation that is much more reliable. In the areas where field observations were not still carried out, we used the empirical model to trace the inundation limit. In these sectors, it will be necessary in the future to validate this limit with field observations and interviews.

Illustration 33 - Limit of the 2004 Tsunami after the retreat (Galle to Welligama road).

- 2004 tsunami limit of destruction layer. This limit has been drawn just after the tsunami by the GSMB during a field survey and corresponds to the limit of destruction of buildings (Illustration 34) The destruction zone is defined as an area where 70% of the buildings are totally destroyed of highly and definitively damaged. According to the height of the waves and to local topography, the destruction area can cover only the first line of building nearest to the shore or on the contrary to

BRGM/RP-55553-FR – Final report 47 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

extend inside the hinterland. The destruction limit is the limit between the destruction zone and others less damaged areas. - 2004 tsunami observation points layer corresponds to the observations points realized by the GSMB after the tsunami. The information linked to these points consists of the district, village name, distance from the sea to the limit of destruction, distance from the sea to the limit of inundation, the maximum wave height, time of the event, number of waves and observation/remarks.

Tsunami hazard layer. This layer corresponds to the hazard in terms of wave height (6 to 3 m high hazard, 3 to 1 m medium hazard, 1-0 m low hazard). The height of water has been computed using an empirical model calibrated on the 2004 Tsunami in the Galle area (Illustration 36). This empirical model has been realized using the different layer of the GIS and using standard functions of the software. The wave height at each point is assumed to be a linear function of the maximum wave height at the coast, inversely proportional to the distance to the coastline knowing the maximum distance of inundation reached by the tsunami in a flat hinterland (backshore). The height obtained in each point has been corrected by the local topography to obtain the inundation value. For the 2004 tsunami we have taken a maximum wave height of 6 m (average value of wave height in the Galle area) and a maximum distance of inundation of 1 000 m. One thousand meters is the average distance between the coast and the inundation limit on flat areas. The result of this empirical model compared to the 2004 Tsunami inundation limit is presented in the Illustration 36. At the scale of the pilot site, this inundation height and the computed inundation limit are in concordance with field data and observations. The difference between the model and the true limit is due to 3 factors: (1) the DEM used is everywhere accurate and shows some error (2) In wetland, lake and rivers, the tsunami wave is propagated more quickly and more easily and thus goes further in the hinterland (3) The real propagation landward of the tsunami inundation is much more complex and can't be completely represented by a very simple model. However, this model integrated in the GIS allows a good assessment of the inundation height and extension for a tsunami of the same amplitude than this one and has been used successfully as a guide for finding the inundation limit. It can also be used for a first evaluation of the tsunami hazard level and inundation height in areas without direct information. The observations carried during field surveys, testimonies collected and the works realized by other authors in this area (Peiris, 2006) or in other countries (Papadopoulos et al., 2006) enabled us to define the 3 classes of hazard according to the height of submersion. We distinguished a high hazard for the submersion higher than 3 m, a medium hazard for the submersion between 3 m and 1m and a low hazard for the immersions less than 1m. The lack of information about the return period of such a tsunami did not make it possible to integrate the return period in the hazard evaluation.

48 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Highly and definitively damaged houses in the houses damaged definitively Highly and . ) destruction area destruction (A: total destruction, only foundations remain in the foreground, church remains church foreground, in the remain foundations (A: total destruction, only destruction limit is located at the road; B: at the road; limit is located destruction intact at the background, the intact at the background,

limit destruction - Within the Illustration 34 Illustration 34 - Within the destruction limit (A: total destruction, only foundations remain in the foreground, church remains intact at the background, the destruction limit is located at the road; B: Highly and definitively damaged houses in the destruction area).

BRGM/RP-55553-FR – Final report 49 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

6 Evaluation of tsunami inundation and hazard with the empirical model

Theorical inundation height (H) 5 Topography (T) Inundation height (H-T) 4 Hazard level (6-3m =>H,3-1m =>M, 1-0m =>L)

3 3 Height

2 2

1 1 Hazard level Hazard

0 0

0 100 200 300 400 500 600 700 800 900 1000 Distance to coastline

Illustration 35 - Schematic representation of the tsunami empirical model

Sea level rise hazard layer results from one simulation of the inundated area in the case of a 0.8 m sea level rise in 2100. On this new sea level we have considered two types of total surge: one of medium class with a 1 m elevation which corresponds to a annual total surge event (storm surge + wave setup + tide) and one of high class with a 2 m elevation which corresponds to an event with a ten years return period (storm surge + wave setup + tide). So we have considered that all the areas which will be submerged with the only sea level of 0.8 m are in the high hazard. Areas which will be only inundated during a medium class storm (0.8 m Sea Level Rise + 1 m total surge) are in the medium hazard and finally areas which will be inundated only during an exceptional storm (0.8 m Sea Level Rise + 2 m total surge) are in the low hazard area. The Sea Level Rise hazard map (Illustration 37) has been elaborated using the DEM of the GIS by selecting points whose altitude is in the range of each hazard level.

50 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

mit of mpared to the 2004 inundation limit in the Galle area, GSMB li limit in the Galle area, inundation to the 2004 mpared inundation and destruction. destruction. and inundation

Illustration 36 - The empirical model results (tsunami hazard) compared to the 2004 inundation co hazard) (tsunami model results - The empirical Illustration 36 limit in the Galle area, GSMB limit of inundation and destruction.

BRGM/RP-55553-FR – Final report 51 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 37 - Map of the 2100 Sea Level Rise hazard. Rise - Map of the 2100 Sea Level Illustration 37

Illustration 37 - Map of the 2100 Sea Level Rise hazard.

52 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Erosion sub-folder. It has been estimated that over 50 to 55 percent of the shoreline of Sri Lanka is subjected to or threatened by coast erosion. The impact of coastal erosion is most severe along Sri Lanka's western and south western coasts (Illustration 38). It has been estimated that 685 kilometers of coast line in the south, south west and the west coast and about 175,000 - 285,000 square meters of coastal land are lost each year (CCD 2004). One of the worst affected shores is the coastal sector from Maha Oya to Lansigama (western coast). Coastal erosion is thus a long-standing problem in Sri Lanka, resulting in the loss or degradation of valuable sandy beaches and coastal lands. Whereas coastal erosion is caused by natural processes, human activities such as mining of beach and river sand, mining of corals and planned maritime structures are major factors contributing significantly to coastal erosion. Fighting Coastal erosion requires considerable public and private expenses (CCD 2004). Moreover, coastal erosion is responsible for the loss of beach and landscape quality, for the damage to or loss of private houses, public buildings, hotels and other infrastructures. It also includes considerable annual expenditure for damage mitigation, control and disaster relief. Approximately 1,520 million SLRs have been invested on erosion management in the Coastal Zone from 1985 to 1999, and 3 billion SLRs have been allocated to coastal stabilisation works through the Coastal Resources Management Project (CRMP) spanning from 2000 to 2005 (CCD 2004).

Main area Local stretches and time periods Yearly erosion rate in m/year Maha Oya Waikkal* (1988 – 1998) 8-10 Lansigama Gin Oya sand bar (1991 – 1999) Wellamankara 10-12 (1994 – 1998) 11-13

Colombo N Mutwal to the Kelani River 0 -1

Dickowita Palliyawatta – Uswetakeiyawa 2-3 Moratuwa Seasonal fluctuations,

Koralawella no overall erosion Wadduwa 0-2 Kalu Ganga 1-3 Payagala Beruwela Beruwela – Bentota 1-2 Bentota – Robolgoda Headland Bentota Hikkaduwa Seenigama – Coral Garden Headland 0-2 Coral Garden Headland – Dodanduwa Galle north 0-1 Illustration 38 - Overview of critical erosion areas and erosion rates along the west and southwest coasts of Sri Lanka (CCD 2004).

BRGM/RP-55553-FR – Final report 53 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Given the lack of geo-referenced data on coastal erosion, a global approach of this process has to be done using permanent vegetation line evolution. It is recommended that the line of “permanent” vegetation be used as the baseline for measurement of coastline changes (Environment and development in coastal regions and in small islands / UNESCO-CSI). This line is the tree line or shrub line and can be easily defined and agreed by different observers; also it shows only slight change except in cases of relatively rare tropical storms and cyclones. Features such as high water mark vary according to the tidal cycle and are very subjective especially when used by untrained observers. In addition, the use of the vegetation line as a baseline provides for the preservation of the most seaward sand dunes (where such dunes exist), these provide the first line of natural defence during a major storm event. The permanent vegetation line (PVL) is thus known to be a good marker of coastline evolution on a scale of decades and can be extracted easily from remote sensing data or acquired directly on the field. In order to evaluate the long-term coastline evolution, we brought this limit directly in the GIS using the aerial photos from 1956. PVL lines have been also acquired for some past years (2000, 2002) by the CCD on a few segments of the pilot site and have been integrated in the GIS as well. In order to have a more complete evaluation of the coastline evolution, we mapped the present day permanent vegetation limit on the field on the same area covered by the 1956 aerial photos. This acquisition has been realized during field work (June 2007) using GPS (821 points) with a spacing of data from 20 to 50 meters depending on the complexity of the limit (50 meters for straight and 20 m for curved segments). These points have been integrated into the GIS after conversion into the Sri Lankan projection (Kandawala datum) and the actual permanent vegetation line has been drawn using these control points (Illustration 39). On the 40 km long line mapped, the coastline appears generally stable at the scale of 50 years. This confirms the value presented for the pilot site area (Hikkaduwa & Galle north) in the Illustration 38 from CCD (2004). Only some specific areas show a more significant evolution in terms of accretion (Illustration 40) or erosion (Illustration 41, Illustration 42). On these figures, the 1956 permanent vegetation line is represented in green and the present day vegetation line inred, while the background uses the 1956 aerial photo. In the maximum coastal erosion area, the average erosion rate over the 50 years duration is significant and reaches 1 m/y. On the last example (Illustration 42), one can notice that some defence works have been realized to prevent coastal erosion (breakwater, groynes and blocks walls). The same order of rate has been found for the maximum of 1mm/y accretion rate (Illustration 40).

54 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

ion line from 1956 (green line) to 2007(red line) to 2007(red line) (green line) line from 1956 ion on the 1956 aerial photo, South of Hikkaduwa. South of Hikkaduwa. photo, aerial on the 1956 Illustration 39 - Evolution of the permanent vegetat the permanent - Evolution of Illustration 39

Illustration 39 - Evolution of the permanent vegetation line from 1956 (green line) to 2007(red line) on the 1956 aerial photo, South of Hikkaduwa.

BRGM/RP-55553-FR – Final report 55 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

34 0m

Beach progression : 50 m

Illustration 40 - Example of coastline progression during the last 50 years (PVL 1956 in green, PVL 2007 in red; South of Hikkaduwa) - 1956 aerial photo.

230m

Beach regression : 30 to 40 m

Illustration 41 - Example of coastline erosion during the last 50 years (PVL 1956 in green, PVL 2007 in red; South of Hikkaduwa) - 1956 aerial photo.

56 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 42 - 57 m of coastline retreat from 1956 to 2007 in the Danuwala area (in green: 1956 permanent vegetation line; in red: 2007 permanent vegetation line).

BRGM/RP-55553-FR – Final report 57 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

The overlay of 2007 permanent vegetation line on Quickbird image extracted from Google Earth allowed to check the accuracy of data acquired on field on the limit between bare sand and vegetation (permanent grass covering soil, bush or trees as coconuts; Illustration 42). Coastal Composite Hazard. This layer corresponds to the crossing of different hazards. It is thus possible to know what type and intensity of hazard can occur at each point (Illustration 38). For example, this layer can be useful for selecting all the building prone to a high level of Sea Level Rise hazard and prone to a high level of tsunami hazard.

3.4.4. Assets folder

The assets layer (Illustration 44) results from a new classification of the Land Use layer (Illustration 31) given by the CCD on the Galle city. The 26 original classes has been regrouped in 6 major classes which are : Communication, Dwelling, Dwelling & economic, Infrastructure (like harbour), religious and tourism. The assets folder contain also every type of assets extracted from others layers like roads, bridge, building etc… It will be also possible to add also environemental assets like natural parks or specific landscapes that must be preserved as environmental resources.

3.4.5. Elements at risk sub-folder

This sub-folder contains 2 layers of specific hazard (tsunami, Sea level Rise). The elements exposed to the risk taken into account are: bridges, communications network (roads and railways) and buildings. Elements at risk were determined by crossing the given level of a hazard with different layers of exposed element like housing, building, road, railways, bridge…. The value of the level of the hazard has been directly affected to each object in a specific field of the attributes table. Once this work has been realized, one can compute the number of buildings, roads, bridges affected by a given level of a specific hazard (low, medium, high). Cartographic representation is presented in the Illustration 45 and in the Illustration 46.

58 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

ap (Sea Level Rise and Tsunami)on the Galle city. the Galle Tsunami)on and ap (Sea Level Rise Illustration 43 - Example of a composite hazard m a composite - Example of Illustration 43

Illustration 43 - Example of a composite hazard map (Sea Level Rise and Tsunami) on the Galle city.

BRGM/RP-55553-FR – Final report 59 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka Illustration 44 - Assets layer (Galle). (Galle). layer - Assets Illustration 44

Illustration 44 - Assets layer (Galle).

60 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 45 - Elements exposed to the tsunami hazard (Galle). (Galle). to the tsunami hazard - Elements exposed Illustration 45

Illustration 45 - Elements exposed to the tsunami hazard (Galle).

BRGM/RP-55553-FR – Final report 61 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka Sea Level Rise risk (Galle). (Galle). risk Sea Level Rise Illustration 46 - Elements exposed to - Elements exposed Illustration 46

Illustration 46 - Elements exposed to Sea Level Rise risk (Galle).

62 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

The Illustration 47 gives an evaluation realized using the GIS of the number of buildings in the pilot site affected by the tsunami hazards and by the Sea Level Rise in 2100. It also gives an evaluation of the number of kilometres of road affected by the 3 levels of these 2 hazards. The accuracy of localisation and identification of buildings and different networks is the base of a correct evaluation.

This evaluation shows that the progressive sea level rise will also significantly affect villages, roads, bridges… near the coastline. This progressive inundation is of course of a different nature than the tsunami inundation but must be taken into account by the decision makers in their projects.

Hazard Tsunami hazard Total Sea Level Rise (2100) Total Level Low Medium High Low Medium High Building (Number) 5 166 7 097 4 821 17 084 11 052 4 419 1 811 17 282 Bridge (Number) 50 120 24 194 201 106 41 348

Communication Network (km) 119 278 130 527 340 203 60 603 Road (km) 105 226 107 439 303 186 55 545 Railways (km) 14 52 22 88 36 17 4 58 Illustration 47 - Quantification of elements at risk in the pilot site for Tsunami and Sea Level Rise.

3.5. THE 100 M LIMIT

Following the 2004 disaster, the sri lankan government delimited a “buffer zone” of 100 meters landward, on which it is strictly forbidden to settle.

Comparison of the limit of “buffer zone” (Illustration 41 to 44) with data from maximum inundation due to the 2004 tsunami, destruction limit (when exists) and tsunami empirical model (high hazard value is in red) shows clearly that this limit of 100 m is not always relevant. In some areas, this 100 m distance corresponds to the limit of destruction of the tsunami and could be satisfying. However, in other places, because of the morphology of the coast, the strip affected by the tsunami is broader while in others cases, this last one is narrower. The following figures (Illustration 51, Illustration 50, Illustration 49, Illustration 48) are examples of different cases existing in the pilot site.

BRGM/RP-55553-FR – Final report 63 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka s red line), the 2004 tsunami inundation limit inundation the 2004 tsunami s red line), and tsunami hazard around Hikkaduwa. Hikkaduwa. hazard around and tsunami Illustration 48 - The “buffer zone” (continuou zone” - The “buffer Illustration 48

Illustration 48 - The “buffer zone” (continuous red line), the 2004 tsunami inundation limit and tsunami hazard around Hikkaduwa.

64 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

s red line), the 2004 tsunami inundation limit inundation the 2004 tsunami s red line), and tsunami hazard around Galle. hazard around and tsunami Illustration 49 - The “buffer zone” (continuou zone” - The “buffer Illustration 49

Illustration 49 - The “buffer zone” (continuous red line), the 2004 tsunami inundation limit and tsunami hazard around Galle.

BRGM/RP-55553-FR – Final report 65 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka red line), the 2004 tsunami inundation limit and inundation the 2004 tsunami red line), tsunami hazard around Koggala. around tsunami hazard Illustration 50 - The “buffer zone” (continuous (continuous zone” - The “buffer Illustration 50

Illustration 50 - The “buffer zone” (continuous red line), the 2004 tsunami inundation limit and tsunami hazard around Koggala.

66 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

s red line), the 2004 tsunami inundation limit inundation the 2004 tsunami s red line), and tsunami hazard around Welligama bay. Welligama hazard around and tsunami Illustration 51 - The “buffer zone” (continuou zone” - The “buffer Illustration 51

Illustration 51 - The “buffer zone” (continuous red line), the 2004 tsunami inundation limit and tsunami hazard around Welligama bay.

BRGM/RP-55553-FR – Final report 67 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

3.6. DIFFICULTIES ENCOUNTERED DURING GIS CONSTRUCTION

3.6.1. Location of data

One of the first problems encountered was to identify each data and institution in charge of this data. The first mission has partially solved this problem. The meetings with the different Sri Lankan institutions have been the occasion to elaborate the list of required data, their location and type.

3.6.2. Data format

Various and numerous data has been acquired by Sri Lankan institutions but this data was generally acquired for the specific needs of each institution without dialogue with the others. As a consequence, the data was stored in different formats and on different media (from paper to database). Sometime, the same types of data were acquired by different institutions with different resolutions and objectives. So meetings organised by GSMB with different institutions allowed identifying the origin of most of the basic data.

3.6.3. Structure of numerical data

Generally, the structure of numerical data is not adapted for integration in a GIS due to the fact that data were integrated in CAD software as a drawing layer. Moreover, we note that the types of objects integrated in a drawing layer are not homogeneous and are of different types. For example, on some areas the same layer contains grids and rivers whereas in other areas it contains rivers and roads. In some files, all types of data are integrated in the same layer with only different graphical attributes. As a consequence, CAD layers were integrated one by one in the GIS, in order to obtain homogeneous information for each layer. For example in the CD files, contour lines were available as line without information on the altitude value. This information was available in another file, as a text. So we integrated all contour lines with the basic information on altitude. Altitude points were also finally added to the GIS layer, in order to obtain a clean final file ready to compute a Digital Elevation Model.

3.6.4. Coordinate system and accuracy in projection system

For GIS purpose, coordinates system of all the data must be the same and computed with the same parameters. Integration of original data coming from the Survey Department, the CCD and GSMB has shown that the superposition of these data was impossible because of the use of different parameters in the coordinate system. Error due to this problem of system coordinates can be of several hundred meters. Some other problems result from the use of non native coordinates acquired with GPS. In this case the error comes from the imprecision of conversion from Lat-Long coordinate to Kandawala system in the GPS device.

68 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

The good projection parameters were finally obtained from Survey Department and all of the data acquired in the different institutes were projected in this projection:

Projection: Transverse Mercator projection:

Everest India: 1830

False Easting: 200000

False Northing: 200000

Central Meridian: 80° 46’ 18.16” E

Central parallel: 7° 00’ 1.729 N

Scale Factor: 0.9999238418

Datum: Kandewala :

Delta A: +860.600

Delta F: 0.28361363

Delta X: -97 m

Delta Y: + 787 m

Delta Z: + 86 m

A part of the work realized has consisted to define for all data the good projection parameters and to compute the coordinates. For most of the data the homogenization and correction was possible but for some others no. In this last case, data are unusable in the GIS and for the project.

BRGM/RP-55553-FR – Final report 69

GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

4. The Risk Scenario

Defence against the effects of a strong coastal event is a major concern for national security in Sri Lanka. The objective is to develop a risk scenario that will model potential damages to the human and built environment. This scenario could provide the basis for updating emergency plans and procedures, as well as developing and conducting training exercise programs for agencies and technical staff to identify shortfall needs. Such results and actions are of fundamental interest for the management/reduction of coastal risk (including presentation activities and crises management).

The modelling mainly relies on three elements: - the exposed elements (built environment and population); - the vulnerability of the built environment; - the damage evaluation algorithms.

It is important that the model allows to upgrade and develop the databases.

The outputs of the model are strongly dependent on the available input data, but it should be able to identify “weak points” in human settlements as well as to test the exposure of the future development.

4.1. VULNERABILITY FUNCTIONS

4.1.1. Field work

Three periods of field investigations on the pilot site were led during the project in order to acquire and check data. The aim was to check the incidence limits of the 2004 tsunami (inundation, destruction…) and to acquire data on buildings which could help to build risk scenarios.

A survey sheet was developed during the first field work and then improved with the help of our sri lankan colleagues and interviewed people‘s reactions. The idea was to find an “easy” way to collect information such as the wave arrival time, the duration and the height of the wave, the number of waves, the type of the building, the damages observed, the place where the inhabitants took refuge. Most of the answers of this form have been stored into a database in order to make the results easier to exploit.

This form is shown in the next page (Illustration 52).

BRGM/RP-55553-FR – Final report 71 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Interview n°… Sex (male/female/family) : Coordinates : Age (<18yo, 18-55yo,>55yo, all): Has been Seen or Heard or Both or Unknown: Body height :

Questions 1 - Where were you at the beginning of the event: A - inside your house, B - at home but outside the house, C - on work, D - in the neighbourhood, E - in a public place (which one), F- others to be precised. 2 - When did you leave the place you were at the beginning of the event: I - immediately or L - later, if later precise when. 3 - Where were you at the end of the event: A - at the same place, B - on a roof, C - in a tree, D - far from the beach, E - far from the beach and in a high place, F – religious place, G - others to be precised 4 – Estimation of wave or progressive sea level increase with types of floatting objects identified: A - branches, B - trunks, C - cars, D - boats and ability of people to E - stand up or F - to swim in the water yes or no. 5 - Sea level increase duration : value in min, or U -unknown or S –sudden. 6 - Sea level decrease duration :value in hour or U – unknown. 7 - Inundation duration. 8 - If you took refuge, where was it: A - inside the house (ground level), B - inside the house (upstairs), C - on the roof, D - in a tree, E - in a religious place, F - far from the beach, G - far from the beach and in a high place, H - others to be precised. 9 - When did you leave it (come back): after the end of the flooding Y - yes or N - no with details. 10 - Wave number (if seen to be precised). 11 - When did each wave arrive? (arrival time of the waves). 12 - Maximal tsunami wave height or water level: A - under the knee, B - under the waist, C - under the chest, D - under the neck, E - under the first floor of houses, or F - higher, then to be precised. Which wave was the highest one ? 13 - Height of first wave : same reference as question 12 or in m. 14 - What sort of floating/flooding rubble or objects did you see: A - rubble, B - branch, C - furniture, D - trunk, E - car, F - boat, G - boulder, H - large pieces of walls, I - bodies, J - others to be precised. 15 - Building type :C - concrete, B - brick, CB1 - ciment block, CB2 - stronger ciment block, L – light. 16 - Fence type :N – None, C - concrete, B - brick, CB - ciment block, L – light. 17 - Do you remember damages on building? : N -no, Y – yes. If yes, which one: 1 - damage limited to chipping of plaster, minor visible cracking, damage to windows or doors, 2 - failure or collapse of parts or whole sections of wall panels, scouring at corners leaving foundations partly exposed, 3 - failure or collapse of masonry wall panels, most parts of structure collapsed, excessive scouring and collapse sections of structure due to settlement, 4 - complete structural damage or collapse, foundations and floor slabs visible and exposed, collapse of large sections of foundations and structures into scour holes. 18 - Was the top of the layer deposit inside the house A – None, B - lower than the ankle, C - lower than the knee or D - higher than the knee? 19 - Did cut road occur? If yes, number and places to be precised. 20 - Did persons succeed to resist to the wave/flooding? Y - yes or N - no or U - unknown. 21 - If a similar event occurs, what would you do: S - the same or E – something else. And if you took refuge, would it be A - in the house, B - on the roof of the house, C - in a tree, D - far from the beach, E - far from the beach and in a high place or F - others to be precised. 22 – Human losses : number of wounded persons, number of dead bodies, I - undefined or U – unknown. 23 – Other comments.

Photos: Illustration 52 - Post-tsunami form.

In Unawatuna Bay, this field work was specifically led in order to build the vulnerability functions. The building layer of the GIS needed to be detailed and completed with information about the type of the buildings. This specific form is a shorten version of the precedent one, with questions only on the water height, the number of floors, the type and the age of the building and the damages observed in 2004. An example of this form is given in the Illustration 53 .

Water Number BUILDING_TYPE- Building Number Height of floor DAMAGES_FIELD Photo Comments FIELD Age - field field 5-6 new in concrete 1 2 CB2+C 1 feet 2004 columns Illustration 53 - Form used in Unawatuna, specific to building damages.

72 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

In parallel to the exploitation of this field work, bibliographic researches were led in order to help us defining the type of building and damages that were observed in Sri Lanka and in particular in our pilot site.

4.1.2. Damage scale

It appears from the observations on the field and the studies led immediately after the 2004 tsunami, that the main damage mode is the out-of-plane failure of load bearing structural walls caused by the fluid pressure from the tsunami flow. Scouring was also responsible for structural damage due to foundation collapses or subsidence most of the time in areas where the upper soils are largely sand.

The level of damage varied along the coastline but it is generally admitted that it diminishes with distance from the coast. Although it is difficult to define the level of structural damage to buildings, some were able to resist the tsunami water pressures but suffered structural damage due to scour, damage scales considering structural and geotechnical failures can be defined. The one we chose to use was created by EEFIT in 2005 (Earthquake Engineering Field Investigation Team). Our choice was justified by the fact that: - their study area included our pilot site; - they used the quite numerous data of the Department of Census and Statistics of Sri Lanka; - the comparison between our field observations and this damage scale was quite good.

The scale is described in the Illustration 54 and illustrated in the next pages (Illustration 55 to Illustration 58).

Damage Scale Description No Damage (D0) No visible structural damage

Light Damage Damage limited to chipping of plaster, minor visible cracking, damage to (D1) windows, doors. Damage minor and repairable. Immediate Occupancy Out-of-plane failure or collapse of parts of or whole sections of wall panels Moderate Damage without compromising structural integrity. Scouring at corners leaving (D2) foundations partly exposed, repairable by backfilling. Masonry walls repairable. Unsuitable for immediate occupancy but suitable after repair. Out-of-plane failure or collapse of masonry wall panels beyond repair, Heavy Damage structural integrity compromised. Most parts of structure collapsed. (D3) Excessive scouring and collapse of sections of structure due to settlement. Requires demolition since unsuitable for occupancy Complete structural damage or collapse, foundations and floor slabs visible Collapse (D4) and exposed. Collapse of large sections of foundations and structures into scour holes Illustration 54 - Damage scale for masonry structures.

BRGM/RP-55553-FR – Final report 73 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 55 - Illustration of structural damage category D1.

Illustration 56 - Illustration of structural damage category D2.

74 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 57 - Illustration of structural damage category D3.

Illustration 58 - Illustration of structural damage category D4.

BRGM/RP-55553-FR – Final report 75 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

This damage scale is the most relevant for masonry structures, but we considered it was suitable too for the other types of buildings one can find in Sri Lanka.

4.1.3. Building typologies

From all the detailed surveys carried out after the 2004 tsunami, we inferred that the predominant building type exposed to the tsunami in Sri Lanka was masonry structures mainly used for residential purposes. The houses are usually single storey constructions, with a maximum of three or four rooms. The height of the walls varies between 2.5 and 3.0 m. The quality of these buildings vary according to the material used: from semi-fired clay blocks for poor quality masonry structures to solid bricks or cement blocks for good quality masonry constructions. The average wall thickness is around 250 mm, but some half-brick thick walls can be found. The foundations are often superficial and consist in either a thick reinforced concrete raft foundation or masonry strip foundations.

Two other types of structures can be found in Sri Lanka: - Timber light frame construction; - Low-rise reinforced concrete construction, common on buildings with up to one storey: structures with solid brick masonry wall panels for exterior and interior walls.

From these observations and the results of the interview led during the fields, we defined a typology with more than three categories of buildings on our pilot site in our forms in order to get precise data: - L: the more fragile type of construction, it contains all the timber light frame constructions, non temporary constructions (made of wood, sheet metal…)…; - B1: buildings made of bricks (light ones); - B2: buildings made of stronger bricks (often 2 rows in a wall); - CB1: buildings made of cement blocks; - CB2: buildings made of stronger cement blocks, sometimes with reinforced columns; - C: reinforced concrete construction; - LB: traditional construction using coral or limestone as material, stronger than “classic” bricks.

We will see in the next chapter that for some practical reasons, only three vulnerability functions have been elaborated, linked with the three original types of construction.

For the needs of risk scenarios, we also defined a random distribution of the three main types of buildings: light constructions (L), masonry constructions (B1 and CB1) and reinforced or stronger constructions (B2, CB2, LB and C).

76 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

The distribution of our field data (a panel of 160 buildings) on Galle and Unawatuna gave us the following distribution (Illustration 59):

Masonry “Reinforced” Building typology Light constructions constructions constructions

Percentage 7 66 27 Illustration 59 - Buildings distribution observed on the field.

Besides, according to the Census of Buildings and Persons affected by the tsunami 2004 realised by the Department of Census and Statistics in 2005, more than 15% of the buildings affected by the tsunami were made of non temporary materials in Galle district.

Therefore we chose to consider that 15% of the buildings of our pilot site were light constructions, which is consistent with the fact that we registered some mixed houses (wood and cement blocks) as masonry constructions and led us to this new distribution (Illustration 60):

Masonry “Reinforced” Building typology Light constructions constructions constructions

Percentage 15 55 30

Illustration 60 - Buildings distribution used during simulations.

4.1.4. Vulnerability functions

For each category of buildings, vulnerability for structural damage can de defined using a set of fragility curves, each of which expressing a particular damage state (D0 to D4) given a certain value of the demand parameter.

According to the FEMA55, all the forces (wave impact, hydrostatic, hydrodynamic, buoyancy, debris impact and scouring) acting on a building due to a tsunami depend on the tsunami water height and the flow velocity. Assuming the fact that the velocity of a tsunami at its arrival on the shoreline as well as on subsequent profiles onshore is difficult to measure whereas the tsunami water height at the shoreline as well as at a certain distance from the coast can be deduced by some physical evidence (silt traces on buildings, debris heights, witnesses…), we decided to use the tsunami water height as the demand parameter to elaborate our vulnerability functions.

BRGM/RP-55553-FR – Final report 77 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

In order to minimize the number of vulnerability functions, we decided to use only three different functions: one for the light constructions (L), one for the masonry constructions (B1 and CB1) and the last one for the reinforced or stronger constructions (B2, CB2, LB and C).

a) Masonry residential properties

For masonry residential properties, we chose to use the vulnerability functions developed by Navin Peiris in his article “Vulnerability functions for tsunami loss estimation” exposed at the First European Conference on Earthquake Engineering and Sesimology” in Geneva (Switzerland) in Septembre 2006 (3 to 8). M. Peiris’ functions were developed for coastal areas in Sri Lanka using the data issued from the house-to- house survey carried out by the Department of Census of Sri Lanka (DCS) in 2005 and the field surveys led by the Earthquake Engineering field Investigation Team (EEFIT) in 2005 in the southwest of Sri Lanka.

Those vulnerability functions are lognormal cumulative distribution functions of tsunami water height as the demand parameter.

For the use of our study we need to know the percentage of buildings affected to a certain state of damage for a certain water height (Illustration 61).

Water height (m) Damage 0 1 2 2.5 3 3.5 4 4.5 5 5.5 6 7 8 9 10 D0-D1 100 90.0 21.3 7.7 2.4 0.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 D2 0 6.0 30.0 23.6 14.9 8.6 5.3 2.9 1.6 1.4 0.7 0.0 0.0 0.0 0.0 D3 0 2.9 26.7 28.7 26.7 20.3 16.0 11.7 7.7 5.3 4.0 2.4 1.3 0.5 0.0 D4 0 1.1 22.0 40.0 56.0 70.4 78.7 85.4 90.7 93.3 95.3 97.6 98.7 99.5 100.0 Illustration 61 - Values of the vulnerability function for masonry constructions

The Illustration 62 presents a characterization of this set of fragility curves.

78 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Vulnerability functions (Peiris) D4 D4+D3 D4+D3+D2 100

90

80

70

60

50

40

damage state probability (%) probability state damage 30

20

10

0 012345678910 highest submerged height (m)

Illustration 62 - Vulnerability functions for masonry constructions.

This vulnerability function is consistent with the small panel of data (around 60 masonry constructions) collected on the field in Galle and Unawatuna.

b) Other constructions

For other constructions, we needed two more vulnerability functions: one for the quite solid constructions, including concrete, reinforced buildings, traditional ones (using coral or limestone for example), and another for light constructions (using non temporary materials: wood, steel sheet…).

We could not find work any study about those kinds of vulnerability functions during our bibliography. Only a few articles mentioned some global data about the behaviour of certain buildings during the tsunami.

Matsutomi et al. (2006) give some information about damage for concrete buildings: - when the height of water remains inferior to 5 m, buildings were very little damaged (equivalent to D0 and D1); - above 5 m of water, building were more damaged (around D2); - very important damage (D3 to D4) occurred only for very high waves (around 20 m).

BRGM/RP-55553-FR – Final report 79 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Papadopoulos et al. (2006) give information on the behaviour of the 3 main types of buildings: - the reinforced concrete structures (slab floors, walls and vertical and horizontal columns) were flooded but endured only light damage (equivalent to D0 and D1); - the less recent buildings, with concrete slab floor, columns and cross beams and unreinforced field stone (coral blocks) fill in the walls, endured various damages (equivalent to D1 to D3); - the oldest buildings made of unreinforced field stone (coral blocks) endured the most extensive forms of damage (equivalent to D2 to D4).

Matsutomi et al. (2001) give relations between inundation depth and degree of damage of cement blocks or stone brick houses following the 1996 Tsunami in Irian Jaya: - if the inundation depth is less than 2.5 m, the degree of damage is equivalent to D0 and D1; - if the inundation depth is between 2 and 3 m, the degree of damage is equivalent to D2; - above 3 m of inundation depth, the degree of damage is equivalent to D3.

For Ghobarah et al. (2006) the impact of the 2004 tsunami on the 3 classes of building is the following: - residential wood houses with tile or corrugated steel sheet roof were washed away by the tsunami (degree of damage equivalent to D4); - non-engineered concrete construction (clay or brick or block masonry walls) were severely damaged or collapsed (degree of damage equivalent to D2 to D4); - engineered reinforced concrete constructions endured minor damage even with high tsunami run-up levels (degree of damage equivalent to D0 to D2).

For the light constructions, our field data (11 light constructions in Galle and Unawatuna) show us 2 main principles: - if the inundation height is superior to 2 m, more than 85% of the buildings are in D4 (complete collapse); - if the inundation height is superior to 1 m, light buildings at least suffer minor damages (D1 or D2).

From these assertions and the shape of the vulnerability function for masonry constructions, we built the vulnerability function for light constructions showed below (Illustration 63).

80 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

D4 Vulnerability functions (L) D4+D3 D4+D3+D2 100

90

80

70

60

50 (%)

40

damage probability state 30

20

10

0 00.511.522.53 highest submerged height (m)

Illustration 63 - Vulnerablity functions for light constructions.

For the reinforced or stronger constructions our field data (24 constructions in Galle and Unawatuna) show us 2 main principles: - if the inundation height is inferior to 1.5 m, more than 85% of the buildings suffer only minor damages (D0 and D1); - important damages (D3 and D4) barely occurred except for very high waves.

From these assertions and the shape of the vulnerability function for masonry constructions, and assuming the fact that reinforced concrete buildings were more solid than others in this category, we built the following vulnerability functions for reinforced constructions (Illustration 64).

BRGM/RP-55553-FR – Final report 81 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

D4 Vulnerability functions (CB2, LB, C) D4+D3 D4+D3+D2 100

90

80

70

60

50

40

30 damage state probability (%) damage state

20

10

0 012345678910 highest submerged height (m)

Illustration 64 - Vulnerability functions for reinforced concrete constructions.

4.2. MODEL AND SCENARIO

4.2.1. The software: ARMAGEDOM

ARMAGEDOM is a BRGM tool for risk assessment on built areas and allows to simulate the impact of a seismic event, a tsunami …on built areas (Sedan & Mirgon 2003).

From the definition of the aggression and the attribution of fragility curves (or vulnerability functions) to the exposed assets, it can simulate the consequences (scenarios) on exposed elements. The results can then be used in tools for synthesis or GIS for example.

Knowing the epicentre location, the source focal depth, the magnitude of earthquake scenario and the empirical attenuation law, ARMAGEDOM can generate the seismic aggression thanks to its “Phenomenon Modulus”, but such a function is not available for tsunami. The aggression has to be calculated by another way (computer simulations for example) and then used under a GIS or raster format in ARMAGEDOM.

82 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

There are three types of exposed elements in ARMAGEDOM: - single buildings; - districts of buildings: groups of homogeneous types of buildings (distribution consistent with respect to vulnerability assessment) or data (number of buildings, inhabitants, percentage of each building type present in the district…); - linear elements: pipelines, lifelines, transportation routes….

The user needs to implement impact and/or vulnerability functions in ARMAGEDOM in order to convert the aggression into an impact. ARMAGEDOM can deal with different types of functions: - numerical function: one aggression value (for tsunami the inundation height for example) for one impact value. The damage functions then give a percentage of damage per known height; - classes functions: one aggression value or descriptive class (moderate, medium and high for example) for a range of values. The function is then a table; - re-affectation: it gives equivalence between values and classes, with the same representation as for the classes function; - probability functions.

There are also different ways to generate those functions. The user can directly create new functions or modify the existing ones in ARMAGEDOM or import existing functions (from ASCII or ARMAGEDOM binary files). In the case of imported files, several functions are allowed in the same file.

There are different ways to assign the impact functions to each exposed element: - deterministic approach: the user can assign one function to one element, all elements at once or by selecting affected elements (graphically or thematically); - random approach: the user assign in a random manner with a statistic treatment of areas with random repartition of exposed elements or by defining the percentage of each impact function type in a given area.

The user can also directly build a new table (ascii text file) with the ID of the exposed element and the corresponding impact function.

ARMAGEDOM is used to simulate the material and structural damage of an event such as a tsunami on built areas by using vulnerability functions, but it can also be used to estimate the cost of such an event by using impact functions. The user will then have to define both vulnerability and impact functions. After having assigned each type or value of damage a cost (direct and/or indirect) and calculated the structural impact on the exposed elements, a second simulation will provide the financial impact of the event.

BRGM/RP-55553-FR – Final report 83 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

ARMAGEDOM has been used on two spots of our pilot site: Galle and Unawatuna.

The aggression used is issued from the empirical model of the 2004 tsunami. The vulnerability functions used are those described above, all of them are probability functions (numerical entry to classes result).

The impact of the tsunami on assets in terms of infrastructures (most specially roads and bridges) was not simulated because of a lack of data needed to build a vulnerability function adapted to such infrastructures. But ARMAGEDOM can simulate impacts on linear elements and this characteristic could be used in the future.

4.2.2. Testing on Unawatuna

Specific data were collected on Unawatuna Bay on buildings, including the type of the building surveyed. Those data (more than 90 buildings) were used as a test panel in order to determine whether the results obtained with the simulations were consistent once compared to the field observations. The aim was to validate first the vulnerability functions and then the random distribution.

a) First test: vulnerability functions

For this first test, we ran the simulation only on the buildings surveyed on the field. Since we knew the type of each building, we assigned the correspondent vulnerability function to each element by selecting thematically the affected elements.

Then we compared the results to the reality observed.

84 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 65 - Distribution of the different types of buildings (Unawatuna test area).

The first comparison between the damage calculated and the damage observed showed many differences due to the fact that vulnerability functions are probability functions. But the positive point was that there was neither optimistic (damage calculated less important than damage observed) nor pessimistic (damage calculated more important than damage observed) trend in the results using the second curve for concrete buildings.

In order to be sure that our results were coherent to our observations, we then compared the calculated and the observed repartition of the buildings affected in each class of damage (Illustration 66).

Damage distribution D0-D1 D2 D3 D4 Calculated occurrences 38 8 6 42 Observed occurrences 32 17 15 30 Calculated percentage 40.4 8.5 6.4 44.7 Observed percentage 34.0 18.1 16.0 31.9 Illustration 66 - Damage repartition: comparison of calculated results and observations.

BRGM/RP-55553-FR – Final report 85 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

The distribution between the buildings damaged (D2, D3 and D4) and those left apart (D0- D1) is quite good according to our results. Among the buildings damaged we can observe a slight over estimation of the amount of D4 damaged houses. Taking into account the facts that this test is realized on a small sample of buildings, and that the vulnerability functions have been built with a restricted amount of data, the result of this comparison is quite correct, which allows us to believe that our vulnerability functions are enough relevant. And that the hazard (inundation height) is approximated with a simple function.

Illustration 67 - Results of the first “test simulation” (Unawatuna test area). b) Second test: random distribution

For this second test, we still ran the simulation only on the buildings surveyed on the field. Since we wanted to validate our random distribution, we used the random approach to assign in a random manner a vulnerability function to the exposed elements. We used the random repartition defined above where the percentage of each impact function type is given (Illustration 68, Illustration 69).

Masonry “Reinforced” Building typology Light constructions constructions constructions Percentage 15 55 30

Illustration 68 - Random repartition of buildings typology.

86 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 69 - Random distribution of the different types of buildings (Unawatuna test area).

Then we compared the results to the reality observed.

In order to be sure that our results were consistent with our observations, we compared the calculated and the observed repartition of the buildings affected in each class of damage (Illustration 70).

Damage distribution D0-D1 D2 D3 D4 Calculated occurrences 30 6 10 48 Observed occurrences 32 17 15 30 Calculated percentage 31.9 6.4 10.6 51.1 Observed percentage 34.0 18.1 16.0 31.9 Illustration 70 - Damage repartition: comparison of calculated results and observations.

The repartition between the buildings damaged (D2, D3 and D4) and those left apart (D0-D1) is still quite good according to our results although there still is a slight over estimation of the amount of D4 damaged houses. Taking into account the fact that this test is realized on a very small sample of buildings, the result of this comparison is quite correct, which allows us to believe that our random distribution is relevant, even if it still can be precised.

BRGM/RP-55553-FR – Final report 87 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 71 - Results of the second “test simulation” (Unawatuna test area).

c) Application to Unawatuna Bay

After those two tests, we ran the simulation on all the buildings with the same random distribution in charge to assign in a random manner a vulnerability function to all exposed elements.

The Illustration 72 presents the assets in terms of buildings of Unawatuna Bay and the aggression used for the simulation.

The first step of the global simulation is to assign the vulnerability functions to the buildings. The following picture shows the result of our random distribution.

It is important to say that ARMAGEDOM will provide a new distribution each time since the user asks for a random distribution. Two factors can determine whether the user has to choose a random distribution to assign a vulnerability functions to the buildings exposed: - the lack of data: in our case, we couldn’t point each building and assign it the relevant vulnerability function depending on its type;

88 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 72 - Unawatuna Bay: assets and aggression (height of water from 6 to 0 m).

- the “quicker” way: if the user only wants to have general ideas on the impact of a specific event, or if he wants to check the consistency of a hypothesis, using a random distribution is the best solution since it needs less work and preparation on the data before using ARMAGEDOM.

We used random distribution because of the lack of data on building types (Illustration 73). With all the information on buildings’ characteristics a more precise simulation could have been run. It is the short term aim for such a project.

ARMAGEDOM then converts the aggression in solicitation on the exposed assets (for us, the buildings, Illustration 74).

And finally the user can see the impact of the 2004 tsunami on Unawatuna Bay on the following picture (Illustration 75). Even if we used a random distribution for this simulation, we can see that the results are consistent with our field observations: the front of the beach is the area the most affected.

BRGM/RP-55553-FR – Final report 89 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 73 - Random vulnerability functions distribution (Unawatuna bay).

Illustration 74 - Aggression converted in height of water (m).

90 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 75 - Unawatuna: Impacts of the 2004 tsunami.

4.2.3. Galle

Assuming the fact that the vulnerability functions and the random distribution tested on Unawatuna could be used on another area, we ran the same type of simulation on Galle, by assigning all the buildings in a random manner a vulnerability function.

The Illustration 76 presents the assets in terms of buildings of Galle and the aggression used for the simulation.

For the same reasons as for Unawatuna the impact of the tsunami on assets in terms of infrastructures (most specially roads and bridges) was not simulated.

The first step of the global simulation is to assign the vulnerability functions to the buildings. The following picture (Illustration 77) shows the result of our random distribution.

ARMAGEDOM then convert the aggression in solicitation on the exposed assets (for us, the buildings; Illustration 78). And finally the user can see the impact in term of damage (D0 to D4) on buildings of the 2004 tsunami on Galle (Illustration 79). Even if we used a random distribution for this simulation, we can see that the results are coherent with our field observations: the front of the beach is the area the most affected.

BRGM/RP-55553-FR – Final report 91 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka from 6 to 0m). 6 to 0m). from Illustration 76 - Galle: assets and aggression (height of water (height and aggression - Galle: assets Illustration 76

Illustration 76 - Galle: assets and aggression (height of water from 6 to 0m).

92 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

(random) - Galle. Galle. (random) - ons distribution ons distribution Illustration 77 - Vulnerability functi

Illustration 77 - Vulnerability functions distribution (random) - Galle.

BRGM/RP-55553-FR – Final report 93 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka Illustration 78 - Aggression applied to buildings (height of water). (height applied to buildings - Aggression Illustration 78

Illustration 78 - Aggression applied to buildings (height of water).

94 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 79 - Galle: Impacts of the 2004 tsunami. of the 2004 - Galle: Impacts Illustration 79

Illustration 79 - Galle: Impacts of the 2004 tsunami.

BRGM/RP-55553-FR – Final report 95 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

4.3. RESULTS OF THE SIMULATION APPLIED TO GALLE

If we want to analyse more in detail the results of our simulations, we can compare them to those of the numerical model on the GIS and to the data collected in 2005 by the Department of Census and Statistics.

For Galle, the comparison with the results of the survey led by the Department of Census and Statistics is exposed in the following table.

Buildings Dept of Simulation affected by Census and Difference (Damage D2 to D4) Tsunami Statistics Galle 1 443 2 066 623 Illustration 80 - Comparison of the number of buildings affected.

For Galle, the results of the empirical model on the buildings give 3 528 buildings affected (§ 3.4.5), but this estimation takes into account all the buildings in the inundation area which correspond to the D0-D1 categories. If we select in the GIS only buildings with a high and medium level of tsunami hazard, we obtain 2 332 buildings affected in the whole Galle district. If using the GIS we select the same area that those used with ARMAGEDON, we obtain around 1900 buildings affected by high to medium level of tsunami hazard.

So we have 2 066 buildings affected for the Dept of Census & Statistics, 2 332 using GIS and tsunami empirical model and 1 443 using ARMAGEDOM.

Knowing the fact that the Department of Census and Statistics considered that only the buildings damaged were affected (like we did in ARMAGEDOM), and that the results obtained with the two methods are quite close, the simulation on Galle is correct. The underestimation we can observe comparing our results to those of the Department and to those coming from GIS is due to: - the fact that in ARMAGEDOM we can not include in the buildings damaged those affected only by minor damages (D1), due to the vulnerability functions we defined; - The area computed with ARMAGEDON is smaller than the administrative Galle limit.

4.4. PROSPECTS

The vulnerability functions derived here could be improved, in particular the repartition between the D2, D3 and D4 damages, and therefore used to develop a component based damage model for tsunami loss estimation to residential property. It is anticipated that the vulnerability functions developed here could be extended to other elements under risk like infrastructure (roads, railways, bridges…).

96 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

With detailed data especially on buildings, the users could run more accurate simulations (without using a random distribution). A precise an updated layer of the buildings could be used to lead an urban planning policy taking into account the assets at risk.

This method provides a tool allowing its user to estimate the damage on the assets and not only the hazard level. Besides since the number of assets, buildings for example, damaged is known it is possible to evaluate the cost of those damages, this tool can be used to predict the economical impact of an event such as a tsunami, which can be useful to predict the cost of a certain tsunami as well as to improve the urban planning policies.

This method also gives the possibility to adapt the tool to the data collected by the user. By using the random or deterministic approach to affect the vulnerability functions to the assets, the user can choose to run a realistic simulation if he has enough data or if his scale study allows it or to run a more global simulation.

The fact that the user can use for an aggression the results of a physical, numerical or empirical model, the data collected after a real event is an evidence of the adaptability of this type of tool.

For all these reasons, this kind of study including the use of such a tool is really important in order to improve the risk management and crisis policies.

BRGM/RP-55553-FR – Final report 97

GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

5. Numerical modelling

This chapter discusses the modelling of the waves which caused the largest damage on the pilot site, namely those generated at the time of the monsoon and of the tsunamis like that of December 2004.

5.1. MONSOON WAVE MODELLING

When one is interested in the coastal risks, it is critical to take into account the action of waves which affect the navigation and the dimensioning of the defence works. Moreover, waves can generate littoral currents (longshore, undertow and rip current) dangerous for the leisure activities (swimmers casualties) and which control the coastline evolution (processes of erosion and accretion). Finally the run-up and the set- up induced by the waves can combine, during storms, with the storm surge (related on the wind and the atmospheric pressure) and thus cause the flooding of coastal areas.

Monsoon waves are seasonal and associated on wind coming from the SW or from NE.

5.1.1. Data inputs

We present hereafter the input data necessary to model as well as results of the numerical simulations around the bay of Galle. Data inputs concern the topography, the bathymetry and waves characteristics. a) Topography and Bathymetry

The topographic data used come from Survey Department and the University of the United Nations. Compilation of: - 5 m interval contours of 1:5000 digital maps; - digitalization of a 0.5 m interval contour of hard copy map covering the Galle District; - and of measurements of high precision (Real Time Kinematics GPS).

The collected data were validated and homogenized (suppression of aberrant values at the land-sea interface).

As for the bathymetric data, those come from the NHO (National Hydrographic Office). They correspond to the various surveys carried out on the zones of Galle (see example Illustration 81) and Welligama. These data were also checked and corrected.

BRGM/RP-55553-FR – Final report 99 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 81 - Example of bathymetric sounding realized by the NHO in the Galle offshore.

Following these checks, the topographic and bathymetric data could be merged in order to generate the DEM necessary to modelling (see for example Illustration 82).

b) Wave and Wind data

Three data sources were consulted: - measurements taken by the CCD (Coastal Conservation Department);. - results of the model WaveWatch 3 by NOAA (National Oceanic and Atmospheric Administration); - results of ERA 40 from ECMWF (European Center for Medium-range Weather Forecast).

100 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 82 - 3D view of the bay of Galle (20 m x 20 m mesh size).

Measurements of the CCD were taken at the offshore of Galle at depth around 70 m. They relate to the period 1989-1995. The format of these data had to be modified in order to make them usable. The Illustration 83 shows the values of significant height, period and direction of the waves. In spite of important gaps for the years 1993 and 1994, one notes that the maximum annual values are between May and September (mode of monsoon of SW) what is in agreement with Swan (1983). The waves can reach then Hs of 5.5 m, for periods around 8s with directions of 240° (see Illustration 83 – June 1991).

To supplement these observations, we also consulted the data base WaveWatch3 archived by NOAA since 1997 from ftp://polar.ncep.noaa.gov/pub/history/waves/. They are data of wind and wave at global scale (1.25 by 1.00 degrees longitude/latitude grid between 78.0° north to 78.0° south). WaveWatch3 is constrained by wind fields from the GFS (Global Forecast System) forecast/analysis system, which integrates the data of the operational Global Dated Assimilation Design (GDAS) and the aviation cycles of the Medium Range Forecast Model. The parameters used and calculated by WaveWatch3 were validated using different data sources (conventional deep-ocean buoy observations, ERS2 scatterometer data and ERS-2 FD altimeter wave height observations). The Illustration 84 shows an extraction of the data at the node 5°N - 78.75°E (SW of Sri Lanka). These data confirm the annual cyclicity and occurrence of maximum values between May and September highlighted previously by the observations. The maximum values of waves heights remain those of 1991.

For the wind data before 1997, we consulted the ERA 40 database. It corresponds to the Re-analyses for the 1957-2002 period with 6 hours of temporal resolution of meteorological parameters realized by ECMWF.

BRGM/RP-55553-FR – Final report 101 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

A

B

Illustration 83 - Measurements of the buoy deployed off Galle harbour at a depth of 70 m. A: during the 1989-1995 period, B: details of the 1991 year.

102 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 84 - Extracted data from WaveWatch3 for the Southwern Sri Lanka.

BRGM/RP-55553-FR – Final report 103 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

5.1.2. Numerical simulation of SW monsoon waves

We were interested in the day of the 06/02/1991 because it corresponds to the largest waves recorded in the sector (Illustration 84): Hs = 5.5 m; Tp=8.33 s; Dp=239°.

Wave propagation is simulated numerically using a non-stationary third generation nonlinear model called SWAN (Simulating WAves Nearshore, Booij et al., 1999). SWAN accounts for most of the linear wave propagation effects: shoaling, diffraction, refraction due to current and depth, frequency shifting due to currents and non- stationary depth, and transmission through and reflection from obstacles. It incorporates nonlinear mechanisms for wave generation by wind, white-capping, bottom friction, depth-induced breaking and wave-induced setup. Robust advection schemes give SWAN the flexibility to describe wave propagation at a wide range of spatial scales, from laboratory up to global, on both regular and a curvy-linear grids in a Cartesian or spherical coordinate system. SWAN has the capability of nested runs, using as input data provided either by SWAN or any of the commonly used deep-water.

In this case we set up 2 nested grids. A first grid of mesh 100x100m located offshore of Galle (starting towards 70m of depth) providing the boundary conditions to a second grid of 20x20m focused on the bay (Illustration 85).

The conditions of wind extracted from ERA40 reveal for this day (06/02/1991) an average wind velocity of 15 m/s coming from the WSW (Illustration 86). These values of wind were also used in simulation.

Illustration 87 shows the transformation of Hs from depths of 70 m to the coast. It is noted that until approximately 30 m depth the height of the waves is preserved (no interaction with the sea bottom). From this depth, friction at the bottom makes fall the height of the waves. One also observes a preferential penetration of the waves on the eastern part of the Galle bay. That is confirmed by the model focusing on Galle (Illustration 88). The height of the waves is around 4 to 4.5 m at the entry of bay. The waves reach close to 3 m at the 5 m isobath in the east of bay. It is half of this value (1.5 m) on the western sector of bay. At the coastline, (contour line 0 m in red) the waves reach 1.5 m except for harbour zone and near the Cricket stadium where the values are lower. The generated setup (Illustration 89) is also important reaching on average 0.30 m on the shore. On the bay of Unawatuna the setup is more important and reaches approximately 0.45 m at the “Unawatuna Hotel”.

104 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 85 - Nested grids used used for modelling.

BRGM/RP-55553-FR – Final report 105 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 86 - Wind field June 06/02/1991 00:00 (constructed with ERA40 database).

Illustration 87 - Hs computed with the medium resolution grid, values in m.

106 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 88 - Hs computed with the high resolution grid, values in m. grid, values resolution with the high computed - Hs Illustration 88

Illustration 88 - Hs computed with the high resolution grid, values in m.

BRGM/RP-55553-FR – Final report 107 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka Illustration 89 - Set-up computed with the high resolution grid, values in m. grid, values high resolution with the - Set-up computed Illustration 89

Illustration 89 - Set-up computed with the high resolution grid, values in m.

108 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

5.2. TSUNAMI MODELLING

The 12/26/2004 event started with a main shock at 0h 58’53” GMT, when the locked fault between the plates ruptured at the megathrust earthquake’s hypocenter, located 3.32 N and 95.85 E, i.e., 160 km west of Sumatra, at a depth of 25-30 km, liberating strain accumulated from subduction since the last large earthquakes occurred in the area, in 1861 and 1881. This is the third largest earthquakes ever recorded, with a moment magnitude (Ammon et al., 2005).

We present hereafter the data used for modelling this event as well as the results of simulations.

5.2.1. Data input a) Bathymetry

For these simulations we used the international database ETOPO2 (Illustration 90) which gives world bathymetric and topographic data according to a grid of 2' x2'. This grid is consistent for the generation and the propagation of the tsunami at the Indian Ocean scale. On the other hand its low density does not allow a precise calculation of the run-up and depth of flood at a local scale.

Illustration 90 - ETOPO2’s bathymetry.

BRGM/RP-55553-FR – Final report 109 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

b) Seismic source

Grilli et al. (2007) determined that one could correctly represent the tsunami of 12/26 2004 if the source were broken up into 5 segments. The inversion method used to determine these segments and the parameters of the seismic source is based on: - many digital simulations of the tsunami; - slipway distributions predicted in seismic inversion models and GPS data (Ammon et al., 2005; Vigny et al., 2005); - the arrival times of successive tsunami waves measured at far distant tide gages; - ocean elevation measured by the Jason 1' S satellite.

The parameters of these segments are presented in Illustration 91.

Illustration 91 - Characteristics of the 5 segments of seismic source after Grilli et al., 2007.

5.2.2. Numerical simulation

For the numerical simulation of the tsunami (generation, propagation and inundation steps) we use the model GEOWAVE. This code is composed by 2 models: TOPICS and FUNWAVE.

TOPICS (“Tsunami Open and Progressive Initial Conditions System”) provides the vertical co-seismic displacements as outputs, as well as a characteristic tsunami wavelength, period and amplitude using Okada’s solution.

110 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

FUNWAVE is a fully nonlinear and dispersive Boussinesq long wave propagation model. FUNWAVE was initially developed for modeling ocean wave transformation from deep water to the coast, including breaking and run-up. FUNWAVE also has a physical parametrization of dissipation processes (including breaking), as well as an accurate moving inundation boundary algorithm, both of which are necessary to correctly estimate coastal tsunami effects and run-up over land.

We simulated the 12/26 2004 tsunami propagation in the Bay of Bengal using GEOWAVE, with the parameters presented in Illustration 91, corresponding to the five rupture segments S1-S5. The total co-seismic seafloor vertical displacement obtained for the 5 combined tsunami sources is depicted in Illustration 92. In this figure, uplift and subsidence contours are plotted at a 1 meter spacing, values range from -6 m to 9 m.

Illustration 92 - The total coseismic seafloor vertical displacement (in m) obtained for the 5 combined tsunami sources.

The Illustration 93 shows some snapshot of the simulation at specific times: during generation, when tsunami reaches Sri Lanka East coast, West coast and at the end of the event.

BRGM/RP-55553-FR – Final report 111 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 93 - Snapshot at 25, 110, 230 and 250 minutes of the tsunami propagation.

The maximum simulated tsunami elevations above sea level are depicted in Illustration 94. The tsunami radiation patterns show high directionality, both because of the source length and in relation with various features of the seafloor. To the west, tsunami propagation depends on the sediment fan that covers most of the Bay of Bengal. To the East, a much more complex pattern emerges due to interference and interactions of multiple wave fronts propagating to and among various shorelines.

112 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Illustration 94 - Maximum simulated tsunami elevations above sea level.

BRGM/RP-55553-FR – Final report 113 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

5.3. MODELLING CONCLUSION

This chapter demonstrates the interest of numerical modelling for evaluation of impact of extreme waves (monsoon and 2004 tsunami) impact in the Galle area. Quantification of wave’s height, wave setup and inundation limits have been realized at various spatial scales. Grid sizes used in these simulations are in the 3 500 m to 20 m range.

Numerical modelling have needed specific works: - topo-bathymeric gridding : acquisition, computing, qualification and homogenization of existing data (Survey Department, NHO, Etopo2 and UNU data); - caraterization of huge monsoon waves : computing and analysis of data provided by the CCD, selection of wave and wind data from the WaveWatch3 (NOAA) and ERA40 (ECMWF) databases.

Next stages for an integration of these simulations results as hazard aggression factors in the scenario approach using ARMAGEDOM or in a GIS could be the following : - validation of the monsoon waves model using near shore in-situ monitoring (at depth less than 10 m); - computing of a tsunami high resolution inundation-limit with a HiRes grid (20 m to 40 m) and results validation using inundation and destruction limits acquired during this project.

These future works would allow: - a better evaluation with the empirical model used in the elaboration of the tsunami hazard layer (§ 1.1.1); - to test new tsunami scenarios with different amplitudes and generated by others geographical seismic sources; - to test others waves scenarios: seasonal, annual average waves etc. in order to have a better understanding of coastline evolution (erosion, accretion…); - to set up a wave forecast system of meteorological waves at 72 h using WaveWatch3 as input data.

114 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

6. Recommendations and conclusion

he objectives of the project “Geographic Information System for Coastal Hazards - T Application in a pilot site in Sri Lanka” were numerous. This project allowed to set up cooperation between Sri Lankan and French teams. It showed the importance of such a cooperation between the different institutes involved in acquisition and processing of data linked to coastal hazards and risks: - Geological Survey and Mines Bureau of Sri Lanka (GSMB, co-leader); - Coastal Conservation Department (CCD); - National Science Foundation (University of Moratuwa – Dept of civil engineering); - National Aquatic Ressources Research and Development Agency (NARA); - Meteorology department; - Survey Department; - Urban Development Authority (UDA).

This report describes the various tasks carried out during the project and illustrates the GIS developed within the framework of this project. The DVD provided with this report includes the Coastal GIS and all the data acquired during the project. The GIS has been created with the ArcGIS 9.1 software from ESRI.

The coastal GIS was set up by building up homogeneous data on the land/sea interface, allowing relevant parameters to be crossed so that studies can be conducted on the exposure of coastal populations to natural hazards.

Most of the data integrated in the GIS were available at the beginning of the project in the different institutions of Sri Lanka (GSMB, NARA, Survey Department, CCD etc.) However, the high number of textual and numerical formats has driven to an important pre-processing work of data before integration in the GIS.

A standardization of the data acquisition and storage processes, associated with a quality control, would facilitate the processing and the integration for future projects. This point is of a major importance in the specific case of natural hazards and disaster management.

We can propose a data management based on the creation of a metadata database maintained by only one institution which would be responsible of this metadata database. Other institutions will be responsible of the updating of the metadata concerning its scientific and technical field.

The data themselves are acquired and maintained by each institution (for example GSMB for geology, NARA for oceanographic data, CCD for coastal data, Survey Department for general geographic data, Meteorology Department for meteorological

BRGM/RP-55553-FR – Final report 115 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

data etc…). With this organisation, it will be possible to know exactly and quickly the data available, the data format, the resolution of data, the date of acquisition, etc. for each scientific field. Moreover, such an organisation would avoid acquiring twice the same data and will optimize the data management.

During the project we noted that the coordinates system and definition of the projection parameters were different between institutions. In order to have an easier and a more efficient integration of data in the GIS, it’s of primary importance that all institutions use the same coordinate system and projection parameters (see § 3.6.4).

Right after the tsunami disaster, the decisions were taken without all of the scientific and technical data needed. Some of this data were not available; others could not be used directly by the decision makers.

This GIS can be used by an expert team in order to better take into account coastal hazards in land-use planning (landscape and built area). Major building projects such as hospital, schools, public service, major communication networks can be planned taking into account their vulnerability to one or several hazard. Decision about the location of these buildings can be justified through references to the level of exposition to a present hazard level (storm surge, tsunami…) or future hazard level (sea level rise, coastal erosion…) Private building projects can also benefit from our tools to set up an appropriate policy at the local scale.

The project showed the capacity of this type of tool to integrate multi-hazards and risks data on a large area and to be a tool for policy-makers and for political and economic decision-makers in the field of risk management.

The project integrated also an important part of technology transfer, on GIS, on remote sensing, on coastal risks management and on modelling.

However, a GIS remains a tool which must be upgraded. This pilot project covered only the south-western part of the island and must be completed on the whole coastline of Sri Lanka. Moreover, the GIS must be updated in the next years in order to take into account the evolution of the vulnerability (new built-up areas, new infrastructures …) and the new scientific data on sea and coastline.

116 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

7. Aknowledgments

This work was carried out within the framework of a project supported by French Ministères des Affaires Etrangères (French Foreign Office) and BRGM (French Geological Survey) in the frame of research activities. The authors wish to thank the CNES (French Space Agency) ISIS program which provides the SPOT images, partly acquired after the Tsunami Disaster in the International Charter “Space and major disasters” framework.

8. Bibliography

Ammon C.J, Ji C., Thio H.K, Robinson D., Ni S., Hjorleifsdottir V. Kanamori H., Lay T., Das S., Helmberger D., Ichinose G., Polet J., Wald D. (2005) - Rupture Process of the 2004 Sumatra-Andaman Earthquake. Science , 308, 5725, pp. 1133 – 1139.

Booij, N., R.C. Ris and L.H. Holthuijsen (1999) - A third-generation wave model for coastal regions, Part I, Model description and validation, J.Geoph.Research C4, 104, 7649-7666.

Coastal Conservation Department (2004) - Coastal Zone Management Plan. 199 p., ISBN 955-9474-03-0, Ministry of Fisheries and Aquatic Resources, Revised Edition.

Garcin M., Preme B., Attanayake N., De Silva U., Desprats J.F., Lenotre N., Pedreros R., Siriwardana C.H.E.R., Weerawarnakula S. (2006) - Geographical Information System for Coastal Hazards – Application in a pilot site in Sri Lanka. BRGM Open File BRGM/RP-54656-FR, 39 p. 18 fig., 10 tables.

Ghobarah A., Saatcioglu M., Nistor I. (2006) - The impact of the 26 December 2004 earthquake and tsunami structures and infrastructures, Engineering structures 28: 312 – 326.

Grilli S.T., Ioualalen M, Asavanant J., Shi F., Kirby J. and Watts P. (2007) - Source Constraints and Model Simulation of the December 26, 2004 Indian Ocean Tsunami. Journal of Waterway Port Coastal and Ocean Engineering, 133(6) (in press).

IPCC (2001) - Climate Change Reports.

BRGM/RP-55553-FR – Final report 117 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Matsutomi H., Nugroho S., Matsuyama M. (2006) - Aspects of inundated flow due to the 2004 Indian Ocean Tsunami, Coastal Engineering Journal, Vol.48, N°2 169-195.

Matsutomi H., Shuto N., Imamura F., Takohashi T. (2001) - Field survey of the 1996 Irian Jaya Earthquake Tsunami in Biak Island, Natural Hazards 24: 199-212.

Papadopoulos GA., Caputo R., McAdoo B., Pavlides S., Karastthis A., Fokaefs A., Orfanogiannaki K., Valhanishi S. (2006) - The large tsunami of 26 December 2004: Field observations and eyewitnesses accounts from Sri Lanka, Maldives Is. And Thailand, Earth Planets and Space 58 (2): 233-241.

Peiris N. (2006) - Vulnerability functions for tsunami loss estimation, First European Conference on Earthquake Engineering and Sesimology, Geneva, 3-8 Septembre 2006, Paper number: 1121.

Sedan O., Mirgon C. (2003) - Application ARMAGEDOM Notice utilisateur, BRGM open file BRGM/RP-52759-FR.

Singh O.P., Tarik Masood, Md Sazedur Rahman (2001)- Has the frequency of intense tropical cyclones increased in the North Indian Ocean? Current Science, 80, 4, 575-580.

Siriwardana C.H.E.R., Weerawarnakula S., Mudunkotuwa S.M.A.T.B. Preme W.K.B.N (2005) - Final report on the tsunami mapping program (TMP) conducted in the Eastern, Southern and Western coastal regions in Sri Lanka. GSMB Report.

Swan B. (1983) - An introduction to the coastal geomorphology of Sri Lanka, National Museums of Sri Lanka, 182 p.

UNU (2005) - Tsunami Survey. CD ROM.

Vigny, C. and X alia (2005) - Insight into the 2004 Sumatra Andaman earthquake from GPS measurements in south east Asia. Nature, 436, 203-206.

Vitart F., Anderson J.L., Stern W.F. (1997) - Simulation of inter-annual variability of tropical storm frequency in an ensemble of GCM Integrations. Journal of Climate, 745- 760.

118 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

9. Appendices

9.1. LOCATION AND TYPE OF DATA

A list of data required for the GIS in the pilot site has been elaborated during the first mission (Tables 5 to 10). These data cover a strip from 2 km North of Bentota to 2 km east of Welligama Bay. Some of these data have been integrated in the GIS at the present time.

9.1.1. Data at the Survey Department

Required Data required Scale Type Format Topographic data (Contour lines and points with altitude) 1/10.000 Vector Arc Info export

As accurate as Scanned topographic maps possible Image Geo Tiff Coastal Rivers/hydrographic network river, lagoons...) 1/10.000 Vector Arc Info export 1/5.000 Multitemporal Aerial photo 1/10.000 Raster GeoTiff Land use and land cover (agricultural land, industry, infrastructure) 1/5.000 Vector Arc Info export Communication networks (roads, railways, electrical network etc.) 1/10.000 Vector Arc Info export

9.1.2. Data at UDA

Required Data required Scale Type Format Land use and land cover (built-up areas) 1/5.000 Vector Arc Gis

BRGM/RP-55553-FR – Final report 119 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

9.1.3. Data at the NARA

Required Data required Scale Type Format Survey of evolution of bathymetric and topographic profiles (multitemporal if available) 1/10.000 Vector ASCII Dxf or shape Mapping of extension of the 12/26/2004 tsunami 1/10.000 Vector file Tidal characteristics (neapest tide, mean neap tide, mean spring tide, springest tide) with location Data Tidal data (temporal records) with location Data ASCII Wave parameters Data Currents: velocity, direction etc./model output Data Tidal circulation model Data High resolution coastline with classification (rocky, sandy, coral reef, ….) Vector

9.1.4. Data at the Meteorology Department

Required Data required Scale Type Format Climate studies of Sri Lanka: monsoon , intermonsoon, cyclones Reports RTF, Word … Wind speed and direction at surface level at the 3 stations around Galle every 3 hours Data ASCII Rainfall at surface level at the 3 stations around Galle every 3 hours Data ASCII Atmospheric pressure at surface level at the 3 stations around Galle every 3 hours Data ASCII Temperature at surface level at the 3 stations around Galle Data ASCII Statistics (monthly, seasonal, annual) for the longest temporal period for wind speed, rainfall, atmospheric pressure, temperature Data ASCII, excel Output parameters from the high resolution Grib meteorological model (10 x 10 km gridded data) (since 2005 and on the South West area of Sri Lanka) Data format

120 BRGM/RP-55553-FR – Final report GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

9.1.5. Data at the Coastal Conservation Department (CCD)

Required Data required Scale Type Format Inundation extension of the tsunami where available (Galle, …) vector shape Contour lines of the permanent vegetation line of the 1994 and 2003 surveys + cross sections (Galle district) 1:2,000 vector shape Mapping of the defense works vector shape 1954 and 1982 scanned and georeferenced aerial photos raster GgeoTiff Pre and post tsunami Quickbird images raster GeoTiff ASCII available or * Nearshore bathymetric survey map * All wave data available (raw and processed data) : offshore of Galle (70 m depth) ASCII or S4 nearshore : Colombo, Galle, SE of Galle) format

* data available from CCD or Prof S.S.L. Hettiarachchi.

9.1.6. Data at the Civil engineering department of Moratuwa University

Required Data required Scale Type Format Sediment transport report report pdf

ASCII available or * Nearshore bathymetric survey vector map

* All wave data available (raw and processed data) : offshore of Galle (70 m depth) ASCII or S4 nearshore : Colombo, Galle, SE of Galle) Data format

9.2. ACRONYMS/ABBREVIATIONS GIS: Geographic Information System RS: Remote Sensing

BRGM/RP-55553-FR – Final report 121 GIS for Coastal Hazards – Application to a pilot site in Sri Lanka

Institutions:

- BRGM: Bureau de Recherches Geologiques et Minières. (France). - CCD: Coastal Conservation Department (Sri Lanka). - Department of Meteorology (Sri Lanka). - DMC: Disaster Management Center (Sri Lanka). - GSMB:Geological Survey and Mines Bureau (Sri Lanka). - HIC: Humanitarian Information Center (Sri Lanka). - LHI: Lanka Hydraulic Institute Ltd (Sri Lanka). - NARA: National Aquatic Resources Research and Development Agency (Sri Lanka). - NHO: National Hydrographic Office (Sri Lanka). - Survey Department (Sri Lanka). - UDA: Urban Development Authority (Sri Lanka). - UNDP: United Nations Development and Humanitary Information Center.

122 BRGM/RP-55553-FR – Final report

Scientific and Technical Centre Development Planning and Risks Division 3, avenue Claude-Guillemin - BP 36009 45060 Orléans Cedex 2 – France – Tel.: +33 (0)2 38 64 34 34