I

/' ~ • I

.... .' ... t

f I

• • .. Southwest Indian Ocean Risk Assessment and Financing Initia tive . : ". -.... -::: ":. :::: : ••:.:: ~iog Disaster ::::::::::Resilience in ,•••••~••...... ":,Stib-Sol'1o.ron Africa • ••••• ••• :·:·:·i:~$ ~- ••0i.\f" p".; ~ _ @WORLDBANKGROUP r,.... I,"J"," ~ ____.._. ..__ . (12016 The World Bank The International Bank for Reconstruction and Development TheWorld Bank Group 1818 HStreet, NW Was hington, D.C. 20433, USA November 2016

Africa Disaster Risk Profiles are co-financed by the EU-funded ACP­ EU NaturalDisaster Risk Reduction Program and the ACP-EUAfrica Disaster Risk Financing Program, managed by the Global Facility for Disaster Reductionand Recovery.

DISCLAI MER This document is the product of work performed by GFDRR staff, based on information provided by GFDRR's partners. Thefindings , analysis and conclusions expressed in this document do not nec­ essarily reflectthe views of any individual partner organization of GFDRR , including,for example, the World Bank, the Executive Directorsof the World Bank, UNOP, the EuropeanUnion, or the governments they represent. Although GFDRR makes reasonable efforts to ensure all the information presented in this document is correct, its accuracy and integrity cannot be guaranteed. Useof any data or information from this document is at the user's own risk and under no circumstances shall GFD RR or any of its partners be liable for any loss, damage, liability or expense incurred or suffered which is claimed to result from reliance on the data con­ tained in this document. The World Bank does not guarantee the accuracyof the data included in this work. The boundaries, colors, denomination, and other information shown in any map in this work do not imply any judgment on the part of TheWorld Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries.

RIGHTS AND PERMISSIONS The material in this work is subject to copyright. BecauseThe World Bank encourages dissemination of its knowledge. this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this workis given. Any queries on rights and licenses. including subsidiary rights. should be addressed to the Office of the Publisher, TheWorld Bank, 1818 HStreet NW, Washi ngton, DC20433, USA; fax: 202-522-2422; e-mail: [email protected]. The SWIO RAFI Project The SWIO RAFI included the collection of existing hazard and exposure data, and the creation of new hazard and exposure data, that were used in the he Southwest Indian Ocean Risk Assessment deve lopmen t ofa risk assessment and risk profiles and Financing Init iative (SWIG RAFI) for The ,, Mauritius, Seychelles, seeks to provide a solid bas is for the futu re T and Zanzibar. imp lemen tation ofdisaster risk financi ng th rough the improved understanding of disaster risks to participa ting isla nd nations. This initiative is The exposure data includes detailed informati on in pa rtne rsh ip with th e Ministries of Finan ce, on building cons truction for a var iety of occupancy National Disaster Risk Management Offices and classes including: resid enti al; commercial; Insurance secto r representatives from The Comoros, indu strial; public facilities such as educational Mad agasca r, Mauritius, Seychelles, and Zanzibar, and facilities and emergency facilities; and infrastructure carr ied out in coordina tion with the Indian Ocean such as roads, airports, ports, and utilities. Finally, Commission (lOC) ISLANDS Project. the United risk information that is determi ned through a Nations Office for Disaster Risk Reduction (UNISDR), combi nation of data on hazard, exposure, and and the French Development Agency (AFD). The vulnerability is provided at the national level and SWID RA FI supports the ISLANDS project's Islan ds at several administration levels for each peril and Fina ncial Protection Program (IFPP), which is also for all pe rils combined, and broke n down int o supported by the European Union (EU), UN ISDR,and occupancy classes. AFD. Africa Disaster Risk Profiles are co-financed by the Elf-funded ACP-EU Natural Disaster Risk In add ition to the information provided in the risk Reduction Program and the ACP-EU Africa Disaster profiles, the hazard and exposure data and the Risk Financing Program, managed by the Global results of the risk analysis will be collated and stored Facility for Disaster Reduction and Recovery. on ope n data geospatial risk information platforms, or GeoNodes, in each country and will be available SWIO RA FI complemented the ongoing wo rk of to a wide ra nge of end-users. The results will be the IOCto reduce vulnerability to natural disasters ava ilable in the form of geospati al files. text files, in accordance with the Mauriti us Strategy for the and de tailed fina l rep orts and can be used for sector Further Implementation of the Program of Action speci fic development pla nning and implementation. for th e Susta ina ble Development of Small Island Developing States (SIDS) 2005-2015. More broadly, this initiative offers support to long-term, core economic, and social developmen t objectives.

The risk mo deling undertaken through SWIO RAF I focused on th ree pe rils: tropical cyclones, floods produced by eve nts other than tropical cyclones, and earthquakes. Three hazards associated with tropical cyclones, wind, flooding and were considered in the risk assessment. In addition, as part of the earthquake risk assessment, tsunami risk zones were identified for each country.

DISASTER RISK PROFILE I INTROOUCTION 1 hiS analys is suggests th at, on average, Comoros experiences nearly US$S.7 mi llion in combined direct Tlosses from earthquakes, floods. a nd tropical cyclones each year. However,a specific event such as a severe can prod uce significantly larger losses. For example. Key Facts results suggest that a lOO-year return period tropical cyclone This ana lysis suggests that: event cou ld pro duce direct losses of $43 million and require approximately $9.9 million in emergency costs. • The average annual direct losses from earthquakes, l100ds, and Tropical cyclones are by far the most significant risk of the tropical cyclones are $5.7 million. three perils in this study, causing approximately 64 percent of the average loss per year. Flooding is the next largest risk. • The tun-year return period loss accounting for nearly 35 percent. It is important to note that an from all perils is over $48 million, I important risk on Grande Comcre, the Karthala volcano, was not or nearly 8% of Comoros's 2014 included in this analysis. • ~ l GOP. e In this analysis the residential sector experiences 80 percent • The 2S0-year return period loss of the combined losses, the public sector over 11 percent, and from all perils could be $150 the commercial sector nearly 6 percent. The highest loss takes million, or 24% ofComoros's 2014 place in , which experiences nearly 80 percent of GDP. the average annual losses from the three perils combined In addition to the direct losses, an annual average of nearly $ 1.3 million is estimated for emergency costs.

Direct Losses from All Perils Average Annual Loss ('*oJ

0 % 0 .1 0 .2 0 .3 0 .4

Avt'rage Annual Loss (%) Direct Losses by Hazard o 0.060.12 0. 18 0.24%

Earthquake Flood Tropical cyclone [1 It ~ ~ ~

$160 M ul $80M ~. + $40M $20 M ~ Ml RPIO RPIOO- RP250 ML RPIO RPIOO RP250 2 DISASTER RISK PROFILE I RISK SUMMARY he population of Comoros in 20 15 build ings and other infrastructure is sector accounts for over 66 percent of was approximately 790,000. The estimated to be nearly $2.6 billion. The the tota l replacem ent value. Tmos t populous island is Gra nde largest con centration of replacement Comore. Nearly 58 percent of Comoros's value is on Grande Cc more . In terms of constructio n type, buildings population lives in metropolitan or with masonry and concrete wall urban areas (that is, areas with more To assess risk be tter, replacement values constr uction account for nearly 83 than 2,000 people per square kilometer) and loss are often categorized according percent of the total replacement value. and slightly more than 29 percent in to occupancy and const ruc tion types. In rural areas (fewer than 1,600 people per terms of occupancy type, the residential square kilometer). In 20 14, Comoros's gross domestic product (GOP) was approximately $6 20 million ($t.3 billion Average Annual Loss 100·Year Return Period Loss in purchasing power parity), and the per capita GDP was $790. Peril Total Direct Emergency Total Direct Emergency Losses Costs Losses Costs For 20 15, the estimated total Earthquakes $99,0 00 $16,00 0 $1.8 million $280 ,0 00 replacement value for all residential, commercial, industrial, and pub lic Floods $2.0 million $460, 00 0 $10 million $2.3 million Tropical Cyclones $3.6 million $830 ,0 0 0 $43 million $9 .9 million _ Direct Losses by Building Type for all Hazards

Residential Commercial/Industrial Average Annual Loss (%) Average Annual Loss (%) 0 .36 % . 0 .36 % _ 0 .27 . 0 .27 0 .18 0 .18 0 .09 0 .09 a a

Infrastructure Public* Average Annual Loss (%) Average Annual Loss (%) _ 0 .36 % _ 0 .36 % - 0 .27 . 0 .27 0 .18 0 .18 0 .09 0 .09 a a o no data

-toucanon. Healthcare, Religion, Emergency

DISASTER RISK PROFILE I RISK SUIollolARY 3 his analysis suggests that , on ave rage, flood loss. On Anjouan, Ouan i has the highes t Comoro s will experience around $2.0 risk ofdirect losses with average annual flood T milli on each yea r in dir ect losses from losses of $66 0,000. Flood risk on Grande floodin g, am ou nt ing to nearly 3S percent of Comcre and Moheli are similar with average the country's total ann ual di rect losses fro m annual flood losses of $35 0,000 and $330,000, earthquakes, flood s, and tropical cyclones. respectively.des pite the large difference in assets at risk ($1.3 billion in Grand e Comore It is estimated that ove r 87 perc ent of the and $170 million in Moheli). dtrect losses from flood ing are from the residential sector and 5 percent from the Significant Ilood lo sses can occur commercial sector. Losses to public assets and frequentjy. For Comoros as a whole, direct industry contribute approximately 6 percent losses from a to-year flood event are and 1 percent, respectively, to the total. Annual estimated to be $4.8 million, and direct losses e me rgency cos ts for floods are estimated at from a tun-year flood event are estimated to over $460,000. on average. be $10 million.

This analysis shows that Anjoua n is at greate st ri sk for flood loss. with an average Return Period Total Modell"d Losses loss per yea r of $ 1.3 million from nontropical AAl _ $2 million cyclone flooding nea rly 66 percent of the total RPI0 _ $5 million Average RPI00 $10 million Annual losses ( $) Exposure ($) RP2S0 $12 million ' 450K . o '350 M 200-450K 0 o 2oo -350 M _200K 0 o _200 M

Modeled Direct Losses

Average Annual loss

$1.5 million Residential RPI0 $4 million RP100AA l ~~~~~~~~~~~::~$8.5 million Key Facts RP250 $9.5 million This analysis suggests th at :

AAl - $90 ,000 The ave rage annual direct loss from Commerciall RPI0 - $250 .000 flooding is $2.0 million . Industrial RPIOO $500 .000 RP250 $60 0 .000 Anjouan has the gr eatest risk of direct loss from flooding with an average annual AAl - $100,000 loss of $1.3 million. RPI0 - $300.000 Public RPloo $800,000 RP250 $9 50.000 • The t nu-year direct loss to Comoros from flooding is $tOmillion. AAl - $20.000 RPI0 - $50 .000 Infrastructure RPl00- $150 .0 00 RP250 - $200 .000 ------

4 DISASTER RISK PROFILE I flOOD Flood extent lOO-yroT RPdota using a rcon threshold n this analysis. the mod eled a nnual , " . average ra infall from non-tropical _ rI I cyclone events is 979 mm with a , minimum of 446 mm and a maximum of • 1,757 mm.

Flood depths are highest in Anjouan and Moheli and . Modeled flood de pths can exceed 1 m for the Fomboni regio n in Moheli and the for the Ouani region on Anjouan.

. . ,. .. ,..p.,h "" ,,¢.$ ..,-

Residen tial Commercial/Industrial Public Infrastructure

~ 'O ~ "' 'q.'o~ ~'O~ "'''> 'o ~ ~., ...... a• 0 O' O' O' O' 0 " c' •> o» ~ .1. '. " . •0 -• •~ •o l l l l • • '" '';:; •0 <; ~ ~ ~ ~ ~ ~ ~ ~ IS:""

___ __ 0 •• _ - _._ -~ ------_...... ------..~ 'OJ ~~~t{ o~..::~ ~~"::l{ ~~,,:: 4 o • 00' 0' 0 ' 0 ' 0 O ' O' c' O ''" 00' 0" (;)' 0 ' 0 0 '0' 0 ' O' 0 - - ~• •> -• ;;• •o •. 0 - ~ ~ C ·

.~ 5- l l • • •> -• e C O ~ <'" ~ ~ ~ ~ "- .... 0 - ---- _. _.. • ....,'OOI ~ ,,>'c0liif' ..., '0 01 iH' .., '0 01 iH' • 0 O ' O ' O' .::- 0 c' c' O ' :: 0 O' c - 0·:) 00' o· 0· ..."" ..> -• -• -o . ~--. ~ ~ I c~ o N - -i! <'" ~ ~ J 0

DISASTER RISK PROFI LE I FLOOD 5 ropica l cyc lones are common in the suggest that almost 77 percent of the loss Sout hwest Indian Ocea n region, and from tropical cyclones orig inates from the o TComoros occasionally expe riences a residential sector and nearly 6 percent from tropical cyclone landfall.The wind, rain, a nd the commercial sector. Losses to the public o stonn surge associated with tropical cyclones sec tor and industry contribute ap proximately all contribute to losses. 13 percent and 1 percent, respecnvefy, to the total of direct losses. Annual emergency costs A recent example ofa tropical cyclone affecting for tropical cyclones are estimated at nearly o Comoros is Hellen, which passed to the south $830.000, on average. in March 20 14. The cyclone caused one fat ality due to storm surge. damaged over 900 Tropical cyclones generate wind, flood. and ...... , houses , and caused flooding and landslides, storm surge hazards. On average in this partirularty on Maheli and Anjouan.l analysis, winds cause over 54 percent of the loss from the three hazards, while storm surge This analysis suggests that. on average, causes around 43 percent and flooding around Comoros will experience around $3 .6 million 2 percent of the remaining damage. in direct losses annually from winds, flooding, and storm surge associated with tropical Return Period Total Modeled losses cyclones. This is 64 percent orthe country's AAl I $4 million total annual direct losses from earthquakes. RPI0 , $5 million floods, and tropical cyclones. The results RPtOO - $40 million RP250 $150 million An rage Annuallosses ($) Exposure ($) >IM . 0 >350M 6001(-IM 0 0 2oo-350M _6001( 0 0 _200M •

Modeled Direct Losses

Avera ge An nual Loss

AAL_ $2.5 million Residential RPI0 _ $3.5 million RP100 $35 million Key Facts RP250 $100 million This analysis suggests th at:

AAL - $250,000 The average annual direct loss from Commerciall RPI0. $300.000 tropical cyclones is $3.6 million. Industrial RPIOO - $3 million RP250 $9.5 million Anjouan has the greatest risk ofdirect loss from tr opical cyclones with an AAL - $500,000 average annual loss of $3.1 million. RPtO- $600.000 Public RPloo $s.s million RP250 $20 million • The tnu-year direct loss to Comoros from tropical cyclones is $43 million. AAL - $70.0 00 Infrastructure RPtO' $20.000 RPloo- $550.000 RP250 - n .5 million ----~------

6 DISASTER RI51( PROFILE I TROPICAl CYClONE Tropical cyclone hazards WmdspeerJs excf'f'dmg 6Jkph. stormsurgeinundolion exceeding 1m,andfloodingexcl'f'ding lCkm hree hazards are prod uced by trop ical cyclones: wind. storm Tsurge, and flooding produced by excess ive rainfall. For Comoros, th is analysis suggests that the greatest wind hazard occurs on the northea st and southeast ends of Anjouan. Winds associated with the 1co-vee r tropical cyclone can exceed 200 kph.

This analysis suggests th at the greatest Wind Inte nsity flood hazard associated with tropical cyclones occurs on Anjoua n and MaheU. Storm surge hazard is greates t along th e southern and easte rn coasts of Comoros. In places storm su rge can exceed 1.5 mete rs . ~.,....

Residen tial Commercial/Industrial Public Infrastructure

~'O~ "' 'q.'o~ ~'O~ ~'o 4: ~., ...... a• 0 O' O' O ' O' 0 o» ~ .1... •" .• -•~-•o l l l l • • '" '';:; •0 <; ~ ~ ~ ~ ~ ~ ~ ~ IS:"" ----_.... _. ------_... "'''l ~ .... "' ''> '1'.. ~ "' ''l ~ 'OJ 00....·0·0·0· 0 c ' c' c ' O' 00....·0·0"' ''l·0· 0 c-,- c- o· c' ~-•-.•> -• ;;• •o •. 0 - ~ ~ ~ C ·

.~ 5- l • • • ~ •> -• C O ~ <' " ~ ~ ~ "- 0 ----- _. _.. • .. ~ .. ~ ~~.. .. o . ~ ~ ~ ~~ • 0 .~ ~~ o • o • ..> -• -• -o . ~--. ~ Q ~ ~ I c~ o N - ti ~ <'" J 0

DIS ASTER RISK PROFI LE I TROPICAL CYClONE 7 arthquakes are com mo n in the expected loss per year of $55,00 0, nearly Southwest Ind ian Ocean region, and 70 percent of the total direct loss from EComoros is somewhat se ismically active earthquakes. Over 75 percent of the loss com pared to ot her islands in the regio n. from earthquakes is from the residential However, damage from earthqua kes is not sector, around 13 percent fro m the public common. The two major sources of seismic asset sector, and around 6 percent from the activity are the Mid-Indian Ridge in the Indian commercial sector. Annual emerge ncy costs Ocean and the East-Africa n Rift system. for earthquakes are estimated at $16,000, on o Earthquakes in these regions are frequent average. but usually oflow to moderate magnitude. A magnitude 5.1 event occurred 88 kilometers Significant losses from earthquakes are south of Moroni on September 29, 2016. expected to occur r elatively infrequently. For example, this analysis suggests that direct This analysis suggests earthquakes account for losses for earthquakes with a SO-year return around 2 p ercent ofComoros's total a nnua l period are -$55,000 and with a t no-year direct losses from earthquakes. floods, and return period an' $1.8 million. tropical cyclones. amounting to an estimated $99,000 on average each year. Return Period Total Modeled Losses The results suggest that the region with AAl • $100,000 the greatest absolute risk of loss from RPIO SO earthquakes is Grande Comore, with an RPtOO - S2 million RP2S0 STmillion Average Annual losses ($) Exposure ($) '20' • o >350 M 10 - 20 K 0 o 2oo -350 M <10' 0 o <200 M o

Modeled Direct Losses

Average Annual Loss

AA L_ $75,000 Residential RPtO' SO RP100 ::::::::~u.s;,;, million:::::,,_ Key Facts RP250 Ss mi llion This analysis suggests th at :

AAL - S6,OOO The ave rage annual direct loss from Commerciall RPtO ' SO earthqua kes is $99,000. Industrial RPIOO - S90,OOO RP2S0 S400 ,OOO Grande Comore has the greatest risk of d irect loss from earthquake with an AAl- SIs,OOO average annual loss of $55,000. RPtO ' SO Public RPlOO- S60,OOO RP2s0 S800,ooO • The t nu-year direct loss to Comoros from earthquakes is $1.8 million. AAl - S4,OOO RPI0 SO Infrastructure RPtOO · 12,000 RP250 - $100 ,000 ------

8 DISASTER RISK PROFILE I EARTHQU AKE his analysis suggests that the Tsunami zone and earthquake hazard Ground molioo froma l50'yeo, RPtof/hquokeand/sunomifisk /Ones gro und motion hazard from Tearthquakes is low throug hout Comoros. Earthquakes tha t would cause significant damage to structures have only Earttlquake Hazard a small probability of occurrence. for2SQRP Tsunamis usua lly result from high­ magnitude. subduction-zone earthquakes. The Sout hwest Indian Ocean region does not experience many high-magnitude earthquakes. nor doe s it contain majo r subduction zones. The entire region is at risk, however, of tsunamis generated by subduction zones elsewhere in the Indian Ocean.

A recent tsunami event that affected the Southwest Indian Ocean regio n was the 2004 India n Ocean ts unami. Some locations along the northeastern tip of Grande Comore experienced inundation of 6.8 meters a bove sea level.

Residen tial Commercial/Indust rial Public Infrast ructure

~'O ~ "' ~'O~ ~'O~ "'~ 'o ~ ~., ...... a• 0 O' O' O' O' 0 0'0' O' O ' 0 O' O' O' O' 0 O' O' O' O' •> o» ~ .1. .. •" .• -•~-•. l l l l • • '" '';:; •e <; ~ ~ ~ ~ ~ ~ ~ ~ IS:"" --- -_ ..... ------_... .. o~,jl...... S- 0> B'o>~ ~ S- O>~ ~ B'o>~~ 0 OO?? 0 9999 0 99?9 tl 0 00.....0 0 e <' " ~ ~ 0 - ---- _. _.. .. /"?, No losses on RP250 • O~ .. .. S-,g-~8' B' 0'0 8' B',g-i5'8' • 0000' 0 0 00'0 ..> 0 0 9999 '1-! -• -.• c . ~--. ~ ~ c ~ I \:t ~ c fi N - -i! ~J ~ ~ <' " ~ ~ J 0

DISASTER RISK PROFILE I EARTHQUAKE 9 in th e SWID region, pa rt icularly for the areas close r to the equator. Flood hazard statistics in this analysis ar e These risk profiles have been developed from a multi­ ultimately based on satellite-derived ra infall estimates hazard risk assess ment using a variety of exposure data from the years 1998-2013. The satellite-derived data are and vulnerability funct ions. Modeled perils ind ude used with a ra infall model to develop a cata log of da ily earthquake, flood, and tropical cyclone. The results for rainfall produced by events other than tropical cyclones. individual and aggregated pe rils ar e available in several A flood model then dynamically distributes the rainfall formats, includin g geos patial data and text files. The risk throughout the affected region and calculates flood depths. profil e results are presented in terms of average loss per year and for selected retu rn pe riods. For details on the development of the risk profiles, see th e final report TROPICAL CYClONE "Southwest Indian Ocean Risk Assess ment Financing Initiative (SWID RAFI): Compon ent 4 - Risk Proffles'; Brief This analysis suggest that the most costly catastrophic explanatio ns of the exposure and hazard data and th e hazard in the SWIObasin is tropical cyclone. The historical vuln erability functions are given below. record of tropical cyclones in the region includes 847 events that took place between the 1950 and 2014. The event catalog is developed using characteristics of the historical catalog. such as annual tropical cyclone frequency, landfall frequency, seasonality, genesis location, Hazard forward speed, central pressure, and radius of maximum This stu dy encompasses four pe rils: earthquake, flood, wind s. Three tropical cydone hazards are considered: land slide, and tro pical cyclone. One or more hazards wind, flooding from rainfall, and storm surge. are associated with each pe ril. For exa mple, the hazards associated with tropical cyclones includ e strong winds, Tropical cyclone wind speeds are calculated using an storm surge, and flood ing. For pe rils other than landslide, equation that includes parameters such as the difference a catalog representing 10,000 yea rs of simulated events between the tropical cyclone's central pressure and the was constructed usin g empirical and th eoretical principles surrounding environment, a storm's forward motio n and and information derived from historical observations. its asymmetry, and account for surface features such as A variety of stat istical characteristi cs derived from the land use . events in th e cata logs are consiste nt with the historical record for each peril. The catalog (which is proprietary) Rainfall produced by modeled tropical cyclones is includes information such as the intensity-for example, calibrated using satellite-derived rainfall estimates an d central pressure for a tropical cyclone and moment used as a boundary condition to force a flood model that magn itude for an earth quake- and location of each peril accounts for factors such as hou rly rainfa ll, elevation, and event. This information Is then cou pled with pe ril-specific soils. emp irical and theoretical considerations to desc ribe the Storm surge is derived from a variety of tro pical cyd one spatial dis tributio n of hazard intensity for each sim ulated characteristics that include central pressure, forward peril event in the catalog. at a grid spaci ng of about one motion of the storm , maxim um wind speed, and radius of kilometer. The information is used to determine the maximum winds. For a tropical cyclone in the Southe rn hazard intensities expected at each return period. Hemisphere, the highes t storm surge generally occurs near th e radius of maximum winds on the left side of the storm track. EAR THQU AKE This analysis suggests tha t there is a low likelihood of earthquakes in the SWIDregion. The catalog ofsynthetic Exposure earthquake events is developed using characteristics The methodol ogy used to develop the exposure data based on the historical record of 1.228 earthquakes wit h is illustrated In figure At. The exact process varies by moment magnitudes 5.0 or greater that occurred in the country because of differences in available da ta. The SWIDbasin between 1901 and 2014 and the slip rates and exposure database for each island nation is constructed geometries of known faults in the region. Ground motion from various data sources, including government prediction equations are used to determine the spatial censuses, local agencies. satellite imagery, publicly distribution of ground motion (su ch as peak ground available spatial sta tistics, and previous regional acceleration. or PGA)produced by ea ch earthquake event. investigations. The end result is datasets that represent the built environment of each Island nation and include nationally appropriate replacement values (that is, the FLOOD estimated cost to rebuild a str ucture as new), construction characteristics, and occupancy classes. The ris k assessment indicates th at floods from rainfall not associated with tropical cyclones are a significant hazard

10 DISASTER RISK PROFILE I METHO DOLOGY The exposure data are divided into eighteen different occupancy classes spanning different types of residential, Vulnerability commercial, industrial. public facility, and infrastructure Vulnerability functions appropriate to the construction assets. The residential occupancy class includes single and and occupancy classes most commonly found in the multifamily residences. The commercial class includes SWIO region are used to estimate loss from a hazard. The general commercial buildings and accommodation. The functions calculate the average level of damage to the exposure groups in the publi c occupancy class are health structures using the hazard intensity and information care services. religion, emergency services, primary on their occupancy and construction. The damage level educational, university educational. and general public represents the fraction of the total building replacement facilities . The infrastructure occupancy classes are road/ value that has been damaged Vulnerability functions highway, bus/rail, airport. maritime port, electrical utility, used in this study have been developed specifically for the and water utility. An ' unknown" occupancy class is also SWIOregion based on research on local building practices, assigned. applicable building codes, engineering analysis, historical damage reports, and expert judgment. In addition to their categorization by occupancy class, the exposure data are categorized according to thirteen Vulnerability functions for earthquake ground shaking. construction classes. Seven of these are specific to non-tropical cyclone flooding. tropical cyclone flooding. infrastructure occupancies and include structures such and tropical cyclone storm surge are assumed to be as roads, railroads, and bridges. Five represent common uniform throughout the SWIO region for all occupancies construction classes, such as single-story traditional other than infrastructure. Except for infrastructure, the bamboo and earthen buildings and single and multistory tropical cyclone wind damage functions for Mauritius traditional wood, wood frame, masonry/concrete, and Seychelles are modified to be less vulnerable than and steel frame buildings. As with occupancy class, an the SWIObase functions used for the other island nations " unkno wn" construction class is assigned. because of their history of more stringent construction practices relative to the other three nations. Alldamage The exposure data for residential, commercial, and general functions for infrastructure occupancy classes are industrial assets are provided on a grid of 30 arc-seconds assumed to be uniform for all perils throughout the SWIO (approximately one kilometer). when high-resolution government and infrastructure data are available, region. these assets are captured at their individual exposure - All dollar amounrs ore US. dollors unless otherwise locat ions. When location- level information is not available, indicated. government and infrastructure assets are distributed to the one-ki lometer grid.

Percentage of Grldded population Averagenumber of buildings by (3Darc-second) buildings per person .. occupancy type ...

Average building floor area by .. .. occupancytype ...

Spanal cost Exposure database of adjustment .. bUildings

Figurt Al. S

DISASTER RISK PROFILE I METHODOLOGY 11 Average Annual Loss Hazard The modeled average ann ual loss (AAL) is equ al Hazard refers to the damaging forces produced by a to the total of all impacts produ ced by a hazard peril, such as inundation associated with flood ing. or (e.g. earthq uake) in a specified time period (e.g. win ds produced by a trop ical cyclone. A single peril 10,000 years) divided by the number of yea rs in th at ca n have multiple hazards associated with it. Those specified tim e period (e.g. 10,000 years). associated with a tropical cyclone, for example, include strong winds, storm surge and flooding. Building Construction Class Building Construction Class is used to classify an Impact asset's construction, wh ich determines an ass et's Impact refers to th e consequences of a hazard vu lnerability to a certain hazard. contributing to affecting th e exposure, give n the exposure's a risk estimate. For example. a traditiona l wood vulnerability. The impact on st ructu res is usually building is more vulnerable (le.likely to be damaged qua ntified in terms of direct monetary loss. or destroyed) by a tropical cyclone than a building made of steel-reinforced concrete. Th us an area with Replacement Value tra ditio nal wood bu ildings is likely to experience Replacement value re fers to the estimated amount it more damage and larger losses fro m a trop ical wo uld cos t to replace physical assets. cyclone th an an area with steel-re info rced concrete buildings. Building Const ruction Class is on e of the factors used to determ ine vu lne rability (see below). Return Period (RP) Th rou ghout th is profile to-year (RP10), 100- Building Type yea r (RP100), and 2S0-year (RP2S0) events are referenced. These eve nts have intensities that (on Building Type, or Occupancy Class, specifies the average) are expected to occur once du ri ng a "return usage of a given building. which contributes to a period'; A return period is based on the probability building's vuln erability.The buildin g types used that an event could happen in a given year. The in these profi les are: resi dential, commercial, larger the return period for an event, the less likely industrial, infrastructu re, and public. its occurrence, and the greater its intensity. The Each building type has subty pes: probability of an eve nt occurring in any given year • Residential: single, multi-family (e.g. apartment) eq uals 1 divided by th e number of years named in the "x-year event'; e.g. for a 10-yea revent (an even t • Commercial: accommodation [e.g. hotel). wit h a t u-year return period), the p robability is commercial (e.g. shop) 1/10 or 10%; for a 100-year eve nt, the proba bility is • Indust rial: general industrial (e.g. factory) 1/100 0r1%. • Infrast ructure: bus terminals, rail termina ls, airports, maritime ports, utilities, roads, Risk highways Risk is a combination of hazard, exposure. and • Public: healt bcare, educati on, religious, vu lnerability. It is qu an tified in probabilistic terms emergency services, general public facilities (for example, average a nnua l loss) using the impacts of all events produced by models. Building Type is one o fth e facto rs used to determine vulnerability (see below) . Vulnerability Exposure I Exposed Assets Vulnerability accounts for the susceptibility of the exposure to th e forces associated with a hazard. Exposure refers to ass ets such as buildings, critical Vulnerability accounts for factors such as the facilities and transportation ne tworks, which could materials used to build the asset (as specified by the be damaged by a hazard. A variety ofattributes Building Construction Class) and the asset's use (as associated with the exposure, such as location speci fied by th e Building Type). and occupancy and structural characteris tics , help determine the vu lnerability of the exposure to a hazard .

12 DISASTER RISK PROFILE I GLOSSARY 1 UNOCHA, "Comoros Islands, Cyclone Hellen," http:// reliefweb.int / sites/reliefweb.int /files/resourcesl Comoros_Hel1en%201 mpact_02%20April%202014_ OCHA% 20ROSA.pdf ACKNOWlEDGNENTS These risk profiles were prepared by a team comprising Alanna Simpson , Emma Phillips, Simone Balog. Richard Murnane, Vivien Deparday, Stuart Fraser. Brenden jcngman. and lisa Ferraro Parmelee. The core teemwishes to acknowledge those that were involved in the production of these risk profiles. First, we would like to thank the financial support from the European Union (EU) in the framework of the African, Caribbean and Pacific (ACPj.[U Africa Disaster Risk Financing Initiative. managed by GfDRR. In the GFDRR secretariat we would like to particularly thank Francis Ghesquiere, Vivien Deparday, Isabelle Forge. Rossella Della Monica. and Hugo Wesley. We would also like to extend our appreciation to the World Bank Africa Disaster Risk Management Team: Christoph Pusch and noekte Wielinga. Thank you 10 the Disaster Risk Financing and Insurance Team: Julie Dana, Samantha Cook, Barry Maher, Richard Poulter. Benedikt Signer, and Emily White. Our thanks to AIR Worldwide for their risk assessment analysis. Finally, we are grateful to Axis Maps and Dave Heyman for creating the data visualizations and these welt-desfgned profiles .

DISASTER RISK PROFILE I ACltNOWUO GMENT5 13