Risks and resilience

Multi-hazard risk analysis of Burundi

Floods, torrential rain, strong winds, earthquakes and landslides: what the first national risk mapping says about the country, where the losses are concentrated, and the corrective actions it recommends.

The five hazards mapped in Burundi. Floods: flood-prone areas at national scale, and floods with return periods of 2 to 500 years in five priority areas. Torrential rain: 3-hour rainfall totals on a 3 km grid, return periods of 2 to 500 years. Strong winds: 10-minute gusts on a 3 km grid, return periods of 2 to 100 years. Earthquakes: ground acceleration for return periods of 100 to 2,500 years, with site effects. Landslides: susceptibility at 100 m resolution and 12 monthly maps depending on rainfall. For climate-related hazards, each map exists for the current climate (1990-2019) and the projected climate (2020-2049); all five analyses are converted into average annual losses in US dollars. Labels on the drawing are in French.

Illustration: ACAPE Burundi. Data: IOM Burundi, Multi-hazard risk assessment and mapping in Burundi (2020-2022), summary of results.

How this page is made

Every figure comes from the multi-hazard risk assessment and mapping carried out for the International Organization for Migration (IOM) in Burundi, funded by the European Union, and published on the national online risk platform. They are model projections: the authors present them as orders of magnitude, without any estimate of uncertainty.

The study uses the administrative division of the time — 18 provinces, 119 communes, 2,637 collines (hills) — which predates the 2025 territorial reform. The province and commune names are therefore those of the study.

The illustrations were drawn for this page by ACAPE Burundi from these data. They are schematic: none of them is an official map. The study’s four official maps, reproduced from the national risk platform, are credited to IOM Burundi beneath each one.

The study was carried out from mid-August 2020 to December 2021 by a consortium coordinated by IDOM, with EURAC Research, Meteosim, the GIIS-Geolyder group and the Burundian association APRN, assisted by Ramboll. It forms the first component of the TUBEHONEZA project for strengthening resilience to natural disaster risks; the next two concern strengthening the capacities of communities and institutions, and then risk mitigation.

Why it matters

Four markers to measure what is at stake

The first two come from IOM’s presentation sheet; the next two from the records of IOM’s Displacement Tracking Matrix (DTM) for 2018-2021, cited by the study.

Displacement

More than 90%

of population displacement in Burundi is caused by climate-related disasters.

Climate

Among the 20 countries

most vulnerable to climate change in the world, according to IOM.

People

About 215,000

people affected between 2018 and 2021, including nearly 80,000 displaced, and about 130 victims or missing persons.

Housing

About 9,500 houses

totally or partly destroyed over the same period, and more than 40,000 houses affected.

The method

From the phenomenon to the loss: four steps

The maps were produced at four levels. Each step answers a simple question.

Step 1

The hazard

Where, how often and how intensely can the phenomenon occur, in the current climate and in the projected climate?

Step 2

Exposure

What lies in its path? Houses classified by type of construction, the population, roads and bridges, schools, health centres, the power network and protected areas.

Step 3

Vulnerability

Who has the fewest means to cope and recover? An index of 3 categories, 11 components and 40 indicators, calculated from 111 variables.

Step 4

Risk

How much do disasters cost, on average, each year? These are the average annual losses (AAL), which add up small frequent events and large rare ones.

Official multi-hazard map of Burundi with province boundaries. Areas are coloured from 1 to 5 according to the legend “number of hazard hotspots”; the most intense colours run down the west of the country, from the Imbo plain to the escarpments, while most of the centre and east show scattered or no hotspots. Labels on the map are in French.

Multi-hazard map: the darker the colour, the more hazards overlap. The west of the country concentrates the most exposed areas. Credit: IOM Burundi, Multi-hazard risk assessment and mapping in Burundi (2020-2022), IDOM consortium, funded by the European Union.

What the study produced

126 risk maps at provincial level (one per hazard and per province), 6 national maps (one per hazard and one multi-hazard map), a risk atlas, a summary of results, an assessment of disaster risk needs and an online platform.

Five priority areas were analysed in greater detail: Bujumbura Mairie, Gitega, Rugombo (Cibitoke), Ngozi and Bukemba (Rutana).

Vulnerability relies on the 2020 DTM survey carried out on 532 collines, that is 20% of the total. Its main lesson: vulnerability is highly localised and can vary greatly between communities that are nonetheless neighbours. No area is the most vulnerable on every aspect at once.

What disasters cost

Nearly 93 million dollars a year

All causes combined, the study puts average annual losses at 92.75 million dollars, about 3.1% of 2019 gross domestic product. According to the authors, this order of magnitude is common in countries exposed to natural disasters.

Projected average annual losses by hazard, in millions of US dollars: river floods in urban areas 33.6 (36.2%), river floods on fields in floodplains 33.1 (35.6%), earthquakes 15.3 (16.5%), torrential rain 5.0 (5.4%), strong winds 4.0 (4.3%), landslides 1.8 (2.0%). Total 92.75 million dollars a year, about 3.1% of Burundi’s 2019 GDP of 3.012 billion dollars. Labels on the drawing are in French.

Illustration: ACAPE Burundi. Data: IOM Burundi, Multi-hazard risk assessment and mapping in Burundi (2020-2022), table 1.

River floods alone account for more than 70% of losses: half in towns, half in the fields of the floodplains. Earthquakes come next, with 15.3 million dollars, more than a third of it in Bujumbura Mairie.

Projected average annual losses by province, in millions of dollars, broken down by hazard: Bujumbura Mairie 32.4, Bujumbura Rural 9.5, Ngozi 8.3, Kirundo 5.1, Rutana 4.4, Ruyigi 4.0, Muyinga 3.8, Makamba 3.4, Gitega 3.2, Cankuzo 2.9, Bubanza 2.8, Kayanza 2.5, Karuzi 2.4, Cibitoke 2.2, Rumonge 2.1, Mwaro 1.3, Muramvya 1.2, Bururi 1.1. In Bujumbura Mairie losses come mostly from floods in the city; in Ngozi, Kirundo or Rutana, mostly from flooded fields. Labels on the drawing are in French.

Illustration: ACAPE Burundi. Data: IOM Burundi, Multi-hazard risk assessment and mapping in Burundi (2020-2022), table 1.

Should flooded fields be counted?

In Ngozi, 6.75 million dollars out of 8.26 come from crops in the floodplains. The authors point out that in many countries periodic floods on agricultural plains are accepted as a natural fact by those who farm this land.

In Burundi, however, many rural households have no other livelihood. The authors leave it to the national authorities to decide, and recommend reserving protective works on these plains for exceptional cases, justified by a cost-benefit analysis.

Where risk is concentrated

The most exposed communes

To compare communes of different sizes, the study divides average annual losses by the area of each commune.

Official multi-risk map of all 119 communes of Burundi (pre-2025 division), each commune coloured by its multi-hazard average annual loss index in US dollars per square kilometre, in classes from 333–500 (yellow) to over 5,500 (dark blue). The darkest communes are around Bujumbura and in the north, around Ngozi and Kirundo, and in the south-east near Bukemba. Commune names are printed on the map.

All communes, from the least exposed (yellow) to the most exposed (dark blue), by average annual losses per km². Credit: IOM Burundi, Multi-hazard risk assessment and mapping in Burundi (2020-2022), IDOM consortium, funded by the European Union.

The ten most exposed communes according to the multi-hazard index, in dollars per square metre. Ranks 1 to 4, Bujumbura and its outskirts: Muha 0.283, Ntahangwa 0.278, Mukaza 0.205 (Bujumbura Mairie), Mutimbuzi 0.060 (Bujumbura Rural). Ranks 5 to 10, the rural north and east, shown on a scale twenty times finer: Marangara 0.0134 (Ngozi), Mpanda 0.0126 (Bubanza), Ntega 0.0121 (Kirundo), Nyamurenza 0.0116 (Ngozi), Bukemba 0.0069 (Rutana), Ngozi 0.0053 (Ngozi). The ranking measures physical economic losses and does not include social vulnerability. Labels on the drawing are in French.

Illustration: ACAPE Burundi. Data: IOM Burundi, Multi-hazard risk assessment and mapping in Burundi (2020-2022), annex 1, table 3.

These ten communes are spread across six provinces: Bujumbura Mairie, Bujumbura Rural, Ngozi, Bubanza, Kirundo and Rutana. The ranking only measures material losses — destroyed houses, lost harvests — and not people’s vulnerability.

The second ranking, without flooded fields

When agricultural losses in the floodplains are removed, the communes of the north and east drop back and the west stands out. The top ten become: Muha, Ntahangwa, Mukaza, Mutimbuzi, Mpanda, Rumonge, Kanyosha, Gatara, Kayanza and Muhuta.

The authors consider this second ranking more consistent with the hazard map, and invite national experts to choose the method most useful for the future national risk management strategy.

Official map of the 50 priority communes, all hazards included, coloured by multi-hazard loss index per square kilometre; the other communes are left white. Many communes in the north (Ngozi, Kirundo) and along the eastern border stand out, along with Bujumbura. Labels on the map are in French.

First ranking: all hazards included. The north and east stand out, because of flooded fields. Credit: IOM Burundi, Multi-hazard risk assessment and mapping in Burundi (2020-2022), IDOM consortium, funded by the European Union.

Official map of the 50 priority communes excluding flooding of agricultural plots, coloured by multi-hazard loss index per square kilometre; the other communes are left white. The west (Bujumbura, Imbo plain, lakeshore) and the north-centre stand out, while most eastern communes drop out. Labels on the map are in French.

Second ranking: without flooded fields. The west, along the lake and the Imbo plain, takes over. Credit: IOM Burundi, Multi-hazard risk assessment and mapping in Burundi (2020-2022), IDOM consortium, funded by the European Union.

The climate ahead

More concentrated rainfall by 2050

Projections from three climate models at 3 km resolution, for 2020-2049 compared with 1990-2019.

Average rainfall

+0.2 to +3 mm a day

Increase in average daily rainfall, mainly during the rainy season.

Imbo plain

+25 to +40%

The largest relative increases in the country. The dry eastern plateaux also see large increases.

Very wet days

+6 to +14%

Of cumulative rainfall on wet and very wet days, and +4 to +10% on extremely wet days, notably in Bujumbura, Bubanza, Kirundo and Ruyigi: more intense rain, concentrated on fewer days.

Winds

Little change

On gusty days, wind speed stays close to today’s, with a slight downward trend.

The simulations use the RCP8.5 scenario, which assumes no emission-reduction policies. The authors justify this choice by the small difference between scenarios over a thirty-year horizon.

A risk of its own

The rising waters of Lake Tanganyika

Among the towns on the Burundian shore, only Bujumbura and Rumonge see the water reach urban areas.

Schematic cross-section of the Lake Tanganyika shore with three water levels. 776.5 m, return period about 60 years (the level observed at the time of the study): below it, an exclusion zone with no new urban development. 777.0 m, about 100 years: below it, no important infrastructure such as roads, water, electricity, shops or industry. 777.5 m, about 200 years: below it, no critical infrastructure such as hospitals, schools or fuel depots. Above 777.5 m, all uses are possible. Estimated average annual losses: 552,732 dollars in Bujumbura and 25,693 dollars in Rumonge. The model does not include climate change and may differ from IGEBU’s models. Not to scale; labels on the drawing are in French.

Illustration: ACAPE Burundi. Data: IOM Burundi, Multi-hazard risk assessment and mapping in Burundi (2020-2022), recommendations 6.c and 6.d.

The model is hypothetical: lake-level records are scarce and sometimes contradictory, and the level depends mainly on what happens across the whole basin, shared with neighbouring countries. These losses are not added to those of river floods, whose dynamics are different.

Corrective actions to take

What the study recommends

The recommendations are designed for a country with limited resources, where simple procedures run by trained staff can be more effective than costly tools.

From the map to action: the study’s recommendations grouped by level. State and institutions: a single official geographical framework for all departments; modernise and maintain weather and river-flow stations; earthquake-resistant building codes and their enforcement; early warning at national and community level. Communes and provinces: include the maps in commune plans (PCDC, PCIA); stop building in flood zones and on unstable slopes; inspect bridges after every major flood; contingency plans and simulation exercises. Schools, clubs, parishes: know the hazards of one’s colline on the platform; reforest slopes with the local authorities; record each event with place, date, extent and damage; relay verified warnings, without rumours. The three levels complement one another: a map only protects if the commune uses it and residents know it. Grouping proposed by ACAPE Burundi; labels on the drawing are in French.

Illustration: ACAPE Burundi. Grouping proposed by ACAPE Burundi from the study’s recommendations (section 1.3) and capacity needs.

Eight cross-cutting priorities

1. A common geographical framework. Agree on a reference geographical framework — official names and boundaries, from collines to provinces — used by every department, and update the baseline data: boundaries, roads, built-up areas, rivers, farmland.

2. Better recording of disasters. Systematise the approach begun by the DTM in 2018: for each event, record the georeferenced location, the extent of the affected area, the intensity of the phenomenon and the damage by building type.

3. Better measuring equipment. Modernise the network of weather stations and river-flow stations, with effective maintenance programmes.

4. Planning towns and countryside. Consistent urban and rural planning is essential to avoid an uncontrolled rise in risk. It may include moving vulnerable households living in flood zones or along overflowing urban ravines to new neighbourhoods.

5. Mitigation with nature. Favour nature-based solutions, such as reforestation against landslides, involving communities in the conservation and use of the forests created.

6. Checking that each structure is worth it. Base every protective work on a sound cost-benefit analysis that validates both the investment and the risk reduction achieved.

7. Build back better. Rebuild houses and infrastructure with better materials and structures, as a priority for the most vulnerable communities who cannot be relocated.

8. Refining the values used. Have a wider group of experts review the building replacement costs and floodplain crop values used in the models.

River floods

Have local civil protection review, commune by commune, the flood zones drawn from the national map, then update the map with each new event.

Improve the management of the catchments of flood-prone rivers.

Improve local maps — geology, land use, aerial photographs — and recover the original course of streams diverted by urbanisation, an observed cause of serious damage in Bujumbura.

Extend the hydrological modelling carried out in the priority areas to other problem rivers.

Inspect bridges after every major flood: scouring at the foot of the piers, slips on the banks.

Install early-warning mechanisms along rivers, with automatic or manual gauging stations.

Torrential rain

Update the model with detailed event records, and use existing automatic rainfall records to better understand the rains that cause damage.

Encourage the resettlement of vulnerable households living along overflowing ravines or in flat areas where water accumulates.

Set up warning systems to evacuate vulnerable residents in time, and build collective shelters in strategic locations.

Strong winds

Also apply most of the measures planned for torrential rain.

Help people build back better, especially households who cannot afford a wind-resistant roof.

Study the value of traditional fired-clay tiles, more common in the mountainous centre than along the lake, for better resistance to strong winds.

Earthquakes

Adopt earthquake-resistant building codes and enforce them.

Reduce self-building: adobe and unreinforced masonry are highly vulnerable and should be strengthened, with support from trained specialists to check designs and supervise building sites.

Inventory and strengthen critical infrastructure: roads, bridges, dams, water, electricity.

Update geological maps, study active faults and carry out microzoning studies in the west of the country, where soft soils predominate.

Use the results for contingency plans and urban planning, and avoid building in high seismic-risk zones.

Landslides

Limit urban development in areas highly susceptible to landslides and avoid any new housing there.

Coordinate river management to take bank erosion into account.

Limit uncontrolled deforestation and land use; reforest or plant crops that hold the soil.

Take risk zones into account when building and maintaining roads and bridges.

Lake Tanganyika

Clarify the gap between the flooded areas observed in Bujumbura and IGEBU’s models, check the gauge at Bujumbura port and install equivalent devices in other shoreline towns.

Review shoreline planning according to the three water levels, and revise plans as knowledge of the lake improves.

In the short term, assist affected households, including by relocation from the most exposed areas.

In the highest-risk areas, consider dykes, vegetation barriers or raised buildings, after technical, social and cost-benefit studies, while preserving the lake’s ecological balance.

Cooperate with the other lakeshore countries to understand the lake’s behaviour and build long-term joint solutions.

Needs expressed by actors on the ground

Information. Commune and provincial planning documents (PCDC, PCIA, SPAT) lack data on land use, hydrology and risk zones. Mapping risk zones would avoid resettling displaced people in other dangerous areas.

Warning and dissemination. An early-warning system at national and community level appears essential. IGEBU, private media and community radio stations need to disseminate information better, with simplified seasonal forecasts accompanied by concrete action messages.

Means. Lack of funding and expertise in communes, lack of protection and drainage works, need to rehearse contingency plans (stressed by the Red Cross), need for vehicles and emergency equipment.

Education. Awareness and education on risk management, and good practice in a crisis, should be spread more widely.

At your level

Learn, act, pass it on

Proposal by ACAPE Burundi for clubs, schools and parishes, drawn from the study’s recommendations.

Learn

Know your commune’s hazards

Look up your province’s and commune’s map on the national risk platform: which hazards, and where? Identify with the elders the places that have already flooded or slipped.

Act

Plant and record

Reforest slopes and riverbanks with the agreement of the local authorities and landowners. Keep a register of events — place, date, extent, damage — and hand it to the commune administration.

Pass it on

Relay what is verified

Relay only official warnings and forecasts, at school, at church or on community radio. Hold a discussion based on your commune’s map.

Safety first

Never go near a river in flood or a bank that has just collapsed to observe or take photographs. Photos of people are taken with their consent, and with their parents’ consent for minors. A warning does not replace the emergency services.

Going further

Using the national risk platform

The study’s maps, layers and reports are freely accessible online.

1

Open the platform

Go to the address 23.239.19.79, from a computer or a phone.

2

Create an account to download

To download maps or layers, register with your email address as username, then log in. The address is not encrypted (http): choose a password you use nowhere else.

3

Browse

The Maps tab lets you explore the maps; the Data tab, the layers and documents. The search bar can be narrowed by keywords, data type and category.

To read

The study’s reports

The summary of results, the final report, the risk atlas, the needs assessment and the preliminary assessment of the rising waters of Lake Tanganyika.

Summary of results on the platform · All documents on the platform

For the course

What this analysis teaches

Proposal by ACAPE Burundi: three ways into the modules of the environmental training course.

Module 4

Climate change

Rain concentrating on fewer days: a quantified Burundian example of what warming changes.

Module 7

Reforestation and restoration

Reforestation is the first nature-based solution recommended against landslides.

Module 10

Practical actions

Keeping a register of events on your colline: a simple action that answers a national recommendation.

Module 4 · Module 7 · Module 10

Sources

Maps: IOM Burundi, reproduced for educational and non-commercial purposes · IOM Burundi, Multi-hazard risk assessment and mapping in Burundi (2020-2022), summary of results, consortium of IDOM, EURAC Research, Meteosim, GIIS-Geolyder, APRN and Ramboll, funded by the European Union, national risk platform, document posted on 21 October 2022 · IOM Burundi and Resilience Programme, factsheet “Multi-hazard mapping in Burundi” · World Bank, Burundi’s 2019 GDP, cited by the study.

See the six illustrations · See the four official maps · All learning resources · Take the environmental training course

The illustrations on this page were produced by ACAPE Burundi from these sources. They may be reused and adapted, provided that ACAPE Burundi and the data sources are cited.

Last modified: Wednesday, 30 September 2026, 3:00 PM