Assessing earthquake risk in East Sub-Saharan Africa

The African continent is prone to a variety of natural and human-induced hazards and disasters. The population in Africa, estimated at 1.1 billion people, is growing at a rate of 3% per annum, resulting in a continuous increase in the number of people exposed to natural hazard threats. The fact that several African countries have elevated levels of poverty makes them ill-prepared to cope with the impacts of these hazards. Physical phenomena such as earthquakes have caused extensive human and economic losses in the past. In the last century, more than 80 destructive earthquakes have occurred, causing approximately 21,000 fatalities, almost a 1,000,000 homeless and economic losses exceeding 12 billion USD.

The vast majority of the studies related with earthquake losses in Sub-Saharan Africa have focused on probabilistic seismic hazard assessment of the active tectonic regions around the African Rift. While such results are certainly useful, in order to effectively develop and implement disaster risk reduction measures, a comprehensive understanding of the possible losses and their geographical distribution is fundamental. In this context, the Global Earthquake Model (GEM), in collaboration with the United States Agency for International Development (USAID), has led an initiative (Sub-Saharan Africa Hazard and Risk Assessment – SSAHARA) to create an open earthquake model for East Sub-Saharan Africa. The figure below highlights the countries covered in this phase of the project.

Countries covered within the SSAHARA project

The East African Rift System (EARS) is the major active tectonic feature of the Sub-Saharan Africa (SSA) region. Although the seismicity level of such a divergent plate boundary can be described as moderate, several damaging earthquakes have been reported over the past few decades, and the seismic risk is exacerbated by the high vulnerability of the local buildings and structures. Formulation and enforcement of national seismic codes is therefore an essential future risk mitigation strategy.

The probabilistic hazard model developed for the Sub-Saharan Africa region within this project contains only distributed seismicity sources. The current model has been calibrated on the most recent and up-to-date information available from scientific literature, global bulletins and local earthquake catalogues, such as those from the partner project AfricaArray. The calculation of seismic hazard was performed using the OpenQuake-engine.

Map of spectral acceleration at 0.1 sec for a probability of exceedance of 10% in 50 years

To fully evaluate the impact of seismic risk in Sub-Saharan Africa, exposure models were developed for the residential building stock and population at national and sub-national levels. The primary sources of population and housing information were the national census surveys from each country. Other sources of data involved peer-reviewed literature, UN-Habitat technical reports, and publicly accessible data from ministries and agencies involved in the housing sector. The information from the housing census of each country allowed the estimation of the number of dwellings and buildings at the first administrative level, as illustrated in figure below.

Distribution of the number of dwellings at the first administrative level

For each country, a mapping scheme was developed to convert the information featured in each housing census into the GEM building taxonomy. This task was performed in close collaboration with local experts, and revised at a local workshop in June 2016 at Addis Ababa. Ten building classes were identified representing concrete, masonry, earthen and wooden buildings. Informal construction constituted 84% of the housing stock in Sub-Saharan Africa as shown by some of the most common types of construction seen in the images below.

Most common types of construction in Sub-Saharan Africa

For each building class, a replacement cost and average built-up area was estimated based on the type of housing, location (i.e. urban or rural), data from the housing and real estate sectors, and urban housing and socio-economic indicators. This approach allowed developing exposure datasets for the eight countries comprising information about the number of dwellings, number of buildings, replacement cost and occupants according to a set of building classes as shown in the case of Ethiopia in the figure below.

Exposure model for Ethiopia at the first administrative level

Due to the lack of data regarding post-earthquake damage in Sub-Saharan Africa, an analytical approach has been employed in the development of these functions. Based on the building classes identified in the exposure component of the project, several numerical models were derived to simulate the expected seismic behavior. These models were tested against a large set of ground motion records selected according to the local seismicity and tectonic environment. In this process, the building-to-building and record-to-record uncertainties were propagated until the final vulnerability functions. These results allow the estimation of damage for specific seismic events, as well as average annual losses considering probabilistic seismic hazard. These functions are the first of their kind defined specifically for the African region. The figure below illustrates the resulting fragility functions for earthen and wooden typologies identified in all the countries.

Fragility functions for wooden (left) and earthen (right) buildings

For the first time in Africa, a systematic approach to seismic risk assessment has been implemented, where all the components of risk (hazard, exposure and vulnerability) were combined in an open-source software for seismic hazard and risk calculations – OpenQuake. Two main approaches for the assessment of losses were followed for this project. Firstly, a deterministic approach was used in which historical earthquakes were repeated to verify and calibrate the various models. In addition, earthquake scenarios provide useful information such as the number of collapsed buildings, casualties, people left homeless and total economic losses for specific events. Secondly, a fully probabilistic approach was followed, in which sets of seismic events generated stochastically (based on the probabilistic seismic hazard model) were utilised to simulate the losses for a large time interval. These losses were used to assess the average annual economic and human losses, losses for specific return periods and risk maps.

Average annual economic losses at the first administrative level

The losses per country have also been aggregated in order to directly determine which nations have higher risk. From the figure below, it is clear that Uganda has the highest economic risk on an annual average basis followed by Rwanda and Ethiopia

Average annual losses for the countries covered within the SSAHARA project

For the region covered in this project, it has been estimated that earthquakes are responsible for an average annual economic loss of about 76.4 million USD. This value is equivalent to the investment in National Agricultural of Burundi for the period of 2012-2014. Finally, in addition to identifying the region where risk mitigation should be prioritized, these analyses also allow understanding which building classes have a higher contribution to the overall economic losses, as illustrated in below. The building classes with a higher risk should be the target of retrofitting campaigns or stricter construction regulations.

Loss contribution due to each building class in Ethiopia at the national level

The collaborative approach adopted by GEM in assessing risk in Africa with transparent and open source tools and datasets facilitated the presentation, discussion and validation of results in Ethiopia. In this event, more than 100 experts from the scientific community, government agencies and international organisations participated. The methodologies and approach of SSAHARA are being extended to the rest of the African countries to assess seismic risk in a holistic manner. The GEM Foundation will continue supporting the African risk reduction community in the assessment of earthquake losses, and development of risk reduction policies. All of the models and datasets will be made available to the general public by the end of 2016.

For additional information about the datasets, methodologies, tools and models described herein, please contact us at