Stochastic Modelling of Seismic Hazard and Risk Assessment Framework for Uganda

Uganda is situated between the eastern and western branches of the world’s largest seismically active continental rift, the East African Rift System. Over the past decades, many catastrophic earthquakes (e.g., 1929 Masaka, 1966 Rwenzori, 1994 Kisomoro and 2016 Bukoba) have had detrimental consequences on the socio-economic welfare and resilience of communities across Uganda. Moreover, the disastrous impacts of earthquakes have been exacerbated by co-seismic and rainfall-induced landslides. For instance, the unprecedented 2010 Bududa landslide which occurred in eastern Uganda killed at least 388 people and over 8500 people were affected. The socio-economic welfare and resilience of the communities located in disaster-prone areas can be directly affected by the catastrophic impacts of earthquakes and landslides. With increasing population, urbanisation and rapid construction, earthquake and landslide risks are escalating fast and is compounded by the high vulnerability of buildings and inadequate disaster prevention and mitigation strategies. Hence, there is urgent need to assess Uganda’s resilience against seismic risks. As a precursor towards building an earthquake risk and resilience framework, this study employs the probabilistic event-based risk calculator of the OpenQuake-engine to holistically assess the potential losses resulting from future earthquakes. In the first part of the dissertation, probabilistic seismic hazard assessment is conducted using a fault-oriented spatially distributed seismicity approach. A logic tree is implemented to better manage the associated epistemic uncertainties and site conditions are modelled using the upper 30 m average shear wave velocity across Uganda. Seismic hazard findings confirm that western Uganda is exposed to the highest level of seismicity where for instance Rwenzori and Kigezi regions can expect bedrock peak ground accelerations increasing up to 0.27 g over a 475-year return period. The second part of this thesis demonstrates the ability of fuzzy logic and geospatial methods in deriving a national co-seismic and rainfall-induced landslide hazard susceptibility framework for Uganda. In addition to the earthquake characteristics, the most prevalent predisposing and geomorphological landslide conditioning factors considered in this study include slope angle, topographic slope position, proximity to major active faults and geology. The final relative co-seismic landslide hazard findings conclude that the Kigezi and Rwenzori sub-regions in western Uganda are highly prone to co-seismic slope failures while landslide occurrences across the other parts are fundamentally driven by rainfall. Detailed absolute landslide hazard analysis is encouraged in regions of moderate landslide regions comprising highly weathered outcropping rocks of Precambrian granites, dominantly metasedimentary, granulites and gneisses geologies within 80 km from active faults: where bedrock PGA ≥ 0.1 g over a 475-year return period, topographic position index ≥ 3.8, slope gradient ≥ 10 degrees, and distance from streams ≤ 1.25 km. The final part of this thesis focusses on assessing potential losses due to future earthquakes in Uganda by combining seismic hazard predictions with the country’s building exposure and global analytical vulnerability models. By and large, the earthquake risk model derived in this study predicts an overall economic mean annual loss of US$ 74.722 million coupled with a mean annual death toll increasing up to 71.3 persons across the whole country. The findings herein will not only be a significantly big step towards the urgent need to kick-start the update of the Uganda Seismic Code, but also strategically contribute to land use planning patterns, optimisation of earthquake insurance pricing and improvement of the National Policy for Disaster Preparedness and Management.