Urban Seismic Risk in Java,Indonesia

Abstract

Indonesia lies in one of the most tectonically active parts of the world and has experienced many damaging earthquakes in the past. We know most of the residential buildings do not have high seismic resilience and do not follow Indonesia's Building Code. The continuing trend of high population and GDP growth in Indonesia means that many residential and large infrastructure projects are expected to be undertaken in the near future. Unless appropriate mechanisms are put in place soon to ensure new construction is seismically resilient Indonesia risks locking itself into a future with high vulnerability to earthquake disasters.

Understanding and quantifying seismic risk is critical to underpin disaster risk reduction and mitigation policies. Detailed knowledge in three key areas is required to quantify seismic risk: earthquake hazard, exposed assets in the built environment, and, the vulnerability of these assets to ground shaking. In spatially small areas such as a small city or urban area it is possible to have extremely detailed information in all three key areas. However at large scales compromises are made on the level of detail, particularly in the characterisation of exposed assets. The uncertainty this introduces into risk assessments is understudied.

I focus on the most populous island in the world, Java, Indonesia. The aim was to use existing data from multidisciplinary sources, to develop a nuanced and robust exposure model, identify key sources of uncertainty, and determine the sensitivity of seismic risk calculations to this uncertainty. Classification of residential houses, which constitute ~70% of the buildings in Java, was identified as the main source of uncertainty in the exposure model. This uncertainty was reduced through: validation studies on the seismic vulnerability of Indonesian houses, scenario damage analysis validation studies using recent historical events in Java, and sensitivity analysis of remaining uncertainty and quantifying the impact that this has on seismic risk and loss calculations. Throughout the study it emerged that there were deficiencies in the hazard model which does not account for the effect active volcanism has on seismic wave attenuation, and has poor characterisation of some active faults located near major cities. Thus, the sensitivity of seismic risks to these factors was also investigated.

My findings show that in Java, the seismic risk is most sensitive to the selection of the ground motion models used. The seismic risk is also significantly sensitive to the classification of houses and the presence of active volcanism. Seismic risk naturally increases with the addition of new faults into the source model, which has implications for several major cities in Java. The economic losses can vary by > 50% depending on the parameters used in a risk simulation, demonstrating the importance of understanding the uncertainty associated with the hazard and exposure models and the impact on risk analysis. The degree of sensitivity of risk results varies spatially at both the province and kabupaten (regency) levels and also vary between risk metrics. In all risk simulations the buildings with the highest economic losses are residential buildings, highlighting the importance of focussing on residential buildings in reducing risk.

The research presented greatly advances our understanding of seismic risk in Java, and can be used in future seismic risk studies, are applicable to impact and risk studies for other natural hazards and can be used to inform disaster risk reduction policy and management. Whilst the focus in this thesis is on Java, the approach used is broadly applicable to the rest of Indonesia and other developing nations where the costs of developing detailed information on the built environment are prohibitive and a serious impediment to understanding and quantifying risks posed by natural hazards. Show more