Macro-level estimation of present and future seismic loss by modelling spatio-temporal dynamics of exposure

• Main Author: Mestav Sarica, Gizem
• Other Authors: Pan Tso-Chien
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2021
• Subjects: Engineering::Civil engineering
• Online Access: https://hdl.handle.net/10356/146135
• Institution: Nanyang Technological University

A society resilient to natural hazards is one that can adapt to the increase in exposure, absorb the impacts of hazardous events, and distribute the risk among different stakeholders. Megacities, cities having more than 10 million inhabitants, are complex, highly interconnected and dynamic systems that are vulnerable hotspots of natural hazards due to high concentration of people, properties, infrastructure and economic activities. Currently, the majority of global megacities are located in Asia, and many Asian megacities are often exposed to extreme seismic events being located on top of or near seismic faults. In fact, Asia has been the region with the largest built-up area exposed to earthquakes, with total area tripled from 1975 to 2015. While historical loss and exposure assessments can explain the past, they do not necessarily shed light on the future. Thus, the main aim of this study is to develop a framework for the estimation of present and future seismic losses by using a macro-level loss estimation approach, and modelling the spatio-temporal dynamics of exposure. For this purpose, three earthquake-prone Asian megacities from developing countries, namely Jakarta, Metro Manila, and Istanbul, were selected as case studies. The proposed framework involves integration of different methodologies in the literature which have previously been used for other purposes. Since hazard component cannot be predicted and controlled, this unique feature of earthquakes highlights the importance of modelling, mapping and quantifying exposure to understand earthquake risk. A common strategy that is used to estimate future exposure is to make temporal predictions that assume all exposed urban grids will be equally affected by possible disasters, although exposure is actually a complex risk component that varies across temporal and spatial dimensions. Thus, projections related to the urban fabric should be done both spatially and temporally in order to analyze trends of exposure, which generally show a nonlinear behavior. In this study, past and present built-up areas of selected Asian megacities were initially extracted from satellite images to obtain high-resolution exposure information. Then, a cellular automata-based urban growth model was utilized to predict the built-up area in 2030, using the mapped past and present built-up areas and several other growth indicators which affect urbanization trends. Projected census populations were disaggregated to grids using a dasymetric mapping technique to obtain present and future population grids. Following that, grid-level wealth information was obtained by multiplying present and future GDP per capita with the corresponding population grids and the exposure correction factor, which accounts for the disparity between national wealth and economic value of assets that are exposed. Finally, probabilistic seismic hazard maps were overlaid with exposure maps to assess spatio-temporal changes in built-up area exposed to seismic hazard, and thus wealth exposed to seismic hazard. An objective understanding of the path towards urban resilience can be provided by measuring loss over time. As such, several scholars have suggested that quantitative risk assessments should not be considered as static analyses which are performed at a certain time, and time-dependent monitoring of risk, which reflect the pace of changes in cities, should be employed. Conventional seismic loss estimate methodologies generally require a detailed inventory of exposed structures that is usually not publicly available especially in developing countries. Several researchers have attempted to simplify this process by using macro-level approaches which eliminate the need of a detailed inventory. These approaches utilize socioeconomic indicators, assuming that there is a direct relation between economic condition of a region and the potential loss resulting from an event in that region. Although macro-level approaches can be effectively used for evaluation of spatio-temporal trends and thus estimation of future seismic loss, they have not been utilized for such an objective before. In addition to this, they have not been integrated into natural catastrophe modelling schemes to estimate primary probabilistic risk metrics, which are widely used by the insurance industry and the authorities. An enhanced macro-level seismic loss estimate methodology was thus developed in this dissertation for probabilistic risk assessment by leveraging on the methods in the literature. The average annual loss (AAL) in grid-level, Admin Level 2 and megacity-level, as well as the probable maximum loss (PML) in grid-level were estimated by considering present and future exposure to seismic hazard of the selected megacities. To achieve this goal, a multi-step framework was developed by utilizing the information on spatio-temporal exposure dynamics. The megacity-level results show that AAL is predicted to increase by 2030 almost twofold, from around 68 million USD to 129 million USD in Jakarta (in 2018 prices), and from around 163 million USD to 333 million USD in Metro Manila (in 2016 prices). The present AAL in Istanbul is predicted to show a high increase of around 57% from around 900 million USD to 1,417 million USD by 2030 (in 2018 prices). The enhanced macro-level loss estimation approach following the proposed framework was observed to produce loss estimations in the same order of magnitude as the conventional loss estimation methods. Therefore, it can be used as a complementary tool in addition to the conventional risk assessment framework. The findings of this research shed light on future exposure and loss assessments by making comparisons among present and future. Since the results obtained are in terms of grid-level maps for built-up area, wealth and seismic loss, this study demonstrates the growth of each grid over time. It was observed that in regions where high levels of exposure and hazard are coupled, high potential of losses are predicted, while there are different trends in each megacity due to their varying historical urbanization patterns. Urban planners can utilize the grid-level present and future exposure to seismic hazard for risk mitigation or zoning purposes. In addition, the AAL maps for Admin Level 2 can aid local authorities in allocation of resources, while megacity-level AAL and PML can provide information to governments and decision-makers about disaster risk financing options to reduce forthcoming risks.