Probabilistic seismic loss estimation for residential buildings in Pakistan

Pakistan and the surrounding areas are located in one of the most seismically active regions of the world. Besides a high seismic hazard, the country is also facing infrastructural challenges associated with a rapidly growing population (currently ranked 5th in the world) and a high rate of urbanization in the last two decades. Several recent studies have highlighted the dire need to develop a national earthquake risk reduction plan by implementing well-researched disaster mitigation policies, pre-planning and pre-financing through disaster insurance programs, and effective rapid response and recovery strategies. One major challenge in the development of such policies and national action plans is the lack of available research and datasets focusing on earthquake losses in a probabilistic framework. This study, therefore, addresses this gap by conducting a detailed probabilistic seismic loss estimation for all major administrative units of Pakistan. For this purpose, first, seismic hazard analysis is conducted based on an updated earthquake catalogue. The seismic sources are modeled using multiple approaches and attenuation models, and the hazard predictions are combined using logic trees to account for epistemic uncertainties. The analysis results showed that peak ground acceleration (PGA) at various sites of the country ranges from 0.1 to 0.65 g corresponding to a 10% probability of exceedance (PoE) in 50 years. The PGA values for a 2% PoE in 50 years are estimated in a range of 0.1–1.4 g. Similarly, the spectral acceleration values at 0.3s period (SA 0.3s) ranges from 0.1 to 1.38 g and 0.2 to 3.0 g for 2% and 10% PoE in 50 years, respectively. Then, an exposure model for various classes of residential buildings in Pakistan is developed based on the data extracted from the recent national census, available past studies, and local expert opinions. This study classifies approximately 31.30 million residential buildings, valued at 211.56 billion USD, into 5 major categories and 9 subcategories. For each building class in this exposure model, a set of fragility curves is adopted from the Global Seismic Risk Model supported by the Global Earthquake Model (GEM) foundation. These fragility curves are converted into vulnerability curves using several damage-to-loss models. Finally, the developed exposure model, hazard curves, and vulnerability curves are combined using the OpenQuake engine to estimate average annual losses and loss exceedance curves for all major districts of Pakistan. These results are reported for several return periods to identify the most vulnerable building classes and their spatial distribution in the country. The study estimates an average annual loss of 401.4 million USD for Pakistan. The study and developed estimates can serve as a consistent basis for developing effective earthquake risk reduction strategies in Pakistan.