Graeme Weatherill

K -means cluster analysis and seismicity partitioning for Pakistan

Pakistan and the western Himalaya is a region of high seismic activity located at the triple junction between the Arabian, Eurasian and Indian plates. Four devastating earthquakes have resulted in significant numbers of fatalities in Pakistan and the surrounding region in the past century (Quetta, 1935; Makran, 1945; Pattan, 1974 and the recent 2005 Kashmir earthquake). It is therefore necessary to develop an understanding of the spatial distribution of seismicity and the potential seismogenic sources across the region. This forms an important basis for the calculation of seismic hazard; a crucial input in seismic design codes needed to begin to effectively mitigate the high earthquake risk in Pakistan. The development of seismogenic source zones for seismic hazard analysis is driven by both geological and seismotectonic inputs. Despite the many developments in seismic hazard in recent decades, the manner in which seismotectonic information feeds the definition of the seismic source can, in many parts of the world including Pakistan and the surrounding regions, remain a subjective process driven primarily by expert judgment. Whilst much research is ongoing to map and characterise active faults in Pakistan, knowledge of the seismogenic properties of the active faults is still incomplete in much of the region. Consequently, seismicity, both historical and instrumental, remains a primary guide to the seismogenic sources of Pakistan. This study utilises a cluster analysis approach for the purposes of identifying spatial differences in seismicity, which can be utilised to form a basis for delineating seismogenic source regions. An effort is made to examine seismicity partitioning for Pakistan with respect to earthquake database, seismic cluster analysis and seismic partitions in a seismic hazard context. A magnitude homogenous earthquake catalogue has been compiled using various available earthquake data. The earthquake catalogue covers a time span from 1930 to 2007 and an area from 23.00° to 39.00°N and 59.00° to 80.00°E. A threshold magnitude of 5.2 is considered for K-means cluster analysis. The current study uses the traditional metrics of cluster quality, in addition to a seismic hazard contextual metric to attempt to constrain the preferred number of clusters found in the data. The spatial distribution of earthquakes from the catalogue was used to define the seismic clusters for Pakistan, which can be used further in the process of defining seismogenic sources and corresponding earthquake recurrence models for estimates of seismic hazard and risk in Pakistan. Consideration of the different approaches to cluster validation in a seismic hazard context suggests that Pakistan may be divided into K = 19 seismic clusters, including some portions of the neighbouring countries of Afghanistan, Tajikistan and India.

Identifying the needs and future directions of seismic hazard for probabilistic infrastructure risk analysis

The vulnerability of urban infrastructure to both ground shaking and geotechnical failure during large earthquakes has been demonstrated by recent earthquakes such as the 2010 2011 Canterbury earthquake sequence (New Zealand, 2010 2011) or 2010 Haiti event. Probabilistic seismic risk analysis to infrastructure systems requires the characterisation of both the transient shaking and permanent ground deformation elements of the hazard, and must do so incorporating both the aleatory and epistemic uncertainties and the spatial correlations and dependencies that are inherent in both of these aspects. Recent developments in characterisation of spatial correlation and cross-correlation in the ground motion uncertainties form the foundations of a comprehensive Monte Carlo-based methodology for analysis of seismic risk to spatially extended systems. New research directions are needed, however, in order to ensure that secondary hazard aspects are incorporated in the same way. These include the treatment of site amplification of the ground shaking, the modelling of permanent ground deformation from slope displacement and liquefaction, and permanent displacement due to coseismic slip on and around the fault rupture. Key considerations for integrated probabilistic framework for physically-realistic characterisation of the ground shaking and permanent ground displacement are illustrated using the example of simulation spatially correlated fault slip on an active fault rupture in a manner that can be integrated within a Monte Carlo-based probabilistic seismic hazard methodology. Probabilistic analysis of seismic risk is the most common methodology used by organisations responsible for maintaining infrastructures to make informed decisions based on the cost to benefit ratios of particular mitigation strategies. But analysis of infrastructural risk presents new challenges to both the hazard and risk modellers to provide models of ground shaking and permanent ground displacement that can be applied to spatially extended and interconnected systems. A single infrastructural system is dependent on many elements over an extended geographical region, and the performance of the system or systems may depend on the location of greatest damage. Further compounding the complexity is the fact that a single infrastructural system is composed of fragile elements that may respond dissimilarly according to different characteristics of the hazard. One example might be a gas network in which mechanical elements may be most adversely affected by high frequency acceleration, storage systems and pumping stations by low frequency ground motion, whilst underground pipes may be most at risk from permanent ground displacement due to geotechnical failures. This paper outlines some of the main challenges facing the seismic hazard modeller in order to provide hazard input into probabilistic seismic risk analysis for interconnected and spatially extended infrastructures. Focus is placed on four critical elements: ground shaking, site amplification, geotechnical failure due to land-sliding and liquefaction, and finally co-seismic displacement due to rupture on the fault surface. A summary of current approaches for characterising each of these elements