Using PCA in evaluating event-controlling attributes of landsliding in the 2005 Kashmir earthquake region, NW Himalayas, Pakistan

Many investigations have been conducted to understand the spatial distribution of earthquake-triggered landslides. This paper presents the event-controlling attributes of landsliding in the 2005 Kashmir earthquake event using principal component analysis (PCA) and regression analysis. The earthquake with magnitude M w 7.6 produced 2930 landslides, distributed in an area of about 3250 km2, in the northern part of Pakistan and the Kashmir region. By utilizing SPOT imageries, reconnaissance field survey and geographic information system, the landslide distribution and density were statistically analysed. Controlling factors such as distance from the epicentre, distance from faults, slope steepness, slope aspect, elevation and lithology were considered for the PCA. In order to determine the relationship between different controlling factors, correlation and regression analyses were also taken account of. The results demonstrate that the major effective factors to control the landslide distribution are distance from the earthquake source followed by the slope steepness. Using PCA, distance from the epicentre and distance from the active Muzaffarabad Fault proved to be key influential factors. There is a significant negative correlation between landslide concentration and distance from the epicentre, Muzaffarabad Fault and Main Boundary Thrust. There is a significant strong positive correlation between landslide concentration and slope steepness. The lithology, however, has no significant relationship with landslide concentration.