‘Breaking New Ground’; An investigation into coseismic ground cracking following the 2016 Mw 7.8 earthquake near Kaikoura, New Zealand

Abstract

Seismic shaking can cause landsliding throughout mountainous topography. Posing a direct hazard to the people and infrastructure that occupy these environments, landsliding receives considerable attention from the scientific community. However, few studies have detailed and analysed another form of earthquake-induced damage – ground cracks. Cracks could be a potential indicator of incipient landsliding and/or a surface expression of the retention of damage by hillslopes. Existing damage makes hillslopes more vulnerable to future failure. As such, ground cracking poses a lingering hazard presenting a need to better understand it – in particular its geomorphological characteristics and most influential controlling factors, and therefore how it can be detected/modelled. In 2016 the Mw 7.8 Kaikoura earthquake in New Zealand resulted in extensive ground cracking, providing an ideal case study. A ground crack inventory was digitally compiled using visual interpretation of post-event aerial photography. A detection attempt using a post-event digital terrain model (DTM) to semi-automatically extract cracks was unsuccessful. However, comparing this with an attempt using higher-resolution sample data emphasizes the necessity to consider the interdependence between feature scale and data resolution when attempting to detect/analyse. Feature analysis found that cracks are preferentially 7 m (~3-8 m) in length. Lack of small features may be due to minimum strain thresholds and strain accumulation. Larger cracks have likely developed into landsliding. Both offer new insight into internal hillslope forcing. Cracks preferentially form in a slope perpendicular direction, indicating a topographic control on propagation. Further potential controls were statistically analysed using Fuzzy Logic, which then informed a spatial prediction. The most influential control is proximity to landsliding, suggesting that in most cases cracking is an expression of incipient landsliding. Cracking preferentially occurs at ridgetop locations and on hillslopes facing the source of shaking. The latter is the inverse of behaviour exhibited by landsliding, highlighting the interdependence between directional shaking, local slope aspect and normal/shear stress. This conforms to and provides a new novel insight into the topographic site effects theory. Whilst quantitatively unsuccessful, the best performing spatial prediction model showed great promise in locating ground cracks in areas of high hazard, providing a solid foundation for improvement through further research so that eventually models like this can better inform ongoing hazard monitoring.