Abstract:
Because of their geographic distribution, New Zealand infrastructure networks are exposed to a range of natural hazards including earthquakes and earthquake induced liquefaction and landslides. Statisti-cal models based on geospatial data can help identifying exposed areas and assessing co-seismic haz-ards across large-scale infrastructure networks. The research investigates the prediction performance of global liquefaction and landslide models for New Zealand application, focussing on the role of global versus region specific input variables on the model outcome. Comparing the predictions with the ob-servational data from the 2010–2011 Canterbury Earthquake Sequence and the 2016 Kaikōura earth-quake suggests that both models benefit from the use of region specific datasets as they provide more updated information and higher spatial accuracy. As opposed to more traditional methods (e.g. in-situ or physics based models), the New Zealand specific geospatial models allow for a rapid estimation of liquefaction and landslide hazards, offering high-resolution outputs for both national- and regional-scale assessments.
The models are applied to the New Zealand State Highway network for 478 earthquake scenarios. Compared to other approaches such as a 1 000 year return period assessment, the multi-scenario ap-proach provides a new perspective on the liquefaction and landslide exposure by identifying network sections that could be recurrently affected by earthquakes. The results are linked to the network criti-cality in order to estimate the liquefaction and landslide impacts across the State Highways, presenting one example of how the geospatial models can be used to evaluate seismic hazards across large-scale infrastructure networks.
Limitations and uncertainties regarding the New Zealand specific hazard models mainly arise from the oversimplification of liquefaction related processes (e.g. no consideration of soil profile stratigraphy) and the landslide runout concepts (e.g. simplified prediction of landslide blockages across roads), which can be addressed by including more observational data and / or incorporating other (e.g. ge-otechnical) methods.
The findings of this research help to better understand the seismic exposure and impacts across large-scale infrastructure networks. They can be used to support decisions regarding mitigation efforts or emergency preparedness and response planning in order to increase infrastructure resilience for future earthquakes in New Zealand.