103. Challenges of Including Groundwater Depth in a Liquefaction Loss Model

The near surface or depth to the first occurrence of groundwater is vital in evaluating liquefaction initiation and therefore a liquefaction loss model. If groundwater is below a susceptible layer, that layer is unlikely to liquefy. This study will discuss the challenges in modeling near surface groundwater depths over a large region, using the Canterbury region of New Zealand as an example, and its importance to loss estimation relative to liquefaction hazard. Estimating the depth to near surface groundwater is an emerging discipline that benefits from geostatistics and hydrology. While accounting for spatial randomness in the groundwater depth, we develop a site-specific mean trend model assimilating secondary datasets correlated with the primary variable and process it through the Kriging system. Secondary datasets considered are a continuous map of distance to a water body (DWB) and ground surface elevation through a digital elevation model (DEM). Resulting estimation improves upon a Kriging method not incorporating the secondary information (Mean Square Error reduction ranges from 47.5 to 73.8 %). The estimation of the groundwater depth at unsampled locations leads to enhanced understanding of liquefaction over a region and thus of spatial distribution in the loss uncertainty. The findings of this study would be of great help to investigate the importance of the location of groundwater monitoring networks near concentrations of building exposures.