LAUSR

Abstract Rainfall-induced slope failures are commonly observed in residual soil. The slopes in Singapore are associated with deep groundwater table. As a result, residual soils in Singapore are commonly observed within unsaturated condition above groundwater table. The unsaturated soil properties (i.e. soil-water characteristic curve) are rarely used as parameters in the development of slope susceptibility map. To incorporate important unsaturated soil properties, such as air-entry value (AEV), in the slope stability analysis, the sampled soil data must be transformed into a continuous field output. Geostatistical methods of ordinary kriging (OK) and simple kriging (SK) were used to estimate the spatial variability of air-entry value for soils in Singapore. The effect of soils from different geological formations in Singapore was considered in the analyses. The first case of analysis assumed Singapore as one region whereas the second case of analysis assumed Singapore as two regions – (i) Jurong Formation (JF); and (ii) Bukit Timah Granite and Old Alluvium (BTG & OA). Leave-one-out cross validation and one additional sampling point were used to assess the performance of the kriging methods and determine a suitable method for digital soil mapping. OK and SK had similar root mean square error (RMSE), of 66.6 and 66.4 respectively, for analyses with the assumption of Singapore as one region. For analyses with the assumption of Singapore as two regions, the RMSE using OK was estimated at 32.6 and 85.6 for BTG & OA and JF, respectively. SK estimated the additional sampling point with greater accuracy and was concluded as the better kriging method to estimate the AEV distribution of soils in Singapore as compared to OK method.