Trials of a hydrogeological model of the Pos Selim landslide

Abstract

Trials have been carried out of a hydrogeological model for the Pos Selim landslide in Malaysia. The objective was to make a post-failure model for the southern area of the landslide (using linear reservoir theory) which generates from rainfall time-series input (Cameron Highlands station) an output of water pressure time-series at the basal rupture surface that corresponds with the measured movement behaviour (the velocity time series from total station monitoring October 2003 to December 2006, Lee 2015 Figure 4.2, which exhibits seasonal peaks). The model has a 5m thick 100m wide confined aquifer along the basal rupture surface (BRS) rising slightly at the toe where discharge takes place, with rainwater inputs from local surface recharge zones through 2m wide flow channels along the geological faults and through 20m wide fractured rock mass zones near the southern margin of the landslide. Various arrangements for south stream catchments and recharge areas at berm 8 to berm 18 were tried in the model. These were decided based on a model of the surface water flow system that was made using aerial photographs. Values for parameters (permeability 1x10-4m/s, effective porosity 10%, rainwater reaching BRS 20% rainfall) were chosen mainly from the literature. BRS geometry was a best-fit by eye circular arc to a BRS constructed by Cruden’s method using displacement dip values (Li 2020). For all cases tested the basal rupture surface water table (BRSWT) generated in the model stays close to toe level (1357m) throughout the test period (October 2003 to December 2006) and seasonal rises in level are limited to a few metres. The BRSTW time series is insensitive to initial value for all cases tested as recession is strong. For the assumed recharge flow paths from berms 11 to 18 through fractured rock mass near the southern margin of the landslide BRSWT spikes are broad and lag behind velocity peaks because of long travel times along the recharge flow paths, up to about 33 days (Series B runs). Halving the permeability value increases BRSWT spikes but also delays and broadens the spike (Series C compared to Series B). With adjustments to the catchments and recharge zones to shorten recharge flow paths (up to about 19 days) the BRSWT time series shows some similarity with the velocity time series but is not a close match. BRSWT spikes still lag behind the two highest velocity peaks, in Oct 2004 and Dec 2005 (Series D). Reducing the thickness of the BRS aquifer from 5m to 3m significantly increases peak BRSWT spikes (Series E compared to Series D). The water pressures are still far below those expected from limit equilibrium analyses by Ip 2014. It is concluded that a recharge model with even shorter travel times and a greater volume of rainwater reaching the BRS aquifer zone is needed to generate a water pressure time-series at the basal rupture surface that corresponds with the measured movement behaviour October 2003 to December 2006.