Abstract:
Landslides are considered one of the recurrent natural disasters spread worldwide, causing
significant property damages and fatalities. Hydraulic gradient is one of the triggering factors
affecting landslide hazards. Depending on the material, landslide bodies can hold a significant
amount of groundwater. The effect of the hydraulic gradient is discussed in terms of decreasing
suction head, rising groundwater table, groundwater exfiltration from bedrock, and hydraulic
uplift pressure below the landslide. Furthermore, Climate change causes changes in rainfall
regime and evapotranspiration, which reduces the amount of water recharge aquifers. As a
result, the hydraulic gradient increases resulting in groundwater flowing faster and being
depleted more quickly. This dynamic maturity of the hydraulic gradient and its spatial variation
can be estimated using a groundwater simulation model.
Kegalle district in Sri Lanka is listed as one of the highly vulnerable areas for landslides by the
National Building Research Organisation (NBRO), Sri Lanka. Heavy rainfall generally occurs
in the area during the southwest monsoon season, causing variably saturated soil conditions
and reducing the shear strength. This study simulated the spatial distribution of hydraulic
gradient in an area covering 1,523 km2 using the United States Geological Survey (USGS)
modular finite-difference flow (MODFLOW) model. The study area was divided into 4,148
active square grids with an approximate 70 m × 55 m grid resolution coverage. The area was
conceptualised as a single-layer aquifer with an average depth of 200 m. The model's upstream
and downstream sides were considered constant head boundaries, and their magnitudes were
estimated by drawing equipotential lines using groundwater level data at nine observation wells
in the study area. Annual total rainfall during the 2007-2017 period at the Holombuwa rain
gauge station was used to estimate the groundwater recharge. The model was run in a steady
state and the hydraulic conductivity, groundwater recharge, and river conductance parameters
were calibrated using observations at nine wells. When projecting future rainfall, outputs from
the CNRM-CM6-1-HR Global ClimateModel (GCM) for two shared socioeconomic pathways
(SSP3-7.0 and SSP5-8.5) were downscaled using the quantile mapping method to the local
scale for the near-term (2020-2050) and the long-term (2070-2100) periods. The increase in
the annual average rainfall for each scenario was calculated and the same percentage of change
was assumed for groundwater recharge in estimating hydraulic gradient in the future.
Past events in the study area categorised as landslides, slope failures, and cutting failures
recorded by the NBRO were obtained for the 2016-2021 period. These observed events were
compared and matched with simulated hydraulic gradient distributions in the study area. A
60% match for the present period was observed between the landslide observations and the
range of hydraulic gradient identified as critical. Future projections indicate a 0.03-1.57%
change in landslide-susceptible areas with the greatest changes estimated for the SSP5 scenario.
