Abstract:
Landslides present a significant peril to lives and economies, notably in Asia where over 18,000
deaths and $8 billion in economic losses occurred due to landslides from 1998 to 2017. These
events stem from a range of factors, including steep terrain, geological features, and extreme
rainfall. Rainfall, influenced by climate change, emerges as a key factor in increasing landslide
vulnerability. The IPCC projects intensified rainfall and droughts due to global warming,
heightening landslide risks. Recognising this, the study focuses on the Kegalle District, Sri
Lanka, to investigate the impact of climate-induced shifts in rainfall patterns on landslide
susceptibility. The findings aim to provide decision-makers with insights for proactive
measures.
The Kegalle District in Sri Lanka, a landslide-prone area, has experienced an increase in
landslides despite a history of fewer occurrences. The study used the HadGEM3-GC31-LL
model from CMIP6 to project potential shifts in future rainfall patterns. The baseline period
selected for analysis is 1975-2015, while two Shared Socioeconomic Pathways: SSP2-4.5 and
SSP5-8.5 were considered for future projections. Statistical downscaling was performed using
the Long Ashton Research Station Weather Generator, and missing values (0.2%) were filled
using the Multivariate Imputation by Chained Equations method. Daily rainfall data from
Ratnapura station was distributed across the study area (Figure 1) using the gridded Climate
Hazards group InfraRed Precipitation with Station (CHIRPS) dataset (with a grid resolution of
5 km x 5 km). A Python code generated bias-correction factors for accurate future rainfall
projections. Historical landslide events categorised by the NBRO in the 2016-2021 period were
correlated with days of excessive rainfall to gauge susceptibility.
Figure 2 shows the average number of days with daily rainfall exceeding 73 mm during the
2016-2021 period. The minimum rainfall threshold for triggering past landslides was identified
using NBRO data. The number of days above this threshold was used to define the range of
days required for triggering landslides, slope failures, and cutting failures (Table 1). Figure 3
shows potential changes in these events from projected rainfall by SSP2-4.5 during the 2031-
2060 period. Results show that the extreme category (4.2-6 days/year) expands throughout the
catchment area in the future compared to the observational period. Similar effects were
observed in different magnitudes for both SSP scenarios in two different periods (Table 2).
The findings of this research concluded that the rainfall effect on landslide susceptibility can
be significant and that climate change effects could exacerbate the likelihood of landslides in
the future. To enhance the accuracy of the analysis, it is recommended to incorporate additional
landslide-triggering factors for the susceptibility analysis.
