Development of empirical rainfall thresholds for landslide forecasting in peninsular Malaysia

Abstract

Peninsular Malaysia recorded a significant number of shallow landslides every year. The occurrences of slope failure cause massive losses in lives, socio-economy, and people's well-being. The disaster becomes more frequent in monsoon or rainy season (November until March) and (May to September) each year, whereby the extreme storm or prolonged rainfall triggers numerous slope failures. Recognizing the rainfall as one of the triggering factors of slope failure, hence, it is essential to develop the empirical rainfall thresholds for shallow landslides forecasting in Peninsular Malaysia. Thirty-seven cases of landslides were gathered from 1993 to 2018 to determine the correlation between the rainfall event and landslide occurrence. By performing the practical method, the important rainfall parameters consist of rainfall intensity, and rainfall duration was acquired from the rainfall series analysis that had triggered the landslide. The obtained rainfall intensity and rainfall duration from the selected cases were utilized to develop empirical Intensity-Duration (I-D) threshold, presented in scattered plots. The threshold line equation of I = αD-β was derived by applying power model regression in double log coordinates. The best-fitted line generated in between the plot was parallelly drawn at the lowest data plot to indicate the minimum rainfall intensity that could trigger the landslide. Two types of thresholds were developed: (Imax -D) and (Imean -D), in which both thresholds acquired identical rainfall duration, but with distinct rainfall intensities. The obtained maximum intensity in the series of rainfall was applied for (Imax -D) while the mean intensity (total cumulative in series of rainfall divide by rainfall duration) was specified for (Imean -D). The proposed (Imax -D) and (Imean -D) thresholds of Peninsular Malaysia are expressed as I = 37.8D-0.114 and I = 17.5D-0.722 , respectively (I = rainfall intensity in mm/hr, and D = duration in an hour). Geology has shown that igneous rock types (granite) are more vulnerable to failure than other rock types, such as metamorphic and sedimentary. Furthermore, the developed I-D thresholds of Peninsular Malaysia dominate the mid-upper position compared to I-D thresholds from various studies worldwide. More rainfall is required to initiate the slope failure, which the thicker layer of Malaysian soils can hugely influence. The curve inclination for the mid-latitude region also resulted in higher value, including the I-D threshold of Peninsular Malaysia (-0.722 to -0.114) in comparison to cold climates region. Moreover, the validation of the developed I-D threshold with the latest landslides resulted in the actual festive event (Imax -D). However, (Imean -D) recognized a single case with false-negative events to the threshold revision. The real-time rainfall monitoring was carried out and justified that the (Imax -D) threshold is more efficient than (Imean -D) for landslide early warning system later in Peninsular Malaysia due to its high threshold value.