Impact of landslide size and settings on landslide scaling relationship: a study from the Himalayan regions of India

The reliable landslide hazard assessment entails a robust understanding of frequency-magnitude analysis of the landslide inventory. Previous studies proposed that the landslide frequency-size distribution follows a power-law distribution even though 75–90% of data deviated from the power-law fit. This deviation from the power-law fit was ubiquitous in various landslide studies irrespective of their triggering factor, spatial and temporal resolution of the landslide inventory and geological and morphological settings of the area. This study conducted a detailed frequency-size distribution of the landslides at four different locations in India’s Himalayan regions to check the validity of power law to represent the landslide data. The landslide inventory of Kathua, Shimla, Pithoragarh and Darjeeling regions are complied, consisting of 1942, 905, 2151 and 393 landslide events, respectively. The landslide frequency distribution is tested for the power-law, exponential and lognormal distribution. After a detailed statistical analysis, our finding suggests that medium-large landslide size events follow a power-law distribution, accounting for 8–25% of the total data; the rest deviates significantly from the power-law fit and follows a lognormal distribution. The cut-off between the lognormal and power-law distributions, or the minimum size for medium-large landslides, is 102.9–104.3 m2. The Kolmogorov–Smirnov (KS) and Lilliefors tests are used to validate the findings. This study provides insight into the landslide size probability distribution in different locations situated in India’s Himalayan regions.