Department of Geological and Mining Engineering and Sciences, Center for Data Sciences
A combined cluster and regression analysis were performed for the first time to identify rainfall threshold that triggers landslide events in Amboori, Kerala, India. Amboori is a tropical area that is highly vulnerable to landslides. The 2, 3, and 5-day antecedent rainfall data versus daily rainfall was clustered to identify a cluster of critical events that could potentially trigger landslides. Further, the cluster of critical events was utilized for regression analysis to develop the threshold equations. The 5-day antecedent (x-variable) vs. daily rainfall (y-variable) provided the best fit to the data with a threshold equation of y = 80.7–0.1981x. The intercept of the equation indicates that if the 5-day antecedent rainfall is zero, the minimum daily rainfall needed to trigger the landslide in the Amboori region would be 80.7 mm. The negative coefficient of the antecedent rainfall indicates that when the cumulative antecedent rainfall increases, the amount of daily rainfall required to trigger monsoon landslide decreases. The coefficient value indicates that the contribution of the 5-day antecedent rainfall is ∼20% to the landslide trigger threshold. The slope stability analysis carried out for the area, using Probabilistic Infinite Slope Analysis Model (PISA-m), was utilized to identify the areas vulnerable to landslide in the region. The locations in the area where past landslides have occurred demonstrate lower Factors of Safety (FS) in the slope stability analysis. Thus, rainfall threshold analysis together with the FS values from slope stability can be suitable for developing a simple, cost-effective, and comprehensive early-warning system for shallow landslides in Amboori and similar regions.
Sajinkumar, K. S.,
Early warning system for shallow landslides using rainfall threshold and slope stability analysis.
Retrieved from: https://digitalcommons.mtu.edu/michigantech-p/940
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.