GSA Annual Meeting in Phoenix, Arizona, USA - 2019

Paper No. 182-28
Presentation Time: 9:00 AM-6:30 PM


OBE, Oluwatosin, Department of Geosciences, MTSU, 1540 Lascassas Pike, Apt 123B, murfreesboro, TN 37130


Landslide is one of the most common hazards occurring worldwide. Rainfall-triggered landslides constitute a serious hazard and an important geomorphic process in many parts of the world. Different approaches have been made to investigate triggering conditions in order to identify patterns in behavior and, ultimately, to define or calculate landslide-triggering rainfall thresholds. This study was carried out in Tennessee and Kentucky. It is aimed at determining the rainfall intensity-duration thresholds (ID Curves) for landslide prediction by considering the effects of antecedent rainfall. Data for the time and location of landslides that occurred in Tennessee from 2008-2014 and Kentucky from 2008-2015 were collected. Overall, 75 landslides histories were used to determine rainfall thresholds. Also, rainfall duration that causes (influenced) landslide events were determined. Triggering rainfall conditions are represented by a combination of rainfall occurring in a period before the event (antecedent rainfall) and rainfall on the day of the event.

The number of antecedent rainfall days that influence the occurrence of landslides was determined by manually fitting a line on the scatter plot of event day rainfall against antecedent rainfall for 3, 5, 10, 15, 20 and 30 days. The line was purposely drawn to separate triggering and non-triggering rainfall events as much as possible. The graph on which the fitted line achieved minimum mixing of triggering and non-triggering rainfall was chosen as being associated with the occurrence of landslides. Results from the empirical model indicate that landslides in the area are generally influenced by 10 days antecedent rainfall and daily rainfall. On the 10 days, antecedent rainfall graph two threshold lines representing minimum triggering rainfall conditions and maximum non-triggering rainfall conditions were drawn. The area below the line crossing minimum triggering rainfall conditions was assigned a 10% probability of landslide occurrence and the area above maximum non-triggering rainfall conditions were assigned 90% probability of landslide occurrence. The accuracy in predicting landslide from the threshold is 70%. Thus, subject to improvement in the future, the threshold derived from this study can be used in a simple early warning system for landslides in the area.