Establishing rainfall frequency contour lines as thresholds for rainfall-induced landslides in Tegucigalpa, Honduras, 1980-2005

In this paper, a method to derive rainfall thresholds based on the relationship between daily and the antecedent rainfall up to 6 days prior to landslide occurrence is proposed for the analysis of 134 landslide days in Tegucigalpa, Honduras, during the years 1980-2005. Based on a simple graphical procedure, rainfall frequency contour lines have been drawn in the daily versus antecedent rainfall plots to connect rainfall combinations relatively having the same frequency of occurrence. A two-bound threshold has been established: Below the lower bound, rainfall events are so frequent that any landslide day may only occur due to a significant anthropogenic disturbance, while, above the upper bound, rainfall alone is capable of inducing landslide days. Contour lines originating at the same daily rainfall value in all plots were then grouped together to form a threshold set, for which the number of well-predicted landslide days and false alarms was determined. It has been determined that 16 and 84 landslide days have fallen below the lower bound and above the upper bound, respectively. In addition, this method has been proven effective in the distinction between days with and without landslides, since it has led to a 23 % reduction in the number of false alarms per well-predicted landslide day when compared to a previously established threshold line for Tegucigalpa.