A Field Verification Based Statistical Approach toward Landslide Susceptibility Assessment

Abstract

In recent time some detrimental effects of climate change has been observed in NE India: landslides and  scale of floods have been increasing annually, misty mornings during Christmas no longer occurs, wet bulb temperatures (TW) during summers have increased, and so on. However, the landslide  susceptibility literature begins with the premise that future hillslope failures are likely to occur under the  conditions of past and present instability. Moreover, most of the land surface parameters considered for  the landslide studies is not possible to make field identification (except slope, aspect, plan and profile  curvature). However, in the study area field the following observations are made - ∙ Earth slides occur only at sites where regolith thickness is between a range, neither too thick nor  too thin.  ∙ The dip angle of the gneissic bedrock surface has a more precise negative correlation with the  regolith thickness than the slope of the ground surface. ∙ Almost all profile curvatures are convex upward, but despite these, earth slides have occurred. ∙ There are a few convergent plan curvatures, but earth slides are not restricted only to these  sites.The reasons why these observations are contradictory to popular belief are under preparation as a full length article, but in short, the conclusion is that statistical learning combined with field observation for  applying some obvious mechanistic knowledge of the process is the key to successful landslide  susceptibility assessment.