Evaluating the impact of road construction on landslide susceptibility-
A case study of Mandi district, Himachal Pradesh, India.
Abstract
Landslide susceptibility mapping has proved to be crucial tool for
effective disaster management and planning strategies in mountainous
regions. The present study is perused to investigate the changes in the
landslide susceptibility of the Mandi district of Himachal Pradesh due
to road construction. For this purpose, an inventory of 1723 landslides
was generated from various sources. Out of these, 1199 (70%) landslides
were taken in the training dataset to be used for modelling and
prediction purposes, while 524 (30%) landslides were taken in the
testing dataset to be used for validation purposes. Eleven landslide
causative factors were selected from numerous hydrological, geological
and topographical factors and were analyzed for landslide susceptibility
mapping using three bivariate statistical models, namely; Frequency
Ratio (FR), Certainty Factor (CF) and Shanon Entropy (SE). Two sets of
LSM maps i.e. landslide susceptibility map natural (LSMN) and landslide
susceptibility map road (LSMR), were generated using the above mentioned
bivariate models and were divided into five landslide susceptibility
classes namely; very low, low, medium, high and very high. These maps
were analyzed for accuracy of prediction and validation using receiver
operating characteristic (ROC) curves and area under curve (AUC)
technique which indicated that all three bivariate statistical models
performed satisfactorily with the SE model had the highest prediction
and validation accuracy of 83-86%. Further analysis LSM maps confirmed
that the percentage area in high and very high classes of land-slide
susceptibility increased by 2.67-4.17% due to road construction
activities in the study area.