Mapping of Landslide Susceptibility in the Central Region of Tajikistan

The project’s objectives were: 1) To analyze the landslide susceptibility in the selected area of Tajikistan by logistic regression approach; 2) To update existing settlements using ALOS satellite images; 3) To assess the level of landslide exposure for villages.

Tajikistan is located in the southeast of Central Asia, where most of the population (93%) live in mountainous areas vulnerable to natural hazards including landslides and mudflows, that caused more than 70% of all deaths recorded from 1997 to 2010. The study area considered in this study covers 11.4% of the total territory of Tajikistan, in which around 58% of the entire population lives.

A statistical approach was used in this study, which uses weighted logistic regressions to calculate the probability of landslide occurrences. The ultimate goal of the logistic regression is to find the best-fitting model that describes the relationship between the dependent variable (landslide inventory) and the independent variables (lithology, land use, slope, elevation, aspect, etc.). Using the location of landslides and topographic factors, this method was used to calculate the weights and the difference between the positive and negative weights for each causative factor. The difference was later used for rating the causative factors. This difference is positive for high-influencing and negative for low-influencing causative factors. Using this methodology, a landslide susceptibility map was produced, which was later categorized into different susceptibility classes to prepare a landslide susceptibility map. The existing settlements layer was updated using the pan-sharpened product obtained by combining the ALOS/PRISM and ALOS/AVNIR-2 data. A proximity analysis was performed to assess the level of landslide exposure to neighboring villages to prepare a landslide exposure.