Landslide susceptibility mapping of the Sera River Basin using logistic regression model

Abstract

Of the natural hazards in Turkey, landslides are the second most devastating in terms of socioeconomic losses, with the majority of landslides occurring in the Eastern Black Sea Region. The aim of this study is to use a statistical approach to carry out a landslide susceptibility assessment in one area at great risk from landslides: the Sera River Basin located in the Eastern Black Sea Region. This paper applies a multivariate statistical approach in the form of a logistics regression model to explore the probability distribution of future landslides in the region. The model attempts to find the best fitting function to describe the relationship between the dependent variable, here the presence or absence of landslides in a region and a set of independent parameters contributing to the occurrence of landslides. The dependent variable (0 for the absence of landslides and 1 for the presence of landslides) was generated using landslide data retrieved from an existing database and expert opinion. The database has information on a few landslides in the region, but is not extensive or complete, and thus unlike those normally used for research. Slope, angle, relief, the natural drainage network (including distance to rivers and the watershed index) and lithology were used as independent parameters in this study. The effect of each parameter was assessed using the corresponding coefficient in the logistic regression function. The results showed that the natural drainage network plays a significant role in determining landslide occurrence and distribution. Landslide susceptibility was evaluated using a predicted map of probability. Zones with high and medium susceptibility to landslides make up 38.8 % of the study area and are located mostly south of the Sera River Basin and along streams.

Dynamics-based regression models for the South Atlantic Convergence Zone

“…For this reason, binary logistic regression has been widely explored more traditionally in health (Arshad and Hide 1992;Brough et al 2015;Lomholt et al 2016), but also in environmental sciences (e.g. : Flantua et al 2007;Guanche et al 2014;Raja et al 2016). Consider we take one of the meridional types of SACZ (AB for instance) and let the SCAZ configuration series be vectors of n days, Y = y 1 , y 2 , … , y n , whose values are coded as y i = 1 for a day with the presence of SACZ and y i = 0 for its absence, where i = 1, 2, … n, and x = x 1 , x 2 , … , x p a matrix of p independent variables of length n (i.e.…”

Exploring the Impact of Analysis Scale on Landslide Susceptibility Modeling: Empirical Assessment in Northern Peloponnese, Greece

“…As it was mentioned above, the LR model allows the integration of both continuous and categorical independent variables. However, combining data with different measuring scales can lead to problems in the interpretation of final results [47]. The common method for resolving this issue is to normalize them.…”

The performance of using an autoencoder for prediction and susceptibility assessment of landslides: A case study on landslides triggered by the 2018 Hokkaido Eastern Iburi earthquake in Japan

“…The sampling process is the key step in constructing landslide (events) and non-landslide points (non-events) for the database used in landslide susceptibility mapping. Several sampling strategies, such as extracting from seed cell (or gridded) points around a polygon of the landslide area (Meusburger and Alewell 2009;Van Den Eeckhaut et al 2010) and increasing the number of nonevents in the non-landslide area (King and Zeng 2001;Raja et al 2017), have been proposed to improve model's performance, predictive capability, and reduction of statistical errors. According to King and Zeng (2001), the non-event sample size must not be large but should be two to five times greater than the events because of the disproportionate cost and effort in acquiring data on many variables, and observations that are not related to the target phenomenon (Heckmann et al 2014).…”