Spatial modelling of gully erosion using evidential belief function, logistic regression, and a new ensemble of evidential belief function–logistic regression algorithm

Abstract

This study aims to assess gully erosion susceptibility and delineate gully erosion-prone areas in Toroud Watershed, Semnan Province, Iran. Two different methods, namely, logistic regression (LR) and evidential belief function (EBF), were evaluated, and a new ensemble method was proposed using the combination of both methods. We initially created a gully erosion inventory map using different resources, including early reports, Google Earth images, and Global Positioning System-aided field surveys.We subsequently split this information randomly and selected 70% (90) of the gullies for calibration and 30% (38) for validation. The method was constructed using a combination of morphometric and thematic predictors that include 16 conditioning parameters. We also assessed the following: (a) potential multicollinearity issues using tolerance and variance inflation factor indices and (b) covariate effects using LR coefficients and EBF class weights. Results show that land use/land cover, lithology, and distance to roads dominate the method with the greatest effect on gully occurrences.We produced three susceptibility maps and evaluated their predictive power through area under the curve (AUC) and seed cell area index analyses. AUC results revealed that the ensemble method presented a considerably higher performance (AUC = 0.909) than did the individual LR (0.802) and EBF (0.821) methods. Similarly, seed cell area index displayed a constant decrease from the ensemble to single methods. The resulted gully erosion-susceptibility map could be used by decision makers and local managers for soil conservation, and for minimising damages in development activities including construction of infrastructures such as roads and the route of gas and electricity transmission lines.

Evaluation of Recent Advanced Soft Computing Techniques for Gully Erosion Susceptibility Mapping: A Comparative Study

“…Given the destructive effects of gully erosion (GE), solutions for managing this phenomenon to achieve sustainable development are essential [2]. Gully erosion-susceptibility mapping (GESM) is one basic method [3] to understand the mechanisms behind gully erosion. To predict the patterns of GE, a gully-erosion inventory and methods to identify and measure pertinent gully-erosion conditioning factors (GECFs) are needed [4].…”

“…A geographic information system (GIS), remote sensing (RS), and statistical data analyses are indispensable tools for examination of multidimensional outcomes like GE. Several factors are potential influences [3]. A variety of GIS-based approaches for GESM have been proposed and they can be classified into three types: multicriteria decision-making (MCDM), statistical modeling, and machine learning (ML) models.…”

“…Statistical models provide a general advantage of working with diverse types of independent variables, like continues, binary, and categorical data [5]. The most successful models may be: information value (IV) [5], conditional probability [11], frequency ratio (FR) [12], evidential belief function [3], index of entropy (IoE) [13], certainty factor [14], weights of evidence (WOE) [15], and logistic regression [16]. The performance of statistical models is low, however.…”

Gully Head-Cut Distribution Modeling Using Machine Learning Methods—A Case Study of N.W. Iran

“…In GESM, testing for collinearity among the effective factors in gullying is very important, because the collinearity reduces the accuracy of the GESM [86][87][88][89]. The variance inflation factor (VIF) and Tolerance (TOL) are very commonly used indicators for checking multicollinearity among parameters [90,91].…”

“…The variance inflation factor (VIF) and Tolerance (TOL) are very commonly used indicators for checking multicollinearity among parameters [90,91]. TOL values less than 0.1 or 0.2 and VIF values greater than 5 or 10 indicate collinearity between the parameters [17,19,86,89,92]. In the present study, the multicollinearity test of gully erosion conditioning factors (GECFs) was done using Equations (10) and (11) in SPSS software:…”