Applying different resampling strategies in machine learning models to predict head-cut gully erosion susceptibility

Fengjie Wang,Mehebub Sahana,Bahareh Pahlevanzadeh,Subodh Chandra Pal,Pravat Kumar Shit,Md Jalil Piran,Saeid Janizadeh,Shahab S Band,Amir Mosavi

doi:10.1016/j.aej.2021.04.026

Abstract

Gully erosion is one of the advanced forms of water erosion. Identifying the effective factors and gully erosion predicting is one of the important tools to control and manage such phenomenon. The main purpose of this study is to evaluate the effect of four different resampling algorithms including cross-validation (5-fold and 10-fold) and bootstrapping (Bootstrap and Optimism bootstrap) on boosted regression tree (BRT), support vector machine (SVM), and random forest (RF) models in spatial modeling and evaluation of head-cut gully erosion in Konduran watershed. For this purpose, based on an extensive field survey, the points of the head-cut of the gully erosion were identified first, and a map of the distribution of head-cut gully erosion in the study area was prepared. Then 18 variable identify and prepare as factors affecting the occurrence of head-cut gully erosion. To assess the efficiency of the models, receiver operating characteristics (ROC) and area under the curve (AUC) were used. <Through the assessment result we indicate that…>The results of the assessment indicated that the use of resampling algorithms increases the efficiency of the models. The integrated optimism-bootstrap-BRT, optimism-bootstrap-SVM, and Optimism-Bootstrap-RF models with AUC 0.85, 0.823 and 0.89 respectively, outperformed the cross-validation 5fold (BRT, SVM, RF), Cross-validation 10fold (BRT, SVM, RF) and Bootstrap (BRT, SVM, RF) integrated algorithms.

Highlights

Gullies are part of the most important processes that result in land degradation
The results indicate that altitude, distance from the river, distance from the road, land use, normalized Difference vegetation index (NDVI), and percentage of clay are of maximum importance, while the rest of the other variables, such as Topographic wetness index (TWI), stream power index (SPI), Topographic position index (TPI), aspect, lithology, and slope are of less importance (Fig. 12)
Variable importance of explanatory factors showed that altitude, distance from the road, distance from the river, percentage of clay, NDVI and land use, had the most significant impact on the occurrence of gully head-cut erosion in the Konduran watershed

Summary

Introduction

Gullies are part of the most important processes that result in land degradation. Gullies cause the loss of 10% to 90% of soil erosion [1]. Gullies form the most fragile land systems and deteriorate the physio-chemical properties of the soils. They are hot spots of soil erosion and an enormous amount of soil is eroded from these lands and transported to low land areas and increasing the risk of sedimentation and flooding in a catchment region [2,3]. Erosion is a frequent event in the arid and semi-arid climatic region and in a dry-wet climatic situation, leads to a large amount of sediment yield [4]. A large rate of soil erosion through head-cut gullies in agriculture is unsustainable and requires a way out in the form of a management strategy. Studying gully erosion and predicting head-cut gully erosion is difficult in complex environments [2,7]

Objectives

Methods

Results

Discussion

Conclusion