Flood susceptibility assessment using extreme gradient boosting (EGB), Iran

Mirzaei, S; Vafakhah, M; Pradhan, B; Alavi, SJ

Flood susceptibility assessment using extreme gradient boosting (EGB), Iran

Mirzaei, S Vafakhah, M Pradhan, B

Alavi, SJ

Permalink

Publisher:: Springer
Publication Type:: Journal Article
Citation:: Earth Science Informatics, 2021, 14, (1), pp. 51-67
Issue Date:: 2021

Open Access

Copyright Clearance Process

Recently Added
In Progress
Open Access

This item is open access.

Adobe PDF

Download Accepted ManuscriptAdobe PDF (3.78 MB)

View on publisher's site

View statistics

Full metadata record

Field	Value	Language
dc.contributor.author	Mirzaei, S
dc.contributor.author	Vafakhah, M
dc.contributor.author	Pradhan, B https://orcid.org/0000-0001-9863-2054
dc.contributor.author	Alavi, SJ
dc.date.accessioned	2021-05-05T00:05:36Z
dc.date.available	2021-05-05T00:05:36Z
dc.date.issued	2021
dc.identifier.citation	Earth Science Informatics, 2021, 14, (1), pp. 51-67
dc.identifier.issn	1865-0481
dc.identifier.issn	1865-0481
dc.identifier.uri	http://hdl.handle.net/10453/148625
dc.description.abstract	Flood occurs as a result of high intensity and long-term rainfalls accompanied by snowmelt which flow out of the main river channel onto the flood prone areas and damage the buildings, roads, and facilities and cause life losses. This study aims to implement extreme gradient boosting (EGB) method for the first time in flood susceptibility modelling and compare its performance with three advanced benchmark models including Frequency Ratio (FR), Random Forest (RF), and Generalized Additive Model (GAM). Flood susceptibility map is an efficient tool to make decision for flood control. To do this, the altitude, slope degree, profile curvature, topographic wetness index (TWI), distance from rivers, normalized difference vegetation index, plan curvature, rainfall, land use, stream power index, and lithology were fed to the models. To run the models, 243 flood locations were detected by field surveys and national reports. The same number of locations were randomly created in the study regions and considered as non-flood locations. The flood and non-flood locations were split in 70% ratio for the training dataset and 30% ratio for the testing dataset. Both flood and non-flood locations were fed into the models and output flood susceptibility maps were produced. In order to evaluate the performance of the algorithms, receiver operating characteristics (ROC) curve was implemented. The results of the current research show that the RF model and EGB have the best performances with the area under ROC curve (AUC) of 0.985, and 0.980, followed by the GAM and FR algorithms with AUC values of 0.97, and 0.953, respectively. The results of variable importance by the RF model show that distance from rivers has an important influence on flood susceptibility mapping (FSM), followed by profile curvature, slope, TWI, and altitude. Considering the high performances of the RF and EGB models in flood susceptibility modelling, application of these models is recommended for such studies.
dc.language	English
dc.publisher	Springer
dc.relation.ispartof	Earth Science Informatics
dc.relation.isbasedon	10.1007/s12145-020-00530-0
dc.rights	This is a post-peer-review, pre-copyedit version of an article published in [Earth Science Informatics, 2021, 14, (1), pp. 51-67]. The final authenticated version is available online at: https://link.springer.com/article/10.1007/s12145-020-00530-0]”
dc.rights	info:eu-repo/semantics/openAccess
dc.title	Flood susceptibility assessment using extreme gradient boosting (EGB), Iran
dc.type	Journal Article
utslib.citation.volume	14
pubs.organisational-group	/University of Technology Sydney
pubs.organisational-group	/University of Technology Sydney/Faculty of Engineering and Information Technology
pubs.organisational-group	/University of Technology Sydney/Strength - CAMGIS - Centre for Advanced Modelling and Geospatial lnformation Systems
pubs.organisational-group	/University of Technology Sydney/Faculty of Engineering and Information Technology/School of Information, Systems and Modelling
utslib.copyright.status	open_access	*
pubs.consider-herdc	false
dc.date.updated	2021-05-05T00:05:33Z
pubs.issue	1
pubs.publication-status	Published
pubs.volume	14
utslib.citation.issue	1

Abstract:

Flood occurs as a result of high intensity and long-term rainfalls accompanied by snowmelt which flow out of the main river channel onto the flood prone areas and damage the buildings, roads, and facilities and cause life losses. This study aims to implement extreme gradient boosting (EGB) method for the first time in flood susceptibility modelling and compare its performance with three advanced benchmark models including Frequency Ratio (FR), Random Forest (RF), and Generalized Additive Model (GAM). Flood susceptibility map is an efficient tool to make decision for flood control. To do this, the altitude, slope degree, profile curvature, topographic wetness index (TWI), distance from rivers, normalized difference vegetation index, plan curvature, rainfall, land use, stream power index, and lithology were fed to the models. To run the models, 243 flood locations were detected by field surveys and national reports. The same number of locations were randomly created in the study regions and considered as non-flood locations. The flood and non-flood locations were split in 70% ratio for the training dataset and 30% ratio for the testing dataset. Both flood and non-flood locations were fed into the models and output flood susceptibility maps were produced. In order to evaluate the performance of the algorithms, receiver operating characteristics (ROC) curve was implemented. The results of the current research show that the RF model and EGB have the best performances with the area under ROC curve (AUC) of 0.985, and 0.980, followed by the GAM and FR algorithms with AUC values of 0.97, and 0.953, respectively. The results of variable importance by the RF model show that distance from rivers has an important influence on flood susceptibility mapping (FSM), followed by profile curvature, slope, TWI, and altitude. Considering the high performances of the RF and EGB models in flood susceptibility modelling, application of these models is recommended for such studies.

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10453/148625