Machine learning methods for landslide mapping studies: A comparative study of SVM and RF algorithms in the Oued Aoulai watershed (Morocco)

Effective management of watershed risks and landslides necessitates comprehensive landslide susceptibility mapping. Support vector machine (SVM) and random forest (RF) machine learning models were used to map the landslide susceptibility in Morocco’s Taounate Province. Detailed landslide inventory m...

Full description

Saved in:
Bibliographic Details
Main Authors: Ladel Latifa, Mastere Mohamed, Kader Shuraik, Spalević Velibor, Dudic Branislav
Format: Article
Language:English
Published: De Gruyter 2025-03-01
Series:Open Geosciences
Subjects:
Online Access:https://doi.org/10.1515/geo-2022-0740
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850072490177462272
author Ladel Latifa
Mastere Mohamed
Kader Shuraik
Spalević Velibor
Dudic Branislav
author_facet Ladel Latifa
Mastere Mohamed
Kader Shuraik
Spalević Velibor
Dudic Branislav
author_sort Ladel Latifa
collection DOAJ
description Effective management of watershed risks and landslides necessitates comprehensive landslide susceptibility mapping. Support vector machine (SVM) and random forest (RF) machine learning models were used to map the landslide susceptibility in Morocco’s Taounate Province. Detailed landslide inventory maps were generated based on aerial pictures, field research, and geotechnical survey reports. Factor correlation analysis carefully eliminated redundant factors from the original 14 landslide triggering factors. As a result, 30% of the sites were randomly chosen for testing, whereas 70% of the landslide locations were randomly picked for model training. The RF model achieved an area under the curve (AUC) of 94.7%, categorizing 30.07% of the region as low susceptibility, while the SVM model reached an AUC of 80.65%, indicating high sensitivity in 53.5% of the locations. These results provide crucial information for local authorities, supporting sound catchment planning and development strategies.
format Article
id doaj-art-1d95c91237e44f8d90f883dec47e4e67
institution DOAJ
issn 2391-5447
language English
publishDate 2025-03-01
publisher De Gruyter
record_format Article
series Open Geosciences
spelling doaj-art-1d95c91237e44f8d90f883dec47e4e672025-08-20T02:47:04ZengDe GruyterOpen Geosciences2391-54472025-03-0117111010.1515/geo-2022-0740Machine learning methods for landslide mapping studies: A comparative study of SVM and RF algorithms in the Oued Aoulai watershed (Morocco)Ladel Latifa0Mastere Mohamed1Kader Shuraik2Spalević Velibor3Dudic Branislav4Geophysics and Natural Hazards Laboratory, Department of Geomorphology and Geomatics, Scientific Institute, Mohammed V University in Rabat, Avenue Ibn Batouta, Agdal, PO Box 703, 10106, Rabat-City, MoroccoGeophysics and Natural Hazards Laboratory, Department of Geomorphology and Geomatics, Scientific Institute, Mohammed V University in Rabat, Avenue Ibn Batouta, Agdal, PO Box 703, 10106, Rabat-City, MoroccoSchool of Engineering and Built Environment, Griffith University, Nathan, QLD 4111, AustraliaBiotechnical Faculty, University of Montenegro, Podgorica, 81000, MontenegroFaculty of Management, Comenius University Bratislava, Bratislava, 81499, SlovakiaEffective management of watershed risks and landslides necessitates comprehensive landslide susceptibility mapping. Support vector machine (SVM) and random forest (RF) machine learning models were used to map the landslide susceptibility in Morocco’s Taounate Province. Detailed landslide inventory maps were generated based on aerial pictures, field research, and geotechnical survey reports. Factor correlation analysis carefully eliminated redundant factors from the original 14 landslide triggering factors. As a result, 30% of the sites were randomly chosen for testing, whereas 70% of the landslide locations were randomly picked for model training. The RF model achieved an area under the curve (AUC) of 94.7%, categorizing 30.07% of the region as low susceptibility, while the SVM model reached an AUC of 80.65%, indicating high sensitivity in 53.5% of the locations. These results provide crucial information for local authorities, supporting sound catchment planning and development strategies.https://doi.org/10.1515/geo-2022-0740managementlandslide susceptibilitymachine learning modelsfactor correlation analysisarea under the curvemorocco
spellingShingle Ladel Latifa
Mastere Mohamed
Kader Shuraik
Spalević Velibor
Dudic Branislav
Machine learning methods for landslide mapping studies: A comparative study of SVM and RF algorithms in the Oued Aoulai watershed (Morocco)
Open Geosciences
management
landslide susceptibility
machine learning models
factor correlation analysis
area under the curve
morocco
title Machine learning methods for landslide mapping studies: A comparative study of SVM and RF algorithms in the Oued Aoulai watershed (Morocco)
title_full Machine learning methods for landslide mapping studies: A comparative study of SVM and RF algorithms in the Oued Aoulai watershed (Morocco)
title_fullStr Machine learning methods for landslide mapping studies: A comparative study of SVM and RF algorithms in the Oued Aoulai watershed (Morocco)
title_full_unstemmed Machine learning methods for landslide mapping studies: A comparative study of SVM and RF algorithms in the Oued Aoulai watershed (Morocco)
title_short Machine learning methods for landslide mapping studies: A comparative study of SVM and RF algorithms in the Oued Aoulai watershed (Morocco)
title_sort machine learning methods for landslide mapping studies a comparative study of svm and rf algorithms in the oued aoulai watershed morocco
topic management
landslide susceptibility
machine learning models
factor correlation analysis
area under the curve
morocco
url https://doi.org/10.1515/geo-2022-0740
work_keys_str_mv AT ladellatifa machinelearningmethodsforlandslidemappingstudiesacomparativestudyofsvmandrfalgorithmsintheouedaoulaiwatershedmorocco
AT masteremohamed machinelearningmethodsforlandslidemappingstudiesacomparativestudyofsvmandrfalgorithmsintheouedaoulaiwatershedmorocco
AT kadershuraik machinelearningmethodsforlandslidemappingstudiesacomparativestudyofsvmandrfalgorithmsintheouedaoulaiwatershedmorocco
AT spalevicvelibor machinelearningmethodsforlandslidemappingstudiesacomparativestudyofsvmandrfalgorithmsintheouedaoulaiwatershedmorocco
AT dudicbranislav machinelearningmethodsforlandslidemappingstudiesacomparativestudyofsvmandrfalgorithmsintheouedaoulaiwatershedmorocco