Evaluating the sensitivity of the landslide event using the support vector machine algorithm

Introduction:

landslide risk assessment provides a systematic framework for evaluating the likelihood and potential consequences of landslides in a given area. It involves the identification and analysis of key factors contributing to landslide occurrence, such as slope characteristics, geological formations, land use patterns, rainfall patterns, and human activities. By integrating these factors into a comprehensive risk assessment methodology, stakeholders can better understand the vulnerability of areas and populations at risk and develop appropriate strategies and measures to mitigate and manage landslide hazards. Advancements in geospatial technologies, such as geographic information systems (GIS), remote sensing, and machine learning algorithms, have significantly enhanced the accuracy and efficiency of landslide risk assessment. These tools enable the integration and analysis of diverse data sources, including topographic data, satellite imagery and historical landslide records, to create detailed landslide susceptibility and hazard maps. These maps provide valuable information for prioritizing risk-prone areas, implementing land-use regulations, designing engineering structures and formulating early warning systems. This study aims to contribute to the field of landslide risk assessment by evaluating the key factors influencing landslide occurrence and developing a comprehensive methodology for assessing landslide risks in the chesb basin of zanjan province. The research findings will provide valuable insights for land managers, policymakers, and stakeholders involved in disaster risk reduction, land-use planning, and infrastructure development. By understanding and effectively managing landslide risks, communities can build resilience, protect lives and property, and ensure sustainable development in landslide-prone regions.



Materials and Methods:

This research was conducted in the catchment area of Chasb, which is located in the city of Eejrud, Zanjan province, between geographical longitudes 36.13 to 36.27 degrees and geographical latitudes 48.1 to 48.41 degrees. To begin, a comprehensive review of literature was conducted to gather existing knowledge and identify influential factors related to landslides. Additionally, field visits were conducted to gather on-site information and observations. Based on the collected information, various data layers were prepared using a geographic information system (GIS). These layers included slope, slope direction, elevation classes, geology, distance from the drainage network to the river, distance from roads, distance from faults, topographic indices (such as stream power index (SPI), topographic wetness index (TWI), and slope length factor (LS)), geomorphological indices (such as topographic position index (TPI), topographic roughness index, and curvature index), land use, normalized difference vegetation index (NDVI), and precipitation.

After data preparation, a total of 81 landslide occurrences were identified in the study area through field surveys and previous studies. For landslide risk modeling, 70% of the landslide points were used to train the support vector machine (SVM) model, while the remaining 30% were used for model validation. Using the SVM model, a sensitivity map for landslides occurrence was generated. The model utilized the prepared data layers to identify areas with varying levels of sensitivity to landslides, ranging from very low to very high.



Results and Discussion:

The results of the study revealed important findings related to landslides and their risk assessment in the Chasb Basin of Zanjan Province. The sensitivity map generated by the support vector machine (SVM) model provided valuable insights into the areas prone to landslides. According to the sensitivity map, approximately 30.63% of the basin area fell into the very low sensitivity class, indicating a lower likelihood of landslides in these areas. The low sensitivity class covered 17.82% of the area, suggesting a relatively lower risk of landslides. The moderate sensitivity class covered 15.43% of the area, indicating a medium level of landslide risk. The high sensitivity class encompassed 17.33% of the area, reflecting a considerable risk of landslides. Lastly, the very high sensitivity class covered 18.5% of the area, representing the highest risk of landslides. The efficiency of the Support Vector Machine (SVM) model was also evaluated using the Receiver Operating Characteristic (ROC) curve, and the area under the ROC curve (AUC) in the validation phase was found to be 0.874. This AUC value indicates a very good capability of the model in classifying and identifying landslide-prone areas in the Chasb catchment area.

These findings were consistent with previous research on landslides and demonstrated the effectiveness of the SVM model in identifying landslide-prone areas. The sensitivity map derived from the model can be instrumental in land-use planning, disaster risk management, and decision-making processes aimed at minimizing the impact of landslides.

Conclusion:

Landslides occur when masses of soil, rocks, and debris rapidly move downhill under the influence of gravity. it can be triggered by various factors, including heavy rainfall, seismic activities, slope instability, geological conditions, and human activities. Landslides can result in devastating consequences such as loss of life, property damage, disruption of transportation networks, and ecological disturbances. To address these challenges, landslide risk assessment provides a systematic framework for evaluating the likelihood and potential consequences of landslides in a given area. It involves the identification and analysis of key factors contributing to landslide occurrence, such as slope characteristics, geological formations, land use patterns, rainfall patterns, and human activities. By integrating these factors into a comprehensive risk assessment methodology, stakeholders can better understand the vulnerability of areas and populations at risk and develop appropriate strategies and measures to mitigate and manage landslide hazards.

The research identified the most influential factors in landslides occurrence and developed a sensitivity map using a support vector machine (SVM) model. The findings highlighted the areas with varying levels of sensitivity to landslides, ranging from low to very high. These results can inform land-use planning strategies, allowing policymakers and stakeholders to better manage and mitigate the risk of landslides in the study area. The outcomes of this study contribute to the broader knowledge on landslides and provide valuable insights for disaster risk reduction efforts in the Chasb Basin. The obtained sensitivity map can guide land managers, decision-makers, and authorities in implementing appropriate mitigation measures and ensuring the safety of the population and infrastructure in the area.