Volume 14, Issue 27 (8-2023)                   jwmr 2023, 14(27): 123-134 | Back to browse issues page


XML Persian Abstract Print


Download citation:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:

Nazariani N, Fallah A. (2023). Landslide Risk Modeling using Data Mining in Hyrcanian Forests. jwmr. 14(27), 123-134. doi:10.61186/jwmr.14.27.123
URL: http://jwmr.sanru.ac.ir/article-1-1212-en.html
Sari University of Agricultural Sciences and Natural Resources
Abstract:   (1138 Views)
Extended Abstract
Introduction and Objective: In recent years, the increases in population and expansion of settlements in hazardous areas have greatly increased the impact of natural disasters in industrialized and developing countries. Landslide risk zoning helps to identify strategic points and geographically prone critical areas. Therefore, measures for rapid, safe mitigation and strategic planning for the future are important. In fact, landslide risk assessment may be a suitable and cost-effective help for land use planning, so in this regard, the aim of the current research is to model the risk of landslides using non-parametric data mining in watersheds of Hyrcanian forests.
Material and Methods: For this purpose, the map of landslide risk according to the Mora and vahrson method with the effect of factors affecting the occurrence of landslides including topographical factors, hydrological and climatic factors, geological factors, land cover factors, human factors, hydrographic network from digital elevation model, geological map It was prepared using the map of the Geological Organization of the country, and to obtain the elevation index and the relative height, first, using the curve of height levels taken from the 1:25000 topographic maps of the region, the map of the elevation classes was prepared. After that, the study area was divided into one-square-kilometer grids, and maps with the lowest and highest elevations in one-square-kilometer grids were obtained; in the last step, by subtracting these two maps, a map was obtained whose information shows the value of the postal index and relative height. Monthly rainfall was also used to obtain the soil moisture index. Finally, algorithms of three random forest models, an artificial neural network, and a decision tree algorithm were used to model the risk of landslides in the STATISTICA 12.0 software environment.
Results: According to the results, the highest distribution of landslide areas belongs to the low-risk class (76%). TP, SL, SR, and TS variables were considered the essential factors of landslide occurrence based on their importance. The results of validation with three algorithms of RF, CART, and ANN showed; According to the coefficient of explanation obtained in modeling the risk of landslides, the artificial neural network model with (R2=0.99) is more accurate than other methods.
Conclusion: The results of the present study showed that data mining methods have a high capability in predicting the risk of landslides. Therefore, the use of the mentioned methods can be considered in reducing the risks associated with landslides and planning for land use.
Full-Text [PDF 2165 kb]   (288 Downloads)    
Type of Study: Research | Subject: ساير موضوعات وابسته به مديريت حوزه آبخيز
Received: 2022/10/24 | Revised: 2023/08/7 | Accepted: 2022/11/29 | Published: 2023/08/7

Add your comments about this article : Your username or Email:
CAPTCHA

Send email to the article author


Rights and permissions
Creative Commons License This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

© 2024 CC BY-NC 4.0 | Journal of Watershed Management Research

Designed & Developed by : Yektaweb