10:40 AM - 11:00 AM
[DM-01] Analysis of disaster risk of slopes along railroad lines using self-organizing maps (SOM)
In Japan, rainfall disasters, which are considered to be caused by severe weather conditions due to global environmental changes, occur frequently. Therefore, accurate risk assessment method for slope disaster is needed.
In this report, the stability of a slope is related to the topographical features of the slope. If such topographic features can be extracted or classified, disaster risk of slopes along railroad lines can be assessed over a wide area.
We used a Self-Organizing Map (SOM), which is one of the competitive neural networks. The topographic features of the slope where the disaster occurred are classified into several clusters by SOM. Slopes with similar features are classified into the same clusters, and the results of SOM calculations are plotted on a two-dimensional map to examine the properties of each cluster.
Data for 57 slopes were used in the calculations. These slopes experienced either rockfall, rock failure, landslide, or debris flow. Data on slope shape, gradient, vegetation, etc. are collected for each slope.
As a result of the analysis, The 57 slopes could be classified into several clusters with similar characteristics. Based on the characteristics of the clusters, the relationship between disaster and topographic features can be discussed.
In the future, we would like to further analyze items related to geology and triggers, and examine non-disaster areas too. Then, we would like to establish a new risk assessment method for disaster slopes.
In this report, the stability of a slope is related to the topographical features of the slope. If such topographic features can be extracted or classified, disaster risk of slopes along railroad lines can be assessed over a wide area.
We used a Self-Organizing Map (SOM), which is one of the competitive neural networks. The topographic features of the slope where the disaster occurred are classified into several clusters by SOM. Slopes with similar features are classified into the same clusters, and the results of SOM calculations are plotted on a two-dimensional map to examine the properties of each cluster.
Data for 57 slopes were used in the calculations. These slopes experienced either rockfall, rock failure, landslide, or debris flow. Data on slope shape, gradient, vegetation, etc. are collected for each slope.
As a result of the analysis, The 57 slopes could be classified into several clusters with similar characteristics. Based on the characteristics of the clusters, the relationship between disaster and topographic features can be discussed.
In the future, we would like to further analyze items related to geology and triggers, and examine non-disaster areas too. Then, we would like to establish a new risk assessment method for disaster slopes.
Abstract password authentication.
The password was sent to the registered participants by e-mail in advance.