5:15 PM - 7:15 PM
[HTT17-P03] Evaluation of Landslide Activity in the 2024 Noto Peninsula Earthquake Using OpenCV Template Matching
Keywords:landslide, The 2024 Noto Peninsula Earthquake, template matching, OpenCV
The 2024 Noto Peninsula earthquake caused many block-glide-type landslides with low-angle slip surfaces. Because many of these landslides move while retaining their original topography, image processing techniques such as template matching have been used to estimate their movements. We used the OpenCV to estimate landslide movements induced by the Noto Peninsula earthquake for the template matching analysis. OpenCV-Python 4.11.0.86 was used for template matching analysis and Google Colab was used as the analysis platform because of its easy management of the python libraries. Air-bone LiDAR DEMs with 0.5m resolution obtained by Ishikawa Prefecture (2022) and Forestry Agency and the GSI (2024) were used for the topographic data. Two shade maps were created from these two topographic DEMs, and template matching analysis was performed for the two grayscale images. In the matching analysis, the template image size was set to 100 pixels (50 m) and normalized cross-correlation was used as a measure of similarity.
The analysis revealed that many landslide blocks with a size on the order of several hundred meters have moved in the mountainous area stretching from the Machino area in the eastern part of Wajima City to Wakayama Town in Suzu City. Because these landslide movements were not accompanied by significant landscape changes, it is considered difficult to identify the movements from aerial photo-decipherment. However, vectors that clearly appeared to be errors were observed around linear features such as continuous ridge lines and roads. In the Kamishoryoku district in Suzu City, a large landslide block approximately 1.3 km long and 300 m wide was found to have moved approximately 3 m to the south. Some new cracks at the head of the landslide and road uplift at the end of the landslide were found in the field survey.
The analysis revealed that many landslide blocks with a size on the order of several hundred meters have moved in the mountainous area stretching from the Machino area in the eastern part of Wajima City to Wakayama Town in Suzu City. Because these landslide movements were not accompanied by significant landscape changes, it is considered difficult to identify the movements from aerial photo-decipherment. However, vectors that clearly appeared to be errors were observed around linear features such as continuous ridge lines and roads. In the Kamishoryoku district in Suzu City, a large landslide block approximately 1.3 km long and 300 m wide was found to have moved approximately 3 m to the south. Some new cracks at the head of the landslide and road uplift at the end of the landslide were found in the field survey.