Scholarlink

Volcano deformation is a continuous cycle, particularly evident at Sinabung volcano. Accurate deformation information is crucial for assessing trends in the volcano’s surface changes and predicting potential future deformations. The aim of this research is to evaluate, compare the application of the time series clustering method TSC using two approaches and to forecast future volcano deformation. This data processed using a geo-artificial intelligence time-series clustering method TSC, incorporating two model based on value, profile correlation, while the Holt-Winters Exponential Smoothing (HWES) method is utilized for forecasting future deformation. The implementation of both clustering models demonstrates their respective advantages due to their distinct approaches to clustering spatial time series data. The value-based model prioritizes the similarity of numerical values, while the profile correlation model emphasizes the similarity of patterns and trends in the spatial data of mountain deformation. The combination of these methods significantly enhances the analysis of deformation assessment. By applying forecasting data, it is projected that the volcano continuing to experience an inflation trend, with an annual maximum inflation of 13.19 cm in 2024, increasing to 16.93 cm in 2025, 21.69 cm in 2026, and reaching a peak of 26.45 cm in 2027. The deformation trend, as revealed by the forecasting results, is significantly influenced by the forecasting time series method utilized. Given the ongoing inflation trend, it is imperative to enhance monitoring efforts and develop more complex models of Mount Sinabung’s activity.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Clustering and Forecasting Vertical Ground Deformation of Sinabung Volcano Based on Deformation Dataset and Holt-Winters Exponential Smoothing Method

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Muhammad Hanif,
  • Sarun Apichontrakul,
  • Ernieza Suhana Mokhtar

摘要

Volcano deformation is a continuous cycle, particularly evident at Sinabung volcano. Accurate deformation information is crucial for assessing trends in the volcano’s surface changes and predicting potential future deformations. The aim of this research is to evaluate, compare the application of the time series clustering method TSC using two approaches and to forecast future volcano deformation. This data processed using a geo-artificial intelligence time-series clustering method TSC, incorporating two model based on value, profile correlation, while the Holt-Winters Exponential Smoothing (HWES) method is utilized for forecasting future deformation. The implementation of both clustering models demonstrates their respective advantages due to their distinct approaches to clustering spatial time series data. The value-based model prioritizes the similarity of numerical values, while the profile correlation model emphasizes the similarity of patterns and trends in the spatial data of mountain deformation. The combination of these methods significantly enhances the analysis of deformation assessment. By applying forecasting data, it is projected that the volcano continuing to experience an inflation trend, with an annual maximum inflation of 13.19 cm in 2024, increasing to 16.93 cm in 2025, 21.69 cm in 2026, and reaching a peak of 26.45 cm in 2027. The deformation trend, as revealed by the forecasting results, is significantly influenced by the forecasting time series method utilized. Given the ongoing inflation trend, it is imperative to enhance monitoring efforts and develop more complex models of Mount Sinabung’s activity.