Pengembangan Platform Otomasi End‐To‐End Untuk Penentuan Mekanisme Fokus Gempa Bumi Berbasis De

Nashrullah, M. Abdul Aziz (2026) Pengembangan Platform Otomasi End‐To‐End Untuk Penentuan Mekanisme Fokus Gempa Bumi Berbasis De. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5017221049-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (24MB) | Request a copy

Abstract

Zona transisi Sunda–Banda di Indonesia bagian timur merupakan salah satu wilayah tektonik paling kompleks di dunia, yang ditandai oleh interaksi antara subduksi oseanik dan tumbukan busur-benua dengan konvergensi oblique serta tingkat seismisitas yang tinggi. Kondisi ini menuntut metode analisis mekanisme fokus yang efisien, konsisten, serta mampu memproses data berskala besar, sementara pendekatan konvensional masih sangat bergantung pada pemilihan fase (picking) secara manual dan alur kerja yang tidak terintegrasi. Penelitian ini mengembangkan platform otomasi end-to-end bernama AutoFocal untuk penentuan mekanisme fokus gempa bumi secara terintegrasi dan reproducible. Pipeline yang dibangun mencakup tahap prapemrosesan waveform, deteksi fase otomatis menggunakan EQTransformer berbasis CNN, LSTM, dan mekanisme attention (Mousavi et al., 2020), estimasi mekanisme fokus menggunakan model berbasis Transformer encoder FocoNet (Song et al., 2025), serta inversi tensor tegangan menggunakan pySATSI (Skoumal et al., 2026). Implementasi dilakukan pada katalog 19.074 kejadian gempa dengan data waveform yang diperoleh melalui fasilitas AUSPASS berdasarkan jaringan seismik YS selama periode 2014–2018. Pipeline AutoFocal mengintegrasikan seluruh tahapan penentuan mekanisme fokusdalam satu alur kerja terpadu tanpa intervensi manual antartahap, yang secara struktural mengatasi keterbatasan utama pendekatan konvensional terutama untuk gempa dengan magnitude <3. Solusi mekanisme fokus yang dihasilkan memetakan variasi tegasan regional zona transisi Sunda–Banda, mencakup dominasi pemendekan kerak (crustal shortening) di segmen tumbukan busur-benua bagian timur (Alor–Timor) serta pola deformasi yang berasosiasi dengan subduksi oseanik kontinu di bagian barat, konsisten dengan studi terdahulu pada wilayah dan periode yang sama (Jiang et al., 2022; Supendi et al., 2022). Kebaruan penelitian ini terletak pada integrasi metode deep learning ke dalam satu pipeline otomasi untuk penentuan mekanisme fokus secara sistematis pada dataset berskala besar di lingkungan tektonik yang kompleks, serta membuktikan bahwa pendekatan end-to-end berbasis deep learning merupakan alternatif yang praktis dan konsisten dibandingkan alur kerja konvensional, serta berpotensi untuk dikembangkan dalam analisis seismisitas skala besar di daerah lainnya.
========================================================================================================================================
The Sunda–Banda transition zone in eastern Indonesia is one of the most complex tectonic regions in the world, characterized by the interaction between oceanic subduction and arc-continent collision with oblique convergence and a high level of seismicity. These conditions demand an efficient and consistent focal mechanism analysis method capable of processing large-scale data, whereas conventional approaches still heavily rely on manual phase picking and non-integrated workflows. This research develops an end-to-end automation platform named AutoFocal for the integrated and reproducible determination of earthquake focal mechanisms. The constructed pipeline includes waveform pre-processing, automatic phase detection using the CNN, LSTM, and attention mechanism-based EQTransformer (Mousavi et al., 2020), focal mechanism estimation using the Transformer encoder-based model FocoNet (Song et al., 2025), and stress tensor inversion using pySATSI (Skoumal et al., 2026). The implementation was carried out on a catalog of 19,074 earthquake events with waveform data obtained through the AUSPASS facility based on the YS seismic network during the 2014–2018 period. The AutoFocal pipeline successfully integrates all stages of focal mechanism determination into a single unified workflow without manual intervention between stages, which structurally overcomes the main limitations of conventional approaches, particularly for earthquakes with a magnitude<3. There sulting focal mechanism solution smap the regional stress variations of the Sunda–Banda transition zone, encompassing the dominance of crustal shortening in the eastern arc-continent collision segment (Alor–Timor) as well as deformation patterns associated with continuous oceanic subduction in the western part, consistent with previous studies in the same region and period (Jiang et al., 2022; Supendi et al., 2022). The novelty of this research lies in the integration of deep learning methods into a complete automation pipeline for the systematic determination of focal mechanisms on large-scale datasets in a complex tectonic environment. It simultan eously proves that an end-to-end deep learning-based approach is a practical and consistent alternative compared to conventional workflows, and holds significant potential to be developed for large-scale seismicity analysis in other regions.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Mekanisme Fokus, Otomasi Seismik, Inversi Tensor Tegangan, Zona Transisi Sunda–Banda, Deep Learning, focal mechanism, seismic automation, Stress Tensor Inversion, Sunda–Banda transition zone
Subjects:	Q Science > QE Geology > QE538.5 Seismic tomography; Seismic waves. Elastic waves
Divisions:	Faculty of Civil, Planning, and Geo Engineering (CIVPLAN) > Geophysics Engineering > 33201-(S1) Undergraduate Thesis
Depositing User:	M. Abdul Aziz Nashrullah
Date Deposited:	22 Jul 2026 09:03
Last Modified:	22 Jul 2026 09:03
URI:	http://repository.its.ac.id/id/eprint/136472

Actions (login required)

View Item