Similarity Analysis on Spatial Point Patterns of Earthquakes in Maluku and Sulawesi Using Siamese Neural Network Discriminant Model

Sarwono, Jessica Zerlina (2024) Similarity Analysis on Spatial Point Patterns of Earthquakes in Maluku and Sulawesi Using Siamese Neural Network Discriminant Model. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5003201071-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 July 2026. Download (12MB) | Request a copy

Abstract

Located in the triple junction where three tectonic plates meet, Maluku-Sulawesi is an earthquake-prone area, with small earthquakes occurring daily. Previous works using point processes have successfully identified that geological factors such as volcanoes, faults, and nearest subduction zones can affect earthquake occurrences in the area. Aside from geological factors, earthquake analyses can be done through their periodical occurrence, as earthquakes occur periodically based on the annual hydrological, atmospheric, thermal, and tidal loadings. As earthquake occurrences can be analyzed using spatial point patterns, similarity analysis of yearly earthquake point patterns can potentially help understand earthquake occurrence in the area. One of the latest advances in spatial point pattern similarity analysis using the Siamese Neural Network yielded better results than intensity and K-function in identifying similarities between point patterns of 130 tree species. Seeing great potential in the Siamese Neural Network through it, this research would use the same architecture to identify similarities between the Maluku and Sulawesi earthquakes’ spatial point patterns between 1993 and 2022. The data for this research is collected from the United States Geological Survey yearly. Results prove One-Shot learning tasks can differentiate and recognize point pattern images quickly. Despite the Siamese Neural Network discriminant analysis showing unclear periodical groupings of yearly earthquake occurrences, consecutive years or years with 3, 6, or 9-year gap between suggested similar patterns. The final dendrogram shows two large clusters, with the Orange-Green Cluster having points that are relatively correlated with lower variance and the Red Cluster having points relatively less correlated with more of an unpredictable pattern.
==================================================================================================================================
Terletak pada triple junction di mana tiga lempengan tektonik bertemu, Maluku dan Sulawesi merupakan daerah rawan gempa dengan adanya gempa-gempa kecil yang terjadi setiap hari. Penelitian terdahulu menggunakan point process berhasil mengidentifikasi bahwa faktor-faktor geologis seperti gunung berapi, sesar, dan daerah subduksi terdekat merupakan pemicu terjadinya gempa. Selain secara geologis, analisis terkait gempa dapat dilihat secara periodik karena kejadiannya disebabkan oleh berbagai faktor hidrologis, atmosperik, termal, dan pasang-surutnya air yang terjadi secara tahunan. Mengingat analisis persebaran gempa bumi dapat dilakukan dengan menggunakan spatial point pattern, analisis kemiripan dari spatial point pattern persebaran kejadian gempa bumi dari satu tahun dengan tahun lain dapat dilakukan. Metode Siamese Neural Network telah terbukti menghasilkan akurasi yang lebih baik dibandingkan analisis intensitas dan K-function untuk mengidentifikasi kemiripan point pattern dari 130 spesies pohon pada waktu yang berbeda. Melihat potensi Siamese Neural Network, penelitian ini akan memanfaatkan model serupa untuk melakukan analisis kemiripan spatial point patterns gempa bumi di Maluku dan Sulawesi dari tahun 1993 hingga 2022. Adapun data pada penelitian ini dikumpulkan secara tahunan dari United States Geological Survey. Analisis dengan model diskriminan Siamese Neural Network menggunakan One-Shot learning telah berhasil mengidentifikasi dan membedakan titik-titik persebaran gempa bumi dalam tahun-tahun yang berbeda. Meskipun hasil belum dapat menunjukkan pengelompokan yang jelas secara periodik untuk kejadiannya, model telah dapat mengidentifikasi bahwa tahun yang berurutan atau memiliki perbedaan 3, 6, dan 9 tahun cenderung memiliki pola persebaran yang mirip. Dendrogram akhir menunjukkan adanya dua klaster besar, dengan Klaster Oranye-Hijau menunjukkan titik-titik yang relatif berkorelasi dengan varians yang lebih rendah dan Klaster Merah dengan titik-titik yang relatif kurang berkorelasi dengan pola persebaran yang acak.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Gempa Bumi, Maluku, Siamese Neural Network, Spatial Point Patterns, Sulawesi; Earthquakes, Maluku, Siamese Neural Network, Spatial Point Patterns, Sulawesi
Subjects:	G Geography. Anthropology. Recreation > G Geography (General) > G70.217 Geospatial data Q Science > QA Mathematics > QA336 Artificial Intelligence Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) T Technology > T Technology (General) > T57.5 Data Processing
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User:	Jessica Zerlina Sarwono
Date Deposited:	13 Feb 2024 07:01
Last Modified:	13 Feb 2024 07:01
URI:	http://repository.its.ac.id/id/eprint/107031

Actions (login required)

View Item