Optimasi Parameter Ground Motion Prediction Equation Berbasis NGA-West2 Dengan Pendekatan Mixed-Effects Untuk Gempa Kerak Dangkal Di Pulau Jawa Menggunakan Algoritma Teaching-Learning-Based Optimization

Chofifah, Chofifah (2026) Optimasi Parameter Ground Motion Prediction Equation Berbasis NGA-West2 Dengan Pendekatan Mixed-Effects Untuk Gempa Kerak Dangkal Di Pulau Jawa Menggunakan Algoritma Teaching-Learning-Based Optimization. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5001221067-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Pulau Jawa merupakan pulau berpenduduk terpadat di Indonesia yang dilalui banyak sesar aktif penghasil gempa kerak dangkal. Meskipun umumnya berkekuatan menengah, gempa kerak dangkal bersifat sangat merusak karena sumbernya dekat permukaan dan kerap tepat di bawah permukiman padat, sehingga menimbulkan guncangan kuat serta kerusakan dan korban jiwa yang besar. Mitigasi terhadap ancaman ini bergantung pada ketepatan prediksi percepatan tanah puncak atau Peak Ground Accelaration (PGA) melalui Ground Motion Prediction Equation (GMPE), namun model GMPE yang tersedia dikembangkan dari basis data wilayah lain sehingga kurang representatif dan menghasilkan estimasi bahaya yang kurang akurat untuk Pulau Jawa. Penelitian ini mengkalibrasi ulang parameter model GMPE Boore et al., (2014) (BSSA14), Campbell&Bozorgnia (2014) (CB14), Chiou&Youngs (2014) (CY14), dan Bindi et al., (2011) (ITA10) menggunakan pendekatan mixed-effects yang dioptimasi dengan algoritma metaheuristik Teaching-Learning-Based Optimization (TLBO), Particle Swarm Optimization (PSO), Whale Optimization Algorithm (WOA), dan Grey Wolf Optimizer (GWO). Hasil penelitian menunjukkan algoritma TLBO berkinerja paling unggul, ditandai konvergensi tercepat, konsistensi tertinggi antar percobaan, dan keunggulan yang signifikan secara statistik terhadap algoritma pembanding lainnya. Kalibrasi ulang parameter terbukti menurunkan kesalahan prediksi secara substansial dibanding penerapan langsung koefisien publikasi asli, sehingga menegaskan bahwa model global tidak dapat diterapkan pada Pulau Jawa tanpa penyesuaian. Model CY14 hasil kalibrasi ulang teridentifikasi sebagai model paling representatif terhadap karakteristik ground motion Pulau Jawa, unggul pada seluruh metrik evaluasi sekaligus memiliki ketidakpastian total terkecil. Analisis residual menunjukkan model tidak mengandung bias terhadap jarak, dengan keterbatasan akurasi pada magnitudo besar akibat ketersediaan data. Penelitian ini menegaskan pendekatan metaheuristik, khususnya TLBO, efektif untuk kalibrasi GMPE terhadap data kegempaan lokal.
====================================================================================================================================
Java is the most densely populated island in Indonesia and is crossed by many active faults that generate shallow crustal earthquakes. Although generally of moderate magnitude, shallow crustal earthquakes are highly destructive because their epicenters are close to the surface and often directly beneath densely populated areas, resulting in strong shaking, extensive damage, and significant loss of life. Mitigation of this threat depends on the accuracy of peak ground acceleration (PGA) predictions using the Ground Motion Prediction Equation (GMPE), however, the available GMPE models were developed using data from other regions and are therefore not representative of Java, resulting in less accurate hazard estimates for the island. This study recalibrates the model parameters of the Boore et al., (2014) (BSSA14), Campbell&Bozorgnia (2014) (CB14), Chiou&Youngs (2014) (CY14), and Bindi et al., (2011) (ITA10) using a mixed-effects approach optimized with the metaheuristic algorithms Teaching-Learning-Based Optimization (TLBO), Particle Swarm Optimization (PSO), Whale Optimization Algorithm (WOA), and Grey Wolf Optimizer (GWO). The results show that the TLBO algorithm performed best, characterized by the fastest convergence, the highest consistency across experiments, and a statistically significant advantage over the other comparison algorithms. Recalibration of the parameters was shown to substantially reduce prediction errors compared to the direct application of the original published coefficients, thereby confirming that global models cannot be applied to Java without adjustment. The recalibrated CY14 model was identified as the most representative of Java’s ground motion characteristics, outperforming all other models on all fit evaluation metric while exhibiting the smallest total uncertainty. Residual analysis showed that the model contains no distance bias, with accuracy limitations at high magnitudes due to data availability. This study confirms that metaheuristic approaches, particularly TLBO, are effective for calibrating GMPE against local seismic data.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	GMPE, Gempa Kerak Dangkal, Mixed-Effects, TLBO, Metaheuristik.,GMPE, Shallow Crustal Earthquake, Mixed-Effects, TLBO, Metaheuristic.
Subjects:	G Geography. Anthropology. Recreation > GB Physical geography Q Science > QA Mathematics > QA9.58 Algorithms Q Science > QC Physics Q Science > QE Geology > QE538.8 Earthquakes. Seismology Q Science > QE Geology > QE539.2.S4 Seismic models
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Physics > 45201-(S1) Undergraduate Thesis
Depositing User:	Chofifah Chofifah
Date Deposited:	04 Aug 2026 02:12
Last Modified:	04 Aug 2026 02:12
URI:	http://repository.its.ac.id/id/eprint/142661

Actions (login required)

View Item