Tofa Mus Rockhim

Magister of Physics, MIPA Faculty, University of Jember, Jl. Kalimantan No. 37 Jember, Indonesia

Stasiun Geofisika Pasuruan, Jl. Sedap Malam 100 Pandaan, Pasuruan, Indonesia

Bowo Eko Cahyono

Physics Department, MIPA Faculty, University of Jember, Jl. Kalimantan No. 37 Jember, Indonesia

Supriyadi

Physics Department, MIPA Faculty, University of Jember, Jl. Kalimantan No. 37 Jember, Indonesia

Pepen Supendi

Meteorology, Climatology and Geophysics Agency, Jl. Angkasa I No.2 Kemayoran, Jakarta, Indonesia

DOI: https://doi.org/10.19184/jid.v27i1.60006

ABSTRACT 

The prediction of Peak Ground Acceleration (PGA) is a crucial parameter in earthquake-resistant structural design, particularly in high-risk regions such as West Java, which are prone to potentially damaging earthquakes. This study aims to evaluate the relationship between epicentral distance, macroseismic intensity based on the Modified Mercalli Intensity (MMI) scale, and horizontal PGA using multiple linear regression and Random Forest approaches. The dataset was obtained from 51 accelerograph stations operated by the Meteorological, Climatological, and Geophysical Agency (BMKG), which recorded a magnitude 4.1 earthquake in Bogor on April 10, 2025. The results indicate that multiple linear regression is capable of capturing the fundamental relationship between distance, PGA, and intensity; however, it has limitations in explaining data variability (R² = 0.3790). In contrast, the Random Forest model demonstrates improved predictive performance, with a 50% reduction in RMSE compared to the linear model The results indicate that machine learning methods can be useful in capturing complex relationships among seismic parameters for PGA prediction. However, the findings are constrained by the limited dataset and the use of a single earthquake event, therefore additional studies with more earthquake records are needed. This study is expected to contribute to the development of locally based seismic hazard maps to support safer structural design.

Keywords: Earthquake, Horizontal PGA, MMI, Multiple Regression, Random Forest, BMKG, West Java.