Click any colored region to explore earthquake risk
✕
Expected Earthquake Size
Mw 6.77.58.08.5+
Dot Size = Magnitude
Zone Overview
Hazard Curve
Return period (years) → Expected earthquake size
Return Level Estimates
TimeframeMwBest Model
5 Largest Recorded Earthquakes
— | — earthquakes shown | USGS ComCat · Indonesia · 1926–2026
Earthquake Catalog
Mainshocks after declustering · Mw ≥ 5.0 · Indonesia 1926–2026
Year Magnitude Depth Latitude Longitude Zone Links
About This Tool
Indonesia Seismic Hazard Explorer · Research Overview

Indonesia Seismic Hazard Explorer

An interactive research tool estimating earthquake return periods across Indonesia using Deep Neural Networks (DNN) for seismic zone classification and Extreme Value Theory (EVT) for statistical modeling — adapting the Ma et al. (2021) framework to one of the world's most active seismic regions.

How to Use This Tool

Choose a Risk Horizon
In the sidebar, select a timeframe (10, 50, 100, or 500 years). Zone colors update to show expected earthquake size within that period.
Click a Zone
Click any colored zone on the map to open a detailed panel with hazard curve, return level table, and the largest historical earthquakes in that zone.
Explore Earthquakes
Click any dot on the map to see earthquake details and links to news coverage, Wikipedia, and the BMKG official database.
Browse the Catalog
Use the Earthquake Catalog tab to browse, sort, and filter all 1,135 mainshocks. Click any row to jump to that earthquake on the map.

Key Findings

Papua
Mw 8.40
Highest 100-year return level — extreme seismic hazard
Java
Mw 6.72
Lowest return level, dominated by deep-focus earthquakes
DNN Accuracy
95.4%
Zone classification accuracy, outperforming K-means clustering
Catalog
1,135 events
Mainshocks after declustering, from 100 years of USGS data

Methodology

📊
Data & Preprocessing
USGS ComCat catalog 1926–2026. Magnitude homogenization via Scordilis (2006). Gardner-Knopoff declustering to isolate independent mainshocks.
🧠
Deep Neural Network Zoning
4-layer fully-connected network classifies each earthquake into one of 6 seismic zones based on location and depth. 95.4% accuracy.
📈
Extreme Value Theory
Four EVT models (GEV, GP, RT-GR, RT-POT) fit to each zone. Best model selected automatically using AIC/BIC criteria.
🎯
Uncertainty Quantification
Bootstrap resampling (500 replicates) produces 95% confidence intervals for all return level estimates.

What is a Return Period?

A "100-year return period" does not mean an earthquake happens exactly every 100 years. It means there is a 1% chance of that earthquake happening in any given year. Engineers use this to design buildings and infrastructure that can withstand rare but potentially catastrophic events.

A 500-year return period = 0.2% annual probability = the standard for critical infrastructure like hospitals and dams.

Primary Reference

Ma, G. et al. (2021). Return Period Evaluation of the Largest Possible Earthquake Magnitudes in Mainland China Based on Extreme Value Theory. Sensors, 21(10), 3519.

Irsyam, M. et al. (2020). Development of New National Seismic Hazard Maps of Indonesia. Earthquake Spectra, 36(1_suppl), 112–136.

USGS Earthquake Hazards Program. ComCat Earthquake Catalog. earthquake.usgs.gov