Japan-wide earthquake intelligence console for infrastructure operators
Real-time seismic monitoring → GMPE intensity computation → infrastructure impact assessment
Namazue is a spatial operations console that turns earthquake events into operational consequences across a living infrastructure map of Japan. When an earthquake occurs, the system computes ground motion intensity using peer-reviewed GMPE equations, propagates P/S wave fronts in real time, evaluates structural fragility across 13 infrastructure asset classes, and surfaces prioritized operator actions.
Built for infrastructure operators (port authorities, rail companies, hospitals, resilience teams) — not general consumers.
Try it live: namazue.dev
M7.2 scenario: intensity field propagation, infrastructure bearing lines, and real-time exposure scoring
- GMPE intensity computation — Si & Midorikawa (1999, revised 2006) with Vs30 site amplification, fault-type corrections, and deep earthquake saturation
- P/S wave propagation — real-time animated wave fronts with depth-corrected velocity
- Nankai Trough scenario engine — multi-segment rupture simulation using SharedArrayBuffer parallel computation
- Intensity grid — per-cell Vs30 microzonation accuracy via Web Workers
- PLATEAU 3D buildings — 3D Tiles for 35 Japanese cities, colored by computed intensity
- AIS vessel tracking — real-time maritime positions via AISstream.io
- Rail network status — ODPT open data feed
- Power grid & telecom — substation and hub locations with exposure scoring
- 766 active fault lines — from AIST active fault database
- Fragility assessment — probit curves calibrated on Kobe 1995, Niigata-Chuetsu 2004, Tohoku 2011, Kumamoto 2016, Noto 2024
- 13 infrastructure asset classes — port, rail hub, hospital, power substation, water facility, telecom hub, nuclear plant, airport, dam, LNG terminal, government EOC, evacuation site, building cluster
- Priority generation — clear / watch / priority / critical severity with domain-specific operator actions
- Multi-source event truth — conflict detection across USGS and JMA with confidence scoring
- Operator-grade dark UI — frosted glass panels, IBM Plex Mono data readouts, JMA severity color scale
- Keyboard-driven workflow —
Cmd+Kcommand palette, full keyboard navigation - Depth cross-section — auto-triggers at 45° map pitch, shows seismological profile
- PWA — offline-capable with stale-while-revalidate cache
- i18n — Japanese, English, Korean
graph TB
subgraph Client["namazue.dev — Cloudflare Pages"]
ML["MapLibre GL JS 5"]
DK["Deck.gl 9 Layers"]
GMPE["GMPE Engine\n(Web Workers)"]
BUILD["PLATEAU 3D Tiles\n(35 cities)"]
ML --> DK
DK --> GMPE
DK --> BUILD
end
subgraph Workers["api.namazue.dev — Cloudflare Workers"]
API["Hono API"]
DO1["SeismicSentinel DO\n(10s JMA poll)"]
DO2["MaritimeHub DO\n(AIS aggregation)"]
DO3["ProjectionProjector DO\n(R2 CDN feed)"]
API --> DO1
API --> DO2
API --> DO3
end
subgraph Data["External Sources"]
USGS["USGS Earthquake API"]
JMA["JMA / P2P地震情報"]
AIS["AISstream.io"]
ODPT["ODPT (Rail)"]
PLATEAU["PLATEAU (3D)"]
JSHIS["J-SHIS (Hazard)"]
end
subgraph DB["Neon Postgres (PostGIS)"]
PG[("57K+ earthquakes\n766 active faults\nSpatial indexes")]
end
Client -- "fetch /api/*" --> Workers
Workers --> DB
Data --> Workers
| Layer | Technology |
|---|---|
| Spatial | MapLibre GL JS 5 + Deck.gl 9 + PMTiles |
| 3D | PLATEAU 3D Tiles via @loaders.gl/3d-tiles |
| Engine | Si & Midorikawa 1999 GMPE, Web Workers, SharedArrayBuffer |
| Frontend | Vanilla TypeScript + DOM (zero frameworks) |
| API | Cloudflare Workers + Hono |
| Real-time | Durable Objects (SeismicSentinel, MaritimeHub, ProjectionProjector) |
| Database | Neon Postgres with PostGIS |
| CDN | Cloudflare R2 snapshot feed |
| Build | Vite 7 + Vitest |
git clone https://github.com/Hybirdss/namazue-console.git
cd namazue-console
npm install
cp .env.example .env # fill in API keys
npm run dev # http://localhost:5173See .env.example for required environment variables.
namazue-console/
├── apps/
│ ├── globe/ # Spatial console (CF Pages → namazue.dev)
│ │ └── src/
│ │ ├── core/ # MapLibre + Deck.gl init, shell, panel system
│ │ ├── data/ # API clients, stores, real-time data managers
│ │ ├── engine/ # GMPE, wave propagation, Nankai model
│ │ ├── i18n/ # Translations (ja, en, ko)
│ │ ├── layers/ # Plugin data layers (seismic, maritime, lifelines, medical, built-env)
│ │ ├── ops/ # Operations intelligence (fragility, exposure, priorities)
│ │ ├── panels/ # Floating panels (22 panels)
│ │ ├── presentation/ # Operator view rendering components
│ │ ├── utils/ # Coordinate math, color scales, formatting
│ │ └── types.ts # Shared type contract
│ └── worker/ # API server (CF Workers → api.namazue.dev)
│ └── src/
│ ├── routes/ # Hono API routes
│ └── lib/ # Durable Objects, tools
├── packages/
│ └── db/ # Drizzle schema + PostGIS queries
├── tools/ # Data pipeline scripts
└── docs/
├── current/ # DESIGN.md (product source of truth)
├── reference/ # GMPE equations, JMA intensity colors, historical presets
└── technical/ # Engine docs, data source reference
| Source | Data | Update Frequency |
|---|---|---|
| USGS Earthquake API | Global earthquake catalog | Real-time |
| JMA / P2P地震情報 | Japan-specific seismic data | 10-second poll |
| AISstream.io | Live vessel positions | Real-time WebSocket |
| ODPT | Rail network status | Periodic |
| J-SHIS | Seismic hazard grid, Vs30 | Static |
| PLATEAU | 3D building models (35 cities) | Static |
| AIST Active Faults | 766 active fault geometries | Static |
| GSI | Elevation, slope data | Static |
The core intensity engine implements the Si & Midorikawa (1999, revised 2006) Ground Motion Prediction Equation — the standard GMPE used by Japan's National Seismic Hazard Maps.
Key parameters:
- Vs30 site amplification — default 270 m/s (Japanese urban alluvial), per-cell grid support
- Fault-type corrections — crustal, subduction interface, intraslab
- Depth saturation — >60 km deep earthquake corrections
- Mw cap — 9.5 for mega-earthquake extrapolation safety
Validation target: within ±1.0 JMA intensity of historical actuals.
See docs/technical/GMPE_ENGINE.md for the full mathematical reference.
See CONTRIBUTING.md for development setup, code style, and PR guidelines.
MIT © 2026 Yunsu Kim