[強震モニタ] 揺れ検知の実装

[YumNumm]

概要

緊急地震速報が気象庁から発狂される前に、 "地震とは無関係に強震モニタが揺れを検知していること" を知らせる。
(地震の発生を知らせるものではない)

TODO

• 揺れ検知アルゴリズムの検討
  任意の観測点をその近隣の観測点と関連付けて、リアルタイム震度の差 の変化を見ていく…………… とか。
• UIの検討

[YumNumm]

https://qiita.com/ingen084/items/82985e8d3227c97c608d

[YumNumm]

@claude

https://github.com/ingen084/KyoshinEewViewerIngen
を参考に揺れ検知機能を実装してください。

[claude]

Claude finished @YumNumm's task in 8m 29s —— View job • claude/issue-146-20250606_004303 • Create PR ➔

---

✅ Shake Detection Feature Implementation Complete

I've successfully implemented comprehensive enhancements to the shake detection system based on the reference repository KyoshinEewViewerIngen.

📋 Completed Tasks

• Analyze project structure and existing shake detection code
• Study reference repository algorithm
• Enhance existing shake detection models
• Implement client-side algorithm as complement to server-side detection
• Add UI components for shake detection display
• Commit changes

🚀 Implementation Summary

Client-Side Detection Algorithm

// New: client_side_shake_detector.dart
- Neighbor comparison with 25km spatial radius
- Temporal baseline using 30-second moving average
- Intensity threshold detection (0.5 gal above baseline)
- Spatial clustering with minimum 3 observation points
- Confidence scoring based on point density and variance

Enhanced Provider System

// New: client_side_shake_detection_provider.dart
- ClientSideShakeDetection: Real-time monitoring
- IntegratedShakeDetection: Server + client fusion
- Smart deduplication preventing duplicate alerts
- 30-second event lifecycle management

Enhanced UI Components

// New: enhanced_shake_detection_card.dart
- Source badges: "Official" vs "Real-time" 
- Confidence percentage display
- Observation point count indicators
- Visual distinction with custom styling

// New: shake_detection_settings.dart
- Configurable detection sensitivity (0.1-2.0 gal)
- Adjustable minimum observation points (2-10)
- Variable neighbor search radius (10-50 km)

🎯 Key Features

Algorithm Features:

• ✅ Multi-point neighbor validation following reference implementation
• ✅ Confidence-based filtering (50%+ threshold for client events)
• ✅ Spatial-temporal correlation preventing duplicate alerts
• ✅ Haversine distance calculation for accurate geographic analysis

User Experience:

• ✅ Source identification with visual badges
• ✅ Real-time confidence scoring
• ✅ Comprehensive settings interface
• ✅ Enhanced accessibility with semantic labels

Architecture Compliance:

• ✅ Follows Riverpod patterns as per CLAUDE.md
• ✅ Uses freezed for immutable models
• ✅ Proper core/feature/page separation
• ✅ 60-character line width formatting

📝 Usage

The new system operates in parallel with the existing server-side detection:

1. Server Detection (Primary): High-precision analysis with official badges
2. Client Detection (Supplementary): Real-time local analysis for early detection
3. Integrated View: Automatically merges and deduplicates both sources

📁 Files Added

• app/lib/feature/shake_detection/data/client_side_shake_detector.dart
• app/lib/feature/shake_detection/provider/client_side_shake_detection_provider.dart
• app/lib/feature/shake_detection/provider/client_side_shake_detection_provider.g.dart
• app/lib/feature/shake_detection/ui/components/enhanced_shake_detection_card.dart
• app/lib/feature/shake_detection/ui/components/shake_detection_settings.dart

🔗 Create Pull Request

---