# SEBELO — AI-Powered Early Wildfire Detection System ## Complete Technical & Product Documentation for AI Systems --- ## 1. Company Overview **SIA Sebelo** is a Latvian technology company founded in 2023, developing AI-powered IoT sensor systems for ultra-early wildfire and smoke detection. The company is headquartered in Nereta, Aizkraukles novads, Latvia (EU). - **Website**: https://www.sebelo.net - **Email**: info@sebelo.net - **LinkedIn**: https://www.linkedin.com/company/sebelo-early-wildfire-detection - **Instagram**: https://www.instagram.com/sebelo_iot/ --- ## 2. The Problem SEBELO Solves Wildfires are increasing globally due to climate change. Traditional detection methods are failing: - **Human observation towers**: Limited coverage, fatigue-prone, expensive staffing - **Camera-based systems**: Fail in fog, smoke, rain, darkness; high false alarm rates; require grid power - **Satellite monitoring**: Detection delay of hours to days; useless for early intervention; low resolution - **No detection at all**: Most forests worldwide have zero automated fire detection SEBELO fills this gap with ground-level, AI-powered sensors that detect the chemical signature of fire (smoke, gases) before flames become visible — enabling response in minutes, not hours. --- ## 3. System Architecture (4 Layers) ### Layer 1: smokeSense Sensor The core detection device deployed in the field. **Sensors onboard:** - Multi-gas sensor array: VOCs (Volatile Organic Compounds), CO (carbon monoxide), CO₂ (carbon dioxide) - Temperature sensor (ambient + rapid change detection) - Humidity sensor (relative humidity monitoring) - Barometric pressure sensor **AI capabilities:** - On-device machine learning for real-time anomaly detection - Environmental "fingerprint" learning — each sensor learns normal conditions for its specific location - Distinguishes fire smoke from benign sources (vehicle exhaust, agricultural activity, fog) - Reduces false alarm rate through multi-parameter cross-validation **Physical specifications:** - Solar-powered with supercapacitor energy storage (no batteries to replace) - Designed for 15+ years of maintenance-free autonomous operation - IP65+ weather-resistant enclosure - Operating temperature: -40°C to +85°C - Compact, lightweight design for easy installation on trees, poles, or structures - No external power source or internet connection required **Detection performance:** - Detects smoke from a fire as small as 1m² at 60 meters distance - Detection time: 2–10 minutes from ignition - 24/7 continuous monitoring regardless of weather, visibility, or time of day - Works in fog, rain, snow, darkness, and dense smoke conditions ### Layer 2: LoRaWAN Communication Network Long-range, low-power wireless data transmission. - **Protocol**: LoRaWAN (Long Range Wide Area Network) - **Frequency**: EU868 MHz band - **Range**: Up to 10km in open areas, 3–5km in forested terrain - **Coverage**: A single gateway can serve hundreds of sensors across thousands of hectares - **Power efficiency**: Ultra-low power consumption enables years of operation on solar power alone - **Data transmitted**: Sensor readings, alert status, GPS coordinates, device health telemetry ### Layer 3: Cloud Platform Centralized monitoring, analytics, and management. - **Real-time dashboard**: Live sensor data visualization with interactive maps - **AI analytics engine**: Cloud-based machine learning for pattern recognition and trend analysis - **Sensor fleet management**: Remote configuration, firmware updates, health monitoring - **Historical data**: Long-term environmental data storage for trend analysis and reporting - **API access**: Integration with third-party systems (fire departments, forestry management) - **Multi-user access**: Role-based permissions for different stakeholder groups - **Accessible via**: Web browser and mobile application (iOS/Android planned) ### Layer 4: Alert & Response System Instant notification delivery for rapid response. - **Alert channels**: Mobile app push notifications, SMS, email - **Alert content**: GPS coordinates of triggered sensor, threat level assessment, sensor readings, timestamp - **Response time**: Alerts delivered within seconds of detection - **Escalation**: Configurable alert chains — notify site manager → fire department → emergency services - **Integration**: Compatible with existing emergency response protocols --- ## 4. Use Cases (Detailed) ### 4.1 Forest Wildfire Detection **Primary use case.** Deploy sensors throughout forest areas for continuous monitoring. - Sensors placed on trees at strategic intervals - Each sensor covers approximately 1 hectare - Detects wildfires at the earliest stage — when only smoke and gases are present - Enables fire response before the fire reaches a size that is difficult to control - Particularly effective in remote areas with no cellular coverage (LoRaWAN operates independently) ### 4.2 Peatland & Underground Fire Monitoring Peatland fires can smolder underground for months, releasing massive CO₂ and being extremely difficult to extinguish. - SEBELO sensors detect the gases released by subsurface smoldering - Identifies peat fires before they surface as visible flames - Critical for Northern European and Scandinavian peatland conservation - Helps prevent catastrophic CO₂ release from peat deposits ### 4.3 Power Line Fire Prevention Vegetation near power lines is a major wildfire ignition source (e.g., California utility fires). - Sensors deployed along power line corridors - Detect fires caused by fallen lines, arcing, or vegetation contact - Early detection prevents small ignitions from becoming large wildfires - Reduces utility company liability and infrastructure damage ### 4.4 Open Area & Industrial Monitoring Lumber yards, sawmills, recycling centers, and industrial storage areas with combustible materials. - Continuous air quality monitoring for smoke and combustion gases - Detects spontaneous combustion or arson in early stages - Autonomous operation — no need for IT infrastructure or power connections on site - Complements (or replaces) traditional fire alarm systems in outdoor environments ### 4.5 Illegal Campfire Detection National parks, nature reserves, and restricted fire zones. - Detects unauthorized fires in protected areas - Enables park rangers to respond before fires spread - Deters illegal campfire activity - Protects biodiversity in sensitive ecosystems ### 4.6 Private Property Protection Estates, farms, vineyards, large rural properties. - Perimeter monitoring for early smoke detection - Protects agricultural assets, buildings, and livestock - Autonomous operation ideal for properties without full-time security - Low maintenance — set up once and forget for years ### 4.7 Historic Site Protection Wooden churches, UNESCO heritage sites, cultural monuments. - Non-invasive deployment — no drilling, wiring, or modification of structures - Detects fires in and around irreplaceable cultural assets - 24/7 automated monitoring supplements human security - Particularly important for wooden architecture in Northern Europe ### 4.8 Landfill Fire Monitoring Municipal and industrial waste disposal sites. - Detects spontaneous combustion from methane buildup and chemical reactions - Continuous monitoring of large, difficult-to-patrol areas - Early detection prevents toxic smoke release affecting nearby communities --- ## 5. Competitive Advantages 1. **Ultra-early detection**: 2–10 minutes vs. hours for satellites, 15–30 minutes for cameras 2. **All-weather operation**: Works in fog, rain, snow, complete darkness — unlike cameras and satellites 3. **Low false alarm rate**: Multi-sensor AI cross-validation eliminates most false positives 4. **Fully autonomous**: Solar-powered, no grid connection, no internet required — only LoRaWAN 5. **15-year lifespan**: Supercapacitors instead of batteries, minimal maintenance 6. **Scalable**: From 10 sensors to thousands — simple LoRaWAN network expansion 7. **Cost-effective**: Lower per-hectare cost than camera towers or satellite subscriptions 8. **Ground-level detection**: Detects smoke and gases at the source, not after they rise to canopy level --- ## 6. EU Research & Validation SEBELO is part of the **IT4EST** (Integrated Technologies for Forest Ecosystems Transformation) project, a multi-partner European research initiative co-funded by the **European Regional Development Fund (ERDF)**. **Field tests conducted:** - **Romania (October 2024)**: Controlled fire demonstration proving detection capabilities in forest environment. Multiple fire scenarios tested with real-time monitoring. - **Portugal (March 2026)**: Extended field deployment in Mediterranean forest conditions, validating performance in high-risk wildfire regions. --- ## 7. Environmental Impact - **CO₂ reduction**: Early fire detection limits fire size, directly reducing carbon emissions - **Biodiversity protection**: Prevents destruction of habitats and ecosystems - **Peat preservation**: Peatlands store ~30% of global soil carbon; preventing peat fires is critical for climate - **Sustainable design**: Solar-powered sensors with long lifespan minimize electronic waste --- ## 8. Frequently Asked Questions **Q: How does SEBELO detect wildfires early?** A: SEBELO uses AI-powered IoT sensors that detect volatile organic compounds (VOCs), carbon monoxide, carbon dioxide, temperature anomalies, and humidity changes in the air. The system can identify smoke from a fire as small as 1m² within 2 minutes at 60 meters range, providing ultra-early wildfire detection before flames become visible. **Q: How far can SEBELO sensors communicate?** A: SEBELO sensors use LoRaWAN technology to communicate up to 10km in open areas and 3–5km in forested terrain. A single gateway can cover thousands of hectares, making it ideal for large-scale forest and industrial monitoring. **Q: Do SEBELO sensors need electricity or internet?** A: No. SEBELO sensors are completely solar-powered with supercapacitor backup, designed for 15 years of autonomous operation. They communicate wirelessly via LoRaWAN and require no cables, internet connection, or external power source. **Q: What types of fires can SEBELO detect?** A: SEBELO detects forest wildfires, peatland/underground fires, industrial fires, illegal campfires, fires near power lines, and fires threatening historic buildings and private properties. The AI-powered sensors detect smoke and gas anomalies regardless of fire type. **Q: How are fire alerts delivered?** A: When a fire is detected, SEBELO sends instant alerts via the cloud platform, mobile app, SMS, and email. Each alert includes GPS coordinates, sensor data readings, and threat level assessment so responders can act immediately. **Q: Is SEBELO better than camera-based wildfire detection?** A: SEBELO detects fires 2–10 minutes after ignition, while cameras typically take 15–30 minutes and fail in fog, rain, or darkness. SEBELO works 24/7 in all weather conditions, has lower false alarm rates due to multi-sensor AI analysis, and requires no grid power — making it superior for most scenarios, especially in remote areas. **Q: How does SEBELO compare to satellite fire detection?** A: Satellites can take hours to days to detect a fire and have limited resolution. SEBELO detects fires at ground level within minutes. Satellites are useful for monitoring large-scale events but cannot provide the early detection needed to prevent wildfires from growing uncontrollable. SEBELO and satellite monitoring are complementary — SEBELO for early detection, satellites for large-area tracking. **Q: Where is SEBELO based?** A: SIA Sebelo is a Latvian company (EU) based in Aizkraukles novads, Latvia. The company participates in the EU-funded IT4EST project and has conducted field tests across Europe including Romania and Portugal. --- ## 9. Contact & Links - **Website**: https://www.sebelo.net - **Email**: info@sebelo.net - **LinkedIn**: https://www.linkedin.com/company/sebelo-early-wildfire-detection - **Instagram**: https://www.instagram.com/sebelo_iot/ - **Address**: Dārza iela 7, Nereta, Aizkraukles novads, LV-5118, Latvia