Floods remain the world's most frequent and deadly natural disaster. Growbud's Flood Warning System fuses in‑situ soil moisture dynamics along riverbanks with real‑time rainfall, river stage, and NOAA forecasts. The result? Actionable alerts minutes to hoursbefore overtopping, protecting lives, property, and critical infrastructure.
Layered Architecture
Complex‑impedance probe measuring volumetric water content, temperature & EC.
8‑ch LoRaWAN concentrator, LTE‑M backhaul, 50 W solar & 20 Ah battery.
Storage, dashboards, AI flood risk model, and multi channel alerting.
Typical Deployment
A river-kilometer is instrumented with 12-24 Dro sensors spread across multiple cross-sectional offsets of the riverbank. Sensors report every five minutes under quiescent conditions and ramp to 60 second intervals when the rate of change in VWC exceeds 2% per minute.
Hardware Specifications
Key performance parameters
Dro Sensor
- VWC range 0-100% ±2%
- Temperature -10°C - 60°C ±0.5°C
- EC range 0-10 dSm⁻¹
- LoRaWAN Class A, 20 dBm TX
- IP68 enclosure; 1 year battery life
Outdoor Gateway
- Semtech SX1302, 8-channel, -137 dBm
- LTE-M / Ethernet with automatic failover
- 50 W mono-crystalline PV + 20 Ah battery
- GNSS time sync & asset tracking
- NEMA-4X aluminum enclosure
Analytics & Alerting Workflow
Growbud's gradient‑boosted ensemble blends Dro sensor features with NOAA rainfall forecasts and USGS stage data to predict flood probability 0–6 h ahead. Alerts trigger when P(flood ≥ minor) > 40 %, instantly pushing SMS, email, siren, or webhook notifications. The model retrains nightly, learning from every event.
Impact & Benefits
Lead‑Time Gain
+12–45 min over legacy stage‑only gauges.
Scalability
Modular network – add sensors & gateways to extend coverage.
Cost Efficiency
~US$18k per kilometer CAPEX <US$1k per year OPEX