A new research effort deploys a dense, crowd-sourced IoT weather-station network to detect volcanic Lamb waves and issue early warnings, integrating physics-driven models, multi-agent AI, and an automated data pipeline to convert thousands of Netatmo records into minutes of advance notice with zero false positives.

The system can localize crater bearing to within about 1.78 degrees at Mt. Shinmoedake and derives wave speeds that match theoretical values, achieving high empirical accuracy (roughly 304.38 m/s in cold air and 313.27 m/s in milder air).

A five-step data pipeline handles raw ingestion, zero-phase filtering, temperature-dependent travel-time inversion, beamforming, and energy yield calculation, complemented by a six-stage, multi-parameter workflow for pre-eruption hydrothermal tracking, filtering, inversion, beamforming, energy flux integration, and ash‑shadow cooling tracking.

The core method ingests 20-minute Netatmo meteorological data via Google Apps Script into Google Drive, maps historical eruptions, and uses physical models (2D Lamb wave equations, Bolton enthalpy, slowness beamforming, cylindrical energy flux) to estimate travel times, crater azimuth, and explosive yields.

The framework is anchored in AI pattern discovery guided by physical laws to avoid spurious correlations and promotes a unified approach to灾害 forecasting across earthquakes, urban floods, and volcanic events.

A four-stage workflow underpins the project: automated live data ingestion, historical event mapping, pre/post-eruption anomaly analysis, and theoretical modeling/validation within a specialized CLI guided by established patterns.

Three real-world applications emerge: aviation safety and urban infrastructure protection through rerouting and HVAC/sealing measures; volcanic tsunami early detection with metropolitan protection via lead-time evacuations; and blind crater localization for dormant or unmonitored volcanoes using pre-eruption tracking and post-eruption triangulation.

The approach is framed as a planetary-defense platform leveraging dense IoT networks to outpace sparse traditional sensing, delivering rapid, multi-stakeholder alerts while maintaining strict false-alarm control.

Empirical results from 29,334 records show precise wave speeds, crater localization within 1.78 degrees, and energy yields for two 2018 eruptions ( Kusatsu-Shirane and Shinmoedake) with a reported 0.0% false-positive rate in control tests.