🌐 INFRAS-CLOUD
Atmospheric Infrasonic Severity Index — Open-Source Planetary Acoustic Weather Intelligence
📖 Overview
INFRAS-CLOUD is the first open-source, physically rigorous framework for converting raw microbarometer recordings into operational atmospheric intelligence. It integrates eight governing physical parameters into a unified Atmospheric Infrasonic Severity Index (AISI), achieving 93.1% accuracy in classifying infrasonic events across six severe weather categories — at detection ranges up to 12,000 km.
The system detects and classifies:
| Source Category | Detection Range | Lead Time / Accuracy |
|---|---|---|
| 🌀 Tropical Cyclones | up to 4,200 km | 18–36 hr before landfall |
| ⛈️ Mesoscale Convective Systems | regional | spectral fingerprint |
| 🌋 Volcanic Explosions | up to 12,000 km | 96.3% classification accuracy |
| 🌍 Large Earthquakes | continental | broad-band discrimination |
| 🌊 Oceanic Microbaroms | global | continuous wave field monitoring |
| 🏭 Anthropogenic Sources | local–regional | SNR-based discrimination |
| 🌪️ Tornadoes | up to 280 km | 12–28 min precursor lead time |
Key result: INFRAS-CLOUD doubles the usable tornado warning window over current Doppler radar systems, with a projected 28% reduction in fatalities per major tornado outbreak.
🔬 The Eight-Parameter AISI Framework
| # | Parameter | Symbol | Physical Role |
|---|---|---|---|
| 1 | Microbarom Amplitude | P_ub |
Ocean–atmosphere coupling energy |
| 2 | Stratospheric Ducting Efficiency | D_str |
Transoceanic wave propagation |
| 3 | Spectral Peak Frequency | f_p |
Event-type discrimination (highest weight: 0.21) |
| 4 | Azimuthal Arrival Angle | θ |
Source localization |
| 5 | Phase Velocity | v_ph |
Stratospheric wind structure |
| 6 | Atmospheric Absorption Coefficient | α_air |
Low-frequency energy loss |
| 7 | Inter-station Coherence | γ² |
Detection validation vs. noise |
| 8 | Signal-to-Noise Ratio | SNR |
Natural vs. anthropogenic discrimination |
$$\text{AISI} = \sum_{i=1}^{8} w_i \cdot \hat{x}_i \quad \text{where} \quad \sum w_i = 1.0$$
AISI Thresholds:
≥ 0.80→ 🔴 Active severe weather — immediate meteorological alert0.55–0.79→ 🟡 Elevated atmospheric activity< 0.55→ 🟢 Background state
📁 Project Structure
infras-cloud/
│
├── 📄 README.md
├── 📄 LICENSE
├── 📄 CHANGELOG.md
├── 📄 CONTRIBUTING.md
├── 📄 pyproject.toml
├── 📄 setup.cfg
├── 📄 requirements.txt
├── 📄 requirements-dev.txt
│
├── 📂 infras_core/ # Core Python package
│ ├── __init__.py
│ ├── processor.py # InfrasProcessor — wavelet CWT pipeline
│ ├── beamformer.py # BeamFormer — f-k analysis, θ & v_ph
│ ├── ducting.py # DuctingAnalyzer — D_str inversion
│ ├── classifier.py # AIEventClassifier — Bayesian AISI
│ ├── microbarom.py # Microbarom amplitude (P_ub) module
│ ├── absorption.py # Atmospheric absorption (α_air)
│ ├── coherence.py # Inter-station coherence (γ²)
│ ├── aisi.py # Composite AISI computation
│ └── utils/
│ ├── io.py # IMS / MiniSEED data I/O
│ ├── filters.py # Bandpass, Dolph-Chebyshev weights
│ ├── geo.py # Azimuth, range, geographic priors
│ └── plotting.py # Scalograms, beams, AISI dashboards
│
├── 📂 data/
│ ├── catalogs/
│ │ ├── validation_1847_events.csv # Full validation catalogue (2005–2025)
│ │ ├── tornado_super_outbreak_2011.csv
│ │ ├── hunga_tonga_2022.csv
│ │ └── hurricane_irma_2017.csv
│ ├── ims_stations/
│ │ ├── ims_47_stations.geojson # Array locations & metadata
│ │ └── sensor_specs.yaml # MB2000/MB2005 specifications
│ └── synthetic/ # Physics-augmented training data
│
├── 📂 models/
│ ├── aisi_weights_v1.json # PCA-regularized logistic regression weights
│ ├── classifier_v1.pkl # Trained AIEventClassifier
│ └── arma_site_params/ # Per-station ARMA(2,1) parameters
│
├── 📂 notebooks/
│ ├── 01_quickstart.ipynb
│ ├── 02_aisi_validation.ipynb
│ ├── 03_case_study_irma_2017.ipynb
│ ├── 04_case_study_hunga_tonga_2022.ipynb
│ ├── 05_case_study_tornado_outbreak_2011.ipynb
│ ├── 06_microbarom_ocean_inversion.ipynb
│ ├── 07_stratospheric_ducting_profiling.ipynb
│ └── 08_climate_change_projections.ipynb
│
├── 📂 scripts/
│ ├── run_realtime.py # Real-time AISI monitoring daemon
│ ├── batch_validate.py # Batch validation over event catalogue
│ ├── download_ims_data.py # IMS IRIS/FDSN data downloader
│ └── export_aisi_report.py # Generate PDF/HTML monitoring report
│
├── 📂 tests/
│ ├── conftest.py
│ ├── unit/
│ │ ├── test_processor.py
│ │ ├── test_beamformer.py
│ │ ├── test_ducting.py
│ │ ├── test_classifier.py
│ │ └── test_aisi.py
│ ├── integration/
│ │ ├── test_pipeline_e2e.py
│ │ └── test_realtime_stream.py
│ └── fixtures/
│ └── synthetic_events/ # Minimal waveform fixtures
│
├── 📂 docs/
│ ├── index.md
│ ├── installation.md
│ ├── quickstart.md
│ ├── api/ # Auto-generated from docstrings
│ ├── theory/
│ │ ├── aisi_framework.md
│ │ ├── wavelet_transform.md
│ │ ├── stratospheric_ducting.md
│ │ └── microbarom_inversion.md
│ └── case_studies/
│ ├── irma_2017.md
│ ├── hunga_tonga_2022.md
│ └── tornado_outbreak_2011.md
│
├── 📂 deploy/
│ ├── docker/
│ │ ├── Dockerfile
│ │ └── docker-compose.yml
│ ├── k8s/ # Kubernetes manifests (optional)
│ └── ansible/ # Provisioning for distributed arrays
│
└── 📂 .gitlab/
├── ISSUE_TEMPLATE/
│ ├── bug_report.md
│ ├── feature_request.md
│ └── new_event_case_study.md
└── merge_request_templates/
└── default.md
⚡ Quick Start
Installation
# Install from PyPI
pip install infrascloud
📦 Package page: pypi.org/project/infrascloud
# Or clone and install in development mode
git clone https://gitlab.com/gitdeeper9/infrascloud.git
cd infras-cloud
pip install -e ".[dev]"
Minimal Example — Single Station AISI
from infras_core import InfrasProcessor, AIEventClassifier
# Load raw microbarometer waveform (MiniSEED or NumPy array)
processor = InfrasProcessor.from_miniseed("IS42.mseed", fs=20.0)
# Run wavelet CWT pipeline
features = processor.extract_features(
freq_band=(0.01, 10.0), # Hz
window_sec=256,
overlap=0.75
)
# Classify and compute AISI
classifier = AIEventClassifier.load_pretrained()
result = classifier.predict(features)
print(f"AISI: {result.aisi:.3f}")
print(f"Event class: {result.event_class}")
print(f"Alert level: {result.alert_level}")
Full Array Beamforming Example
from infras_core import BeamFormer, DuctingAnalyzer, aisi
# Multi-station array (requires ≥3 stations)
bf = BeamFormer(
stations=["IS42_H1", "IS42_H2", "IS42_H3", "IS42_H4"],
coords=station_coords_dict
)
theta, vph, coherence = bf.fk_analysis(waveforms, freq_band=(0.1, 2.0))
# Stratospheric duct inversion
duct = DuctingAnalyzer()
d_str = duct.invert_from_phase_velocity(vph, elevation_profile)
# Full 8-parameter AISI
index = aisi.compute(
pub=microbarom_amplitude,
d_str=d_str,
fp=spectral_peak_freq,
theta=theta,
vph=vph,
alpha_air=absorption_coeff,
gamma2=coherence,
snr=snr_db
)
🧪 Running Tests
# Unit tests
pytest tests/unit/ -v
# Integration tests (requires sample IMS data)
pytest tests/integration/ -v --ims-data ./data/fixtures/
# Full test suite with coverage
pytest --cov=infras_core --cov-report=html
🐳 Docker
# Build image
docker build -t infras-cloud:latest -f deploy/docker/Dockerfile .
# Run real-time monitoring daemon
docker-compose -f deploy/docker/docker-compose.yml up infras-realtime
# Run batch validation
docker run --rm \
-v $(pwd)/data:/app/data \
infras-cloud:latest \
python scripts/batch_validate.py --catalogue data/catalogs/validation_1847_events.csv
📊 Validation Results
| Event Class | AUC | Precision | Recall |
|---|---|---|---|
| Tropical Cyclone | 0.984 ± 0.018 | 0.961 | 0.974 |
| Mesoscale Convective System | 0.961 ± 0.022 | 0.938 | 0.947 |
| Volcanic Explosion | 0.976 ± 0.014 | 0.963 | 0.969 |
| Large Earthquake | 0.958 ± 0.025 | 0.931 | 0.942 |
| Oceanic Microbarom | 0.971 ± 0.019 | 0.952 | 0.958 |
| Anthropogenic | 0.943 ± 0.031 | 0.918 | 0.926 |
| Macro Average | 0.966 ± 0.021 | 0.944 | 0.953 |
Validated on 1,847 confirmed events from 47 IMS stations (2005–2025), stratified 10-fold cross-validation.
📡 Key Case Studies
🌀 Hurricane Irma (2017)
AISI crossed elevated-alert (0.55) 9 days before Irma reached Category 5. Peak AISI = 0.94 on record peak intensity day. Azimuthal accuracy ±4.3°.
🌋 Hunga Tonga-Hunga Ha'apai (January 2022)
Infrasonic signals recorded at all 53 operational IMS stations, circumnavigating Earth four times. Source energy estimated at 38 ± 4 megatons TNT (within 10% of independent estimates) from acoustic data alone.
🌪️ 27 April 2011 Super Outbreak
41 of 47 tornadoes detected within range (87.2% detection rate). Mean precursor lead time: 16.4 ± 8.2 min before NWS confirmed touchdown. All four EF5 tornadoes flagged ≥ 22 min before touchdown.
🤝 Contributing
We welcome contributions from the atmospheric science, signal processing, and open-source communities.
- Fork the repository
- Create a feature branch:
git checkout -b feature/your-feature-name - Commit your changes with conventional commits:
git commit -m "feat: add gravity wave detection module" - Push and open a Merge Request using the MR template
Please read CONTRIBUTING.md for coding standards, test requirements, and the physics documentation policy.
Reporting Issues: Use the GitLab issue tracker with the appropriate template (bug report / feature request / new case study).
📚 Citation
If you use INFRAS-CLOUD in your research, please cite:
@article{baladi2026infrascloud,
title = {INFRAS-CLOUD: A Unified Eight-Parameter Atmospheric Infrasonic
Severity Index for Global Severe Weather Classification},
author = {Baladi, Samir},
journal = {Journal of Geophysical Research — Atmospheres},
year = {2026},
month = {March},
doi = {10.5281/zenodo.18952438},
url = {https://doi.org/10.5281/zenodo.18952438}
}
If you are citing the Python package specifically:
@software{baladi2026infrascloud_pypi,
title = {infrascloud: Atmospheric Infrasonic Severity Index — Open-Source Python Package},
author = {Baladi, Samir},
year = {2026},
url = {https://pypi.org/project/infrascloud},
note = {PyPI package. pip install infrascloud}
}
APA style:
Baladi, S. (2026). INFRAS-CLOUD: A Unified Eight-Parameter Atmospheric Infrasonic Severity Index for Global Severe Weather Classification. Journal of Geophysical Research — Atmospheres. https://doi.org/10.5281/zenodo.18952438
APA style (package):
Baladi, S. (2026). infrascloud [Python package]. PyPI. https://pypi.org/project/infrascloud
📜 License
This project is licensed under the MIT License — see LICENSE for details.
The IMS infrasound archival data used for validation is subject to CTBTO data access policies. See docs/data_access.md.
📬 Contact & Acknowledgements
- Author: Samir Baladi — gitdeeper@gmail.com
- ORCID: 0009-0003-8903-0029
- Affiliation: Ronin Institute / Rite of Renaissance — Independent Interdisciplinary AI Researcher
- GitHub: github.com/gitdeeper9/infrascloud
- GitLab: gitlab.com/gitdeeper9/infrascloud
- Dashboard: infrascloud.netlify.app
- Zenodo: doi.org/10.5281/zenodo.18952438
- IMS Data Access: CTBTO Virtual Data Exploitation Centre (vDEC)
- Funding: Independent researcher — no external funding declared
"INFRAS-CLOUD converts the inaudible acoustic language of the atmosphere into operational meteorological intelligence — the first time the sky has been systematically listened to rather than merely watched."
Metadata
Release files for infrascloud 1.0.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| infrascloud-1.0.0.tar.gz | 68.7 kB | Without provenance | Details for infrascloud-1.0.0.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| infrascloud-1.0.0-py3-none-any.whl | Python 3 | none | any | Details for infrascloud-1.0.0-py3-none-any.whl |
Total release size: 97.7 kB
Release files / infrascloud-1.0.0.tar.gz
| Download URL | infrascloud-1.0.0.tar.gz |
|---|---|
| Size | 68.7 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
dc98d3ae04fa47a5469f6788c83f8bece768dc001ee1e8931a2b98b6cabda8aa
|
|
BLAKE2b-256 checksum How to use checksums |
4ed7d8ff7d078974ca7b333b3973a1a900c8e33ea80109b6954287e5104fa138
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
INFRASCLOUD-Uploader/1.0
|
Release files / infrascloud-1.0.0-py3-none-any.whl
| Download URL | infrascloud-1.0.0-py3-none-any.whl |
|---|---|
| Size | 29.0 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
e20267f27eade91c8a44c4002cf70c0e813d521982f34e912a427b3d93f5d291
|
|
BLAKE2b-256 checksum How to use checksums |
2ae35643298d0762c3764068eb2f07b1dbc0cf6c2cd2ee0ec4ce72678bf6c42e
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
INFRASCLOUD-Uploader/1.0
|
Security for infrascloud 1.0.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.