Live on TestFlight

Every iPhone
a weather station. Every city a sensor grid.

WorldPulse Unified turns the iPhone in your pocket into a precision atmospheric sensor node — fusing barometric pressure, motion, GPS, and on-device AI into a planetary-scale monitoring network.

Join the Beta See the Technology
r = 0.97 Pearson correlation vs. NOAA ASOS
24 Environmental fields per reading
5 km Spatial resolution grid
0 Bytes of personal data transmitted
How It Works

Precision sensing, already in your pocket.

Modern iPhones contain barometric pressure sensors accurate to ±0.1 hPa. WorldPulse orchestrates these — alongside GPS altitude, accelerometer, microphone, network, and optional HealthKit data — into calibrated, science-grade atmospheric readings.

📡
Sense
iPhone sensors sample barometric pressure, altitude, motion, and ambient conditions every second. GPS altitude calibration eliminates elevation-induced pressure error.
🧠
Process On-Device
CoreML anomaly detection and sensor fusion run entirely on your device. Raw sensor data never leaves your phone — only anonymized, aggregated environmental readings.
🌐
Contribute to the Grid
Your node contributes to a 5km spatial grid with Laplace noise differential privacy. The network grows denser with every device, improving resolution for everyone.
📊
Power Real Science
Aggregated data feeds ADCIRC storm surge models, NOAA research pipelines, and real-time urban heat and air quality monitoring at resolutions no fixed sensor network can match.
🛡️ No audio recorded · No location stored · Anonymous grid-cell data only
Dashboard
Environment
Environment
Monitoring Active
audio · barometric · compute ·
health · location · motion ·
network · seismic
0
events
1016.0
hPa
Normal range
Pressure
46%
level
Loud
Acoustic (Est.)
Pressure Trend
Stable
Sea Level Pressure
1015.98 hPa · 30.00 inHg
Station Pressure
1012.93 hPa
Raw · not altitude corrected
🌍 2
Planet
📈 My Data
📡 Global Events
🏙️ Cities
🔒 Privacy
Core Technology

AI-first from the sensor up.

🔬
Sensor Fusion Engine
SensorOrchestrator fuses barometric, motion, acoustic, network, GPS, and optional HealthKit signals into calibrated environmental readings. GPS altitude normalization removes elevation-induced pressure bias.
24 fields · 1Hz sampling · CSV + NetCDF export
⚡
On-Device AI
CoreML anomaly detection identifies atmospheric events — pressure drops, seismic signatures, thermal anomalies — with zero data leaving the device. 99% bandwidth reduction vs. raw sensor streaming.
CoreML · Federated learning roadmap · Privacy-first
🌊
Scientific Data Pipeline
Pressure readings formatted for ADCIRC storm surge models (fort.221/222 compatible). NetCDF export ready for NOAA, FEMA, and academic workflows. Kriging interpolation planned for Phase 2.
ADCIRC-ready · NetCDF · 5km grid · WGS84
🔐
Differential Privacy
Laplace noise injected at the mesh aggregation layer. No raw readings transmitted. No device fingerprinting. GDPR and HIPAA architecture built in from day one, not bolted on later.
Laplace noise · AES-256 at rest · Keychain tokens
📡
Distributed Architecture
Each iPhone is a sovereign sensor node. Background BGAppRefreshTask keeps readings consistent without draining battery. Bluetooth mesh aggregation in development for dense urban deployment.
BGAppRefreshTask · BLE mesh (roadmap) · Edge-first
🏙️
Urban Intelligence
City-level pressure gradient mapping, real-time air quality API integration, and population-level health signal detection. Designed for emergency management, urban planning, and climate resilience.
AQI integration · Heat island detection · Population signals
Pilot Study — 10-Device Network 0.97
Pearson correlation coefficient against
official NOAA ASOS weather station data
Independent 10-device pilot · Palo Alto, CA · 2024
Scientific Validation

NOAA-grade data from consumer hardware.

A 10-device WorldPulse pilot network achieved Pearson r = 0.97 against co-located NOAA Automated Surface Observing System (ASOS) stations — validating that calibrated smartphone sensor fusion can match scientific instrumentation at a fraction of the cost.

✓
Peer-reviewed methodology
Calibration approach follows Mass & Madaus (2014), Madaus & Mass (2017), and McNicholas & Mass (2018) — the foundational literature on smartphone pressure sensing for meteorology.
✓
ADCIRC model integration pathway
Data formatted for direct ingestion by ADCIRC, the storm surge prediction model used by NOAA, FEMA, and the U.S. Army Corps of Engineers for hurricane and flood modeling.
✓
Fills a real observational gap
Fixed ASOS stations average 10–30 miles apart. A dense smartphone network provides sub-kilometer resolution in urban areas where severe weather damage and human risk are highest.
Applications

One network. Planetary-scale impact.

Emergency Management
Storm Surge Prediction
Real-time pressure gradient data fed to ADCIRC storm surge models could improve hurricane track and intensity forecasting by hours. Every hour of additional warning time saves lives and reduces evacuation failure rates.
Reinsurance & Climate Risk
Parametric Risk Intelligence
Hyper-local atmospheric data for parametric insurance models. Real-time pressure readings provide ground truth for storm loss estimation, enabling faster claims processing and more accurate risk pricing.
Urban Planning
Heat Island & Air Quality Monitoring
City-wide sensor density reveals urban heat islands and air quality gradients invisible to sparse fixed networks. Actionable data for city planners, public health officials, and climate resilience programs.
Scientific Research
Citizen Science at Planetary Scale
WorldPulse democratizes atmospheric observation. Academic researchers gain access to sensor density that would cost billions to replicate with traditional fixed infrastructure. NOAA SBIR Phase I research in progress.
Privacy Architecture

Your phone. Your data. Full stop.

Privacy isn't a policy — it's an engineering constraint. WorldPulse is architected so that meaningful participation in a planetary sensor network requires contributing nothing personal.

🔒
On-Device Processing
All sensor fusion and anomaly detection runs locally. Raw readings never leave your device. Only anonymized environmental aggregates contribute to the network.
🎲
Differential Privacy
Laplace noise is applied at the aggregation layer before any data is transmitted, making it mathematically impossible to trace a reading back to an individual device.
🔑
No Identity Required
No account creation. No email. No location history stored. Device tokens are ephemeral and stored only in the iOS Keychain — not in any cloud database.
🛡️
AES-256 Encryption
All locally stored data is encrypted at rest with AES-256. Health data from HealthKit never leaves the device under any circumstance.
📋
GDPR & HIPAA Architecture
Privacy compliance is built into the data model, not added after the fact. The system is designed to be compliant by construction with both GDPR and HIPAA frameworks.
🔍
Open Privacy Controls
A dedicated Privacy tab shows exactly what sensors are active, what data is stored locally, and what (if anything) has been contributed to the network. Full user control, always.
Development Roadmap

Building the planetary nervous system.

WorldPulse is live on TestFlight with validated sensor fusion. Each phase adds a new layer of intelligence, privacy, and scale.

✓
Current · Live
Sensor Fusion & Validation
5-tab iOS app on TestFlight. 24-field environmental CSV export. GPS altitude calibration. r = 0.97 NOAA correlation validated in 10-device pilot.
1
Phase 1 · 6–8 weeks
Bluetooth Mesh + Backend API
BLE mesh networking for dense urban sensor aggregation. FastAPI backend with NetCDF ingestion pipeline. ADCIRC fort.221/222 export.
Python FastAPI · SQLite → PostgreSQL · Bluetooth LE
2
Phase 2 · 4–6 weeks
CoreML On-Device Anomaly Detection
Replaces static thresholds with trained CoreML models. 99% bandwidth reduction. Real-time event classification for storm signatures, seismic activity, and air quality alerts.
CoreML · Create ML · Bias correction (McNicholas & Mass 2018)
3
Phase 3 · 3–4 months
Federated Learning
MLUpdateTask on-device model training. Flower FedAvg gradient aggregation — no raw data ever leaves the device. Minimum 50-device rounds. GDPR-compliant by design.
Flower (flwr) · MLUpdateTask · FedAvg
4
Phase 4 · 3–4 months
Physics-Informed Neural Network
Storm surge prediction PINN trained on ADCIRC outputs and real-time WorldPulse pressure gradients. Sub-kilometer resolution forecasting for coastal emergency management.
PyTorch PINN · ADCIRC integration · NOAA SBIR Phase II

Be a node in the network.

WorldPulse Unified is in active beta on TestFlight. Research partners, emergency managers, and curious citizens welcome.

Request Beta Access Research Partnership

iOS · TestFlight · Privacy-first · No account required