🌧️ weather-ml-edge

Edge client for ML-based rain prediction and automated shutter control.

This package connects local devices (Raspberry Pi, servers, PCs) to the weather-ml machine learning system and enables real-time weather monitoring and automation — without exposing private API keys or location data.

✨ Features

• 🤖 ML-based rain probability monitoring
• 🪟 Smart shutter / relay control support
• ⚡ Lightweight CLI tool
• 🔐 Secure environment-based configuration
• 🌍 Uses your own Visual Crossing account
• 🌧️ CHIRPS-aware model metadata status (from upstream trained model)
• 📦 No bundled datasets or credentials
• 🖥️ Designed for edge / IoT deployment

🔒 Security Notice

⚠️ This package does NOT provide API access.

You must supply:

• Your own Visual Crossing API key
• Your own location

The author does not proxy, store, or distribute credentials.

All API usage is billed to your own account.

📦 Installation

Global Install (Recommended)

npm install -g weather-ml-edge

Local Install

npm install weather-ml-edge

⚙️ Configuration

Create a .env file in your project directory:

VISUAL_CROSSING_KEY=your_api_key_here
VISUAL_CROSSING_LOCATION=lat,lon
WEATHER_ML_MODEL_META_URL=https://raw.githubusercontent.com/rotsl/weather-ml/main/models/hgb_D_next_6h_current_meta.json

Example:

VISUAL_CROSSING_KEY=abc123xyz
VISUAL_CROSSING_LOCATION=lat,lon

WEATHER_ML_MODEL_META_URL is optional; the default points to the main weather-ml model metadata artifact.

Never commit this file.

🚀 Usage

CLI Mode

After installing globally:

weather-ml

Output:

🌧️ Weather Status
Temp: 26.4°C
Rain %: 12
Humidity: 81
CHIRPS training: Enabled
CHIRPS features: 17
Model horizon: D_next_6h
Model trained at: 2026-03-08T11:15:46.848136

Programmatic Usage

import { getChirpsStatus, getWeather, loadConfig, loadModelMeta } from "weather-ml-edge";

const cfg = loadConfig();

const weather = await getWeather(cfg.apiKey, cfg.location);
const meta = await loadModelMeta(cfg.modelMetaUrl);
const chirps = getChirpsStatus(meta);

console.log(weather.precipprob);
console.log(chirps.enabled, chirps.count);

🪟 Raspberry Pi Integration

Designed for use with:

• Servo motors
• Relay boards
• GPIO buttons
• Smart window systems

Typical setup:

Prediction → Threshold → GPIO → Motor

Example:

if (rainProb > 0.5) {
  closeShutters();
}

See main project:

https://github.com/rotsl/weather-ml

📊 Data Sources

Weather data provided by:

• Visual Crossing Weather API https://www.visualcrossing.com/

Model metadata may include CHIRPS-derived features from:

• CHIRPS Daily (Google Earth Engine) https://developers.google.com/earth-engine/datasets/catalog/UCSB-CHG_CHIRPS_DAILY

Subject to their terms and pricing.

📁 Project Structure

weather-ml-edge/
├── src/
│   ├── index.js
│   ├── predictor.js
│   ├── modelLoader.js
│   └── config.js
├── package.json
└── README.md

🛠️ Requirements

• Node.js ≥ 18
• Internet connection
• Visual Crossing API account

🚧 Limitations

• No offline forecasting
• Requires external API
• CHIRPS data is not downloaded directly by this package
• No bundled ML models
• No centralized service

This package is a client, not a server.

🗺️ Roadmap

Planned features:

• MQTT output
• Home Assistant integration
• Camera verification
• Local caching
• Edge retraining hooks
• Mobile notifications

🤝 Contributing

Contributions are welcome.

Steps:

1. Fork repository
2. Create feature branch
3. Submit PR

See main project for guidelines.

📜 License

MIT License

See LICENSE for details.

👤 Author

Rohan (rotsl)

GitHub: https://github.com/rotsl Project: https://github.com/rotsl/weather-ml

📄 License

MIT © Rohan R.

🌱 Philosophy

"Build systems that predict, adapt, and protect — automatically."

Minimal • Secure • Edge-first • ML-powered