Pull IoT data from live weather balloon contracts using Hyperion

Data flow

• Listen for datapoints on recent launch to be 15 minutes apart
• Query all datapoints from the launch with Hyperion query
• Send raw data to CSV file
• Call python script to pre-process the data

Python Pre-Process

1. Convert all the data into relevant ML-ingest format (wind instead of lat/lon, etc.)
2. Calculate any derived quantities
3. Choose which ML models are to be run (rain/flood, 300/500hpa, wind speed included/not included)
4. Interpolate the data to fit defined grid (sfc , 1000hpa, 950hpa, etc.)

Python Process

1. Call machine learning models with the data

Python Post-process

1. Give forecast percentage
2. Output to SQL file

Prerequisites:

• Python >= 3
• Nodejs

Usage:

$ npm install & node listen_and_forecast.js