I Built a 170-Year Caribbean Hurricane Observatory, Alone, in Python
How CaribScope turns 170 years of NOAA hurricane data, USGS earthquake feeds and a Random Forest into a free, open-source Caribbean climate risk observatory.
Build story
CaribScope
Open source observatory of Caribbean hurricanes
- Status
- Live
- My role
- Design, development
- Online at
- caribscope.fwicss.fr
01
What it is
Hurricane tracks since 1851, earthquakes and climate trends in the Caribbean, based on NOAA and USGS data. 1,376 cyclonic systems and a category prediction using machine learning.
02
What it does
- Tracks of 1,376 cyclonic systems from 1851 to 2025, filterable by name, year or category
- Explorer by hurricane or by island, comparison of two hurricanes
- Map and statistics of Caribbean earthquakes
- Air and ocean temperature trends
- Open code, MIT license
03
The technologies
- Python
- Streamlit
- Plotly
- pandas
- scikit-learn