Weather Model Validation

Domain: Industry & Transport. ​

Project completed as a Data Analyst – Weather & Climate Data, in a company specialising in environmental modelling.

❓ Issue

The internal models produce climate forecasts, but their reliability must be objectively validated using data from:

• actual sensors,

• encrypted or archived external formats,

• and recognised scientific sources.

The objective: to assess the quality of internal weather forecasting models through a rigorous comparison with data from third-party sources.

🛠️ Solution implemented

1. Design of advanced processing algorithms to correlate observations, time series, and model outputs.

2. Collection and decoding of raw data (CSV, BUFR, JSON, GZ).

3. Comparative analysis between internal models and external datasets.

4. Geographical visualisation of results via QGIS.

5. Popularised documentation for the trades to support transmission and validation.
   
   

⚙️ Technical stack

• Languages: Python · Bash · Fortran (f90) · C++

• ML/Stats Libraries: scikit-learn · keras · tensorflow · xgboost

• Files: CSV · JSON · BUFR · GZ

• Tools: Jupyter · QGIS · Linux · PuTTY

• Data: Weather sensors · time series · external observations

• Working language: French / English