Bise wind at Zurich airport causes systemic delays of aircrafts due to reduction of airport slot capacity by over 30%. Ability to accurately predict Bise on a timescale of hours enables better planning capacities, resulting in saved fuel, CO2 emissions and money. A novel approach has been explored, where data from a national network of weather sensors from MeteoSwiss can be used as an input for ML model that predicts wind conditions in short - term horizon. Here we present results of POC conducted in collaboration with Propulsion Data Science Academy and Google.

Download the slides for this talk.Download ( PDF, 2378.12 MB)