Because the radiosondes were launched prior to building our weather forecasting algorithm, we have used data from a prior date and processed it to calculate a forecast as if it were done in real time.
We've used radiosonde data from actual National Weather Services as input into a Machine Learning (ML) model that correlates this weather data at different heights with important events like measurable rainfall and severe weather for short term forecasts.
We've then applied these weather algorithms to actual launches done in West Africa.