Modeling Error Types in Aviation Weather Forecasts: Machine Learning Approach

Abstract

The accurate prediction of aviation weather conditions is crucial for the safety and efficiency of air travel. METAR (Meteorological Aerodrome Reports) and TAF (Terminal Aerodrome Forecasts) serve as key tools for aviation professionals to assess actual and forecasted weather conditions at airports. This study explores the application of machine learning techniques to evaluate the accuracy of weather forecasts using data from aviationweather.gov. Historical weather data are utilized to train and assess various machine learning models, including random forests, naïve Bayes, artificial neural networks, gradient boosting, support vector machines, and k-nearest neighbors to classify weather forecasts as correct, false alarms, or missed detections. The primary objective of this research is to identify the most influential features, such as specific airports or weather characteristics, that impact the accuracy of forecasts.