Using LiDAR Output to Identify Atmospheric Rotors: A Convolutional Neural Network Approach

ABSTRACT The identification and characterization of atmospheric turbulence, particularly rotors, is crucial for aviation safety due to their small scale and turbulent nature. Traditional methods of detecting rotors rely on manual inspection and reports, which are limited in temporal coverage and require skill when observing. This study explores the novel use of machine learning methods, focusing on convolutional neural networks (CNNs), to identify rotors within Light Detection and Ranging (LiDAR) output. LiDAR technology provides high‐resolution, three‐dimensional wind field data, enabling detailed analysis of atmospheric phenomena. Unlike previous applications of machine learning to LiDAR observations of engineered wake flows, this study focuses on atmospheric rotors, a meteorological turbulence phenomenon with less constrained forcing and greater spatial variability, and demonstrates the value of simple, ensemble‐based CNN architectures under realistic operational constraints. By leveraging this LiDAR data, annotated by operational meteorologists, we developed a robust CNN model capable of detecting rotors with high accuracy. The model was trained on labeled LiDAR data, with a comprehensive hyperparameter search conducted to optimize its performance. The results indicate that the CNN models trained effectively, achieving high performance on the training dataset, though there was a tendency to overfit. An ensemble approach was utilized to mitigate against this with relatively simple CNN architectures allowing relatively fast training on CPU infrastructure. The ability to correctly classify rotor images, even with an over predictive bias, remains valuable for operational meteorologists and for generating larger, consistently labeled datasets to support future research into rotor dynamics and turbulence prediction. This study demonstrates the potential of machine learning techniques to advance turbulence detection in the meteorological domain, ultimately contributing to safer aviation practices.