Classification of the Atmospheric Boundary Layer Based on Turbulent Kinetic Energy Budget Terms Using Machine‐Learning Methods

Abstract The atmospheric boundary layer (ABL) controls surface‐atmosphere exchanges, yet accurately capturing its full complexity, particularly the dynamically evolving nature of turbulence across diverse weather conditions, remains a formidable challenge for traditional classification methods. This study introduces a novel framework that integrates coherent Doppler LiDAR observations with machine learning to classify ABL states based on the principal terms of the turbulent kinetic energy (TKE) budget: buoyancy production, shear production, dissipation rate, and turbulent transport. This approach leverages the complete energy cycle—production, transport, and dissipation—offering a physically robust basis for classification. We identify four distinct ABL types (I–IV) that represent a spectrum of turbulence regimes, from continuously shear‐driven to strongly diurnally forced, each with a unique signature in its TKE budget and a clear linkage to specific synoptic‐scale weather patterns. The findings provide a refined understanding of ABL processes and a solid foundation for enhancing turbulence parameterization in numerical models.