Modification of Similarity Relationships and Parameterization of Submesoscale Motions Under Spectral Regimes Over Uniform Flat Terrain

Abstract The presence of submesoscale motions and turbulence intermittency introduces nonturbulent variability and nonstationarity, leading to significant deviations when similarity theory is applied for flux estimation in the atmospheric boundary layer. Using high‐resolution atmospheric turbulence observations from relatively uniform flat terrain, Horqin station, this study identified four distinct regimes through spectral analysis, each representing different impacts of submesoscale motions and turbulence intermittency. Regime 1 conforms to the classical energy spectrum. Regime 2, dominated by submesoscale motions that lead to large‐scale intermittency, occurs most frequent (>70%). Regime 3, characterized by deviations in the inertial subrange, suggest small‐scale intermittency. Regime 4, where the full‐range energy spectrum departs from the classical form, implying full‐scale intermittency. Marked differences are observed in the turbulence time series characteristics and the performance of similarity relationships across the four regimes. Revised surface‐layer similarity relationships are established under Regimes 2 and 4 by removing submesoscale motions with a Hilbert‐Huang transform‐based separation, encompassing the connections among the normalized velocity and temperature standard deviations (), the flux‐profile functions (), and the stability parameter . In this filtered framework, countergradient transport is substantially reduced, and the positive skill scores demonstrate that the new formulations achieve significant improvements. Furthermore, an empirical parameterization relationship is proposed linking the dimensionless submesoscale wind speed standard deviations to the bulk Richardson number, with validation from five independent data sets demonstrating its robustness and applicability. These findings provide both methodological and theoretical foundations for improving boundary layer turbulence representation in high‐resolution atmospheric models.