Parameter estimation for the YSU boundary-layer turbulence scheme

Planetary boundary layer (PBL) parameterizations depend on spatially and temporally invariant empirical parameters. These are commonly set by comparing parameterization output with large-eddy simulations (LES) and seeking for the parameter values that minimize the differences. Model errors are caused not only by the oversimplified closure assumptions (structural error), but also by the suboptimal specification of spatially and temporally invariant empirical parameters (parameteric error).

We seek to improve the accuracy of PBL parameterization schemes by making parameters adaptive to atmospheric conditions, and therefore spatially and temporally dependent, using ensemble-based parameter estimation (PE). To achieve this, we utilize an idealized modelling environment implemented with WRF and DART. We conduct Observing System Simulation Experiments (OSSEs), which involve an LES serving as the virtual truth and an ensemble of single-column models (SCM), where the only model error source is the PBL parameterization. Based on previously published parameter identifiability studies, we focus on global parameters influencing the parameterized vertical turbulent mixing. We assimilate vertical profiles from LES using the Ensemble Adjustment Kalman Filter (EAKF), in order to objectively adjust empirical turbulence parameters.

Specifically, we focus on the YSU PBL scheme, which implements a first-order turbulence closure. The empirical parameters in this scheme were originally determined through subjective comparison with a set of dry LES, which represent various wind speed and sensible heat flux regimes. We feed synthetic observations from these LES into the PE algorithm and demonstrate that adjusting turbulence parameters using ensemble-based methods outperforms experiments that estimate the state alone. Moreover, we address the limitations imposed by the EAKF’s linearity assumption. Finally, we discuss how the estimated parameters are affected by environmental conditions.