UTLS Turbulence Forecasting with NWP Models at 1-km Grid Spacing: The 'Unexpected' True Consequences of PBL Diffusion

Operational numerical weather prediction (NWP) over the U.S. is currently available at horizontal grid spacings of Δ_h = 3 km, as is the example of the High-Resolution Rapid Refresh (HRRR) product based on the Weather Research and Forecasting (WRF) model. Even beyond that grid spacing, a prototype of the HRRR running at Δ_h = 1 km is being exercised for specific field campaigns over regional-scale domains, and with the increasing trend in computational power, such fine-resolution mesoscale forecasts are expected to become a standard relatively soon. NWP models run at Δ_h = 1 km present the advantage of being able to explicitly resolve some of the turbulence instabilities relevant to commercial aviation at cruising altitudes in the upper-troposphere lower-stratosphere region (UTLS), and therefore have the potential to provide enhanced turbulence forecasts for the aviation community. However, there are modeling aspects associated with ‘grey-zone’ parameterization and the onset of three-dimensional resolved eddies that need to be carefully understood when Δ_h approaches 1 km. We demonstrate these aspects by examining a specific clear-air turbulence (CAT) event that took place on 30 April 2017 in the northwestern portion of South Dakota, where many pilot reports and in-situ eddy dissipation rate measurements reported significant turbulence levels. Full-physics WRF nested large-eddy simulations (LES) at an unprecedented fine grid spacing (Δ_h = 250 m, Δ_v = 100 m) are preformed to systematically investigate the impact of UTLS turbulence modeling on 1-km forecasts by analyzing turbulence-related quantities such as kinetic energy spectra and probability distributions of velocity fluctuations. We demonstrate that the choice of the planetary boundary layer (PBL) parameterization has a strong impact on the presence, structure and intensity of both the modeled gravity waves and turbulent features. An alternative approach that eliminates the model’s vertical diffusion arising from the PBL parameterization in the free troposphere is implemented, which results in a more consistent depiction of turbulence across different PBL parameterizations. Finally, the LES results are used to gain insight into these intricate phenomena, which combine convectively induced turbulence and other type of CAT instabilities related to large-scale forcing developing several hundred kilometers away from the most deep and intense convection.

This research is in response in part to requirements and funding by the Federal Aviation Administration (FAA). The views expressed are those of the authors and do not necessarily represent the official policy or position of the FAA.