Boundary layer parameterizations
–Develop turbulence parameterizations for high-wind boundary layers to advance the forecast skill of hurricane models (e.g., NOAA's HAFS)
A framework for simulating TC boundary layers using large-eddy simulation
Current planetary boundary layer (PBL) parameterizations are generally designed for low-wind boundary layers, and their performance in hurricane boundary layers is not well understood, mostly due to very scarce in-situ turbulence measurements. To overcome this limitation, we developed a modeling framework based on large-eddy simulations to derive profile information of turbulence quantities at hurricane-force wind speeds using specifically designed large-eddy simulations (LES). The novelty of this framework includes the usage of a few input parameters to represent the TC vortex and the addition of a simple nudging term for temperature and moisture to account for the complex thermodynamic processes in TC boundary layers. With this special setup, the LES profiles match well with existing turbulence measurements. The utility of this framework is further highlighted by evaluating a first-order PBL parameterization, suggesting that an asymptotic turbulence length scale of 40 m produces a good match to LES results.
How do PBL schemes perform in hurricane conditions?
This special setup of the developed modeling framework mentioned earlier allows an "apple-to-apple" comparison of different PBL schemes under the same controlled situations. Using this framework and LES output, we evaluated two types of PBL schemes under realistic hurricane conditions. The framework reveals the pros and cons of each PBL scheme.
Key takeaways:
K-profile parameterization schemes are inherently flawed in hurricane boundary layers
The turbulence-kinetic-energy (TKE)-based Mellor–Yamada–Nakanishi–Niino (MYNN) scheme performs well in hurricane conditions due to its high-order closure and sophisticated parameterizations of mixing length
The TKE-based Eddy-Diffusivity Mass-Flux (EDMF-TKE) scheme, used in NOAA's HAFS model, produces excessive vertical mixing in hurricane conditions compared to LES results
Develop PBL parameterizations for hurricane models
We used the developed modeling framework to address the existing issues of the EDMF-TKE scheme in hurricane conditions. Specifically, we improved the EDMF-TKE scheme by matching the surface layer and PBL parameterizations, by using a smaller maximum allowable mixing length based on LES results and observations, and by adopting a new definition of boundary layer height that works better in high-wind conditions. The modified EDMF-TKE can reproduce the LES profiles of effective eddy viscosity and turbulence kinetic energy; it also shows promise to improve forecast skill of rapid intensification (RI) (see the left example of ensemble HAFS forecasts of Hurricane Michael 2018).
Verification of earth-relative wind speeds at 2-km height (upper) and azimuthal-mean tangential wind (lower) from HAFS against tail Doppler radar (TDR) data from a NOAA P-3 mission into Hurricane Ida on 29 August 2021
Improved PBL schemes advance the skill of NOAA's next-generation hurricane forecast model
Beyond the case study, we examined the original (experiment HAFA) and modified (experiment HAFY) EDMF-TKE schemes in NOAA’s Hurricane Analysis and Forecast System (HAFS) during the entire 2021 North Atlantic hurricane season.
Key takeaways:
Intensity and structure forecasts from HAFY were 10-15% better than those from HAFA, especially early in each forecast. This was because the boundary layer inflow was stronger and more realistic in HAFY than in HAFA.
HAFY was also able to better forecast rapid intensification and produced a deeper and more realistic circulation (left figure) than HAFA.
The modified EDMF-TKE scheme has been implemented into the operational HAFS in 2023.
Eyewall turbulence mixing is important as well
Current PBL schemes assume horizontal homogeneity and neglect vertical motions; thus, advection of turbulence kinetic energy (TKE) by grid-scale mean flow is typically neglected. However, TC boundary layers are inhomogeneous, especially in the eyewall. We investigated the effect of TKE advection on TC simulations using an improved MYNN scheme, where we could turn the advection on and off.
Key takeaways: TKE advection should be included in TC modeling using a high-order PBL scheme
It helps produce a more realistic, deep eyewall "TKE column" (left figure) against observations
Large TKE values above the PBL in the eyewall are predominantly due to vertical advection of TKE, rather than buoyancy production of TKE.
The enhanced eyewall vertical mixing leads to a slightly stronger and smaller TC.
Effect of PBL parameterizations on TC modeling in the gray zone
Conventional planetary boundary layer (PBL) parameterization schemes start to violate basic design assumptions when horizontal grid spacings of numerical models approach O(1) km and become comparable with the dominant length scales of boundary-layer flows (i.e., “gray zones”). We investigated the effect of a new scale-aware (Shin-Hong – SH) PBL scheme on TC modeling compared to the traditional non-scale-aware version (Yonsei University – YSU) it is built upon by performing subkilometer-resolution WRF simulations.
Key takeaways:
The SH PBL scheme tends to forecast a stronger TC with a smaller core than the YSU PBL scheme
The scale awareness in SH comes into effect when the PBL depth is bigger than the horizontal grid spacing. When this happens, the strength of the turbulent mixing (especially the nonlocal fluxes) in the PBL decreases.
The scale-aware effect is most prominent in the nonprecipitation regions radially outside of a vortex-tilt-related convective rainband, and from the early stage through the RI phase.
Comparison of near-surface inflow angle (upper) and inflow layer depth (lower) between simulations with wind-based (Orange) and stratification-based (B/G/R) approaches and observations (gray).
Unifying turbulence parameterizations across different PBL regimes
Eddy-diffusivity mass-flux (EDMF) type PBL schemes are widely used in global and regional models (e.g., GFS and HAFS). Surface-buoyancy-driven mass fluxes (MF) represent thermal plumes in convective boundary layers (CBL) and how to properly parameterize MF in nearly neutral hurricane boundary layers (HBL) remains a challenge. Strong shear in HBLs can distort and dampen thermal plumes and thereby weaken the MF. To mimic this effect and consider the fact that convective plumes are intrinsic to CBLs, this study proposes a new approach by turning off the MF in non-convective boundary layers based on stratification (i.e., the ratio of shear forcing to the buoyancy forcing).
Key takeaways:
Compared to a traditional approach of MF tapering based on 10-m wind speeds, the new approach is physically more appealing as both shear and buoyancy forcings are considered and the width of the effective tapering zone responds to diurnal variations of surface buoyancy forcing.
Using either approach of MF tapering can lead to a stronger and more compact inner core due to enhanced boundary layer inflow outside the RMW; nevertheless, the radius of gale-force wind and inflow layer depth are only notably reduced using the new approach
Comparison with observations shows some support for this new approach. [This new approach has been adopted in HAFS-B and helped suppress the model's overreaction in TC genesis or RI onset. ]