betaFeedback

WorksPhysics and Astronomy

Turbulent Boundary Layer Height Scales in Hurricanes

Abstract

Boundary layer processes drive the air-sea exchange of momentum, heat, and moisture that powers and shapes hurricanes. The height of the boundary layer is a critical parameter in engineering and meteorological models of hurricane wind speed, turbulence intensity, and storm strength. Existing models rely on a height scale derived with the assumption of a constant eddy viscosity, a strong simplification that limits physical accuracy. This work proposes formulae for the turbulent boundary layer height in hurricanes outside the eyewall. The proposed scalings are u⋆/βu_\star/β for neutral stratification, and u⋆/βNu_\star/\sqrt{βN} for stable stratification, where u⋆u_\star is the friction velocity, ββ is the absolute fluid vorticity and N is the Brunt-Vaisala frequency of the background stratification. These scalings are analogous to those used in the literature for neutrally and stably stratified turbulent atmospheric boundary layers. The formulae are backed by analytical derivation and validated against velocity profiles from large-eddy simulations and field observations. They are predictive to within 2.5% relative error on average and yield a good collapse of the simulated and observational velocity profiles away from the surface. The results further enable quantitative relationships between boundary layer height and other characteristic scales, including the height of maximum wind speed and the depth of the inflow layer. The proposed expressions offer a practical basis for interpreting observational data, informing mesoscale simulations, and specifying turbulent flow statistics in wind engineering and coastal resilience.

Corresponding author: M.G. Giometto, [email protected]

Introduction

Boundary layer processes in hurricanes are of interest for applications in wind engineering, wind energy, coastal resilience and meteorology. A key parameter in the functional form of wind profiles and turbulent statistics is the boundary layer (BL) height. Analytically, it is well established for the hurricane boundary layer (HBL) idealized with a constant eddy viscosity KK that the HBL height scales as 2K/I\sqrt{2K/I} where II is the inertial stability parameter [45, 32, 62].

However, there remains a significant gap in the literature when attempting to connect this scaling with models that are practically applicable. In the context of weather forecasting, predictions of hurricane intensity are strongly dependent on turbulence closure models. There are therefore several papers discussing the intricacies of the closure model [22, 50, 66, 35, 27, 26, 76, 78, 11, 12, 10, 14, 60, 43]. As an initial step toward a more accurate closure, Zhang et al. [77] and Kepert [35] contend that a dynamical boundary layer model offers a more representative description than a thermodynamical formulation, with the latter continuing to underpin many mesoscale closure models. Almost all popular dynamical models used in practice are a function of the height of the boundary layer. Chen [10] studied the simulation of various features of HBL winds such as the velocity profiles, eddy viscosity and total stress using four popular mesoscale closure models. It was found that the primary source of variation is due to the differences between the models in the prediction of the HBL height. In wind engineering too, proposed wind profile models, for example by Vickery et al. [72], use HBL height based on a constant eddy viscosity. It would be desirable to inform these models with hurricane-relevant parameters such as radial distance, and replace the existing reliance on a constant eddy viscosity with more practical parameters such as the friction velocity u⋆u_\star.

The literature addressing the atmospheric boundary layer (ABL) is grounded in a comparatively more developed framework. The original proposition by Ekman [19] for the constant eddy viscosity scaling has been extended to a commonly accepted turbulent scaling u⋆/fu_\star/f for neutrally stratified flows [4, 70]. Since then, there have also been well-accepted modifications to the height scale to include an inversion layer, and for stable and unstable stratification. [79, 56, 31, 80] (Kitaigorodskii 1988; Kitaigorodskii and Joffre 1988). A good review of the historical progress is given in the introduction section of Narasimhan et al. [49].

This gap in the literature for the HBL can be attributed to the historical inability to perform turbulence-resolving simulations for the HBL. The sensitivity analyses in such studies require numerous turbulence-resolving simulations. However, to simulate the entire hurricane system across its large range of scales, even a single simulation would take several months on a dedicated cluster [7]. To circumvent this, Nakanishi and Niino [48] proposed a method in which large-eddy simulations (LESs) of the HBL could be conducted in a channel-flow-like setup to resolve a small but representative part of the HBL by including the large-scale rotational effects in the governing equations. Modifications to this methodology were proposed by Green and Zhang [28] and an influential paper by Bryan et al. [8]. This has spurred several follow-up papers that target various related applications [73, 11, 12, 47, 15, 61, 59] (Ma and Sun 2021). While this technique is limited to regions outside the eyewall, as it neglects average vertical advection, it now allows for LES of the HBL to be conducted at a reasonable computational cost. In particular, one can obtain reliable predictions of the friction velocity and the HBL height, as well as sensitivities to input parameters.

In this work, we draw inspiration from the literature on neutrally and stably stratified turbulent atmospheric boundary layers, as well as the analytical work on the HBL, to propose a scaling for the HBL height. The scaling is informed and validated using a suite of process-resolving LES of the HBL in the above framework and with available observations.

§ 2 documents the scaling obtained analytically. § 3 compares the proposed formulae with the HBL height obtained directly from the simulations and with wind profiles from field observations of hurricanes. § 4 establishes relations between standard metrics of the boundary layer height, such as radial and tangential jet heights, and discusses further scaling implications. A summary and conclusions are presented in § 5.

Analytical Model for Height Scaling

Background

Most observational profiles of velocity and potential temperature in the HBL exhibit characteristics of stable stratification outside the eyewall and outer rainbands [77, 75, 46]. Stable stratification must hence be accounted for when interpreting observational and mesoscale simulation data. The primary analytical derivation and validation are performed with simulations and observations possessing stable stratification. Validation of the formula for neutral stratification is described in the Appendix.

In the presence of stable stratification, it is necessary to account for the competing influence of buoyancy, which acts to inhibit vertical motion and oppose inertial forces. Stable stratification is typically quantified by a surface cooling flux BsB_s [79], a constant cooler surface temperature, or a surface cooling rate [49]. In a hurricane, however, very near the surface, there is unstable stratification due to the presence of warm ocean water below cooler air, and stable stratification above due to latent heat release from cloud formation coupled with a gradual subsidence outside the eyewall. This switch in stratification from unstable to stable, therefore, involves complex coupling between surface heat flux, evaporation and condensation, radiation, etc. Additionally, there is the practical difficulty of obtaining accurate measurements of heat flux near the surface during a hurricane. Stratification strength is therefore often characterized by the profile of potential temperature alone. For example, Chen et al. [13] nudge to a known potential temperature profile to circumvent modeling the complex physics of radiation and moisture. We therefore attempt to analytically characterize stratification in a similar manner.

Monin-Obukhov theory (MOST) states that Wyngaard [74]

κzθ⋆dθdz=−κzu⋆Hdθdz=ϕh(ζ)=1+7.8z−z0Ls(1)\frac{\kappa z}{\theta_\star}\frac{d\theta}{dz}=-\frac{\kappa z u_\star}{H}\frac{d\theta}{dz}=\phi_h(\zeta)=1+7.8\frac{z-z_0}{L_s} \tag*{(1)}

where the Obukhov length is

Ls=−u⋆3Bs=−u⋆3θ0κgH,(2)L_s=-\frac{u_\star^3}{B_s}=-\frac{u_\star^3\theta_0}{\kappa g H}, \tag*{(2)}

BsB_s is the dynamic heat flux and HH is the kinematic heat flux. If we choose a sufficiently large zz, we can write

−κzu∗Hdθdz=−7.8zκgHu∗3θ0.(3)-\frac{\kappa z u_*}{H}\frac{d\theta}{dz}=-7.8\frac{z\kappa gH}{u_*^3\theta_0}. \tag*{(3)}

Solving for HH and hence for LsL_s, we have

Ls=u∗κ7.8θ0g dθ/dz=u∗N7.8κ,(4)L_s=\frac{u_*}{\kappa}\sqrt{\frac{7.8\theta_0}{g\,d\theta/dz}}=\frac{u_*}{N}\frac{\sqrt{7.8}}{\kappa}, \tag*{(4)}

where NN is the Brunt-Väisälä frequency. Thus, the Obukhov length can be expressed roughly as ∼u∗/N\sim u_*/N, where we assume NN is a constant for simplicity. Kitaigorodskii (1988) uses a similar characterization and refers to this in his work as “imposed stable stratification” or “background stratification”.

The time-averaged, linearized HBL equations (Kepert [32]; Sathia and Giometto [62]) are

α(Vg−v)=ddz(τxz),(5)\alpha(V_g-v)=\frac{d}{dz}(\tau_{xz}), \tag*{(5)}
βu=ddz(τyz),(6)\beta u=\frac{d}{dz}(\tau_{yz}), \tag*{(6)}

where uu and vv are the radial and tangential components of the velocity, τxz\tau_{xz} and τyz\tau_{yz} are the components of the stress along these directions, VgV_g is the gradient wind speed, ff is the Coriolis frequency, nn is the normalized radial derivative of the gradient wind −RVg∂Vg∂r-\frac{R}{V_g}\frac{\partial V_g}{\partial r} and

α=f+2VgR,β=f+(1−n)VgR.(7)\alpha=f+\frac{2V_g}{R},\qquad\beta=f+\frac{(1-n)V_g}{R}. \tag*{(7)}

α\alpha is twice the absolute angular velocity and β\beta is the absolute fluid vorticity (Smith and Montgomery [65]; Sous et al. [67]).

A typical approach to determining the boundary-layer height in the ABL relies on a mixing-length argument. Consider representative length scales

∣Vg−v∣∼v~,∣u∣∼u~.(8)|V_g-v|\sim\tilde{v},\qquad|u|\sim\tilde{u}. \tag*{(8)}

Stresses are modeled using an eddy viscosity KmK_m, which is parameterized as Km∼u⋆lTK_m \sim u_\star l_T, where u⋆u_\star is the friction velocity, and lTl_T is the mixing length, a turbulent length scale representative of the largest eddies within the boundary layer [80]. This gives us

αv~∼u⋆lTh2u~,βu~∼u⋆lTh2v~.(9)\alpha\tilde{v} \sim\frac{u_\star l_T}{h^2}\tilde{u}, \qquad\beta\tilde{u} \sim\frac{u_\star l_T}{h^2}\tilde{v}. \tag*{(9)}

Typical choices for lTl_T are hh under neutral stratification, and −u⋆3/Bs∼u⋆/N-u_\star^3/B_s \sim u_\star/N under stable stratification [80]. Substituting and simplifying, we obtain h∼u⋆/Ih \sim u_\star/I under neutral stratification and h∼u⋆/INh \sim u_\star/\sqrt{IN} under stable stratification, where II is the inertial frequency [32].

I=αβ=(f+2VgR)(f+(1−n)VgR).(10)I = \sqrt{\alpha\beta} = \sqrt{\left(f+\frac{2V_g}{R}\right)\left(f+\frac{(1-n)V_g}{R}\right)}. \tag*{(10)}

Recall that the constant eddy-viscosity scaling for the HBL is 2K/I\sqrt{2K/I} and that the turbulent ABL scalings are u⋆/fu_\star/f and u⋆/fNu_\star/\sqrt{fN}. The formulae obtained above from the mixing-length argument thus seem to be a reasonable, natural extension to the constant eddy viscosity expression. These, however, are inconsistent with our findings. The formulae we propose instead are

h=CRu⋆β(11)h = C_R\frac{u_\star}{\beta} \tag*{(11)}

under neutral stratification, and

h=CSu⋆βN(12)h = C_S\frac{u_\star}{\sqrt{\beta N}} \tag*{(12)}

under stable stratification. The following subsection derives the expression under stable stratification. While we do not possess a derivation for the neutral stratification case, the proposed formula is the same as the empirical suggestion by Sous et al. [67] for a spin-down flow in a rotating tank. It is also provides accurate predictions of the boundary layer height obtained from our LES runs, and is described further in the Appendix.

Derivation

We follow the derivation of Pollard et al. [56] (henceforth P73). P73 had derived an expression for the deepening of the upper ocean mixed layer, due to an imposed wind stress, against quiescent, stably stratified ocean water beneath. We derive a similar expression for the deepening of the stratified HBL due to the gradient wind and compare it to the one in P73.

Starting from (5) and (6) but retaining the unsteady terms, we have

dudt+α(Vg−v)=ddz(τxz),\frac{du}{dt}+\alpha(V_g-v)=\frac{d}{dz}(\tau_{xz}),
dvdt+βu=ddz(τyz).\frac{dv}{dt}+\beta u=\frac{d}{dz}(\tau_{yz}).

Integrating from 0 to hh to obtain a slab model [33], the equations become

d(huˉ)dt−u∣hdhdt+αh(Vg−vˉ)=τxz∣h−τxz∣0\frac{d(h\bar{u})}{dt}-\left.u\right|_h\frac{dh}{dt}+\alpha h(V_g-\bar{v})=\left.\tau_{xz}\right|_h-\left.\tau_{xz}\right|_0
d(hvˉ)dt−v∣hdhdt+βhuˉ=τyz∣h−τyz∣0\frac{d(h\bar{v})}{dt}-\left.v\right|_h\frac{dh}{dt}+\beta h\bar{u}=\left.\tau_{yz}\right|_h-\left.\tau_{yz}\right|_0

Here we have used the Leibniz rule

ddt∫0h(t)u(t,z) dz=∫0h(t)∂u∂t dz+u(t,h)dhdt.\frac{d}{dt}\int_0^{h(t)}u(t,z)\,dz=\int_0^{h(t)}\frac{\partial u}{\partial t}\,dz+u(t,h)\frac{dh}{dt}.

At the surface, we use a standard drag parameterization [33]

τxz∣0=CDuˉuˉ2+vˉ2,\left.\tau_{xz}\right|_0=C_D\bar{u}\sqrt{\bar{u}^{2}+\bar{v}^{2}},
τyz∣0=CDvˉuˉ2+vˉ2.\left.\tau_{yz}\right|_0=C_D\bar{v}\sqrt{\bar{u}^{2}+\bar{v}^{2}}.

We first linearise about uˉ=0\bar{u}=0 and vˉ=Vg\bar{v}=V_g to obtain

τxz∣0=CDVguˉ,\left.\tau_{xz}\right|_0=C_DV_g\bar{u},
τyz∣0=CDVg2+2CDVg(vˉ−Vg).\left.\tau_{yz}\right|_0=C_DV_g^2+2C_DV_g(\bar{v}-V_g).

The terms CDVguˉC_DV_g\bar{u} and 2CDVg(vˉ−Vg)2C_DV_g(\bar{v}-V_g) correspond to linear Rayleigh damping. For analytical convenience, we drop the damping terms and model the surface stresses as

τxz∣0=0,τyz∣0=CDVg2.\left.\tau_{xz}\right|_0=0,\qquad\left.\tau_{yz}\right|_0=C_DV_g^2.

We later integrate numerically, retaining the surface stresses in the form of (18) and (19) to show that this is a robust approximation for this derivation.

At the top of the HBL, the stress is due to the frictional forces from the momentum exchange at the interface between the HBL and the ambient gradient wind region [29]. We neglect this stress.

τxz∣h=0,τyz∣h=0.(13)\left.\tau_{xz}\right|_{h}=0,\qquad\left.\tau_{yz}\right|_{h}=0. \tag*{(13)}

Substituting, we obtain

du~dt−αv~=0,(14)\frac{d\tilde{u}}{dt}-\alpha\tilde{v}=0, \tag*{(14)}
dv~dt+βu~=−CDVg2.(15)\frac{d\tilde{v}}{dt}+\beta\tilde{u}=-C_DV_g^2. \tag*{(15)}

where u~=huˉ\tilde{u}=h\bar{u} and v~=h(vˉ−Vg)\tilde{v}=h(\bar{v}-V_g). Solving with initial conditions u~=0\tilde{u}=0, v~=0\tilde{v}=0, we get

huˉ(t)=−αβCDVg2I(1−cos⁡(It)),(16)h\bar{u}(t)=-\sqrt{\frac{\alpha}{\beta}\frac{C_DV_g^2}{I}}\left(1-\cos(It)\right), \tag*{(16)}
h(vˉ−Vg)(t)=−CDVg2Isin⁡(It).(17)h(\bar{v}-V_g)(t)=-\frac{C_DV_g^2}{I}\sin(It). \tag*{(17)}

For the stratified ABL, we have α=β=f\alpha=\beta=f, and these reduce to the same expressions obtained in P73.

To close the system, we require an additional equation, for which we use the bulk Richardson number

Ri⁡b=gθ0dθdxh2(Δu)2.(18)\operatorname{Ri}_{b}=\frac{g}{\theta_0}\frac{d\theta}{dx}\frac{h^2}{(\Delta u)^2}. \tag*{(18)}

Shear is the main driver of turbulent mixing, whereas stable stratification suppresses it through buoyancy. When Ri⁡b\operatorname{Ri}_{b} reaches a critical value Ri⁡c\operatorname{Ri}_{c}, shear production balances buoyancy destruction. Once this balance is attained, turbulent entrainment ceases, and the height of the turbulent boundary layer (HBL) no longer increases. In the analytical model, following P73, we impose Ri⁡b=Ri⁡c\operatorname{Ri}_{b}=\operatorname{Ri}_{c} throughout the growth phase, so that the evolving boundary layer is treated as remaining in this marginally balanced state up to the point where growth terminates. This closure condition yields

h2=RicN2(uˉ2+(vˉ−Vg)2).(19)h^2 = \frac{\mathrm{Ri}_c}{N^2}\left(\bar{u}^2 + (\bar{v} - V_g)^2\right). \tag*{(19)}

Squaring and adding (16) and (17), and substituting in (19), we obtain

h2uˉ2+h2(vˉ−Vg)2=N2Rich4=CD2Vg4(1β2(1−cos⁡(It))2+1I2sin⁡(It)2).(20)h^2\bar{u}^2 + h^2(\bar{v} - V_g)^2 = \frac{N^2}{\mathrm{Ri}_c}h^4 = C_D^2V_g^4\left(\frac{1}{\beta^2}(1-\cos(It))^2 + \frac{1}{I^2}\sin(It)^2\right). \tag*{(20)}

When It=πIt = \pi, the HBL stops growing and (20) reaches its maximum value

h=(2Ric0.25)CDVg2βN.(21)h = \left(\sqrt{2}\mathrm{Ri}_c^{0.25}\right)\sqrt{\frac{C_DV_g^2}{\beta N}}. \tag*{(21)}

CDVg2C_DV_g^2 can be written equivalently as u⋆2u_\star^2, and so we have

h=CSu⋆βN.(22)\boxed{h = C_S\frac{u_\star}{\sqrt{\beta N}}}. \tag*{(22)}

where CSC_S is a tuning constant. We show in § 3 that, from LES results, CS=1.2C_S = 1.2. Thus, the parameter in the denominator ends up being β\beta instead of I=αβI = \sqrt{\alpha\beta}.

Once again, as a sanity check, we confirm that as R→∞R \to\infty, we have β→f\beta\to f and (22) reduces to u⋆/fNu_\star/\sqrt{fN} which is the well accepted expression for the stratified ABL [56, 80]. The two main approximations introduced above for analytical tractability are (a) the simplification of the stress terms, going from (18), (19) to (22), and (b) the assumption that the Richardson number remains fixed at its critical value. To verify that these approximations do not substantially affect the resulting scaling, we numerically integrate the following equations.

duˉdt+α(Vg−vˉ)=−uˉhdhdt−CDuˉhuˉ2+vˉ2,(23)\frac{d\bar{u}}{dt} + \alpha(V_g - \bar{v}) = -\frac{\bar{u}}{h}\frac{dh}{dt} - \frac{C_D\bar{u}}{h}\sqrt{\bar{u}^2+\bar{v}^2}, \tag*{(23)}
dvˉdt+βuˉ=−vˉ−Vghdhdt−CDvˉhuˉ2+vˉ2,(24)\frac{d\bar{v}}{dt} + \beta\bar{u} = -\frac{\bar{v}-V_g}{h}\frac{dh}{dt} - \frac{C_D\bar{v}}{h}\sqrt{\bar{u}^2+\bar{v}^2}, \tag*{(24)}
dhdt=1τ(RicN2(uˉ2+(vˉ−Vg)2)−h).(25)\frac{dh}{dt} = \frac{1}{\tau}\left(\frac{\mathrm{Ri}_c}{N^2}\left(\bar{u}^2 + (\bar{v}-V_g)^2\right)-h\right). \tag*{(25)}

Boundary layer growth is modelled using a simple relaxation equation ([69]), in which h→heqh \to h_{eq}, and heqh_{eq} is defined by (19). We set the characteristic timescale τ=343 s\tau= 343\ \mathrm{s}, corresponding to h/u⋆h/u_\star diagnosed from the LES for the parameter set used in the numerical integration. A representative choice of τ\tau is sufficient since we seek the peak value of hh.

We plot two curves in Figure 1. The purple curve plots hh obtained from (20). The green curve is obtained from the numerical integration. The green curve follows, as expected, roughly the same amplitude and frequency as the purple one, albeit with damping.

Figure 1. Comparison between analytical solution (20) and numerical integration of (23) and (24). Constants used are f=10−4 s−1f = 10^{-4}\ \mathrm{s}^{-1}, N=1.28×10−2 s−1N = 1.28 \times10^{-2}\ \mathrm{s}^{-1}, CD=10−3C_D = 10^{-3}, Vg=45 m s−1V_g = 45\ \mathrm{m\,s}^{-1}, R=40 kmR = 40\ \mathrm{km}, n=0.5n = 0.5, τ=343 s\tau= 343\ \mathrm{s}.

We can attempt to combine (11) and (22) into a single formula. Similar to the approach in Zilitinkevich et al. [80], we propose

h=u⋆((1CRβ)4+(1CSβN)4)1/4.(26)h = \frac{u_\star}{\left(\left(\frac{1}{C_R}\beta\right)^4+\left(\frac{1}{C_S}\sqrt{\beta N}\right)^4\right)^{1/4}}. \tag*{(26)}

Validation for this formula against LES results is provided in the Appendix.

Model Validation

In deriving the scaling in § 2, we did not specify a precise definition of the boundary layer height. We now define the boundary layer height hh explicitly as the height at which the turbulent kinematic stress becomes negligible. Boundary layer height is obtained from the simulations as the height of the first grid point at which the kinematic stress (−u′w′‾+τxzSGS)2+(−v′w′‾+τyzSGS)2\sqrt{\left(-\overline{u'w'}+\tau_{xz}^{SGS}\right)^2+\left(-\overline{v'w'}+\tau_{yz}^{SGS}\right)^2} achieves a value <0.02 u⋆2<0.02\,u_\star^2.

Numerical Simulations

The governing equations solved in the LES are

∂u∂t+(u×(∇×u))x=−∂Φ∂x−f(Vg−v)+(⟨u⟩2R+⟨v⟩2R−Vg2R)−(∇⋅τSGS)x(27)\frac{\partial u}{\partial t}+\left(\mathbf{u}\times(\nabla\times\mathbf{u})\right)_x=-\frac{\partial\Phi}{\partial x}-f(V_g-v)+\left(\frac{\langle u\rangle^2}{R}+\frac{\langle v\rangle^2}{R}-\frac{V_g^2}{R}\right)-\left(\nabla\cdot\boldsymbol{\tau}^{SGS}\right)_x \tag*{(27)}
∂v∂t+(u×(∇×u))y=−∂Φ∂y−fu−(⟨u⟩⟨v⟩R−n⟨u⟩VgR)−(∇⋅τSGS)y(28)\frac{\partial v}{\partial t}+\left(\mathbf{u}\times(\nabla\times\mathbf{u})\right)_y=-\frac{\partial\Phi}{\partial y}-fu-\left(\frac{\langle u\rangle\langle v\rangle}{R}-n\langle u\rangle\frac{V_g}{R}\right)-\left(\nabla\cdot\boldsymbol{\tau}^{SGS}\right)_y \tag*{(28)}
∂w∂t+(u×(∇×u))z=−∂Φ∂z+g(θ−⟨θ⟩)θ0−(∇⋅τSGS)z(29)\frac{\partial w}{\partial t}+\left(\mathbf{u}\times(\nabla\times\mathbf{u})\right)_z=-\frac{\partial\Phi}{\partial z}+g\frac{\left(\theta-\langle\theta\rangle\right)}{\theta_0}-\left(\nabla\cdot\boldsymbol{\tau}^{SGS}\right)_z \tag*{(29)}
∂θ∂t+u⋅∇θ=(θr−⟨θ⟩)τr−(∇⋅πSGS)(30)\frac{\partial\theta}{\partial t}+\mathbf{u}\cdot\nabla\theta=\frac{\left(\theta_r-\langle\theta\rangle\right)}{\tau_r}-\left(\nabla\cdot\boldsymbol{\pi}^{SGS}\right) \tag*{(30)}

where xx is the outward radial, yy is the tangential and zz is the vertical direction.

The primary validation is performed with simulations possessing stable stratification. For these stratified cases, we use the simulation set Sathia and Giometto [63] uploaded to the NHERI DesignSafe Database. Details of the numerical simulations are provided in the Appendix. We use a linearly increasing reference profile θr=300+(5×10−3)z\theta_r=300+(5\times10^{-3})z Kelvin, as described in Chen et al. [13]. Additional simulations are conducted with neutral and mild stratification. These are also described in the SI Appendix.

Note here that the governing equations of the LES include the nonlinear terms, whereas the derivation of the height scales was performed using the linearized form of the equations. We assume that the nonlinearity contributes only a correction to the solution of the linearized equations and follows the same height scale (Kepert and Wang [34]; Foster [22]; Sathia and Giometto [62]).

Observations

To validate the proposed scaling against observations, one requires not only mean velocity profiles but also the corresponding values of RR, nn, and NN. In practice, even obtaining robustly averaged HBL profiles is nontrivial, let alone those conditioned on specific RR or nn. As a preliminary assessment, we therefore use the composite hurricane profiles reported in Bryan et al. [8] and Chen et al. [13], adopting the values of RR and nn provided therein. The vertical shear of the gradient wind, GzG_z, is estimated as the slope of a linear fit to vv above the wind maximum. The gradient wind GG is then computed as the mean of v−Gzzv-G_z z over the same region. These estimates are consistent to leading order with the values reported in the original studies.

The friction velocity is obtained by fitting logarithmic profiles to the near-surface winds (below 50 m), namely u⋆,ylog⁡(z/z0)/κu_{\star,y}\log(z/z_0)/\kappa to vv and u⋆,xlog⁡(z/z0)/κu_{\star,x}\log(z/z_0)/\kappa to uu, and then forming u⋆=u⋆,x2+u⋆,y2u_\star=\sqrt{u_{\star,x}^2+u_{\star,y}^2}. For all five cases, we find u⋆≈u⋆,yu_\star\approx u_{\star,y}.

For the Chen et al. composites, the reported potential temperature profiles correspond to dθ‾/dz≈5 K km−1d\overline{\theta}/dz\approx5\ \mathrm{K}\,\mathrm{km}^{-1}. Although dθ‾/dzd\overline{\theta}/dz is not explicitly available for the Bryan et al. cases, semi-logarithmic plots of vv reveal comparable slopes to those in Chen et al., suggesting similar stratification and thus comparable NN across all profiles. We find that a slightly larger CS=1.44C_S=1.44 better fits the data and use this value for the observational analysis. The parameter values used to estimate hh for each profile are summarized in Table 1.

GG (ms−1^{-1})RR (km)nnff (10−410^{-4} s−1^{-1})GzG_z (10−310^{-3} s−1^{-1})NN (10−210^{-2} s−1^{-1})u⋆u_\star (s−1^{-1})hpredh_{\mathrm{pred}} (m)
Chen_V25 (∗\ast)421100.70.501.21.221212
Chen_V35 (□\square)58750.70.5−4-41.21.361031
Chen_V45 (∘\circ)68400.750.5−3-31.21.34781
Bryan_V40 (∙\bullet)40400.80.5−2.2-2.21.20.93749
Bryan_V60 (△\triangle)59400.730.5−3.3-3.31.21.17710

Table 1. List of parameter values and predicted BL height for observations

Validation

Figure 2 shows the tangential velocity time- and horizontally spatially-averaged (denoted by ⟨⋅⟩t\langle\cdot\rangle_t) from both simulations and observations, demonstrating the dimensional variation to GG, RR, z0z_0, nn and ff. Figure 3 shows the tangential velocity profiles for all 216 LES runs and the observations in normalized form, with the height scaled by CSu⋆/βNC_Su_\star/\sqrt{\beta N}. We see that the LES profiles collapse well under the proposed scaling. The primary variation near the ground for both the tangential (Figure 3) and radial velocities (Figure 4) is due to that of z0z_0. The blue curves are for z0=10−3z_0=10^{-3} m, the green curves for z0=10−2z_0=10^{-2} m and the purple curves for z0=10−1z_0=10^{-1} m. The remaining variation within a color is due to the variations in RR and GG. These variations arise from changes in the friction velocity, driven—in order of decreasing influence—by z0z_0, RR and GG. The parameters ff, nn and GzG_z have a negligible effect on the friction velocity, and therefore collapse almost perfectly into a single profile.

Figure 2. Sensitivity of tangential velocity to input parameters in dimensional form. Purple curves correspond to simulations with z0=10−1z_0 = 10^{-1} m, green to z0=10−2z_0 = 10^{-2} m and blue to z0=10−3z_0 = 10^{-3} m. Symbols correspond to observations (see Table 1).

Tangential velocity in normalized form

Figure 3. Tangential velocity in normalized form.

Figure 4. Radial velocity in normalized form. Dashed lines denote the average of the curves of the same color.

The observations in normalized form also fall within the range of variation of the simulated profiles, and collapse well in the outer region. The dimensional V45V_{45} profile exhibits a pronounced nose and a non-zero radial velocity uu in the outer layer. Both features are characteristic of flow in, or proximate to, the eyewall region—a regime not accounted for in the present derivation. Despite this mismatch in underlying assumptions, the profile nevertheless collapses satisfactorily with the others, and the proposed scaling continues to perform well.

Figure 5 shows a plot of LES-derived HBL height against the prediction from (22). The proposed scaling is highly predictive, with an average relative error of ∼2.5%\sim2.5\% and an average absolute error of ∼18\sim18 m.

Figure 5. Parity plot comparing LES-derived HBL height against prediction from (22). R2=0.99R^2 = 0.99, bias = 1.04 m and RMSE = 23.50 m. Colors encode error magnitude - blue denotes a small error, and red denotes a large error.

Discussion

Having established a robust scaling for the boundary layer depth hh, defined as the height at which the total stress becomes negligible, we can now express other characteristic heights of the velocity profiles in terms of hh.

The height of peak tangential velocity (from Figure 3) and that of the velocity magnitude (not shown) is around 80% of the HBL height, with mild sensitivity in the normalized form to variation in GG and RR. This finding is consistent with the observational study by Zhang et al. [77]. Figure 4 plots the normalized radial velocity variation. There is considerable variation near the radial inflow peak, with the peak height varying between 6–20% of the HBL height. Figure 6 plots a histogram showing the variation of the height of peak tangential and radial velocities, normalized by the boundary layer height hjet/hh_{jet}/h. This ratio, for the tangential velocity peak, lies roughly between 0.65 and 0.85, with a clear peak near 0.8. The curves collapse better near the HBL height. The inflow depth, defined as the first grid point at which the velocity profile attains a positive value, varies between 0.86−0.98 h0.86 - 0.98\ h. A similar characterization can also be made for the tangential velocity. The depth of the tangential velocity profile, defined as the first grid point above its peak where the profile attains a value less than GG, varies between 1.08−1.11 h1.08 - 1.11\ h.

Histograms of radial and tangential jet peak heights

Figure 6. Histograms of radial (purple) and tangential (green) jet peak heights, normalized by the predicted HBL height.

The proposed scaling also allows us to analyse the overall variation of these quantities to radial distance. From (22) we have

h∼u⋆N(f+(1−n)GR).(31)h \sim\frac{u_\star}{\sqrt{N\left(f+(1-n)\frac{G}{R}\right)}} . \tag*{(31)}

Assuming a constant NN and nn (in particular, which do not vary with RR) and neglecting ff, the expression simplifies to

h∼u⋆RG.(32)h \sim u_\star\sqrt{\frac{R}{G}} . \tag*{(32)}

Typically, u⋆∼Gu_\star\sim G. u⋆u_\star is, as discussed in § 3, somewhat sensitive to z0z_0 and to a lesser extent to RR, and is relatively insensitive to all other parameters. We assume that z0z_0 stays constant with RR, and neglect the small dependence of u⋆u_\star to RR. Additionally, we substitute the standard assumption G∼R−nG \sim R^{-n} ([8]). Substituting these in (32), we obtain that the boundary layer height varies with RR as

h∼R1−n2.(33)h \sim R^{\frac{1-n}{2}} . \tag*{(33)}

Since the radial and tangential jet heights are simply fractions of the HBL height, this scaling must also hold true for them. This scaling can also be contrasted against that for neutral stratification where h∼u⋆/β∼Rh \sim u_\star/\beta\sim R.

We can similarly interpret the variation of eddy viscosity with RR. The eddy viscosity typically scales as Km∼u⋆lTK_m \sim u_\star l_T where lTl_T is the mixing length. Typically, the background stratification (characterized by u⋆/Nu_\star/N) puts a stronger limit on lTl_T than the rotation (characterized by u⋆/βu_\star/\beta) (Zilitinkevich and Esau 2003). We therefore have

Km∼u⋆2N∼G2∼R−2n.(34)K_m \sim\frac{u_\star^2}{N} \sim G^2 \sim R^{-2n} . \tag*{(34)}

Thus KmK_m decreases with radius at a rate that is twice as fast as the gradient wind speed. This is in contrast to neutral stratification in which Km∼u⋆h∼G2/β∼R1−nK_m \sim u_\star h \sim G^2/\beta\sim R^{1-n} and KmK_m therefore increases with radius.

Observations of the HBL ([68, 77, 75]) typically show that eddy viscosity is large near the eyewall and decreases further away, i.e., decreasing KK as RR increases. This further emphasizes the need to consider a scaling that accounts for stratification when interpreting observational data.

Summary and Conclusions

In this work, we propose a scaling for the hurricane boundary layer height under neutral and stable stratification. These simple scalings draw inspiration from the literature on the analytical treatment of the HBL and on stratified atmospheric boundary layers. We find that the HBL scales as h=CRu⋆/βh = C_R u_\star/\beta under neutral stratification, and CSu⋆/βNC_S u_\star/\beta\sqrt{N} under stable stratification.

We conducted a comprehensive set of turbulence-resolving HBL simulations spanning physically realistic ranges of input parameters to validate the proposed scaling. The LES results suggest that CR=0.58C_R = 0.58 and CS=1.2C_S = 1.2. With these values, the formulae predict the LES-derived HBL heights to within ∼18\sim18 m on average.

The proposed formulae accurately collapse the LES-derived profiles for the nonlinear equations and available observations. This also enables the characterization of other height scales, such as the height of peak wind speed and the depth of the inflow layer. We find that the inflow jet peak occurs at 6–20% of hh and the tangential jet peak occurs at 65–85% of hh. One can further obtain scaling relationships for the HBL height and the eddy viscosity in terms of input parameters such as RR.

The proposed scalings provide a quantitative basis for interpreting observational measurements of hurricane boundary layers, where the depth is often difficult to define consistently due to limited vertical resolution and differing diagnostic criteria. By relating boundary layer height directly to measurable quantities such as surface stress, gradient wind speed, and stratification, the scalings enable a consistent normalization of observed velocity profiles across storms and radial locations, and offer a more practical method for inferring boundary layer depth.

The results also have direct implications for HBL parameterizations in mesoscale and operational models. In such models, boundary layer depth and mixing lengths are typically empirically tuned; the proposed formulae provide physically motivated constraints that parameterizations must satisfy. The scalings provide a basis for prescribing mean wind profiles, eddy viscosity distributions, vertical momentum transport, and stratification effects.

Finally, these findings are significant for wind engineering and coastal resilience applications, where accurate specification of mean wind profiles and turbulence characteristics is essential. The predicted boundary layer height and associated velocity structure are used to define inflow conditions in turbulent inflow generators for studying wind-induced structural response, including tall buildings and wind turbines. The simple yet powerful scalings will help decrease uncertainty in simulated wind structure and intensity in reduced-order and surrogate models used in wind energy assessment and coastal hazard analysis.

More broadly, these results unify the roles of rotation and stratification in hurricane boundary layer structure, and connect theory, simulations, and observations across a wide range of conditions.

Acknowledgments. This research was funded by the National Institute of Standards and Technology (ror.org/05xpvk416) under Grant No. 70NANB22H057. This work also used the Anvil supercomputer at Purdue University through allocation ATM180022 from the Advanced Cyberinfrastructure Coordination Ecosystem: Services & Support (ACCESS) program, National Science Foundation which supports grants #2138259, #2138286, #2138307, #2137603, and #2138296.

Data availability statement. The database of stably stratified simulations is uploaded to the NSF NHERI DesignSafe repository. Profiles of velocity and second-order moments are present therein. Scripts and other datasets generated as part of this study are available from the corresponding author upon reasonable request.

Governing Equations and Numerical Setup

The equations solved in the large-eddy simulations (LES) are

∂u∂t+v(∂u∂y−∂v∂x)+w(∂u∂z−∂w∂x)=−∂Φ∂x−f(Vg−v)+(⟨u⟩2R+⟨v⟩2R−Vg2R)−(∇⋅τSGS)x(35)\frac{\partial u}{\partial t}+v\left(\frac{\partial u}{\partial y}-\frac{\partial v}{\partial x}\right)+w\left(\frac{\partial u}{\partial z}-\frac{\partial w}{\partial x}\right)=-\frac{\partial\Phi}{\partial x}-f(V_g-v)+\left(\frac{\langle u\rangle^2}{R}+\frac{\langle v\rangle^2}{R}-\frac{V_g^2}{R}\right)-(\nabla\cdot\boldsymbol{\tau}^{SGS})_x \tag*{(35)}
∂v∂t+u(∂v∂x−∂u∂y)+w(∂v∂z−∂w∂y)=−∂Φ∂y−fu−(⟨u⟩⟨v⟩R−n⟨u⟩VgR)−(∇⋅τSGS)y(36)\frac{\partial v}{\partial t}+u\left(\frac{\partial v}{\partial x}-\frac{\partial u}{\partial y}\right)+w\left(\frac{\partial v}{\partial z}-\frac{\partial w}{\partial y}\right)=-\frac{\partial\Phi}{\partial y}-fu-\left(\frac{\langle u\rangle\langle v\rangle}{R}-n\langle u\rangle\frac{V_g}{R}\right)-(\nabla\cdot\boldsymbol{\tau}^{SGS})_y \tag*{(36)}
∂w∂t+u(∂w∂x−∂u∂z)+v(∂w∂y−∂v∂z)=−∂Φ∂z+gθ−⟨θ⟩θ0−(∇⋅τSGS)z(37)\frac{\partial w}{\partial t}+u\left(\frac{\partial w}{\partial x}-\frac{\partial u}{\partial z}\right)+v\left(\frac{\partial w}{\partial y}-\frac{\partial v}{\partial z}\right)=-\frac{\partial\Phi}{\partial z}+g\frac{\theta-\langle\theta\rangle}{\theta_0}-(\nabla\cdot\boldsymbol{\tau}^{SGS})_z \tag*{(37)}
∂θ∂t+u∂θ∂x+v∂θ∂y+w∂θ∂z=θr−⟨θ⟩τr−(∇⋅πSGS)(38)\frac{\partial\theta}{\partial t}+u\frac{\partial\theta}{\partial x}+v\frac{\partial\theta}{\partial y}+w\frac{\partial\theta}{\partial z}=\frac{\theta_r-\langle\theta\rangle}{\tau_r}-(\nabla\cdot\boldsymbol{\pi}^{SGS}) \tag*{(38)}

where tt is time, xx is the cross-streamwise or radially outward direction, yy is the streamwise or tangential direction, and zz is the vertical direction. uu is the filtered radial velocity component, vv is the filtered tangential velocity component, and ww is the filtered vertical velocity component. We use the Boussinesq approximation to account for buoyancy. θ\theta is the potential temperature, θ0\theta_0 is a constant reference temperature, which we set to be 300 K and g=9.81 m s−2g=9.81\ \mathrm{m\,s^{-2}}. τSGS\boldsymbol{\tau}^{SGS} is the subgrid-scale (SGS) tensor for velocity, and πSGS\boldsymbol{\pi}^{SGS} the SGS vector for potential temperature. Φ\Phi is a modified filtered pressure field, namely Φ=pρ+13τiiSGS+12uiui\Phi=\frac{p}{\rho}+\frac{1}{3}\tau_{ii}^{SGS}+\frac{1}{2}u_i u_i where ρ\rho is a reference constant density and pp is the pressure deviation from the mean pressure field imposed from the hydrostatic balance and geostrophic forcing. ff is the Coriolis frequency. ⟨⋅⟩\langle\cdot\rangle denotes a horizontal spatial average. The terms in black represent the standard equations of the atmospheric boundary layer, while the terms in blue represent the corrections introduced to account for the large-scale structure of the hurricane.

The equations are solved with the convective terms in their rotational form [53] to ensure the conservation of energy in the inviscid limit. We use the same governing equations as in Momen et al. [47], but with horizontally-averaged terms so as to prevent the growth and transport of spurious oscillations [48, 8]. As in these previous works, we assume that the domain is situated sufficiently far from the hurricane eyewall so that average vertical convection is negligible, and that the horizontal domain extent is small compared to RR.

Following Chen et al. [13], we add a nudging term for potential temperature. The role of the potential temperature is simply to modulate the velocity to its correct value; we do not seek to solve for the potential temperature directly, as this would involve the explicit resolution of complex physics, such as radiation. The reference potential temperature θr\theta_r we specify as a linear function 300+0.005z300 + 0.005z [77, 13]. τr\tau_r is a nudging time scale, which we choose to be 1 minute, following Bryan et al. [8]. We also ran additional simulations increasing and decreasing τr\tau_r by a factor of 5, and did not see any change. We do not explicitly solve for moisture. Its effects could be incorporated by interpreting θ\theta as the virtual potential temperature rather than just the potential temperature.

The lateral boundary conditions are periodic for all variables. A free-lid boundary condition ∂u∂z=∂v∂z=w=∂θ∂z=0\frac{\partial u}{\partial z} = \frac{\partial v}{\partial z} = w = \frac{\partial\theta}{\partial z} = 0 is applied at the top boundary. At the bottom boundary, the vertical velocity is set to zero w=0w = 0, and θ=302\theta= 302 Kelvin is set following Chen et al. [13]. Boundary conditions for the horizontal components and the evaluation of tangential SGS stresses are based on an algebraic wall-layer model based on the equilibrium logarithmic law assumption [16]. It is generally well-accepted that the log-law exists for hurricanes near the surface and sufficiently away from the eye. The roughness length z0z_0 is varied in the simulations to model ocean and land.

High-fidelity Numerical Database

The gradient wind Vg=G+GzzV_g = G + G_z z drives the flow, analogous to the geostrophic wind in the Ekman layer. The domain is centered at distance RR from the hurricane centre. nn is the magnitude of the non-dimensional radial derivative of gradient wind dVgdrRVg\frac{dV_g}{dr}\frac{R}{V_g}, treated as a constant. Together with ff and z0z_0, these comprises 6 input parameters.

Based on preliminary sensitivity analyses, we found sensitivity to VgV_g, RR and z0z_0 greater and hence assigned them three values each. nn, GzG_z and ff are assigned two values each. Based on the physical range of these parameters and their sensitivity, we use the values listed in Table A1. These comprise 216 simulations. The simulation data and averages of the first- and second-order moments have been uploaded to the NSF DesignSafe Data Depot ([63]).

ParameterChosen Values
VgV_g (ms−1^{-1})30, 45, 60
nn0.25, 0.5
GzG_z (s−1^{-1})-0.02, -0.04
ff (s−1^{-1})5×10−55 \times10^{-5}, 1×10−41 \times10^{-4}
RR (km)40, 80, 120
z0z_0 (m)1.00e-03, 1.00e-02, 1.00e-01

Table A1. List of parameters for database with stratification

Numerical Setup

Simulations are performed in a regular rectangular domain of size (2π×2π×2.5)×1000(2\pi\times2\pi\times2.5) \times1000 km using a grid of 256×256×512256 \times256 \times512 collocation nodes in the radial, tangential and vertical directions respectively. The grid size is consistent with the recommendation in Chen et al. [13].

The LES algorithm was initially developed by Albertson and Parlange (Albertson and Parlange [1]). Equations are solved in strong form on a regular domain, and a pseudo-spectral collocation approach (Orszag [51, 52]) based on truncated Fourier expansions is used in the xx, yy coordinate directions. A second-order accurate centered finite differences scheme is adopted in the vertical direction using a staggered grid; the variables uu, vv, Φ\Phi are stored at (j+1/2)Δz(j+1/2)\Delta z, with j=1,nzj=1,n_z. Nonlinear terms are fully dealiased via the 3/2 rule, to avoid piling up energy in the high wavenumber range ([39, 9]). SGS stresses in the bulk of the flow are parameterized using the scale-dependent Lagrangian dynamic Smagorinsky model (Bou-Zeid et al. [5]). Over the past two decades, this solver has been used to develop a series of algebraic SGS closure models for the bulk of turbulent flows [44, 58, 57, 5] (Lu and Porté-Agel 2010), wall-layer models [30], and immersed-boundary methods to accurately represent solid-gas interfaces [71, 17, 20] (Li et al. 2016). It has also been extensively validated against field and laboratory measurements and used to gain insight into a range of applications involving different flow phenomena, spanning from atmospheric boundary layer flow over flat surfaces to flow over urban areas and forests [71, 5, 6, 20, 64, 55, 21, 3, 25, 54, 23 b]. In the context of the hurricane boundary layer (HBL), the solver has been used by Momen et al. [47] and Sabet et al. [61] where it has also been validated against observational data.

Simulations are initiated using the uniform fields VgV_g and θr\theta_r plus random noise. Flow is assumed to be in a fully rough aerodynamic regime, and viscous stresses can be safely neglected. The turbulent Prandtl number is set to Pr=0.7Pr = 0.7 (although this has a negligible effect since we are using potential temperature nudging). A sponge layer with a Rayleigh damping coefficient of 0.01 s−10.01\ \mathrm{s}^{-1} is used above 2000 m for velocity and temperature.

Time integration is performed via a fully explicit second-order accurate Adams-Bashforth scheme, and a fractional step method is adopted to compute the pressure field [18, 36]. A time-step based on CFL=0.075CFL = 0.075 is used. Simulations are run for 1.5TI1.5T_I where TI=2π/IT_I = 2\pi/I is the inertial period and II is the inertial frequency [32]. 1000 instantaneous velocity and potential temperature snapshots are collected (skipping the first 0.5TI0.5T_I to achieve statistical stationarity).

Validation of proposed formula for neutral and mildly stable stratification

In addition to the stably stratified simulations Sathia and Giometto [63], we run simulations with neutral and mildly stable stratification. These are conducted with the aim of validating the proposed scaling for neutral stratification and to propose a formula that connects the two scalings for mildly stable stratification.

Simulations

We run cases with small n=0n = 0, 0.25 for which the total stress profile the Nieuwstadt profile (1−z/h)1.5(1-z/h)^{1.5}. For mild and neutral stratification, the stresses follow this profile closely up to z/h≈0.8z/h \approx0.8, above which they approach zero very slowly. These simulations, therefore, require a much taller domain, and so the remaining parameters were chosen such that the predicted HBL height is relatively small. Additionally, this means that the criterion to obtain HBL height used for the stratified simulations (i.e. 2% of u⋆2u_\star^2) does not work well. Instead we use the approach described in Narasimhan et al. [49], where we assume the Nieuwstadt profile holds for the stress, and fit a straight line to (1−τtotal/u⋆2)1/1.5(1-\tau_{\mathrm{total}}/u_\star^2)^{1/1.5}. The HBL height is then obtained as the inverse of the slope. We verified that for the stratified cases with small n=0.25n = 0.25, this criterion works equally well.

We ran eight simulations with neutral stratification and eight simulations varying the stratification strength NN. The values of NN are chosen such that they are in geometric progression from 10−410^{-4} to 10−210^{-2} s−1^{-1}. A summary listing the various simulations conducted is given in Table B1. All the numerical details are the same as for simulations with stable stratification Sathia and Giometto [63].

GG (ms−1^{-1})RR (km)z0z_0 (m)nnff (s−1^{-1})GzG_z (10−310^{-3} s−1^{-1})NN (s−1^{-1})Total Runs
Set A45, 6030, 350.0010, 0.2510−410^{-4}−2-2–8
Set B45350.001010−410^{-4}−2-210{−4.00, −3.71, −3.43, −3.14, −2.86, −2.57, −2.29, −2.00}10^{\{-4.00,\,-3.71,\,-3.43,\,-3.14,\,-2.86,\,-2.57,\,-2.29,\,-2.00\}}8

Table B1. List of neutrally and mildly stably stratified LES runs

Validation

Sous et al. [67] had performed experiments of spin-up and spin-down in a rotating tank and suggested a revised scaling of the boundary layer height. They recommended u⋆/βu_\star/\beta as a simple empirical modification to account for the fluid rotation, since the fluid feels the absolute vorticity β\beta instead of the wall vorticity ff. Figure B1 shows the profiles for the 8 curves of Set A in Table B1 in normalized form with the height scaled by the prediction hpred=CRRu⋆/βh_{\mathrm{pred}}=C_R R u_\star/\beta, with CR=0.58C_R = 0.58. Figure B2 shows a parity plot of the simulations with neutral stratification against the predictions. We see from these figures that the predicted height scale works well, with an average error of 6%.

Figure B1. Tangential and radial velocity in normalized form for neutrally stratified simulations (Set A of Table B1)

Parity plot comparing LES-derived HBL height against prediction

Figure B2. Parity plot comparing LES-derived HBL height against prediction CRu⋆/βC_R u_\star/\beta. R2=0.99R^2 = 0.99, bias = 2.93 m and RMSE = 6.92 m.

To fit a curve that connects the neutrally and stably stratified scalings, we use the approach described in Zilitinkevich et al. [80].

(1h)p=(βCRu⋆)p+(βNCSu⋆)p(39)\left(\frac{1}{h}\right)^p = \left(\frac{\beta}{C_R u_\star}\right)^p + \left(\frac{\sqrt{\beta N}}{C_S u_\star}\right)^p \tag*{(39)}

Figure B3 plots with symbols the boundary layer heights for various values of NN obtained from the simulation Set B in Table B1. The purple line is the prediction from (B1), using p=4p = 4. While

Figure B3. LES-derived (symbols) and predicted (line) HBL heights for mildly stably stratified simulations.

Zilitinkevich et al. [80] suggested p=2p = 2 for the ABL, the figure indicates that a larger value, p=4p = 4 provides a better fit. This is possibly because ff is much smaller than typical NN whereas β\beta is more comparable to NN.

References

  1. [1]Albertson, J. D., and M. B. Parlange, 1999a: Natural integration of scalar fluxes from complex terrain. Advances in Water Resources, 23, 239–252.DOI
  2. [3]Anderson, W., Q. Li, and E. Bou-Zeid, 2015: Numerical simulation of flow over urban-like topographies and evaluation of turbulence temporal attributes. Journal of Turbulence, 16, 809–831.DOI
  3. [4]Blackadar, A. K., and H. Tennekes, 1968: Asymptotic similarity in neutral barotropic planetary boundary layers. Journal of Atmospheric Sciences, 25 (6), 1015–1020.doi.org/10.1175/1520-0469(1968)025<1015:asinbp>2.0.co;2
  4. [5]Bou-Zeid, E., C. Meneveau, and M. Parlange, 2005: A scale-dependent lagrangian dynamic model for large eddy simulation of complex turbulent flows. Physics of Fluids, 17, 025 105.
  5. [6]Bou-Zeid, E., J. Overney, B. D. Rogers, and M. B. Parlange, 2009: The effects of building representation and clustering in large-eddy simulations of flows in urban canopies. Boundary-Layer Meteorology, 132, 415–436.
  6. [7]Bryan, G. H., N. A. Dahl, D. S. Nolan, and R. Rotunno, 2017a: An eddy injection method for large-eddy simulations of tornado-like vortices. Monthly Weather Review, 145 (5), 1937–1961.
  7. [8]Bryan, G. H., R. P. Worsnop, J. K. Lundquist, and J. A. Zhang, 2017b: A simple method for simulating wind profiles in the boundary layer of tropical cyclones. Boundary-Layer Meteorology, 162 (3), 475–502.DOI
  8. [9]Canuto, C., M. Y. Hussaini, A. Quarteroni, and T. A. Zang, 2006: Spectral methods. scientific computation. Springer-Verlag, Berlin.
  9. [10]Chen, X., 2022: How do planetary boundary layer schemes perform in hurricane conditions: A comparison with large-eddy simulations. Journal of Advances in Modeling Earth Systems, 14 (10), e2022MS003 088.DOI
  10. [11]Chen, X., and G. H. Bryan, 2021: Role of advection of parameterized turbulence kinetic energy in idealized tropical cyclone simulations. Journal of the Atmospheric Sciences, 78 (11), 3593–3611.DOI
  11. [12]Chen, X., G. H. Bryan, A. Hazelton, F. D. Marks, and P. Fitzpatrick, 2022: Evaluation and improvement of a tke-based eddy-diffusivity mass-flux (edmf) planetary boundary layer scheme in hurricane conditions. Weather and Forecasting, 37 (6), 935–951.DOI
  12. [13]Chen, X., G. H. Bryan, J. A. Zhang, J. J. Cione, and F. D. Marks, 2021a: A framework for simulating the tropical cyclone boundary layer using large-eddy simulation and its use in evaluating pbl parameterizations. Journal of the Atmospheric Sciences, 78 (11), 3559–3574.DOI
  13. [14]Chen, X., A. Hazelton, F. D. Marks, G. J. Alaka Jr, and C. Zhang, 2023: Performance of an improved tke-based eddy-diffusivity mass-flux (edmf) pbl scheme in 2021 hurricane forecasts from the hurricane analysis and forecast system. Weather and Forecasting, 38 (2), 321–336.DOI
  14. [15]Chen, X., M. Xue, B. Zhou, J. Fang, J. A. Zhang, and F. D. Marks, 2021b: Effect of scale-aware planetary boundary layer schemes on tropical cyclone intensification and structural changes in the gray zone. Monthly Weather Review, 149 (7), 2079–2095.DOI
  15. [16]Chester, S., C. Meneveau, and M. B. Parlange, 2007a: Modeling turbulent flow over fractal trees with renormalized numerical simulation. J. Comput. Phys., 225, 427–448.
  16. [17]Chester, S., C. Meneveau, and M. B. Parlange, 2007b: Modeling turbulent flow over fractal trees with renormalized numerical simulation. Journal of Computational Physics, 225, 427–448.
  17. [18]Chorin, A. J., 1968: Numerical solution of the navier-stokes equations. Mathematics of Computation, 22, 745–762.
  18. [19]Ekman, V. W., 1905: On the influence of the earth’s rotation on ocean-currents. Arkiv för Matematik, Astronomi och Fysik.
  19. [20]Fang, J., M. Diebold, C. Higgins, and M. B. Parlange, 2011: Towards oscillation-free implementation of the immersed boundary method with spectral-like methods. Journal of Computational Physics, 230, 8179–8191.DOI
  20. [21]Fang, J., and F. Porté-Agel, 2015: Large-eddy simulation of very-large-scale motions in the neutrally stratified atmospheric boundary layer. Boundary-Layer Meteorology, 155, 397–416.
  21. [22]Foster, R. C., 2009: Boundary-layer similarity under an axisymmetric, gradient wind vortex. Boundary-Layer Meteorology, 131, 321–344.DOI
  22. [23]Giometto, M., G. Katul, J. Fang, and M. Parlange, 2017a: Direct numerical simulation of turbulent slope flows up to grashof number. Journal of Fluid Mechanics, 829, 589–620.
  23. [25]Giometto, M. G., A. Christen, C. Meneveau, J. Fang, M. Krafczyk, and M. B. Parlange, 2016: Spatial characteristics of roughness sublayer mean flow and turbulence over a realistic urban surface. Boundary-Layer Meteorology, 160, 425–452.DOI
  24. [26]Gopalakrishnan, S., A. Hazelton, and J. A. Zhang, 2021: Improving hurricane boundary layer parameterization scheme based on observations. Earth and Space Science, 8 (3), e2020EA0001 422.DOI
  25. [27]Gopalakrishnan, S. G., F. Marks Jr, J. A. Zhang, X. Zhang, J.-W. Bao, and V. Tallapragada, 2013: A study of the impacts of vertical diffusion on the structure and intensity of the tropical cyclones using the high-resolution hwrf system. Journal of the Atmospheric Sciences, 70 (2), 524–541.DOI
  26. [28]Green, B. W., and F. Zhang, 2015: Idealized large-eddy simulations of a tropical cyclone–like boundary layer. Journal of the Atmospheric Sciences, 72 (5), 1743–1764.DOI
  27. [29]Haiden, T., and C. D. Whiteman, 2005: Katabatic flow mechanisms on a low-angle slope. Journal of Applied Meteorology, 44 (1), 113–126.DOI
  28. [30]Hultmark, M., M. Calaf, and M. B. Parlange, 2013: A new wall shear stress model for atmospheric boundary layer simulations. Journal of the Atmospheric Sciences, 70, 3460–3470.DOI
  29. [31]Joffre, S., M. Kangas, M. Heikinheimo, and S. Kitaigorodskii, 2001: Variability of the stable and unstable atmospheric boundary-layer height and its scales over a boreal forest. Boundary-Layer Meteorology, 99 (3), 429–450.DOI
  30. [32]Kepert, J., 2001: The dynamics of boundary layer jets within the tropical cyclone core. part i: Linear theory. Journal of the Atmospheric Sciences, 58, 2469–2484.doi.org/10.1175/1520-0469(2001)058<2469:tdoblj>2.0.co;2
  31. [33]Kepert, J., 2010: Slab-and height-resolving models of the tropical cyclone boundary layer. part i: Comparing the simulations. Quarterly Journal of the Royal Meteorological Society, 136, 1686–1699.DOI
  32. [34]Kepert, J., and Y. Wang, 2001: The dynamics of boundary layer jets within the tropical cyclone core. part ii: Nonlinear enhancement. Journal of the Atmospheric Sciences, 58, 2485–2501.doi.org/10.1175/1520-0469(2001)058<2485:tdoblj>2.0.co;2
  33. [35]Kepert, J. D., 2012: Choosing a boundary layer parameterization for tropical cyclone modeling. Monthly Weather Review, 140 (5), 1427–1445.DOI
  34. [36]Kim, J., and P. Moin, 1985: Application of a fractional-step method to incompressible navier-stokes equations. Journal of Computational Physics, 59, 308–323.DOI
  35. [39]Kravchenko, A., and P. Moin, 1997: On the effect of numerical errors in large eddy simulations of turbulent flows. Journal of Computational Physics, 131, 310–322.
  36. [43]Matak, L., and M. Momen, 2023: The role of vertical diffusion parameterizations in the dynamics and accuracy of simulated intensifying hurricanes. Boundary-Layer Meteorology, 188 (3), 389–418.
  37. [44]Meneveau, C., T. S. Lund, and W. H. Cabot, 1996: A lagrangian dynamic subgrid-scale model of turbulence. Journal of Fluid Mechanics, 319, 353–385.DOI
  38. [45]Meng, Y., M. Matsui, and K. Hibi, 1995: An analytical model for simulation of the wind field in a typhoon boundary layer. Journal of Wind Engineering and Industrial Aerodynamics, 56 (2-3), 291–310.DOI
  39. [46]Ming, J., J. A. Zhang, and R. F. Rogers, 2015: Typhoon kinematic and thermodynamic boundary layer structure from dropsonde composites. Journal of Geophysical Research: Atmospheres, 120 (8), 3158–3172.DOI
  40. [47]Momen, M., M. B. Parlange, and M. G. Giometto, 2021: Scrambling and reorientation of classical atmospheric boundary layer turbulence in hurricane winds. Geophysical Research Letters, 48 (7), e2020GL091 695.DOI
  41. [48]Nakanishi, M., and H. Niino, 2012: Large-eddy simulation of roll vortices in a hurricane boundary layer. Journal of the Atmospheric Sciences, 69 (12), 3558–3575.DOI
  42. [49]Narasimhan, G., D. F. Gayme, and C. Meneveau, 2024: Analytical model coupling ekman and surface layer structure in atmospheric boundary layer flows. Boundary-Layer Meteorology, 190 (4), 16.DOI
  43. [50]Nolan, D. S., J. A. Zhang, and D. P. Stern, 2009: Evaluation of planetary boundary layer parameterizations in tropical cyclones by comparison of in situ observations and high-resolution simulations of hurricane isabel (2003). part i: Initialization, maximum winds, and the outer-core boundary layer. Monthly Weather Review, 137 (11), 3651–3674.DOI
  44. [51]Orszag, S. A., 1969: Numerical methods for the simulation of turbulence. Physics of Fluids, 12, II–250.
  45. [52]Orszag, S. A., 1970: Transform method for the calculation of vector-coupled sums: Application to the spectral form of the vorticity equation. Journal of Atmospheric Sciences, 27, 890–895.doi.org/10.1175/1520-0469(1970)027<0890:tmftco>2.0.co;2
  46. [53]Orszag, S. A., and Y.-H. Pao, 1975: Numerical computation of turbulent shear flows. Advances in Geophysics, Vol. 18, Elsevier, 225–236.
  47. [54]Pan, Y., M. Chamecki, and H. M. Nepf, 2016: Estimating the instantaneous drag–wind relationship for a horizontally homogeneous canopy. Boundary-Layer Meteorology, 160, 63–82.DOI
  48. [55]Pan, Y., E. Follett, M. Chamecki, and H. Nepf, 2014: Strong and weak, unsteady reconfiguration and its impact on turbulence structure within plant canopies. Physics of Fluids, 26, 2003–2017.DOI
  49. [56]Pollard, R. T., P. B. Rhines, and R. O. Thompson, 1973: The deepening of the wind-mixed layer. Geophysical Fluid Dynamics, 4 (4), 381–404.
  50. [57]Porté-Agel, F., 2004: A scale-dependent dynamic model for scalar transport in large-eddy simulations of the atmospheric boundary layer. Boundary-Layer Meteorology, 112, 81–105.
  51. [58]Porté-Agel, F., C. Meneveau, and M. B. Parlange, 2000: A scale-dependent dynamic model for large-eddy simulation: application to a neutral atmospheric boundary layer. Journal of Fluid Mechanics, 415, 261–284.DOI
  52. [59]Richter, D. H., G. H. Bryan, J. Dennis, J. Sun, and S. Voelz, 2025: Large-eddy simulation of the spray-laden hurricane boundary layer: I. spray transport and statistics. Journal of Geophysical Research: Atmospheres, 130 (17), e2025JD044 054.DOI
  53. [60]Romdhani, O., J. A. Zhang, and M. Momen, 2022: Characterizing the impacts of turbulence closures on real hurricane forecasts: A comprehensive joint assessment of grid resolution, horizontal turbulence models, and horizontal mixing length. Journal of Advances in Modeling Earth Systems, 14 (9), e2021MS002 796.
  54. [61]Sabet, F., Y. R. Yi, L. Thomas, and M. Momen, 2022: Characterizing mean and turbulent structures of hurricane winds via large-eddy simulations. CTR Summer Research Briefs.
  55. [62]Sathia, K. R., and M. G. Giometto, 2025: Analytical solution to a nonlinear single column model of hurricane boundary layer wind profiles outside the eyewall. Boundary-Layer Meteorology, 191 (11), 1–28.DOI
  56. [63]Sathia, K. R., and M. G. Giometto, 2026: Hurricane boundary layer flow simulations, in hurricane boundary layer wind database. Designsafe-CI, https://doi.org/10.17603/DS2-5TC9-XR68.
  57. [64]Shah, S., and E. Bou-Zeid, 2014: Very-large-scale motions in the atmospheric boundary layer reduced by snapshot proper orthogonal decomposition. Boundary-Layer Meteorology, 153, 355–387.DOI
  58. [65]Smith, R. K., and M. T. Montgomery, 2020: The generalized ekman model for the tropical cyclone boundary layer revisited: The myth of inertial stability as a restoring force. Quarterly Journal of the Royal Meteorological Society, 146 (732), 3435–3449.DOI
  59. [66]Smith, R. K., and G. L. Thomsen, 2010: Dependence of tropical-cyclone intensification on the boundary-layer representation in a numerical model. Quarterly Journal of the Royal Meteorological Society, 136 (652), 1671–1685.DOI
  60. [67]Sous, D., J. Sommeria, and D. Boyer, 2013: Friction law and turbulent properties in a laboratory ekman boundary layer. Physics of Fluids, 25 (4).DOI
  61. [68]Stern, D. P., and D. S. Nolan, 2009: Reexamining the vertical structure of tangential winds in tropical cyclones: Observations and theory. Journal of the Atmospheric Sciences, 66 (12), 3579–3600.DOI
  62. [69]Stull, R. B., 2012: An introduction to boundary layer meteorology. Springer Science & Business Media.
  63. [70]Tennekes, H., and J. L. Lumley, 1972: A first course in turbulence. MIT press.
  64. [71]Tseng, Y.-H., C. Meneveau, and M. B. Parlange, 2006: Modeling flow around bluff bodies and predicting urban dispersion using large eddy simulation. Environmental Science & Technology, 40, 2653–2662.
  65. [72]Vickery, P. J., D. Wadhera, M. D. Powell, and Y. Chen, 2009: A hurricane boundary layer and wind field model for use in engineering applications. Journal of Applied Meteorology and Climatology, 48 (2), 381–405.DOI
  66. [73]Worsnop, R. P., G. H. Bryan, J. K. Lundquist, and J. A. Zhang, 2017: Using large-eddy simulations to define spectral and coherence characteristics of the hurricane boundary layer for wind-energy applications. Boundary-Layer Meteorology, 165, 55–86.DOI
  67. [74]Wyngaard, J. C., 2010: Turbulence in the Atmosphere. Cambridge university press.
  68. [75]Zhang, J. A., and W. M. Drennan, 2012: An observational study of vertical eddy diffusivity in the hurricane boundary layer. Journal of the Atmospheric Sciences, 69 (11), 3223–3236.DOI
  69. [76]Zhang, J. A., D. S. Nolan, R. F. Rogers, and V. Tallapragada, 2015: Evaluating the impact of improvements in the boundary layer parameterization on hurricane intensity and structure forecasts in hwrf. Monthly Weather Review, 143 (8), 3136–3155.DOI
  70. [77]Zhang, J. A., R. F. Rogers, D. S. Nolan, and F. D. Marks Jr, 2011: On the characteristic height scales of the hurricane boundary layer. Monthly Weather Review, 139 (8), 2523–2535.
  71. [78]Zhang, J. A., R. F. Rogers, and V. Tallapragada, 2017: Impact of parameterized boundary layer structure on tropical cyclone rapid intensification forecasts in hwrf. Monthly Weather Review, 145 (4), 1413–1426.DOI
  72. [79]Zilitinkevich, S., 1972: On the determination of the height of the ekman boundary layer. Boundary-Layer Meteorology, 3 (2), 141–145.
  73. [80]Zilitinkevich, S., I. Esau, and A. Baklanov, 2007: Further comments on the equilibrium height of neutral and stable planetary boundary layers. Quarterly Journal of the Royal Meteorological Society, 133 (622), 265–271.DOI

Paper details

vlnc:W_mn54gv6j
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents