12.1 Atmospheric boundary layer

Atmospheric boundary layer characteristics

The atmospheric boundary layer (ABL) is the lowest portion of the troposphere that responds directly to forcing from Earth's surface. It's where we live, where pollutants disperse, and where surface friction and heating shape the wind and temperature profiles that drive local weather. Understanding the ABL matters because its turbulence controls the vertical transport of heat, moisture, momentum, and contaminants.

The ABL's structure is governed by three interacting factors: turbulence generated by shear and buoyancy, the roughness and thermal properties of the underlying surface, and a pronounced diurnal cycle driven by solar heating and radiative cooling.

Turbulence in the atmospheric boundary layer

Turbulence in the ABL comes from two main sources. Mechanical (shear) turbulence arises when wind interacts with the rough surface and when velocity gradients create instabilities. Buoyancy-driven (convective) turbulence develops when the surface heats the air from below, making parcels rise.

Turbulent eddies span a huge range of scales, from millimeters up to the full depth of the boundary layer (order of kilometers). These eddies are the primary mechanism for vertical transport of heat, moisture, and momentum. The intensity of turbulence at any point depends on height, atmospheric stability, and surface roughness.

Turbulent kinetic energy (TKE) quantifies turbulence intensity:

$TKE = \frac{1}{2}(\overline{u'^2} + \overline{v'^2} + \overline{w'^2})$

where $u'$, $v'$, and $w'$ are the fluctuating (departure from mean) velocity components and the overbars denote time or ensemble averages. A TKE budget equation tracks how shear production, buoyancy production/destruction, transport, and dissipation balance in different parts of the ABL.

Velocity profile near Earth's surface

Wind speed in the ABL increases with height because the drag exerted by the surface diminishes as you move away from it. In the surface layer (roughly the lowest 10% of the ABL), the wind profile under neutral stability follows a logarithmic law:

$U(z) = \frac{u_*}{\kappa} \ln\!\left(\frac{z}{z_0}\right)$

• $U(z)$ is the mean wind speed at height $z$
• $u_*$ is the friction velocity, a velocity scale defined by the surface shear stress ($u_* = \sqrt{\tau_0 / \rho}$)
• $\kappa \approx 0.4$ is the von Kármán constant
• $z_0$ is the aerodynamic roughness length (discussed below)

This log profile is strictly valid only for neutral stability. Under unstable conditions (daytime convection), enhanced mixing reduces the vertical wind shear, so the profile curves below the neutral log law. Under stable conditions (nighttime cooling), suppressed mixing steepens the shear, and the profile curves above it. Monin-Obukhov similarity theory (covered later) provides the correction functions for these non-neutral cases.

Diurnal cycle of the atmospheric boundary layer

The ABL undergoes a repeating daily cycle tied to the surface energy balance:

1. Morning transition. After sunrise, solar heating warms the surface. A shallow mixed layer begins to grow upward, eroding the remnant of the previous night's stable layer.
2. Daytime convective boundary layer (CBL). Strong surface heating drives vigorous thermals. The CBL is well-mixed in potential temperature, moisture, and wind, and can grow to 1–2 km (or more in arid regions). A sharp capping inversion at the top limits further growth.
3. Evening transition. As solar input drops, surface heating weakens. Turbulence decays, and the mixed layer stops growing.
4. Nocturnal stable boundary layer (SBL). Radiative cooling at the surface creates a temperature inversion near the ground. Turbulence is weak and intermittent, and the SBL is typically only 100–300 m deep. Above it, the residual layer retains the properties of the previous day's CBL but is no longer turbulently connected to the surface.

These transitions are among the hardest parts of the diurnal cycle to model because they involve rapid, nonlinear changes in turbulence regime.

Stability effects on the boundary layer

Atmospheric stability determines which turbulence source dominates and therefore controls ABL depth, mixing intensity, and vertical gradients.

A useful bulk indicator is the gradient Richardson number, $Ri_g$, which compares buoyancy suppression to shear production of turbulence. When $Ri_g$ exceeds a critical value (often taken as ~0.25), turbulence tends to be suppressed.

Surface roughness impact

The character of the underlying surface exerts a strong control on ABL turbulence and structure. Rougher surfaces (forests, cities) generate more mechanical turbulence and stronger vertical mixing, producing a deeper ABL. Smoother surfaces (open water, flat grassland, ice) produce less drag and weaker mixing.

Aerodynamic roughness length

The aerodynamic roughness length $z_0$ is the height at which the log-law wind profile extrapolates to zero. It's not a physical height you can measure with a ruler; it's an effective parameter that encapsulates the drag properties of the surface.

Typical values:

Over water, $z_0$ itself depends on wind speed (through the Charnock relation), which adds a feedback between the wind and the surface drag.

Displacement height for obstacles

When the surface is covered with tall, densely packed obstacles like trees or buildings, the effective ground level for the wind profile is shifted upward. A displacement height $d$ is introduced:

$U(z) = \frac{u_*}{\kappa} \ln\!\left(\frac{z - d}{z_0}\right)$

The displacement height is typically about 2/3 to 3/4 of the mean obstacle height. For a forest canopy averaging 20 m tall, $d$ would be roughly 13–15 m. The log law then applies only for $z > d + z_0$.

Urban vs. rural surface roughness

Urban areas have much larger $z_0$ and $d$ values than surrounding rural land. This leads to:

• Stronger mechanical turbulence and deeper urban boundary layers
• Reduced mean wind speeds near the surface but enhanced gustiness
• Complex flow patterns including street canyon vortices (recirculating flow between buildings) and wake effects behind tall structures
• The urban heat island effect, where waste heat, reduced vegetation, and altered surface energy balance keep cities warmer than their surroundings, further modifying ABL stability

Monin-Obukhov similarity theory

Monin-Obukhov similarity theory (MOST) extends the neutral log-law profile to non-neutral conditions. It provides a systematic framework for relating mean gradients of wind and temperature to turbulent fluxes in the surface layer.

MOST rests on the idea that surface-layer turbulence statistics, when properly nondimensionalized, depend on only a few governing parameters: the friction velocity $u_*$, the surface buoyancy flux, and the height $z$.

Dimensionless parameters in similarity theory

The central scaling variable is the Obukhov length:

$L = -\frac{u_*^3}{\kappa \frac{g}{\theta_v} \overline{w'\theta_v'}}$

where $g$ is gravitational acceleration, $\theta_v$ is the virtual potential temperature, and $\overline{w'\theta_v'}$ is the surface kinematic buoyancy flux. $L$ represents the height at which buoyancy production of TKE equals shear production. Its sign tells you the stability regime:

• $L < 0$: unstable (positive buoyancy flux, surface heating)
• $L > 0$: stable (negative buoyancy flux, surface cooling)
• $|L| \to \infty$: neutral

The stability parameter is $\zeta = z / L$. MOST then defines dimensionless gradient functions:

• Wind shear: $\phi_m(\zeta) = \frac{\kappa z}{u_*} \frac{\partial U}{\partial z}$
• Temperature gradient: $\phi_h(\zeta) = \frac{\kappa z}{\theta_*} \frac{\partial \theta}{\partial z}$

where $\theta_* = -\overline{w'\theta'}/u_*$ is the surface temperature scale.

Under neutral conditions ($\zeta = 0$), both $\phi_m$ and $\phi_h$ equal 1, recovering the standard log law.

Flux-profile relationships

Empirical fits to field data (most famously from the 1968 Kansas experiment) give the stability correction functions:

Unstable ($\zeta < 0$):

$\phi_m = (1 - \gamma_m \zeta)^{-1/4}$

$\phi_h = (1 - \gamma_h \zeta)^{-1/2}$

Stable ($\zeta > 0$):

$\phi_m = 1 + \beta_m \zeta$

$\phi_h = 1 + \beta_h \zeta$

Commonly used values are $\gamma_m \approx 16$, $\gamma_h \approx 16$, and $\beta_m \approx \beta_h \approx 5$ (Businger-Dyer formulations), though different datasets yield somewhat different constants.

These relationships are powerful because they let you estimate turbulent fluxes of momentum and heat from routine mean-profile measurements (wind speed and temperature at two or more heights), or conversely, predict the mean profiles from known fluxes.

Limitations of similarity theory

MOST works well in the idealized conditions it was designed for, but several real-world situations push beyond its assumptions:

• Horizontal homogeneity. MOST assumes a flat, uniform surface extending far upwind. Over patchy land use, coastlines, or complex terrain, internal boundary layers develop and MOST breaks down locally.
• Stationarity. The theory assumes steady-state conditions. During the morning and evening transitions, fluxes change rapidly and MOST predictions become unreliable.
• Surface layer only. MOST applies to roughly the lowest 10% of the ABL. Above that, Coriolis effects, entrainment at the ABL top, and large-eddy structure require different scaling.
• Very stable conditions. In strongly stable boundary layers, turbulence becomes intermittent and patchy. The smooth, continuous flux-gradient relationships of MOST often fail here, and this remains an active area of research.

Boundary layer height estimation

The boundary layer height $z_i$ (also called the mixing height) is one of the most important single parameters for air quality, weather prediction, and climate modeling. It sets the volume into which surface emissions are diluted and defines the depth over which surface fluxes are distributed. Typical values range from ~100 m (stable nighttime conditions) to ~2 km or more (strongly convective daytime conditions over arid land).

Daytime convective boundary layer height

The most intuitive method for estimating CBL height is the parcel method:

1. Take the surface (virtual) potential temperature.
2. Follow a dry adiabat upward on a thermodynamic sounding.
3. The height where this adiabat intersects the environmental temperature profile is $z_i$. At that level, a rising parcel becomes neutrally buoyant.

The CBL top is typically marked by a capping inversion where potential temperature jumps sharply. Other estimation approaches include identifying the height of maximum negative buoyancy flux (the entrainment zone) or using profiles of turbulence intensity, humidity, or aerosol concentration to locate the mixed-layer top.

Nocturnal stable boundary layer height

The SBL is shallower and harder to define precisely because the transition from the turbulent SBL to the quiescent residual layer above is often gradual rather than sharp.

A common approach uses the Richardson number criterion: compute the bulk or gradient Richardson number as a function of height, and define the SBL top as the level where $Ri$ exceeds a critical value (often $Ri_c \approx 0.25$). Diagnostic formulas relating SBL height to friction velocity and stability (e.g., $z_i \sim c \, u_* / f$, where $f$ is the Coriolis parameter and $c$ is an empirical constant) are also used, though they carry significant uncertainty.

Boundary layer height measurement techniques

Several remote sensing and in-situ platforms are used to observe $z_i$:

• Radiosondes provide vertical profiles of temperature, humidity, and wind. The ABL top shows up as a sharp inversion in potential temperature or a drop in humidity.
• Lidar (e.g., ceilometers, Doppler lidar) measures aerosol backscatter profiles. Because aerosol concentration is usually higher inside the ABL, a sharp gradient in backscatter marks the ABL top.
• Sodar (acoustic sounder) detects turbulence-related temperature fluctuations via acoustic backscatter. It's especially useful for resolving the shallow nocturnal SBL but has limited range (typically < 1 km).
• Wind profilers (radar) measure wind speed and direction aloft. Changes in the refractive-index structure constant at the ABL top produce enhanced radar returns, allowing $z_i$ estimation.

Each technique has strengths and limitations in range, resolution, and the atmospheric conditions under which it performs best. Combining multiple instruments gives the most reliable estimates.

Numerical modeling considerations

Accurate ABL representation in numerical weather prediction (NWP) and climate models is essential because the ABL mediates the exchange of energy, water, and trace species between the surface and the free atmosphere. Since turbulent eddies in the ABL are far smaller than typical model grid cells, their effects must be parameterized.

Boundary layer parameterization schemes

ABL parameterization schemes translate the net effect of unresolved turbulence into tendencies for the resolved model variables (wind, temperature, moisture). The main families are:

• Local (first-order) closure schemes (e.g., MYJ, or Mellor-Yamada-Janjić) compute turbulent fluxes at each grid level using local gradients and an eddy diffusivity $K$. They work reasonably well in stable and near-neutral conditions but can underestimate mixing in strongly convective boundary layers.
• Nonlocal closure schemes (e.g., YSU, Yonsei University scheme) add a counter-gradient correction term that accounts for transport by large convective eddies spanning the full CBL depth. This better represents the well-mixed character of the daytime CBL.
• Higher-order closure schemes (e.g., MYNN, Mellor-Yamada-Nakanishi-Niino) carry prognostic equations for TKE and sometimes higher-order turbulence moments, providing a more physically detailed representation at the cost of added complexity.
• Large-eddy simulation (LES) explicitly resolves the energy-containing eddies (grid spacing of order 10–100 m) and parameterizes only the smallest scales. LES is the gold standard for process studies but remains too expensive for operational forecasting or climate runs.

Surface energy balance in models

The lower boundary condition for ABL schemes comes from a land surface model (LSM), which solves the surface energy balance:

$R_n = H + \lambda E + G$

where $R_n$ is net radiation, $H$ is sensible heat flux, $\lambda E$ is latent heat flux, and $G$ is ground heat flux. LSMs account for soil heat conduction, soil moisture, vegetation transpiration, snow cover, and canopy radiative transfer. Errors in any of these surface fluxes propagate directly into the ABL simulation, making the coupling between the LSM and the ABL scheme a critical link in the modeling chain.

Challenges in simulating boundary layer processes

Despite steady progress, several persistent challenges remain:

• Morning and evening transitions. The rapid shift between convective and stable regimes is hard to capture because parameterizations are typically tuned for quasi-steady conditions.
• Stable boundary layers. Very stable conditions produce intermittent, weak turbulence that most schemes either over-mix (erasing the inversion) or under-mix (decoupling the surface from the atmosphere).
• Heterogeneous surfaces. Real landscapes include patchworks of land use, terrain slopes, and coastlines. Grid-averaged surface parameters can misrepresent the actual forcing.
• Grey zone resolution. As NWP models approach ~1 km grid spacing, the largest ABL eddies become partially resolved. Neither traditional parameterization nor full LES is appropriate, and hybrid approaches are still under development.
• ABL-cloud feedbacks. Shallow cumulus and stratocumulus clouds are tightly coupled to ABL turbulence. Getting the cloud fraction, liquid water path, and entrainment rate right depends on accurate ABL representation, and vice versa.

Ongoing research targets improved turbulence closures, better observational constraints (including from satellite-based lidar and dense surface networks), and scale-aware parameterizations that adapt as model resolution changes.