Skip to content

Boundary Layer, Turbulence and the Ekman Layer

CSIR NET Part C meteorology: boundary layer structure, Reynolds averaging, K-theory, log wind profile, Obukhov length, Ekman spiral, pumping and spin-down.

  • 19 min read
  • Updated
On this page · 14 sections
  1. Prerequisites
  2. The Hook: The Wind That Crosses the Isobars
  3. The Planetary Boundary Layer
  4. Turbulence and Reynolds Averaging
  5. The Surface Layer and the Log Wind Profile
  6. Stability in the Boundary Layer
  7. The Ekman Layer
  8. Memory Tricks
  9. Don’t Confuse
  10. Common Mistakes
  11. Practice
  12. Quick Revision Table
  13. Key Takeaways
  14. What’s Next

Prerequisites

The Hook: The Wind That Crosses the Isobars

The Planetary Boundary Layer

The PBL is the lowest part of the atmosphere directly influenced by the surface, responding to surface forcing on time scales of about an hour or less. Its depth varies from tens of metres (calm, clear nights) to 2–3 km or more (hot, dry afternoons over land); about 1 km is typical.

Structure and diurnal cycle over land

LayerDepthCharacter
Surface (constant-flux) layerlowest about 10% of the PBLvertical fluxes nearly constant with height; strong gradients; log profile
Mixed (convective) layerdaytime bulk of the PBLbuoyant thermals mix \( \theta \), \( q \) and momentum nearly uniform with height
Entrainment zonetop of mixed layercapping inversion; free-tropospheric air drawn down into the PBL
Stable (nocturnal) boundary layertens to a few hundred m at nightradiative cooling from below, weak intermittent turbulence
Residual layerabove the stable layer at nightremnant of the day’s mixed layer, neutral, decaying turbulence
Free atmosphereabovenearly geostrophic, little turbulence
graph LR
    A[Sunrise: surface heats] --> B[Convective mixed layer grows, entrains warm air from above]
    B --> C[Afternoon: deep well-mixed layer, cumulus at its top]
    C --> D[Sunset: surface cools radiatively]
    D --> E[Stable nocturnal layer forms near ground; residual layer above]
    E --> F[Nocturnal low-level jet above the inversion]
    F --> A

Over the ocean, the surface temperature hardly changes in a day, so the marine boundary layer has a weak diurnal cycle; its depth is controlled more by large-scale subsidence and by stratocumulus cloud-top processes.

Turbulence and Reynolds Averaging

Why turbulence

The PBL Reynolds number is of order \( 10^8 \), far above the critical values of laboratory flows, so its flow is turbulent: three-dimensional, rotational, irregular eddies spanning scales from the PBL depth down to the Kolmogorov microscale (about a millimetre), where viscosity dissipates kinetic energy as heat. Energy cascades from large to small eddies.

Reynolds decomposition

Split every variable into a mean (time or ensemble average) and fluctuation: \( u = \bar u + u' \), with \( \overline{u'} = 0 \). Averaging the momentum equation produces extra terms from the nonlinear advection:

$$\frac{\partial\bar u}{\partial t} + \bar{\mathbf U}\cdot\nabla\bar u = -\frac{1}{\rho}\frac{\partial\bar p}{\partial x} + f\bar v - \frac{\partial\overline{u'w'}}{\partial z} + \dots$$

The Reynolds stress (kinematic momentum flux) \( \overline{u'w'} \) behaves like a stress gradient. Similarly \( \overline{w'\theta'} \) is the kinematic heat flux and \( \overline{w'q'} \) the moisture flux. Under the boundary-layer approximation, vertical derivatives of fluxes dominate (horizontal homogeneity).

The closure problem

Each equation for a mean involves fluxes (second moments); equations for fluxes involve third moments; and so on. There are always more unknowns than equations: the closure problem. Practical models close the system at some order.

First-order (K-theory) closure relates fluxes to mean gradients by analogy with molecular diffusion:

$$\boxed{\overline{u'w'} = -K_m\frac{\partial\bar u}{\partial z},\qquad \overline{w'\theta'} = -K_h\frac{\partial\bar\theta}{\partial z}}$$

\( K_m \) and \( K_h \) are eddy viscosity and eddy diffusivity, typically 1–100 m\(^2\) s\(^{-1}\) in the PBL, enormous compared with molecular \( \nu \approx 1.5\times 10^{-5} \) m\(^2\) s\(^{-1}\). Unlike molecular values they are properties of the flow, not the fluid. K-theory fails in the convective mixed layer, where large thermals carry heat even against (or with zero) local gradient (“counter-gradient” transport).

Mixing-length theory (Prandtl): an eddy displaced a distance \( l \) carries a velocity fluctuation \( u' \approx -l\,\partial\bar u/\partial z \), giving

$$K_m = l^2\left|\frac{\partial\bar u}{\partial z}\right|$$

Near a surface, eddies are limited by distance from the ground: \( l = \kappa z \) with the von Kármán constant \( \kappa \approx 0.4 \).

The Surface Layer and the Log Wind Profile

In the surface layer, the momentum flux is approximately constant with height and equal to its surface value. Define the friction velocity

$$u_* = \left(\frac{\tau_0}{\rho}\right)^{1/2} = \left(\overline{u'w'}^2 + \overline{v'w'}^2\right)_0^{1/4}$$

Under neutral stability, dimensional reasoning (the only length scale is \( z \)) or mixing length gives \( \partial\bar u/\partial z = u_*/(\kappa z) \), which integrates to

$$\boxed{\bar u(z) = \frac{u_*}{\kappa}\ln\frac{z}{z_0}}$$

where \( z_0 \) is the aerodynamic roughness length, the height at which the extrapolated log profile reaches zero. It is typically of order a tenth of the height of the roughness elements. Over tall vegetation and cities a displacement height \( d \) is added: \( \ln[(z-d)/z_0] \).

SurfaceRoughness length \( z_0 \) (order of magnitude)
Open sea (depends on wind and waves)\( 10^{-4} \)–\( 10^{-3} \) m
Short grass, flat open countryabout 0.01–0.05 m
Crops, scattered bushesabout 0.1 m
Forest, suburbsabout 0.5–1 m
City centresabout 1 m or more

Stability in the Boundary Layer

Turbulence kinetic energy budget

The turbulence kinetic energy per unit mass \( \bar e = \tfrac{1}{2}(\overline{u'^2} + \overline{v'^2} + \overline{w'^2}) \) obeys, for horizontally homogeneous conditions,

$$\frac{\partial\bar e}{\partial t} = \underbrace{-\overline{u'w'}\frac{\partial\bar u}{\partial z} - \overline{v'w'}\frac{\partial\bar v}{\partial z}}_{\text{shear production}\ (>0)} + \underbrace{\frac{g}{\bar\theta_v}\overline{w'\theta_v'}}_{\text{buoyancy}} - \underbrace{\frac{\partial}{\partial z}\left(\overline{w'e} + \frac{\overline{w'p'}}{\rho}\right)}_{\text{transport}} - \varepsilon$$
  • Shear production is (almost always) positive: mean-flow energy feeds the eddies.
  • Buoyancy produces TKE when the heat flux is upward (unstable, daytime) and destroys it when downward (stable, night).
  • Dissipation \( \varepsilon \) is always a sink.

Three Richardson numbers

NameDefinitionUse
Flux Richardson number\( R_f = \dfrac{(g/\bar\theta_v)\,\overline{w'\theta_v'}}{\overline{u'w'}\,\partial\bar u/\partial z} \)ratio of buoyant destruction to shear production; turbulence cannot be sustained (with no other source) if \( R_f > 1 \)
Gradient Richardson number\( Ri = \dfrac{(g/\bar\theta_v)\,\partial\bar\theta_v/\partial z}{(\partial\bar u/\partial z)^2} \)from mean profiles; \( R_f = (K_h/K_m)Ri \); laminar flow becomes turbulent when \( Ri \) falls below about 0.25
Bulk Richardson number\( R_b = \dfrac{(g/\bar\theta_v)\,\Delta\bar\theta_v\,\Delta z}{(\Delta\bar u)^2 + (\Delta\bar v)^2} \)from finite differences across a layer; used to diagnose PBL depth in models

Signs: negative = unstable (convective), zero = neutral, positive = stable.

Monin–Obukhov similarity

The Obukhov length

$$\boxed{L = -\frac{u_*^3\,\bar\theta_v}{\kappa g\,\overline{w'\theta_v'}_0}}$$

is the height at which buoyant production (or destruction) of TKE is comparable with shear production.

\( z/L \)StabilityBehaviour
\( L < 0 \) (\( z/L < 0 \))unstable (upward heat flux)above (
(L\to\infty )
\( L > 0 \)stable (downward heat flux)turbulence suppressed above about \( L \)

Monin–Obukhov similarity theory states that in the surface layer, suitably scaled gradients are universal functions of \( z/L \):

$$\frac{\kappa z}{u_*}\frac{\partial\bar u}{\partial z} = \phi_m\!\left(\frac{z}{L}\right)$$

with \( \phi_m = 1 \) for neutral conditions. The functions are fitted empirically (for stable conditions a commonly used form is approximately linear, \( \phi_m \approx 1 + 5z/L \)). Wind increases with height faster than logarithmically in stable conditions and more slowly in unstable conditions.

The Ekman Layer

The three-way balance

Above the surface layer, the mean horizontal flow is steady and balances three forces: pressure gradient, Coriolis and the divergence of turbulent stress. With constant \( K \) and geostrophic wind \( (u_g, 0) \) independent of height:

$$K\frac{d^2u}{dz^2} + f(v - v_g) = 0,\qquad K\frac{d^2v}{dz^2} - f(u - u_g) = 0$$

with \( u = v = 0 \) at \( z = 0 \) and \( (u, v)\to(u_g, 0) \) as \( z\to\infty \).

The Ekman spiral

The solution (combine \( u + iv \)) is

$$\boxed{u = u_g\left(1 - e^{-\gamma z}\cos\gamma z\right),\qquad v = u_g\,e^{-\gamma z}\sin\gamma z,\qquad \gamma = \left(\frac{f}{2K}\right)^{1/2}}$$
  • Plotted as a hodograph, the wind vector traces the Ekman spiral: at the surface it points 45° to the left of the geostrophic wind (NH), across the isobars towards low pressure, and turns clockwise and strengthens with height.
  • Ekman depth: \( D_e = \pi/\gamma = \pi(2K/f)^{1/2} \), where the wind first becomes parallel to \( \mathbf V_g \). There it is slightly supergeostrophic: \( u = u_g(1 + e^{-\pi}) \approx 1.04\,u_g \).
  • Observed surface cross-isobar angles are smaller than 45°, roughly 10–20° over the sea and larger over land, because \( K \) is not constant and the surface layer is better represented by a log profile. A modified Ekman layer joins a surface layer to an Ekman layer above it.

Ekman transport and Ekman pumping

Integrating the cross-isobar component through the layer gives a net mass transport towards low pressure, perpendicular to the geostrophic wind. Around a cyclone, this inflow converges; continuity then demands vertical velocity at the top of the layer:

$$\boxed{w(D_e) = \frac{\zeta_g}{2\gamma} = \zeta_g\left(\frac{K}{2f}\right)^{1/2} = \frac{D_e}{2\pi}\,\zeta_g}$$

This is Ekman pumping: upward over cyclonic (\( \zeta_g > 0 \)) circulations, downward (subsidence) over anticyclones.

Spin-down

The Ekman pumping drives a secondary circulation: the rising air at the top of the PBL diverges outward in the free atmosphere, and by the stretching term this outflow reduces the vorticity of the whole column (outflow means divergence, which spins down). For a barotropic atmosphere of depth \( H \):

$$\frac{d\zeta_g}{dt} = -\frac{f}{H}\,w(D_e) = -\left(\frac{fK}{2H^2}\right)^{1/2}\zeta_g \quad\Rightarrow\quad \boxed{\tau_e = H\left(\frac{2}{fK}\right)^{1/2}}$$

With \( H = 10 \) km, \( f = 10^{-4} \) s\(^{-1}\), \( K = 10 \) m\(^2\) s\(^{-1}\): \( \tau_e = 4.5\times 10^5 \) s \( \approx 5.2 \) days. Ordinary diffusion with molecular viscosity alone would take \( H^2/\nu \sim 10^{13} \) s, hundreds of thousands of years. Rotation plus turbulence makes friction act on the whole troposphere in days, through the secondary circulation rather than through direct diffusion. In a stratified atmosphere the secondary circulation is confined to a shallower depth, and spin-down above it is weaker.

Memory Tricks

Don’t Confuse

PairDistinction
Molecular vs eddy viscosityabout \( 10^{-5} \) vs 1–100 m\(^2\) s\(^{-1}\); fluid property vs flow property
Surface layer vs Ekman layerconstant flux, log profile, no turning vs three-way balance, wind turns with height
Mixed layer vs residual layeractively convecting (day) vs leftover neutral layer (night)
\( Ri \) vs \( R_f \) vs \( R_b \)gradients vs fluxes vs finite differences
\( z_0 \) vs \( d \)roughness length vs displacement height
Obukhov length vs Ekman depthstability scale of the surface layer vs rotational scale of the whole PBL
Atmospheric vs oceanic Ekman transporttowards low pressure vs 90° to the right of the wind stress (NH)
Ekman pumping vs spin-downvertical velocity at PBL top vs resulting decay of the column’s vorticity

Common Mistakes

Practice

All questions below are original practice questions written for this page. They are not from any CSIR NET paper.

Level 1: Foundation

  1. Write the neutral logarithmic wind profile.
  2. Write the Ekman depth in terms of \( K \) and \( f \).
  3. What does a negative Obukhov length indicate?
Answers
  1. \( \bar u = (u_*/\kappa)\ln(z/z_0) \), \( \kappa \approx 0.4 \).
  2. \( D_e = \pi(2K/f)^{1/2} \).
  3. Upward surface buoyancy flux: an unstable (convective) surface layer.

Level 2: Part B Style

Practice Q1. For \( f = 10^{-4} \) s\(^{-1}\) and eddy viscosity \( K = 10 \) m\(^2\) s\(^{-1}\), the Ekman depth is closest to:

(a) 140 m (b) 450 m (c) 1.4 km (d) 4.4 km

Answer: (c) 1.4 km

\( D_e = \pi\sqrt{2\times 10/10^{-4}} = \pi\times 447 = 1{,}405 \) m. Option (b) omits \( \pi \).

Practice Q2. Neutral winds are 4.0 m s\(^{-1}\) at 2 m and 5.5 m s\(^{-1}\) at 10 m. The friction velocity and roughness length are closest to:

(a) 0.37 m s\(^{-1}\), 0.027 m (b) 0.15 m s\(^{-1}\), 0.27 m (c) 0.37 m s\(^{-1}\), 0.27 m (d) 0.93 m s\(^{-1}\), 0.0027 m

Answer: (a)

\( u_{10} - u_2 = (u_*/\kappa)\ln 5 \), so \( u_* = 0.4\times 1.5/1.609 = 0.373 \) m s\(^{-1}\). \( z_0 = 10\exp(-\kappa u_{10}/u_*) = 10\exp(-0.4\times 5.5/0.373) = 10\exp(-5.90) = 0.027 \) m.

Practice Q3. Above a cyclonic geostrophic vorticity of \( 2\times 10^{-5} \) s\(^{-1}\), with \( K = 8 \) m\(^2\) s\(^{-1}\) and \( f = 1.2\times 10^{-4} \) s\(^{-1}\), the Ekman pumping velocity is:

(a) 0.37 mm s\(^{-1}\), downward (b) 3.7 mm s\(^{-1}\), upward (c) 3.7 mm s\(^{-1}\), downward (d) 3.7 cm s\(^{-1}\), upward

Answer: (b)

\( w = \zeta_g\sqrt{K/2f} = 2\times 10^{-5}\times\sqrt{8/2.4\times 10^{-4}} = 2\times 10^{-5}\times 182.6 = 3.65\times 10^{-3} \) m s\(^{-1}\), upward because the vorticity is cyclonic.

Practice Q4. In the TKE budget, which term changes sign between a sunny afternoon and a clear night over land?

(a) Shear production (b) Buoyancy production (c) Dissipation (d) None

Answer: (b)

The surface heat flux is upward by day (buoyancy produces TKE) and downward at night (buoyancy destroys TKE). Shear production is normally positive and dissipation is always a sink.

Level 3: Part C Style

Practice Q5. A mid-latitude barotropic vortex of depth 10 km sits above a boundary layer with \( f = 10^{-4} \) s\(^{-1}\). Which statements are correct?

  1. With \( K = 5 \) m\(^2\) s\(^{-1}\), the spin-down e-folding time is about 7.3 days.
  2. Doubling \( K \) halves the spin-down time.
  3. Without rotation, turbulent diffusion through the depth with the same \( K = 5 \) m\(^2\) s\(^{-1}\) would take of order \( H^2/K \approx 230 \) days, much longer than Ekman spin-down.
  4. Spin-down acts through a secondary circulation: outflow above the boundary layer reduces vorticity by the stretching term.

(a) 1, 3 and 4 only (b) 1 and 2 only (c) 2, 3 and 4 only (d) 1, 2, 3 and 4

Answer: (a) 1, 3 and 4 only
  • 1 true: \( \tau_e = 10^4\sqrt{2/(10^{-4}\times 5)} = 10^4\times 63.2 = 6.32\times 10^5 \) s = 7.3 days.
  • 2 false: \( \tau_e\propto K^{-1/2} \); doubling \( K \) reduces \( \tau_e \) by \( \sqrt 2 \) (to 5.2 days).
  • 3 true: \( 10^8/5 = 2\times 10^7 \) s \( \approx 231 \) days.
  • 4 true: Ekman pumping into the free atmosphere forces outward flow aloft, i.e. divergence, which reduces absolute vorticity.

Practice Q6. Surface-layer observations at night over a grassland give \( u_* = 0.2 \) m s\(^{-1}\), \( \bar\theta_v = 290 \) K and kinematic heat flux \( -0.02 \) K m s\(^{-1}\). Which statements are correct?

  1. The Obukhov length is about +30 m.
  2. At \( z = 10 \) m the surface layer is stable, with \( z/L \approx 0.34 \).
  3. Compared with a neutral log profile having the same \( u_* \), the wind increases more rapidly with height.
  4. Since \( L > 0 \), the turbulence kinetic energy must be increasing.

(a) 1, 2 and 3 only (b) 1 and 4 only (c) 2 and 3 only (d) 1, 2, 3 and 4

Answer: (a) 1, 2 and 3 only
  • 1 true: \( L = -(0.008\times 290)/(0.4\times 9.81\times(-0.02)) = 2.32/0.0785 = 29.6 \) m.
  • 2 true: \( 10/29.6 = 0.34 > 0 \).
  • 3 true: \( \phi_m > 1 \) in stable conditions, so \( \partial\bar u/\partial z \) exceeds \( u_*/(\kappa z) \).
  • 4 false: a downward heat flux means buoyancy destroys TKE; \( L > 0 \) says nothing about TKE increasing.

Practice Q7. For a constant-\( K \) Ekman layer in the NH, which statements are correct?

  1. The wind at the surface (as \( z\to 0 \)) is directed 45° to the left of the geostrophic wind.
  2. At \( z = D_e \) the wind is parallel to the geostrophic wind and about 4% stronger.
  3. The net Ekman mass transport in the layer is directed towards low pressure.
  4. In the oceanic Ekman layer under a wind stress, the net transport is 90° to the left of the wind in the NH.

(a) 1, 2 and 3 only (b) 1 and 4 only (c) 2 and 3 only (d) 1, 2, 3 and 4

Answer: (a) 1, 2 and 3 only
  • 1 true: as \( z\to 0 \), \( u\approx v\approx u_g\gamma z \), so the angle is 45°, towards low pressure (left of \( \mathbf V_g \) in the NH).
  • 2 true: \( \gamma D_e = \pi \): \( v = 0 \), \( u = u_g(1 + e^{-\pi}) = 1.043\,u_g \).
  • 3 true: the integrated cross-isobaric flow is towards low pressure.
  • 4 false: oceanic Ekman transport is 90° to the right of the wind stress in the NH.

Quick Revision Table

IdeaFact or formula
PBLabout 1 km; surface layer about 10%; mixed, entrainment, stable, residual layers
Reynolds stress\( \overline{u'w'} \); closure problem
K-theory\( \overline{u'w'} = -K_m\,\partial\bar u/\partial z \); \( K \sim 1 \)–100 m\(^2\) s\(^{-1}\)
Mixing length( K_m = l^2
Log profile\( \bar u = (u_*/\kappa)\ln(z/z_0) \); \( u_* = \sqrt{\tau_0/\rho} \)
TKEshear production + buoyancy − dissipation + transport
\( Ri \), \( R_f \), \( R_b \)gradient, flux, bulk; critical about 0.25 (\( Ri \)), 1 (\( R_f \))
Obukhov length\( L = -u_*^3\bar\theta_v/(\kappa g\,\overline{w'\theta_v'}) \); \( L < 0 \) unstable
Ekman spiral\( u = u_g(1 - e^{-\gamma z}\cos\gamma z) \), \( v = u_ge^{-\gamma z}\sin\gamma z \), \( \gamma = \sqrt{f/2K} \)
Ekman depth\( D_e = \pi\sqrt{2K/f} \approx 1 \) km; surface angle 45° (theory)
Pumping\( w = \zeta_g\sqrt{K/2f} \)
Spin-down\( \tau_e = H\sqrt{2/fK} \approx 5 \) days

Key Takeaways

What’s Next