Atmospheric Boundary Layers:
An introduction and model intercomparisons
Bert Holtslag
Lecture for Summer school on “Land-Atmosphere Interactions“,
Valsavarenche, Valle d'Aosta (Italy), 22 June, 2015
Meteorology and Air Quality Department
Atmospheric Boundary layer (ABL)
Daytime ABL:
500-2000 m
Nighttime ABL:
10-500 m
(Stull, 1988)
The lower layer of the Atmosphere influenced by the
presence of the earth’s surface due to:
Friction, Surface Heating (Convection) or Cooling
Important characteristics are Turbulence and Diurnal Cycle!
2
Why is the boundary layer important?
We live in this layer!
Weather Forecasting:
Surface temperatures, Wind, Fog, ...
Climate and Earth System Studies:
Impact of changing conditions: Land use, Greenhouse gases?
Relevant for Agriculture, Energy use, Health, Traffic,
Urbanization, Dispersion of pollutants, et cetera
3
The interaction of the land-surface with the atmospheric boundary
layer includes many processes and complex feedback mechanisms
(Figure by Mike Ek, see also Ek and Holtslag, 2004)
History of day and night time temperature errors
Monthly averages over Europe
Standard
deviation
Mean
error
New land surface
scheme
Increased turbulent
diffusion + soil
moisture freezing
ECMWF/GABLS
workshop
7-10 November
2011 (5)
Courstesy
Anton
Beljaars,
ECMWF
Reduced outer layer
diffusion in stable
situations
Daytime Clear Convective Boundary Layer over Land
Why does an inversion form at ABL top?
1500
free atmosphere
height [m]
1000
thermal inv ersion
500
conv ectiv e
boundary layer
Convection over land
0
285
290
temperature [K]
295
6
Intermezzo
2900 m
Potential temperature
7
Temperature that a parcel
will have if it is moved (dry)
adiabatically to a reference level
(P0=1000 hPa or z=0)
 P0 
  T 
P
Rd / C p
Poisson’s equation
For dry adiabatic process the
potential temperature is constant!
Actual temperature decrease (increase)
for rising (sinking) parcel
due to air expansion (compression)
Example: Cabauw tower observations 4 November 2006
(diurnal cycle of potential temperatures at 2 to 200 m)
Intense daytime turbulent mixing gives same potential temperature,
but nighttime surface cooling leads to strong vertical gradients!
www.knmi.nl/~bosveld
8
9
Temperature structure of
standard atmosphere
with turbulent ABL and
capping inversion
(Much more background on Atmospheric
Boundary Layers is given by Roland Stull in
Chapter 9 of the “Atmospheric Science”
book by Wallace and Hobbs, 2006)
Convective Boundary Layer (CBL) or Mixing Layer
Turbulence is dominated by buoyancy
Entrainment
Strong Mixing
Surface Heat
Strong mixing leads to rather uniform profiles,
in particular for potential temperature
(Figure by John Wyngaard, 1985)
10
The typical 3 phases of boundary layer evolution
under fair weather over land (Fig 9.24)
11
Cumulus clouds
How does the land surface impact on the onset of Cumulus?
Role of soil moisture and atmospheric processes?
weaker inversion
stronger inversion
dry soil
wet soil
Courtesy Michael Ek (NCEP)
dry soil
How to quantify?
(Ek and Mahrt, 1994; Ek&Holtslag 2004)
Consider Relative Humidity (RH) tendency at boundary layer top
γθ
Δθ
θ
Use mixed-layer bulk model
for the tendency terms
γq
Δq
q
Analytical result
(Ek & Holtslag 2004, JHM)
Evaporative fraction
ne: non-evaporative processes, including
entrainment and boundary layer growth
Confirmed with Cabauw (wet soils) and AMMA observations (dry soils)
(Westra et al, 2012, JHM)
Boundary layer over heterogeneous terrain
15
Stable boundary layer
Turbulence by friction, surface cooling stabilizes
Large gradients in temperature and wind
(Figure by John Wyngaard, 1985)
16
At night, besides of turbulence also radiative cooling…
© Vincent van Gogh
and waves
17
Sketch of ABL processes depending on stability
dust devils
(After Holtslag and Nieuwstadt 1986 and Nieuwstadt and Hunt 2003)
18
Diurnal cycle over land
Free Atmosphere
Residual Layer
Convective
Mixed Layer
Stable (nocturnal) Boundary Layer
Noon
Sunset
unstable
Midnight
Sunrise
Noon
stable
19
Mean profiles


Free Atmosphere
U
U
Ugeo
Low Level Jet
Residual Layer
Convective
Mixed Layer
Ugeo
Stable (nocturnal) Boundary Layer
Noon
Sunset
unstable
Midnight
Sunrise
Noon
stable
20
Turbulence in boundary layer
1) Wind shear (S)
Day
+
U
2) Buoyancy (B)
+
Day:
S>0
B>0
Night:
S>0
B<0
Night
21
Observations of temperature fluctuations at several
heights over flat terrain
22
Spectrum of turbulence kinetic energy
and the turbulence cascade
“Big whirls have little whirls, That feed
on their velocity; And little whirls have
smaller whirls, and so on to viscosity”
(L.F. Richardson, 1928)
Eddies size of boundary layer depth to smallest scale of about 1 mm!
(Kolmogorov scale)
23
Turbulent heat flux and high frequency data
Reynolds
decomposition

w  w  w'
    '
thus
w'  0,  '  0
and
w'  0, w '  0
24

'
w
w
w'
  
w  w'  w 'w' ' 
w  w  w'    ' 
w  w' '
Flux is a co-variance
The origin of turbulent vertical heat flux
25
The origin of turbulent vertical heat flux
26
Turbulent heat flux
27



w  w  w'    '  w  w' '
w0
near surface, thus:
w  w' '
H  C p w' '
Turbulent heat flux
W
kg J m
J
 3
K 2
2
m
m kg K s
sm
28
Break
Progress in the Atmospheric Sciences: Connecting scales
Meteosat observations versus ECMWF predictions (T1279 ~ 15 km)
(Courtesy MeteoSat and ECMWF)
Atmospheric Budget Equations
Sub-grid processes
Discretization
Budget equation for potential temperature
after Reynolds decomposition
d  



 w' '

U
V
W

...
dt
t
x
y
z
z

w' '   K
 ....
z
K: Eddy diffusivity for heat
Similar equations for wind and any ‘conserved’ quantity C
(specific humidity, tracers,…)
Can be solved for given initial and surface conditions,
and if the diffusivities (K) are known
31
How to deal with turbulent mixing
and eddy diffusivity?
K  f ( height , stability, ABL depth ,...)
Strategy
Distinguish between stable and unstable conditions
Use theoretical concepts, observations and
numerical simulations (LES and DNS) as a guidance
Compare model results with independent data
Many turbulent mixing descriptions in the literature!
32
Non-local mixing in convective boundary layers
Use profile function for eddy-diffusivity and
non-local (‘counter-gradient’) corrections
 

w   K 
 
 z

Deardorff (1966, 1972)
Troen and Mahrt (1986)
Holtslag and Boville (1993)
Large et al (1994)
Hong and Pan (1996)
Beljaars and Viterbo (1998)
Lock et al (2000)
et cetera
This approach is now widely adopted (e.g., NCAR, NCEP, UKMO, ECMWF,...),
but details differ...
Flux-Gradient Theory (first order closure)
C
w' c'   K
z
U 2
K
l Fm ,h ( Ri )
z
l : length scale
g 
Ri 
 z
U
z
2
Turbulent Flux
Diffusivity K depends on
characteristics of turbulent flow
Richardson number
(measure for local stability)
34
Stable boundary
layer mixing
Enhanced friction
needed in models
Fm
Three alternatives for
stability functions of
heat and momentum
LTG
MO
Revised LTG
LTG: Louis et al (1982)
MO: Extended MoninObukhov type, in
agreement with Cabauw
Fh
tower observations
(Beljaars and Holtslag,
1991)
Revised LTG
MO
LTG
35
Mean model difference in 2 meter temperature for
January 1996 using two different stability functions in
ECMWF model (Courtesy A. Beljaars)
36
Turbulent Kinetic Energy (TKE) closure scheme
(applied in many atmospheric models…)
TKE
 Adv  Shear  Buoy  Transport  Dissipation
t
K  l TKE
TKE represents the major characteristics of turbulence
How to do the length scale?
Is local gradient assumption valid?
Many proposals have been made for various boundary layers
37
Does more complexity help?
38
Modeling Atmospheric Boundary Layers:
It is still a challenge!
Atmospheric models do have problems in representing
the stable boundary layer and the diurnal cycle
Sensitivity to details in mixing formulation
Strategy
Enhance understanding by benchmark studies
over land and ice in comparison with observations
and fine scale numerical model results
So far focus on clear skies!
GEWEX Atmospheric Boundary Layer Studies (GABLS)
provides platform for model intercomparison and development to
benefit studies of Climate, Weather, Air Quality and Wind Energy
GABLS1
GABLS2
GABLS3
LES as reference
Data (CASES99)
Data (CABAUW)
Academic set up
Idealized forcings Realistic forcings
Prescribed Ts
Prescribed Ts
Full coupling (SCM)
Prescribed Ts (LES)
No Radiation
No Radiation
Radiation included
Turbulent mixing
Diurnal cycle
Low levet jet +
transitions
LES: Large Eddy Simulation; SCM: Single Column Model
GABLS3: SCM and LES model studies
GABLS3-SCM
12 Z
01 July
2006
12 Z
02 Jul
GABLS3-LES
00 Z
02 Jul
09 Z
02 Jul
Initialization Profiles
Cabauw tower, Profiler, De Bilt Sounding
Geostrophic Wind (time-height dependent)
Similar for both SCM and LES
Large-scale Advection (time-height dependent)
Similar for both SCM and LES
Cabauw tower
(KNMI, NL)
Surface Boundary Conditions
Cabauw tower
41
GABLS3
intercomparison of
Single Column versions
(SCM) of operational
and research models
(Coordinated by
Fred Bosveld, KNMI)
Note:
Each SCM uses its
own radiation and land
surface scheme
interacting with the
boundary layer scheme
on usual resolution!
(Nlev is number of
vertical levels in whole
atmosphere)
Name
Institute
Nlev
BL.Scheme
Skin
ALADIN
Meteo France
41
TKE
No
AROME
Meteo France
41
TKE
No
GLBL38
Met Office
38
K (long tail)
Yes
UK4L70
Met Office
70
K (short tail)
Yes
D91
WUR
91
K
Yes
GEM
Env. Cananda
89
TKE-l
No
ACM2
NOAA
155
K+non-local
No
WRF YSU
NOAA
61
K
No
WRF MYJ
NOAA
61
TKE-l
No
WRFTEMF
NOAA
61
Total E
No
COSMO
DWD
41
GFS
NCEP
57
K
Yes
WRF MYJ
NCEP
57
TKE-l
Yes
WRF YSU
NCEP
57
K
Yes
MIUU
MISU
65
2nd order
MUSC
KNMI
41
TKE-l
No
RACMO
KNMI
80
TKE
Yes
C31R1
ECMWF
80
K
Yes
CLUBB
UWM
250
Higher order
No
Surface fluxes
noon
midnight
noon
Surface fluxes
noon
midnight
noon
Surface fluxes
noon
midnight
noon
Surface fluxes
noon
midnight
noon
noon
midnight
noon
Surface parameters
noon
midnight
noon
Low level jet
noon
midnight
noon
Novelty: Process diagrams, here for the influence of mixing
Sensitivity runs
with RACMO-SCM
using various
parameters and
forcings
GABLS3 Large Eddy Simulation intercomparison
(coordinated by Sukanta Basu, NC State Univ)
Initialized at midnight (02-jul-2006 00:00 UTC) and run for 9 hours (11 LES models)
Prescribed temperature at lowest model level from observations!
Wind speed magnitude
Potential Temperature
Mean profiles 03-04 UTC
(Red dots: Tower; Blue squares: Wind Profiler)
Green dashed line: 1m LES run (Courtesy Siegfried Raasch)
ISARS-2010, Paris, 28-30 June 2010
GABLS3 Large Eddy Simulation intercomparison
(coordinated by Sukanta Basu, NC State Univ)
Sensible Heat flux (W/m2)
Momentum flux (N/m2)
Flux profiles 03-04 UTC
(Red dots: Cabauw Tower observations)
Green dashed line: 1m LES run (Courtesy Siegfried Raasch)
ISARS-2010, Paris, 28-30 June 2010
Temporal evolution
53
53
Holtslag et al, 2013, BAMS
Holtslag et al, 2013, BAMS
Diurnal cycles of temperature and wind –
A challenge for weather and climate models!
Significant variation are seen in models which can be
related to relevant atmospheric and land surface
processes
Sensitivity to details in mixing formulation,
interaction with the land surface,
the representation of radiation (divergence),
et cetera
Overview of results and citations in
Holtslag et al, 2013, Bulletin American Meteorological
Society (BAMS, online)
56
At present GABLS4:
Antarctic summer case (Dome-C)
(Eric Bazile, Fleur Couvreux et al)
http://www.cnrm.meteo.fr/aladin/meshtml/
Helsinki meeting 14/15 Feb 2013
GABLS4/GABLS4.html
GABLS basic publications
(plus many conference and invited presentations)
GABLS1:
Special issue Feb 2006, Boundary Layer Meteorology (7 papers)
Svensson and Holtslag, 2009, BLM (wind turning issue)
GABLS2:
Steeneveld et al, 2006, JAS (SCM) and 2008, JAMC (Mesoscale study)
Holtslag et al, 2007, BLM (Coupling to land surface)
Kumar et al, 2010, JAMC (LES study)
Svensson et al, 2011, BLM (SCM intercomparison)
GABLS3:
Baas et al 2010, QJRMS (set up case and SCM tests)
Special issue of BLM 2014, including intercomparison
papers by Bosveld et al (2014, SCM), Edwards et al (2014, LES +
Radiation)....
GABLS overview paper in 2013 (Holtslag et al, BAMS, online)
58