Atmospheric Dispersion

1
Title
Convective boundary layers Jordi
Vilà Collaboration and discussions David Pino,
Olaf Vellinga, Reinder Ronda, Harm Jonker, Bert
Holtslag UCLA, Lake Arrowhead, June 2005
2
Forum on modeling the atmospheric boundary layer
Forum on modeling the atmospheric boundary
layer (summary in BAMS (2005) by R. Pielke and T.
Vukicevic) The third conclusion is that
parameterizations can be replaced with lookup
tables or analytical formulations... With
this approach the existing parameterizations of
sub-grid scale fluxes, radiative fluxes,
... would be replaced by lookup tables or
analytical formulations.
3
My approach to the talk Critical user of a
research/operational mesoscale model with
particular interest on the performance of the
planetary boundary layer. Here, the majority
of the results are based on runs carried out by
the mesoscale MM5 (PSU-NCAR), but the results
are similar to other mesoscale models.
4

• Advantages of using MM5
• Several parameterizations/boundary layer schemes
  are implemented.
• Possibility to investigate and to teach
• the impact of physical assumptions on
• the representation of the CBL
• c) Possibility to study feedbacks (but
• sometimes this is also a disadvantage)

5
Essential processes to be represented in the CBL
(3) Entrainment (exchange fluxes)
(1) Turbulent mixing (turbulent fluxes)
(2) Land/surface processes (surface fluxes)
6

• (1) Turbulent mixing
• Two main schools/approaches to represent
• turbulent mixing in the CBL
• Non-local convective scaling
• Local prognostic TKE equation

7
Non-local approach Main goal to include the
influence of coherent structure (organized
eddies) which fill the entire boundary layer
Argot parcel method, countergradient,
transilient, ...
8
Local approach Main goal to represent the
turbulent characteristics of the CBL by
calculating its main relevant variable the
TURBULENT KINETIC ENERGY Argot master
turbulent length scale, Mellor-Yamada levels,...
9
Motivation a) Study of the capabilities of a
widely used mesoscale model (NWP and
regional climate modeling) to simulate the
evolution of a convective boundary layer b)
Evaluating with observations the performance of
several boundary layer schemes derived with
different physical assumptions
10
Case study Synoptic situation (2/3 May
1995) High pressure over Denmark Sunny day
(cloudless) Low horizontal easterly winds
Suitable conditions for the formation of a well
mixed convective boundary layer
11
Numerical experimental setup
Four domains (27, 9, 3, 1 km grid length) Two
way nesting Initial and boundary
conditions provided by ECMWF (every 6
hours) Available observations at the Cabauw mast
and De Bilt.
12

• MM5 model has several options to
• parameterize the turbulent flux in the
• ABL (boundary layer schemes)
• Two main categories
• 1-order non local (MRF, Blackadar)
• 1-1/2-order local (ETA, BRT)

13
Vertical profiles of potential temperature
Non-Local
Local
14
Vertical profiles of specific humidity
Non-local
Local
15
Vertical profile of wind
Local
Non-local
16
Time evolution of the convective boundary layer
height
Non-local
Non-local
Local
17
Other MM5/RAMS research studies which evaluated
the boundary layer height
MM5 o RAMS
Observations
Model
Model
Observations
Zhong and Fast (2003)
Hanna and Yang (2001)
18
Discussion points (I) The first-order non local
schemes reproduce better the vertical
structure of the convective boundary layer
and the boundary layer growth. Wind is poorly
represented by all schemes
19

• Discussion points (II)
• All the BL schemes reproduce normally colder and
  moister boundary layer compared to observations
• soil moisture content
• lower mixing height

20
(2) Land/surface processes Is the surface
forcing represented by the surface turbulent
fluxes consistent with the previous results?
21
Time evolution of the surface fluxes
Sensible heat flux
Non-local
Local
22
Sensible heat flux MM5 results compare to
observations (Berg and Zhong, 2004)
Model
(non-local)
(non local-TKE)
(non-local)
Observed
23
Time evolution of the surface fluxes
Latent heat flux
Local
Non-local
24
Time evolution of the surface fluxes
Friction velocity
Non-local
Local
25
Discussion points (III) Normally there is an
overestimation of the momentum, sensible and
latent heat flux compared with observations
26
Discussion points (IV) Still (at least for me)
not well understood the interaction and feedbacks
between the surface schemes and boundary
layer schemes on 3D models SL scheme
BL scheme
27
(3) Entrainment The third essential process is
entrainment
Top-down diffusion of an scalar.
Laboratory experiment by Harm Jonker (TUD)
28
How do we represent the entrainment processes in
the boundary layer schemes? Majority of the
analyzed and discussed boundary layer schemes
represent implicitly the entrainment of warm and
dry air from the free troposphere
29
Mixed-layer models prescribed or calculated
explicitly the entrainment flux at the top of the
boundary layer. Commonly,
Free convection case ? 0.2
30
The ECMWF model also prescribes explicitly that
the entrainment flux is equal to -20 the
surface buoyancy flux.
Beljaars and Betts (1993)
31

• Is it possible to improve/generalize
• this closure assumption?
• To allow departures from the constant 0.2
• To introduce other relevant processes
• at the entrainment zone (shear, dissipation)

32
Entrainment process is closely related to the
energy that is made available from turbulence to
overcome the buoyancy forces at the inversion

• analysis by scaling
• the terms in theTKE equation at
• the entrainment interface

33
TKE at the entrainment interface
TE
B
S
TP
D
34
Previous research studies Lilly (1968) Tennekes
(1973) Betts (1973) Carson (1973) Zeman
(1975) Mahrt and Lenschow (1976) Driedonks
(1981) ..... Can we add something?
35
LES can provide new insight and quantification
B buoyancy only BG buoyancy and shear
(initial UgVg10 m/s)
36
TKE budget in an ABL only driven by buoyancy (B)
37
TKE budget in an ABL driven by buoyancy and
shear (BG)
38
Terms taking into account in the TKE equation at
the entrainment interface (so far)
TE B S P T D
39
Ratio of the entrainment to the surface flux
Entrainment parameterization implemented in a
mixed-layer model and compared to LES.
N
P
LES MXL
40
Boundary layer schemes have also an impact on
atmospheric chemistry studies!!!
41
Some considerations of the influence of boundary
layer schemes on atmospheric chemistry
studies a) Boundary layer controls the diurnal
variability of atmospheric compounds Example
transition cloudless BL to cloudy BL -
Enhancement of vertical transport - As a result,
reactant concentration are more
diluted
42
Enhancement of vertical transport.
Higher dilution of atmospheric compounds.
Scatter clouds appear
Cloudless
Wind profile measurements H. Klein-Baltink (KNMI)
43
b) Turbulent mixing can control/limit the
reactivity of atmospheric compounds Testing the
impact and validity of assumptions such as -
Homogeneous mixing (asymmetry of the transport in
the CBL) - Validity of K-theory
44
Turbulent reacting flows (first-order
reaction) LES (Jonker et al., 2004)
Da lt 1 slow chemistry Da 1 moderate chemistry
Da gt 1 fast chemistry
Mixing ratio reactant
45
Last discussion/summary points Non-local schemes
perform better than local-TKE schemes. Need to do
a systematic evaluation. Study the
feedback/interaction of BL schemes with surface
schemes The entrainment processes should
be treated explicitly or implictly.
46
Further Answering the Forum. It is still
necessary to parameterize the essential processes
that drive the CBL evolution and
characteristics Need of strong interaction
between developers of BL-schemes and
users (mesoscale, regional climate
models,...) The parameterizations must have a
balance between relevant processes and
retaining simplicity (easy to say it than to do
it)