AEROSOL INDIRECT EFFECT
Download
Report
Transcript AEROSOL INDIRECT EFFECT
AEROSOL INDIRECT EFFECT:
THE ELUSIVE COMPONENT OF
CLIMATE CHANGE
Athanasios Nenes
HDGC Seminar, November 20, 2002
photo: G.Roberts
Global warming vs. climate change.
Global warming is only one
aspect.
We are really looking at is
climate change.
Air pollution is much more
than greenhouse gases.
Aerosols (suspended particles)
are a major component: why?
2
Research focus: climate change.
Aerosols and their interaction with clouds play a major role in the climate system.
J.T. Houghton: “The science of climate change”
3
Why are we interested in aerosols?
Aerosols are emitted together with
greenhouse gases.
Pollution plume 1500 km south of India.
Outside of the plume. Pristine state.
4
Why are we interested in aerosols?
Pollution plumes off of Asia: they are of continental proportions.
5
Why are we interested in aerosols?
Biomass burning in the Amazon.
6
Research focus: aerosol effects on clouds.
Clouds play a major role in the climate system.
A small change in cloud properties can strongly affect climate.
7
Question: How do clouds form? Answer: from aerosols.
Clouds form in regions of the atmosphere where water vapor is supersaturated.
We focus on liquid water clouds.
Water vapor supersaturation is generated by cooling (primarily through
expansion in updraft regions and radiative cooling)
Cloud droplets form from pre-existing particles found in the atmosphere
(aerosols). This process is known as activation.
Aerosols that can become droplets are called cloud condensation nuclei (CCN).
Cloud
Aerosol particle
that does not activate
CCN that activates
into a cloud drop
8
Example of cloud droplet formation in an updraft.
log10(concentration)
movie
drop growth
activation
aerosol
log10(size)
9
What is the aerosol indirect effect?
It is the change in cloud properties caused by a change in the CCN population.
Cloud properties are a strong function of droplet concentration.
Two kinds of indirect effects lead to climatic cooling :
• Increase in cloud reflectivity
• Increase in cloud lifetime & coverage.
Less polluted
More polluted
Smaller droplets: clouds reflect more and last longer
10
Observational evidence of indirect effect
“Ship tracks”: linear features of high cloud reflectivity embedded in
marine stratus clouds, resulting from aerosols emitted by ships.
M.Kulmala: “Nucleation and Atmospheric Aerosols, 1996”
11
Anthropogenic indirect forcing: least understood.
• Potentially of large magnitude. (comparable to greenhouse gas warming)
• Cooling effect (counteracts greenhouse gas warming).
• Large uncertainty.
My focus
12
Why is the indirect effect poorly understood?
Indirect forcing uncertainties arise because:
• Aerosol-cloud interactions take place at smaller spatial scales than climate
models can resolve, and must be parameterized.
• Aerosol-cloud interactions are complex; many aspects are unknown or poorly
understood.
• Climate models provide limited information about clouds, and aerosols.
Central problem of indirect effect:
Determine the relationship between aerosol and cloud radiative properties,
using the limited information available by climate models.
This problem has historically been reduced to finding the relationship between
aerosol mass concentration and cloud droplet number concentration.
13
Droplet Concentration
Current understanding: empirical
(Boucher & Lohmann, 1995)
Aerosol mass concentration
• Very large variability.
• Unresolved meteorology, cloud microphysics, aerosol chemical factors are
responsible for the variability.
• Need for physically-based parameterizations.
14
Desired approach: from first principles.
Cloud droplet number balance in each grid box of the model:
dN drop
dt
Qactivation Qevap Qadvection ...
Activation is the direct aerosol-cloud droplet link.
Embedding a numerical activation model is too slow; must use
parameterization of activation.
Parameterizations of aerosol activation have appeared in the literature
over the years (from 1959). They are derived assuming: idealized cloud
dynamics, aerosol composition and size distribution.
Are current parameterizations good enough? (Hint: No).
15
Parameterizations: prescribed size distribution bias
Fitting ambient size distributions to prescribed functional form
introduces biases which can be important for indirect effect.
This aerosol is
“shifted” to
larger sizes.
16
Current parameterizations: other weaknesses
Lack of explicit treatment of mass transfer limitations in droplet growth; this
has been shown to be important for polluted conditions (Nenes et al., 2001).
Empirical correlations are used in many. They are derived from numerical
simulations and can introduce biases when used outside their region of
applicability.
They lack important chemical effects that can influence cloud droplet
formation. Such effects are the presence of :
• slightly soluble species in the aerosol (Shulman et al., 1996)
• water soluble gas-phase species (Kulmala et al., 1993)
• surface tension changes from surface-active species in the aerosol
(Facchini et al., 1999)
• changes in water vapor accommodation coefficient from the presence of
film-forming compounds (Feingold & Chuang, 2002)
17
Currently unaccounted “chemical” effects.
Slightly soluble compounds:
They add solute to the drop as it grows; this facilitates their ability to activate.
Examples: organics (succinic acid), CaSO4.
Soluble gases:
They add solute to the drop as it grows; this facilitates their ability to activate.
Examples: HNO3, HCl, NH3.
A(g)
A(aq)
A(g)
A(g)
A(aq)
A(aq)
18
Currently unaccounted “chemical” effects
Surface-active soluble compounds:
They decrease surface tension of droplets; this facilitates their ability to activate.
Examples: organics (succinic acid, humic substances).
Their solubility can be large.
75
Data of surface tension of
concentrated samples
from Tenerife clouds
and Po Valley fogs
-1
(kN m )
70
65
60
(Charlson et al., 2001)
55
50
1e-4
1e-3
1e-2
1e-1
-1
C(mol l )
19
Currently unaccounted “chemical” effects
Film-forming compounds:
They can slow down droplet growth. Once the film breaks, rapid growth
is resumed:
Film breaks
water
molecule
water
molecule
Slow
Rapid
Examples: hydrophobic organics.
Such substances do not alter droplet thermodynamics; kinetics of droplet
growth are affected.
If present, such substances can strongly affect droplet number.
20
Cloud droplet formation with film-forming compounds.
log10(concentration)
movie
drop growth
activation
aerosol
log10(size)
21
Chemical effects: assessment of their importance.
Calculate the potential change in cloud properties when a chemical effect
is present.
0.1 m s-1
0.3 m s-1
1.0 m s-1
3.0 m s-1
insoluble
organic
no D
organic
with D
5 ppb
HNO3
x2
conc.
marine aerosol
22
Chemical effects: summary, implications for parameterizations
Chemical effects are seen to be important for many conditions.
They can even be more effective than doubling aerosol concentrations (i.e.
more effective than the “Twomey” effect).
Chemical effects can be synergistic. One effect can be important for low
updrafts (e.g. soluble gas effects) and another at higher updrafts (e.g. surface
tension effects). This would lead to a systematic increase in droplet number
for almost any cloud type.
Lack of including them in activation parameterizations can lead to important
uncertainties in indirect forcing.
What does all this mean for current aerosol activation parameterizations?
They are not adequate.
We need to develop a new approach.
23
New parameterization: Underlying ideas
section i
d(Naerosol)/d(Size)
Use sectional representation of aerosol chemistry and size distribution.
Each section can:
• have its own chemical composition
• i-th section characterized by (i-1, i) boundaries
• piecewise linear profiles between boundaries
Size
Multiple populations with their own distributions can co-exist and
compete for water vapor. Modified Köhler theory for computing CCN
properties.
24
New parameterization: conceptual framework
Properties calculated from energy, mass
balances. Adiabatic parcel model used
to calculate droplet number.
• Lagrangian framework of reference
• Parcel properties are uniform
t
drop growth
Smax
activation
• Constant updraft velocity
• Parcel pressure is equal to ambient
• Explicit treatment of mass transfer
limitations of water vapor to droplet
phase.
aerosol
S
25
New parameterization: Formulation
Input: P,T, updraft velocity (cooling rate), RH, aerosol characteristics.
Output: Droplet number
How:
Solve the algebraic equation (numerically) :
wGSmax
2
aV
Spart
C1
0
f1 ( s)ds C2 f 2 ( s)ds 1 0
S p a rt
S max
i
Sci
2 part N i Ai
i
ln i 1
i 1
3 i 1 Sc Sc Sc
G
aV
1/ 2 i part
Ni
x 2
2
S
x
max
i
i 1
2
i 1 S c S c
RT
M w L2
P sat M
c p PM aT
w
1/ 2
x Si
S max
x
Arc sin
2
S x Si1
RT
L M wL
G sat
1
P Dv M w kaT RT
1
26
Performance of new parameterization (200 test cases)
1.000
Activation ratio (New Parametrization)
SM1
SM2
SM3
SM4
0.100
SM5
SM6
TM1
TM2
0.010
0.001
0.001
0.010
0.100
1.000
Activation ratio (Parcel Model)
27
Performance of existing parameterization (Ghan et al, 2000)
1.000
Activation ratio (Ghan Parametrization)
SM1
SM2
SM3
SM4
0.100
SM5
TM1
TM2
0.010
0.001
0.001
0.010
0.100
1.000
Activation ratio (Parcel Model)
28
New parameterization: Marine aerosol with surfactants
0.70
Numerical Simulation
Parameterization (with s.t.)
Parameterization (without s.t.)
Activation Fraction
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0.1
1
10
Updraft Velocity (m/s)
29
New parameterization: assessment.
A powerful activation parameterization for climate models has been
developed for aerosol of:
• “arbitrary” (non-ideal) size distribution,
• complex chemical and size-dependant composition (surfactants,
slightly soluble substances present).
Furthermore, it
• is fast (> 1000 times quicker than full numerical parcel model).
• uses minimal amount of empirical information.
• is more robust and accurate than other parameterizations in use.
Some future directions:
• incorporate other activation effects.
• use it in a global model (GISS GCM with TOMAS).
30
New effect: Black Carbon heating
Black carbon exists in polluted aerosol; it absorbs visible sunlight and
heats the surrounding air. This can leads to decreased cloud coverage,
and climatic warming.
If black carbon is included in cloud droplets, the heat released can
increase the droplet temperature enough to affect the droplet
equilibrium. This is a new effect.
drop
BC core
Absence of heating
Presence of heating: droplet and gas phase get heated
31
New effect: Black Carbon heating
BC heating can have an important effect on CCN of large dry
diameter only. This results in three effects:
1. Inhibition of the activation of large CCN. This would make water
vapor available for the activation of smaller (and more numerous)
CCN. This tends to increase droplet concentrations and cool climate.
2. Release of heat into the gas phase drops parcel supersaturation.
This tends to decrease droplet concentrations, and warm climate.
3. Decreased size of low Sc CCN can decrease the chance of drizzle
formation (Giant CCN can initiate drizzle formation). This would
tend to increase cloud lifetime and cool climate.
Which of the mechanisms can prevail?
32
Black Carbon heating: effect on cloud albedo
Consider a case where Mechanism 1,2 are strongest.
0.005
Simulation is for urban
aerosol, 20% BC, 0.1 m s-1
updraft.
Albedo difference
0.000
Mechanism 1 alone.
-0.005
Conditions were chosen so
that the heating released from
the BC in each curve is the
same.
-0.010
Mechanism 1+2.
-0.015
Albedo calculated by a twostream approximation.
Differences are compared to
absence of BC heating.
Mechanism 2 alone.
-0.020
-0.025
0
50
100
150
Cloud depth (m)
Droplet changes are alone not significant but they can affect the
magnitude of albedo change from BC heating.
33
Black Carbon heating: effect on drizzle formation.
Assess the size of Giant CCN with and without BC heating, for a
marine stratocumulus cloud. We use a Large Eddy Simulation of nonprecipitating marine stratocumulus (Stevens et al., J.Atmos.Sci, 1996).
500 Lagrangian trajectories derived from the LES are used to “drive” a cloud
parcel model and grow the giant CCN in the cloud. Ensemble 1-hr averages of the
parcel properties yield average cloud properties.
34
Black Carbon heating: 50 selected LES trajectories.
Height
Cloud
movie
Horizontal extent
35
Activation along one of the 500 trajectories.
log10(concentration)
movie
log10(size)
36
Black Carbon heating: potential effect on drizzle
900
BC can effectively
decrease the probability
for drizzle formation.
500 parcel average
800
700
Cloud top
A heating mechanism
can lead to climatic
cooling!
Height (m)
600
500
Cloud base
This effect can be
parameterized (not
shown).
400
300
200
Is it important?
We don’t know yet.
0% BC, Pristine
10% BC, Pristine
20% BC, Pristine
100
0
0
10
20
30
GCCN average size (m)
40
50
37
CCN INSTRUMENTS: What do they do?
Purpose: Measure the number of cloud droplets that can form for a variety of
water vapor supersaturations (“CCN spectrum”), at a given point in space and
time.
Operation principle: Expose particle sample to a known water vapor
supersaturation, and measure those that become droplets.
Desired range: 0.01% - 1.0% supersaturation
38
CCN INSTRUMENTS: The need for theoretical analysis
Objective: Assess the performance and limitations of current CCN
measurement methodologies.
Few designs available; we focus on spectrometers:
• Most attractive for aircraft missions.
• Minimal theoretical assessment of their capabilities.
Aerosol
sample
Instantaneously
CCN
spectrum
Approach:
• Develop comprehensive models that express optimum instrument behavior.
• Simulate instrument response, assuming that inlet aerosol is monodisperse.
39
Fukuta & Saxena (1979) CCN Spectrometer (FCNS)
Outlet, droplet detection
Metal envelope
Insulation
Inlet
Metal envelope
Hot tip
Thot
Insulation
Tcold
Smax
Flow
Wetted walls
Cold tip
Property
(cross section)
Thot-Tcold
Smax
Distance from tips
40
FCNS: How is it used?
Inlet
Characteristics:
Each streamline has different S
S constant along a streamline
Droplet concentration
Outlet
Detection must distinguish
droplets from unactivated aerosol.
Usage:
Measure droplet concentration at
a (centerline) streamline.
CCN = droplet concentration.
Scan for all streamlines.
Streamline supersaturation
41
FCNS: Mathematical model
}
Growth “driving force”
Droplet phase:
Gas phase:
dD p
dt
S S
eq
(Lagrangian)
Dp
div u div S
t
transient
advection
diffusion
(Eulerian)
source
Conservation
Law
S
Continuity
1
0
0
x1 -momentum
u
x 2 -momentum
v
T
ka
cp
C
D v
Heat
Water vapor
x 2a P x 2a
x1
x1
u
x1
P
x 2a
x 2a
x2
x1
x2
u
x2
D H vap
cp
a v
x 2
x1
a v
x 2
x 2
x2
J buoy
av
x2
J cond
J cond
42
FCNS: Activation along different streamlines
log10(concentration)
movie
Each color represents a
different streamline
CCN activate as they flow
through the instrument
log10(size)
43
FCNS: Activation at outlet of instrument
poor
Note:
log10(concentration)
fair
Droplets separate
well from aerosol at
high S.
good
good separation
aerosol
Separation degrades
as S decreases.
droplets
log10(size)
44
FCNS: Why does performance degrade as S decreases?
Centerline supersaturation (%)
1.5
Outlet
1.0
High s: 1.05% (good)
Inlet
0.5
0.0
-0.5
-1.0
-1.5
0.0
0.1
0.2
0.3
0.4
0.5
0.6
y-position
Distance from
entrance(m)
of instrument (m)
0.7
45
FCNS: Why does performance degrade as S decreases?
Centerline supersaturation (%)
1.5
Outlet
1.0
Inlet
Medium s: 0.60% (good)
0.5
0.0
-0.5
-1.0
-1.5
0.0
0.1
0.2
0.3
0.4
0.5
0.6
y-position
Distance from
entrance(m)
of instrument (m)
0.7
46
FCNS: Why does performance degrade as S decreases?
Centerline supersaturation (%)
1.5
Outlet
1.0
Inlet
0.5
Low s: 0.14% (fair)
0.0
-0.5
-1.0
-1.5
0.0
0.1
0.2
0.3
0.4
0.5
0.6
y-position
Distance from
entrance(m)
of instrument (m)
0.7
47
FCNS: Why does performance degrade as S decreases?
Low S takes longer to develop than high S, and
CCN are exposed to S > Sc for less time.
Growth at low S is much slower than at high S;
activation takes longer than at high S.
Separation between droplets and non-activated
aerosol is lost at low S.
48
CCN INSTRUMENTS: Assessment
Fukuta CCN spectrometer:
• sensitivity problems under low supersaturations.
• optimum Sc range ~ 0.2% to 1.1%
• not sensitive to aerosol chemical characteristics.
All instruments displayed sensitivity problems for Sc < 0.08-0.1%.
Some display large sensitivity to chemical composition.
Model captures instrument behavior very well. Can be reliably used.
Conclusion:
• Instrumentation needs to be further improved. This is currently
being done (Caltech, UCSD).
• A deep theoretical understanding of any instrument is necessary for
reliable CCN measurements.
49
GENERAL SUMMARY
The indirect effect of atmospheric aerosols is one of
the most important and challenging aspects of climate prediction science.
A variety of aerosol activation effects need to be included in
parameterizations of aerosol-cloud interactions.
There are possibly many (counterintuitive) mechanisms to be discovered.
Parameterizations are being developed, and included within a comprehensive
climate model system.
CCN instrumentation is a key tool for untangling the aerosol-cloud puzzle.
Current instrumentation is not adequate in fulfilling its task. Source of
problems identified. New CCN instruments under construction will.
50
ACKNOWLEDGMENTS
People
John Seinfeld, Caltech
Rick Flagan, Caltech
Bill Conant, Caltech
Tracey Rissman, Caltech
Graham Feingold, NOAA
Bob Charlson, University of Washington
Christina Facchini, Instituto ISAO, C.N.R.
Steven Ghan, PNNL
Peter Adams, Carnegie Mellon
Greg Roberts, UCSD
Funding
EPA
NASA
ONR
51