Simulation of Below-cloud and In-cloud Aerosol Scavenging in ECHAM5-HAM

1
Simulation of Below-cloud and In-cloud Aerosol
Scavenging in ECHAM5-HAM

• Betty Croft, Ulrike Lohmann, Philip Stier, Sabine
  Wurzler, Sylvaine Ferrachat, Hans
  Feichter, Randall Martin, and
  Ulla Heikkilä
• ETH Group Retreat Presentation
• Einsiedeln, Switzerland
• February 6, 2008

2

• Motivation
• Below-cloud scavenging should depend on aerosol
  size and
• precipitation rates, as opposed to fixed
  scavenging ratios for
• each aerosol mode.
• In-cloud scavenging should be linked to the cloud
  microphysics
• and depend on cloud droplet (or ice crystal)
  number concentrations,
• cloud droplet size and aerosol size, as opposed
  to fixed scavenging
• ratios for each aerosol mode.
• Project Goals
• Size-dependent below-cloud scavenging
• Microphysically-dependent in-cloud scavenging

3
Wet scavenging of aerosols

• In-cloud scavenging processes
• (aerosol ? droplets or crystals)
• Nucleation
• Impaction

• Below-cloud scavenging processes
  (precipitation-aerosol collisions)
• Inertial impaction and interception
• Brownian motion
• Thermophoresis and diffusionphoresis
• Turbulence
• Electrostatic attraction

4
ECHAM5-HAM has 7 lognormal aerosol modes and
includes black carbon, particulate organic
matter, sulfate, sea salt and dust.
All results shown are from 5-year simulations
after 3-month spin-up.
5
Below-cloud scavenging coefficients for rain
Present-day GCMs use typically mean scavenging
coefficients (solid red steps). This study
selects mass (solid lines) and number (dashed
lines) below-cloud scavenging coefficients from a
look-up table based on aerosol size and rainfall
rate.
6
The scavenging coefficients are found assuming
both a raindrop (or cloud droplet) distribution
and a log-normal aerosol distribution
Below-cloud N(Dp) Marshall-Palmer
distribution In-cloud N(Dp) Gamma distribution
Then,
7
Below-cloud scavenging coefficients for snow
(Slinn, 1984) (normalized by precipitation
rate)
Previously, ECHAM5-HAM used 5x10-3 mm-1 for all
aerosol sizes (dashed line).
8
Global and annual mean deposition budgets (black
carbon) Below-cloud scavenging (BCS) is
increased with the new parameterization.
9
Simulated Geographic distribution of wet
deposition (SO4)
Changes in the annual mean wet deposition near
source regions can be above 10 as compared to
the simulation with mean coefficients. Scavenging
is increased for rain rates near 1mm/hr and
higher, but decreased for rain rates below 1mm/hr
. All scavenging by snow is increased.
10
Validation with MODIS-MISR (global zonal mean
optical depth comparison)
Compare the control simulation (MEANC) and
revised below cloud scavenging (ASDS-RS) with
solid red (observations from MODIS-MISR).
11
Validation with NADP data (observed sulfate
wet deposition from US)
Sea salt deposition has improved correlation
coefficients and slope-offset parameters in
simulation ASDS-RS as opposed to the MEANC
simulations. Sulfate deposition is more within
factor of 2.
12
Part II - In-cloud scavenging
1) Impaction scavenging (aerosol-cloud droplet
collisions)
Project goal Introduce aerosol size dependent
in-cloud impaction scavenging.
Look-up table is a function of mean cloud droplet
size, aerosol size and CDNC.
Mass (dashed) and number (solid) coefficients
13
In-cloud scavenging 2) Nucleation scavenging
parameterization
Standard ECHAM5-HAM uses fixed in-cloud
stratiform scavenging ratios for each of the 7
modes. These are 0.1, 0.25, 0.85, 0.99 for the
NS, KS, AS, and CS modes, respectively. Revised
scavenging parameterization is consistent with
the Lin and Leaitch cloud droplet activation
scheme. Assume CDNC total number of aerosols to
be scavenged . Scavenging ratio for ith mode is,

Where Na is the sum over all soluble modes of
the number of aerosols gt 35nm, and xfracni is
the fraction of aerosol number gt35 nm in the
ith mode.
14
We use the cumulative lognormal function to find
a critical radius where Ci is the number in the
lognormal tail if r gt rcrit. Scavenge all aerosol
mass above rcrit.
For mixed clouds, same approach but CDNCICNC
total number of aerosols scavenged. Ice clouds,
do not use same activation, so assume ICNC
total number scavenged and scavenge from largest
to smallest mode progressively and find rcrit for
the partially scavenged mode.
Alternatively, Tost et al., 2006 gave the
scavenging ratio as a function of aerosol radius.
We also tested this parameterization.
15
Predicted scavenging ratios (normalized
frequency of occurrence)
Warm stratiform clouds
Warm convective clouds
Generally, unity for AS and CS modes and greatest
variability in KS mode, zero for NS mode.
Greater variability in predicted convective cloud
scavenging ratios.
16
Example Dust deposition budgets - Higher
stratiform in-cloud scavenging and lower
convective in-cloud scavenging, comparing IC-ALL
with CTL.
17
Validation (global SO4 wet deposition
Dentener et al, 2006)
Correlation coefficients are improved by
revisions to in-cloud scavenging.
18

• Summary and future work
• Aerosol size-dependent below-cloud scavenging
  was introduced to ECHAM5-HAM and is a more
  physical representation of below-cloud scavenging
• Microphysically dependent in-cloud scavenging
  was implemented in all stratiform, and warm
  convective clouds and results are comparable with
  simulations using fixed coefficients and the
  method of Tost et al. (2005). This approach is
  desirable since the scavenging physics are now
  more consistent with the cloud parameterizations.
• Convective ice cloud scavenging will be
  implemented.
• The sensitivity of the below-cloud scavenging to
  the assumptions about the raindrop distribution
  will be investigated.
• Global validation of vertical profiles of
  extinction will be conducted with CALIPSO data to
  better examine influence of the scavenging
  parameterizations on the vertical aerosol
  profiles.

19
(No Transcript)
20
Simulations MEANC uses the existing mean
below-cloud scavenging coefficients ASDS-RS
revised below-cloud scavenging by both rain and
snow (aerosol size dependent scavenging). ASDS-R
revisions only for rain ASDS-RS-PF as ASDS-RS
but uses the old precipitation fraction
parameterization labelled as method 1 in the
subsequent slides. ASDS-RT as ASDS-R but add
the thermophoretic effects so that scavenging
also depends on the below-cloud relative
humidity IC-ALL revised in-cloud impaction and
nucleation scavenging IC-WARM revised in-cloud
scavenging only in warm clouds IC-WARM-T
applies the parameterization of Tost et al (2006)
for warm clouds IC-STRAT revised in-cloud
scavenging only for stratiform clouds
21
Collection efficiency for snow (Slinn 1984)
where,
Scavenging coefficient, normalized by
precipitation rate is,
Parameters are varied for different types of snow
(powder, rimed and dendrites).
22
Precipitation fraction parameterization
Methods for finding the fraction of grid box
that is raining (PF), using cloud fraction (CF)
all simulations use method 2 except ASDS-RS-PF
Method 1 (Stratiform)
Where CF(k) follows Tompkins, 2001
Method 2 (Stratiform) If
then
else
Weighting similar to method 1 only if CF(k) gt PF
(k-1).
23
Sensitivity to precipitation fraction
parameterization
Method 1 (Convective)
Based on updraft mass flux and velocity.
Method 2 (Convective)
Kiehl et al. 1996 Xu and Krueger (1991) PF (k)
is limited to be within the range of 0.05 to 0.8