Learning Center
Plans & pricing Sign in
Sign Out

ppt presentation Influences of In cloud


									          Influences of In-cloud Scavenging
            Parameterizations on Aerosol
         Concentrations and Deposition in the
                 ECHAM5-HAM GCM
Betty Croft - Dalhousie University, Halifax, Canada
Ulrike Lohmann - ETH Zurich, Zurich, Switzerland
Randall Martin - Dalhousie University, Halifax, Canada
Philip Stier - University of Oxford, Oxford, U.K.
Johann Feichter - MPI for Meteorology, Hamburg, Germany
Sabine Wurzler – LANUV, Recklinghausen, Germany
Corinna Hoose - University of Oslo, Oslo, Norway
Ulla Heikkilä - Bjerknes Centre for Climate Research, Bergen, Norway
Aaron van Donkelaar –Dalhousie University, Halifax, Canada

CAFC Winter Meeting, Toronto, February 1, 2010
Do global models agree on predicted aerosol profiles?

  Koch et al.
  (2009), ACP
  Black carbon
  profiles differ
  by 2 orders of
  among global

 Deficits in our understanding of the processes involved and their interactions
The aerosol-cloud-precipitation interaction puzzle:

AEROSOLS              CLOUDS                 PRECIPITATION

  This problem involves many processes, and isolating the
  effects of one or the other is difficult.
Aerosol Scavenging Processes:

                                                  Sedimentation and
                                                  dry deposition

                                                          (Figure adapted
Wet scavenging accounts for 50-95% of                     from Hoose et al.
aerosol deposition, and strongly controls                 (2008))
aerosol 3-dimensional distributions, which
influence climate both directly and indirectly.
  Aerosol wet scavenging processes:

  Aerosols  Cloud Droplets / Ice Crystals  Precipitation
    Processes:                      Processes:
    1) Nucleation of                In-cloud: (tuning parameters)
                                    1) Autoconversion
    2) Impaction with
                                    2) Accretion
                                    3) Aggregation
We examine the relative             Below-cloud:
contributions of nucleation and
impaction to in-cloud scavenging.   1) Impaction with rain/snow
Modeling of Aerosol In-Cloud Scavenging:
Methodologies -
1) Prescribed scavenging ratios (e.g., Stier et al. (2005))
2) Diagnostic - cloud droplet and ice crystal number
   concentrations are used to diagnose nucleation
   scavenging + size-dependent impaction scavenging
   (e.g. Croft et al. (2009))
3) Prognostic - in-droplet and in-crystal aerosol
   concentrations are prognostic species that are passed
   between model time-steps (e.g., Hoose et al. (2008))
  Using the ECHAM5-HAM GCM, we compare the
  strength/weaknesses of these 3 fundamental approaches,
  and examine the sensitivity of predicted aerosol profiles to
  differences in the parameterization of in-cloud scavenging.
ECHAM5 GCM coupled to HAM (Hamburg Aerosol Module):
(Stier et al. (2005))

All results shown are for a 1-year simulation of the ECHAM5-HAM global
aerosol-climate model, at T42 resolution, nudged to the meteorological
conditions of the year 2001, and following a 3 months spin-up period.
SU:sulfate; BC:black carbon; POM:particulate organic matter; DU:dust;
SS:sea salt
1) Prescribed in-cloud scavenging ratios:
standard ECHAM5-HAM (nucleation+impaction)
   Prescribed In-Cloud Scavenging Ratio



                                                                                   Liquid   T>273K
                                          0.6                                      Mixed    238<T<273K
                                                                                   Ice      T<238K


                                                NS   KS   AS   CS   KI   AI   CI
2) Diagnostic scheme: Size-Dependent Nucleation Scavenging

Assume each cloud droplet or ice crystal
scavenges 1 aerosol by nucleation, and apportion
this number between the j=1-4 soluble modes,

based on the fractional contribution of each mode
to the total number of soluble aerosols having radii
>35 nm, which are the aerosols that participate in
the Ghan et al. (1993) activation scheme.
                                                       Find rcrit that contains Nscav,j in
                                                       the lognormal tail.
 From the cumulative lognormal size-distribution,

 Scavenge all mass above this radius for nucleation scavenging. Thus, we typically
 scavenge a higher fraction of the mass versus number distribution.
Size-Dependent Impaction Scavenging by Cloud Droplets:

                                                   Example for
                                                   CDNC 40 cm-3,
                                                   assuming a

                                                   coefficients of
                                                   Hoose et al.
                                                   scheme are
                                                   shown with
                                                   red steps

   Solid lines: Number scavenging coefficients
   Dashed lines: Mass scavenging coefficients
   Data sources described in Croft et al. (2009)
Impaction Scavenging by Column and Plate Ice Crystals:

                                                                         coefficients of
                                                                         Hoose et al.
                                                                         (red steps)

Assume plates for 238.15<T<273.15 K      Assume columns for T<238.15K

(Data from Miller and Wang, (1991), and following Croft et al. (2009))
3) Prognostic scheme: Aerosol-cloud processing approach
(Hoose et al. (2008))

Stratiform in-droplet and
in-crystal aerosol
concentrations are
additional prognostic

Two new aerosol modes 
In-droplet (CD)
In-crystal (IC)
Histograms of diagnosed vs. prescribed scavenging ratios:

  mode 

mode 

mode 

                T>273 K       238<T<273 K      T<238 K
Uncertainty in global and annual mean mass burdens:




  20                                          F100-CTL
[%] 0


        SO4   BC    POM     DUST    SS
Uncertainties in Aerosol
Mass Mixing Ratios:
 Zonal and annual mean black
carbon mass is increased by
near to one order of magnitude
in regions of mixed and ice
phase clouds relative to the
simulation with prescribed
scavenging ratios.
Uncertainties in Accumulation
Mode Number:
Assuming 100% of the in-cloud
aerosol is cloud –borne reduces the
accumulation mode number burden
by up to 0.7, but the diagnostic and
prognostic scheme give increases
up to 2 and 5 times, respectively
relative to the prescribed fractions.
Uncertainties for Nucleation
Mode Number:
 Increased new particle
nucleation is found for the
simulation that assumes 100% of
the in-cloud aerosol is cloud-

       Uncertainties in
       Aerosol Size:
       The size of the
       mode particles
       changes by up to
Contributions of nucleation vs. impaction to annual and
global mean stratiform in-cloud scavenging: Diag. scheme



  60                                                                     Nuc (Warm)
  50                                                                     Nuc (Mixed)
                                                                         Nuc (Ice)
 [%]                                                                     Imp (Warm)
  30                                                                     Imp (Mixed)
  20                                                                     Imp (Ice)


        SO4        BC       POM        Dust     SS         Number
>90% of mass scavenging by nucleation (dust:50%); >90% of number scavenging by impaction.
Influence of impaction on dust scavenged mass:
Influence of impaction on black carbon scavenged mass:
Observed black carbon profiles from aircraft (Koch et al. 2009)
Observations of MBL size distributions (Heintzenberg et al. (2000))
Observations of AOD from MODIS MISR composite (van Donkelaar et al., subm.)
Observations of sulfate wet deposition (Dentener et al. (2006))
Observed 210Pb and 7Be concentrations and deposition (Heikkilä et al. (2008))
  Current work: Coupled Stratiform-Convective Aerosol Processing:
           Stratiform Clouds                  Convective Clouds


CDVC and ICCV will not be prognostic variables since the convective clouds
entirely evaporate or sublimate after the above processes for each GCM timestep.
Aerosol Processing by Convective Clouds:

               CN: Solid                              CCN0.6/CN

               CCN0.6: Dotted

Red: 12 hours before convective system       Evidence for dust coating by
                                             sulfate above the boundary
Blue: 12 hours after convective system       layer as a result of cloud

Figure from Crumeyrolle et al. (2008), ACP
- case study from Niger.
 Summary and Outlook:
 1) Mixed /ice phase cloud scavenging was most uncertain between the
    parameterizations. Middle/upper troposphere black carbon
    concentrations differed by more than 1 order of magnitude between the
    scavenging schemes. Recommend:  understanding nucleation and
    impaction processes for cloud temperatures T<273K.
 2) In stratiform clouds, number scavenging is primarily (>90%) by
    impaction, and largely in mixed and ice phase clouds (>99%). Mass
    scavenging is primarily (>90%) by nucleation, except for dust (50%).
    Recommend:  understanding of impaction processes for cloud
    temperatures <273K, and for dust at all cloud temperatures.
 3) Better agreement with black carbon profiles for diagnostic and
    prognostic schemes.  ↓ prescribed ratios for mixed phase clouds.
 4) Recommend diagnostic and prognostic schemes over the prescribed
    ratio scheme, which can not represent variability of scavenging ratios.
 5) Recommend further development of the prognostic aerosol cloud
    processing approach for convective clouds.

Acknowledgements:                                          Thanks!

To top