Integrating global chemical and transport models and satellite observations: Implications for the Ar

1
Integrating global chemical and transport models
and satellite observations Implications for the
Arctic aerosol burden
Sylvia Generoso, Isabelle Bey
LMCA, École Polytechnique Fédérale de Lausanne,
Switzerland
Matthieu Labonne, François-Marie Bréon
LSCE / IPSL, CEA - CNRS, Gif-sur-Yvette, France
Jean-Luc Attié
LA, Toulouse, France
Contact sylvia.generoso_at_epfl.ch
2
Our approach to reduce present uncertainties
Satellites
Field experiments
Merging information from models and observations
Limitations no detection in the presence of
clouds, no information on the chemical nature
Limitations localized in time and space
Need of complete 4D aerosol distributions in
order to study their impacts
3
1. Assimilation of POLDER Aerosol Optical
Thickness (AOT) into the LMDz-INCA model
Implications for the Arctic aerosol burden 2.
Impact of boreal fires on the Arctic aerosol
burden using the GEOS-Chem model and satellite
observations (POLDER, MODIS, MOPITT) 3.
Evaluation of aerosol vertical distribution in
GEOS-Chem Comparisons with CALIPSO (preliminary
results)
4
Part. 1
Assimilation method (Kalman filter)
At each time step
Assimilated fields


Obs
Model
xb K(y0-Hxb) xa
K BHt ( HBHt O)-1
B (a priori) O (observations)


In this study
B and O are diagonal matrices
LMDz-INCA 96 x 73 x 19 (lon x lat x vert) 9
aerosol tracers
POLDER AOTfine (ocean land) AOTcoarse (ocean
land)
Results 1996-97 8 months 2003 7 months
POLDER (Polarization and Directionality of the
Earth Reflectance aboard ADEOS)
5
Part. 1
Comparison to ground-based measurements
AERONET
a posteriori
a priori
(Total AOT)
6
Part. 1
A polar view of the results
AOT
a posteriori - a priori
(increment)
fine mode
7
Part. 1
Impact on the Arctic aerosol load
Comparison to ground-based measurements
AERONET
a posteriori
a priori
Realistic seasonal variations within the Arctic
circle as observed for instance by Herber et al,
JGR, 2002
Generoso et al , J. Geophys. Res, 2007
8
Part. 2
The 2003 Russian fires
NASA / MODIS AOTfine May 2003
A potential impact for the Arctic Scientific
goals 1. The representation of boreal fires in a
CTM (fire emission strenght, time resolution,
injection height) 2. The impact of the 2003
Russian fires to the Arctic aerosol burden
Aerosol satellite products - POLDER AOTfine_865
(rlt0.5 µm) - MODIS AOTfine_550 (rlt0.5 µm)
CO satellite products - MOPITT
GEOS-Chem - resolution 2x2.5, 30 vertical
levels, 50 tracers, GEOS-4 meteorological
fields - biomass burning emissions for 2003
interannual and seasonal variations based on
AATSR fire counts Generoso et al, 2003 - AOT of
the fine mode (AOTfine) including carbonaceous
aerosols, sulfates and sea salt - CO columns
applying MOPITT averaging kernel
9
Part. 2
Model / satellites comparison in the North Pacific
(May-June-July-August 2003 seasonal mean)
POLDER
MODIS
MOPITT
R 0.65 B-1066
R 0.72 B-2961
R 0.78 B-99
10
Part. 2
Daily results averaged over the North Pacific

• Set-up for the Best Estimate (BE) simulation
• Fire emissions are
• increased by a factor 3 in May
• prescribed on a daily-basis
• Injected up to 4.5 km in July and August

11
Ratio between the Best Estimate (BE) and a
simulation with the Russian fire sources turned
off (woBB)
Contribution of the 2003 Russian fires to the
Northern Hemisphere
AOT
BC deposition
12
Part. 2
Contribution of the 2003 Russian fires to the
Arctic Haze events
Number of days (during MJJA 03) for which the
AOT550gt 0.094 threshold to define Arctic haze
events e.g. Herber et al, J. Geophys. Res., 2002
Contribution of the 2003 Russian fires to the AOT
during the days of Arctic Haze events
Generoso et al , J. Geophys. Res, in press
13
Part. 2
Ratio between the Best Estimate (BE) and the
standard (STD) simulations
AOT
BC deposition
14
Part. 3
The CALIPSO mission Cloud-Aerosol Lidar and
Infrared Pathfinder Satellite Observation
Quick overview
- CALIPSO is a joint U.S. (NASA) and French
(Centre National dEtudes Spatiales/CNES)
satellite mission with an expected 3 years
lifetime. - Launched on April 28, 2006 aboard
CloudSat satellite - Data available since
mid-June 2006 - Part of the A-train
constellation - Includes 3 instruments, which
measure aerosol and cloud properties
Cloud-Aerosol Lidar with Orthogonal Polarization
(CALIOP) Imaging Infrared Radiometer (IIR)
Wide Field Camera (WFC)
In this study We use CALIOP attenuated
backscatter at 550 nm (Level 1 products)
15
Part. 3
CALIOP measures attenuated backscatter - giving
information on the vertical structure of the
atmosphere - including information from aerosols
and clouds - depending on the extinction and the
phase function
16
Part. 3
17
June 30, 2006
Part. 3
CALIOP
GEOS-Chem
18
July 1, 2006
Part. 3
CALIOP
GEOS-Chem
19
July 2, 2006
Part. 3
CALIOP
GEOS-Chem
20
July 3, 2006
Part. 3
CALIOP
GEOS-Chem
21
July 4, 2006
Part. 3
CALIOP
GEOS-Chem
22
July 5, 2006
Part. 3
CALIOP
GEOS-Chem
23
July 6, 2006
Part. 3
CALIOP
GEOS-Chem
24
Conclusions
- Using a simple approach for data
assimilation, we -gt reduce the uncertainties
in terms of aerosol distributions in most of the
regions (e.g. the Mediterranean Sea, Eurasia,
Arctic) -gt provide an observation-based
estimate of AOT for the Arctic -gt reproduce
typical seasonal variations of aerosol load in
the Arctic
- Using a chemical and transport model and multi
satellite observations -gt we estimate that
the 2003 Russian fires contribute to 16-33 to
the monthly mean AOT averaged north of 75N and
to 40-56 to the mass of BC deposited. They
contribute to more than 30 of the AOT during the
days of Arctic haze events in spring and summer.
- Using the CALIPSO data -gt we show high
potential for evaluating aerosol vertical
distribution in models
25
Additional slides
26
POLDER-1 Nov 96 to Jun 97 POLDER-2 Apr 03 to
Oct 03 Similar instrument PARASOL currently
flying (2006)
(aboard ADEOS)
Fine mode AOT
AOT (total) (Aerosol Optical Thickness)
(over ocean and land)
(over ocean)
May2003
27
LMDz GCM
Resolution 96 x 73 (3.75 in lon. / 2.5 in
lat.) x 19 (vertical) Relaxed toward the ECMWF
reanalysis (6 hours)
INCA (Hauglustaine et al, JGR, 2004)
tropospheric chemistry and transport model which
transports 9 aerosol tracers
DUST
SEA SALT
BC and OC (carbonaceous)
Sulfates
- Coarse

• Fine
• Coarse
• - Super coarse

• Fine Soluble
• Fine Insoluble
• Coarse Soluble
• - Coarse Insoluble

• Fine

Emissions computed on-line from ECMWF winds
Emissions computed on-line from ECMWF winds
Emissions prescribed from monthly inventories
Emissions prescribed from monthly inventories
28
Error characterization in O and B
AERONET Sites
We assumed that all model grid boxes within a
given region have similar errors
Model and observation error standard deviation
are estimated by comparison to ground-based
measurements
Region Near-Ocean
29
Assimilation procedure
We assimilate AOT but the model transports
aerosol mass
Mbefore,3D,fine Mbefore,3D,coarse
AOTbefore,2D,fine AOTbefore,2D,coarse
The vertical profile and the distribution per
species within a mode is not modified
CALL assimilation
Mafter,3D,fine Mbefore,3D,fine x
AOTafter,2D,fine/ AOTbefore,2D,fine Mafter,3D,coa
Mbefore,3D,coa x AOTafter,2D,coa/
AOTbefore,2D,coa
AOTafter,2D,fine AOTafter,2D,coarse
30
Results for the FINE mode
AOT
a posteriori - a priori
(increment)
Positive
Negative
Increment
31
Results for the coarse mode
AOT
a posteriori - a priori
(increment)
Positive
Negative
Increment
32
Simulations conducted
Standard simulation from July 2002, and analyzed
from May to August 2003
Russian fire sources Biomass Burning
Emission Injection Height (EIH) Deposition
processes
Sensitivity simulations from April 2003, and
analyzed from May to August 2003

• Results show
• aerosol distributions highly sensitive to
  daily resolution of BB emission inventories and
  to injection above the PBL in the late fire
  season (July and August)
• CO distributions sensitive to injection above
  the PBL in the late fire season (July and August)
  
• Biomass Burning source strenght underestimate
  of our emissions in May 2003

33
Surface CO simulated by the BE compared to
measurements (CMDL)
ppbv
34
Injection Mode
David Lavoué, personnal communication
35
A summary of the sensitive analysis
36
Contribution of dust to the AOT in the NPac
window
37
Monthly versus daily ATSR-derived biomass burning
emission inventories
38
Conclusions

• - In the standard configuration, the model
  significantly underestimates the MOPITT CO
  columns, POLDER and MODIS AOT over the north
  Pacific during May to August 2003.
• - Increasing the amount of the biomass burning
  emissions does not necessarily result in
  significant improvements (except in May).
• - In contrast, the use of daily biomass burning
  emission inventories improved significantly the
  aerosol simulations, without any significant
  effect on the CO columns.
• The injection of particles up to 4.5 km also
  impacts our simulation but only in the late fire
  season (July and August).
• - According to our improved simulation, the 2003
  Russian fires contribute to 16-33 to the monthly
  mean AOT averaged north of 75N and to 40-56 to
  the mass of BC deposited. They contribute to more
  than 30 of the AOT during the days of Arctic
  haze events in spring and summer.
• - Very encouraging preliminary results from the
  comparisons between GEOS-Chem and CALIOP data !