Global Pattern and Transport of Aerosols Rudolf B. Husar CAPITA, Washington University, St. Louis

1
Global Pattern and Transport of AerosolsRudolf
B. Husar CAPITA, Washington University, St. Louis
Synthetic satellite image
Intercontinental Transport and Climatic Effects
of Pollutants EPA Workshop Climate Change and
Air Quality Research Triangle Park, NC, Dec 3-5,
2001
2
Contents

• Global aerosol pattern multiple satellite and
  surface observations
• Satellite sensing of aerosol features color,
  spatial texture
• Fire location and smoke relationship problems,
  problems
• Smoke aerosol transport Central American fires,
  May 1998
• Long range dust transport Asian dust transport
  to N. America
• Sahara impact on SE US a quantitative estimate
• Summary much progress in the 90s, much to go
• Outlook bright, funif we share, collaborate,
  integrate!!

3
Industrial Sulfur Emission Density and Trends

• The regional hot-spots for industrial sulfur
  emissions are in
• E. North America,
• Europe and
• E. Asia

S emission over N America and Europe have N.
America have leveled off since 1980. Regional
sulfur emission has been increasing over East
Asia.
Lefohn et al, 1999
4
AVHRR satellite optical depth data over the
oceans
Husar et al, 1997.

• The oceanic aerosol pattern is highly regional
  and and seasonal
• The highest oceanic aerosol optical thickness
  (AOT, 1989-91) is over the tropical regions
• The oceanic AOT around N. America, Europe and E.
  Asia is small compared to Africa and Asia

5
Visibility on Ships, 1938
McDonald 1938

• Ship observations cataloged in 1938 indicate
  qualitatively similar pattern to the 1990 AVHRR
  values

6
POLDER Aerosol Polarization Index
In January, the aerosol index is highest over
West Africa, India and China.
In June, high aerosol levels occur over Central
Africa, India and China.
Deuzé J.L. et al., 2001 Remote sensing of
aerosols over land surfaces from POLDER-ADEOS 1
polarized measurements. JGR, vol.106, NO. D5, pp
4913-4926, March 16 2001
7
TOMS Absorbing Aerosol Index
???
Dust
Smoke

• July

• The TOMS absorbing aerosol index shows the
  qualitative pattern of dust and smoke.
• The average July aerosol index, is highest over
  the Sahara, Saudi and Takla Makan deserts

8
Surface Visual Range over the Continents

• Visibility is recorded at 7000 synoptic stations
  hourly
• NOAA NCDC archives the Global Summary of the Day
  (SOD) data

The surface extinction coefficient can be
estimated from Koschmieder formula. The data are
filtered for fog, precipitation, RHgt90 and bad
stations.
9
Surface Extinction Coefficient for SE Asia

• Throughout the year, the most intense surface
  haze is found over India.
• In Thailand and Laos, the haze peak is in DJF
• The Indonesian haze is highest in SON
• The hazy regions of China include the Sichuan
  Basin and the East Coast.

10
Continental Surface Extinction Coefficient
Climatology
High bext (gt0.5km-1) India year round
Indonesia, Bolivia in Aug,Sep,Oct W. Africa
winter Bolivia fall
Husar et al, 2000
India
Dec, Jan, Feb
Mar, Apr, May
Sep, Oct, Nov
Jun, Jul, Aug
11
Tentative AVHRR-Visibility Data Fusion

• Highest JJA AOT values (gt0.5) are over India, W.
  Africa.
• The AOT over the industrial NAM, Europe and Asia
  are low compared to Africa, S. Asia

12
Summary of Observed Global Aerosol Pattern

• The global aerosol pattern indicates regional
  emission hot-spots with plumes extending
  5-10,000 km from the source regions.
• The most pronounced aerosol regions are Africa
  (Sahara dust and sub-Saharan smoke), SE Asia and
  S. America each region shows a pronounced
  seasonality.
• Interestingly, the industrial regions of E. N.
  America, Europe and East Asia show low annual
  aerosol values (PM2.5 lt 30 mg/m3) compared to
  the dusty/smoky parts of Africa, SE Asia and S.
  America (PM2.5 gt 50 mg/m3)
• Tentative Conclusions
• The global industrial aerosol signal is
  superimposed on a larger (10x ??) non-industrial
  aerosolbackground (smoke and dust).
• The non-industrial aerosol has larger
  year-to-year variation due to climatic effect
  (e.g. El Nino).
• Hence, determining the (controllable) industrial
  aerosol impact,requires the characterization and
  subtraction of the non-industrial aerosol.

13
Satellite Detection of Aerosols

• Aerosols add to the reflectance and sometimes
  reduce the reflectance of surface objects
• Aerosols always diminish the contrast between
  dark a bright surface objects
• They change the color of surface objects
• Haze adds a bluish while dust adds yellowish tint
  to the surface color of surface objects.

14
Hazy regions of SE Asia, Dec 28/29 1999
Husar et al, 2000

• SeaWiFS reflectance (412, 550 and 670 nm) of SE
  Asia, Dec., 28/29, 1999
• Intense haze is seen N. India, Sichuan Basin and
  around the Yellow Sea

15
Haze over ChinaSeaWiFS

• The winter haze over China is gray
• Note haze accumulation in valleys

16
Haze over the Po River Valley and Central
CaliforniaSeaWiFS satellite data
Husar et al, 2000

• Bluish industrial and agricultural haze over the
  Po Valley and the San Joaquin Valley, CA
• Note the confinement of haze by topography The
  Alps and the Sierras are major barriers to
  transport

17
Blue, Yellow and White Forest Fire Smoke over N.
California
Bluish aged smoke

• White fresh smoke

Yellow fresh smoke
Yellow aged smoke
18
Industrial Haze over the US East Coast
Bluish color of industrial (sulfate) haze arises
from higher backscattering at 0.42 mm compared to
0.67 mm.
Ocean
Haze Excess
Quantitative AOT retrieval over water and land
Over the ocean with thick haze, the haze
correction removes over 90 of the signal
Haze Excess
Land
Retrieved aerosol optical thickness over water
and land Range 0-1.2
19
Vertical Distribution of Aerosols LITE Space
Lidar
click on figures to enlarge
20
Satellite Detection of FirePixelsMODIS

• In AMJ, the most dense fire pixel locations
  were in Mexico an Central America

Apr, May, Jun 20011
In SON, the dense fire locations shift to the SE
US, from Mississippi to Georgia and the Carolinas
Sep, Oct, Nov 20011
21
Smoke Plumes over SE US

• Smoke plumes are clearly visible on SeaWiFS,
  MODIS, TOMS, etc satellite sensors
• Chemical fingerprinting of smoke aerosol is not
  possible
• Hence, the smoke fraction of PM2.5 is not known.

R 0.68 mm G 0.55 mm B 0.41 mm

• The influence of the smoke is to increase the
  reflectance ant short wavelength (0.4 mm)
• At longer wavelength, the aerosol reflectance is
  insignificant.

0.41 mm Strong backscattering
Florida
Mississippi
0.87 mm No backscattering
22
Summary of Satellite Aerosol Detection

• In the 1990s, there was a sensory revolution,
  particularly in satellite remote sensing of
  aerosols.
• Currently, 2-4 real-time, public satellite
  products allow the daily monitoring of the
  global aerosol pattern at 1km resolution or less.
• Under favorable conditions, dust (yellow) and
  sulfate (blue) haze can be distinguished by their
  spectral reflectance. The multi-colored biomass
  smoke is difficult to distinguish.
• Fire locations as fire pixels and the
  associated smoke plumes are being detected by
  several satellites, but the quantitative
  relationship between the fire pixels and smoke
  emission rate is not known.
• The vertical distribution of global/regional
  aerosol cannot be derived from current satellites

23
Central American Smoke Over E. N. America
Falke et al, 2001

• During May 7-17, 1998, smoke from numerous fires
  in Central America drifted northward

3D rendering indicates that the low-level smoke
was below the high elevation Mexican Plateau
24
Central American Smoke Plume over Eastern N.
America
Falke et al, 2001

• Overlay of SeaWiFS, TOMS (green) and surface
  extinction coefficient, Bext
• On May 15, the smoke plume extends from
  Guatemala to Hudson Bay
• The high Bext shows that the smoke is present at
  the surface

25
Hourly PM10 Concentration During the May 98 Smoke
Event
Husar et al 2002 (?)

• The May 1998 Central American fires caused PM10
  concentration gt 150 mg/m3 at many locations
• Several airports were closed during the smoke
  passage

26
Smoke Aerosol and Ozone During the Episode
Inverse Relationship ?
Extinction Coefficient (visibility)
Surface Ozone
Husar et al 2002 (?)
The surface ozone is generally depressed under
the smoke cloud
27
Asian Dust Transport Across the Pacific April
1998

• The dust cloud crossed the Pacific, subsided to
  the surface and impacted severely the PM
  concentration over West Coast of N America
• The dust was observed and interpreted by an
  ad-hoc international web-based virtual community.
• As of Dec 2001, 12 papers were published by the
  community on the April 98 Dust Event

28
The April 19th Dust Storm

• txt

Husar et al, 2001
29
April 19th Dust Storm Transport Across
Mongolia, ChinaYellow coloration of white clouds
by dust

• Husar et al 2001

30
Asian Dust Cloud Over North America
Husar et al 2001
By April 27th, the dust cloud rolled into North
America and split with one branch heading
southward along the CA coast and the another
branch continuing eastward across the Canadian
Rockies.
During the dust event, the avg. PM10
concentration over the West Coast has increased
from about 15 to 65 mg/m3.
31
Summary of Long Range Transport of Smoke and Dust

• Dust and smoke events originating from Asia,
  Africa and Central America have been shown to
  impact the PM concentration over N America.
• Intercontinental smoke and dust transport is
  highly episodic due to the sharpness of the
  emission pulses in dust storms and forest fires.
• During smoke and dust events, intercontinental
  transport exceeding 10000 km from the source has
  been seen in satellite and surface data.
• Extreme smoke and dust events visualize
  inter-continental aerosol transport and other
  processes.
• The industrial emission rates are more uniform in
  time and intercontinental transport of industrial
  haze is more difficult to detect.

32
Seasonal Attribution of Local and Sahara Fine Dust

• Based on IMPROVE chemical fingerprints, the
  Sahara and Local fine dust can be distinguished.
• In July, over the Southeast, Sahara dominates the
  fine dust
• The Spring and Fall fine dust is evidently of
  local origin

33
Quantitative Sahara and Local Dust Apportionment
The Sahara and Local dust was apportioned based
on their respective Al/Si ratios.

• The maximum annual Sahara dust contribution to
  PM2.5 is about 1 mg.m3
• In Florida, the local and Sahara dust
  contributions are about equal but at Big Bend,
  the Sahara contribution is lt 25 of the local
  fine dust.

• In July the Sahara dust contributions are 4-8
  mg/m3
• Throughout the Southeast, the Sahara dust exceeds
  the local source contributions by a factor of 2-4.

34
Sahara PM Events over the Eastern US

• Speciation data (IMPROVE) show the intrusion
  Sahara dust over the SE US.
• The EPA AIRS PM10 database (600 stations) shows
  the spatial extent and concentration during the
  Sahara dust incursions (gt 80 mg/m3)
• Three such Sahara episodes over the Gulf states
  are shown on the right for July 5, 1992, June 30,
  1993 and June 21, 1997.
• The TOMS aerosol index in July also shows the
  tail of the dust plume entering NAM at the UsMex
  boarder.

PM10 July 5, 1992
PM10 June 30 1993
TOMS, July Aerosol Index
Sahara Dust
PM10 June 21, 1997
Sahara Dust
35
Long Range Transport of PM2.5 to the SE US
G.Smoky Mtn.
Chemical tracer data from IMPROVE indicate
significant (extra jurisdictional) long range
transport PM2.5 contribution to the PM2.5 over
the SE US. Sahara dust incursions occurs in
July, while Central American smoke occurs in
May. Conjecture Without Sahara dust and Mex.
Smoke, the area of estimated PM2.5 nonattainment
(annual gt 15 mg/m3 brown in map) would be
significantly smaller in the Southeaster US (how
much??)
Everglades, FL
Big Bend, TX
Mex. Smoke-May
Sahara Dust - July
36
Summary

• The gross features of global aerosol pattern are
  now available from multiple complementary
  satellite and surface observations. Dust and
  biomass smoke over Africa, Asia and South-Central
  America dominate the global aerosol distribution.
• Satellite sensors also reveal the fine-scale
  spatial aerosol features, aerosol type by
  reflectance color, fire locations etc. However,
  due to the extreme multidimensionality and
  complex dynamics, none of the current monitoring
  systems provides a complete characterization of
  the global aerosol system.
• Intercontinental dust and smoke transport events
  from of Asia, Africa and Central America have
  been shown to produce significant episodic
  impacts on the US PM2.5 and PM10 levels.
• It is suggested, that the area of PM2.5
  non-attainment in the SE US is increased
  substantially by the contributions from African
  and C. American aerosol sources.

37
Outlook

• Ultimately, the aerosol monitoring data will need
  to be continually assimilated into predictive
  dynamic aerosol models for the same reason that
  weather prediction models assimilate current
  weather data.
• The strong data-model symbiosis for aerosol is
  further necessitated by the fact that many
  unpredictable aerosol sources (e.g. fires,
  volcanoes and partly dust storms) require
  real-time monitoring data for the source term in
  dynamic models.
• The massive data processing and analysis
  associated with global aerosol data integration
  can only be accomplished through
• data/knowledge exchange using the internet
  infrastructure
• international collaboration through increased,
  open data and knowledge sharing
• building distributed scientific value-adding
  chains for the enhanced production of
  actionable knowledge.