Transcript Descriptive Analysis Map Progress

Causes of Haze Assessment
Presented at the
RPO National Technical Workgroup Meeting
November 5, 2003
Dave DuBois
Desert Research Institute
Acknowledgements
• Funding from WRAP and CENRAP
• WRAP Air Monitoring & Reporting Forum Chair
– Marc Pitchford
• DRI Faculty
–
–
–
–
Mark Green, Principal Investigator
Jin Xu
Dan Freeman
Vic Etyemezian
• DRI students and hourly employees
–
–
–
–
Aleksandra Nikolic
Karl Graham
Anthony Deleon
Eric Peters
1
COHA Status Report
•
•
•
•
•
•
•
•
Review goals and objectives
COHA approach
Virtual report
Aerosol descriptions
Meteorological descriptions
Emissions descriptions
Trajectory analysis
Episode analysis
2
Goals & Objectives
• Assess causes of haze for all study sites
• Systematic study of 128 Class I areas in WRAP and
CENRAP, 8 tribal sites and 10 CENRAP protocol sites
• Encourage broad-based stakeholder participation
throughout the assessment process
• Enhance the utility and accessibility of the results for:
–
–
–
–
–
SIP & TIP development,
Regional air quality model evaluation & interpretation,
Identification of monitoring gaps,
Improved methodology for setting natural haze levels,
Tracking effectiveness of emission control programs
3
COHA Study Data
• Began analysis of 1997 to 2002 IMPROVE
and protocol database
• Primarily using IMPROVE and protocol
sites with full speciation data in the study
region (118 sites by December 2002)
• Using nationwide network of 158 sites (end
of 2002) to establish continental and
regional setting
4
COHA Approach
Determine causes of haze at WRAP and CENRAP
Class I areas, tribal and selected CENRAP IMPROVE
protocol sites
5
COHA Approach
• Systematic study—many sites, many
questions to answer
• Processing reports in batch mode to
facilitate timely completion
• Give each site individual attention once
batch processing is completed
6
COHA Approach
• Virtual report—no paper report
• A virtual report designed as a tool to:
– guide us in the causes of haze
– communicate results
– help users to interpret causes of haze
• Virtual report gives us the ability to mix text,
graphics, animations and links to external
web sites in addition to timely updates
7
The Causes of
Haze web site is
online now in a
DRAFT, password
protected form:
http://coha.dri.edu
Username: dri-coha
Password: hazeyweb
Much of the web site
is a shell ready to
receive data and
causes of haze
information that we
generate
8
View reports
by state, area,
tribal area or
protocol site
This interface
is under
construction
and may
change
View animations
of IMPROVE
measurements
9
Aerosol Descriptive Analysis
Provides answers to the questions:
• For the years 1997-2001, how many measurements are available
for the site in each month of each year, and what are the
contributions of the major aerosol components to light extinction in
each month of each year?
• What is the overall average light extinction at the site, and what are
the contributions of the major aerosol components to the light
extinction?
• What are the light extinction contributions by the major aerosol
components for best, worst and average days and how do they
compare?
• What percentage of the sampling days are the worst days in each
month & how variable are the chemical components?
10
Aerosol Descriptive Analysis
Aerosol descriptions available on web page now:
BIBE1, Big Bend National Park, TX
BOWA1, Boundary Waters Canoe Area, MN
DENA1, Denali National Park, AK
GRCA2 , Hance Camp at Grand Canyon NP, AZ
HAVO1, Hawaii Volcanoes National Park, HI
JARB1, Jarbidge Wilderness, NV
MORA1, Mount Rainier National Park, WA
SAGO1, San Gorgonio Wilderness, CA
SAGU1, Saguaro National Monument, AZ
UPBU1, Upper Buffalo Wilderness, AR
11
Aerosol Descriptive Analysis
Pages designed for users to
copy and paste text and figures
into their own reports
Example: San Gorgonio
Wilderness Area, California
Both charts and text to describe
the 20% best, worst and
middle 60%
Printer friendly and black &
white versions of pages
12
Meteorological & Emissions
Descriptive Analysis
• Archived monitoring network locations, climate,
emissions, wildfires, census, political, physical, and
image databases
• Information from these databases are helping us
build conceptual models and answer descriptive
analysis questions by visualizing data (e.g. map
emissions densities)
• Assist us in the general and detailed description of
the meteorological setting of each site
• Over 120 GB of spatial data archived at DRI
13
Descriptive Analysis Map Progress
• Standard maps to support descriptive analysis text
–
–
–
–
–
Terrain (labels on rivers, lakes, major features, peaks)
Nearby met/air quality networks
Urban boundaries, roads, landmarks
Landuse map
Landsat image (qualitative landuse, with some features
labeled)
– Emissions map (fires, WRAP point inventory, urban
areas, roads)
• Specialized maps (as needed basis)
– 3D terrain map
– Meteorological flow map to illustrate transport
14
Standard Analysis Map Progress
50
50
50
80
40
80
40
50
25
80
75
75
50
80
Maps for web site
Approx. percent completed
25
50
33
30
60
90
60
80
Hawaii: 0, Alaska: 0
15
Cucamonga Wilderness Area
Example “20 km” terrain map
16
Cucamonga Wilderness Area
Example “2 km” terrain map
17
Meteorological Description Update
• Describe meteorological influences by defining
regions:
– Hawaii and So. Calif. Coastal are posted on web
• Detailed meteorological descriptions:
– San Gorgonio and Jarbidge are posted on web
• Terrain description
– Access database tool developed to assist descriptive text
• Identify nearby meteorological measurements for use
in interpreting aerosol data
18
Cucamonga Wilderness Area
Example “20 km” met/air quality network map
19
Trajectory Analysis Status
• Three years (2000-2001), three heights (10, 500,
1500m), every three hours
• EDAS for continental sites and FNL for Hawaii and
Alaska
• HYSPLIT v4.6 model calculations done for all sites
• Trajectory output being processed and stored in
database
• Trajectory tool being developed to produce ASCII
summary files and convert trajectories into shapefiles
• Generate summary maps
20
Emissions Description Status
• Developing standard template for web site
• Creating maps of emissions surrounding
each site at two scales: 2 km and 20 km
• Include table of surrounding point sources
ranked by distance and emission rate
21
Cucamonga Wilderness Area
“20 km” emissions map
22
Episode Analysis
• Use combination of backtrajectory,
synoptic, mesoscale meteorological
analysis, aerosol and emissions data to
conceptually understand single site and
regional or sub-regional episodes of high
aerosol component concentrations
• Systematic survey of episodes from the
1997 to 2002 IMPROVE database
23
Episode Analysis
• Created animated maps of IMPROVE and protocol
measurements for entire network
• Choose episodes base on sites classified with 20%
worst light extinction
• Noting duration, frequency, regional extent, season
and components that contributed to light extinction
• Assemble case studies and classify into episode
types
• Create database of these episodes
• Combine results of episode analysis with cluster
analysis to develop conceptual models
24
“Hazagon” Analysis
• The hazagon provides a way to visualize speciated
extinction for those sites in the 20% worst category
25
Wrap up
•
•
•
•
Phase I analyses completed by summer 2004
Phase II analyses completed by 2005
Possibilities of adding other RPO's
Continue to maintain and update the website
information plus the need to repeat the
assessment periodically (e.g. 5 year cycle).
26
Example Episodes
•
•
•
•
April 16, 2001 Asian Dust over Western US
August 2001 wildfires
October 16, 2001 Arizona dust
September 3, 1997 Eastern sulfate transport
to Colorado Plateau
27
4/16/01 Asian Dust Episode
28
GOES View of the Dust Streak Across North America,
April 17
GOES10 view of dust streak on the
morning of April 17
GOES8 view of dust streak on the
evening of April 17
29
Transport of the Asian dust to the United States
The common weather conditions are usually associated
with the upper low pressure trough / cut-ff low and
surface low pressure system (low formed by a strong
cyclonic vortex) over northeast China and north Korea
[Kim et al., 2002]. Under this weather conditions, Asian
dust can move fast along the zonal wind distribution
due to the jet streak [Kim et al., 2002].
30
Large Area Regional Haze on April 16, 2001
•
•
45 of the 68 WRAP IMPROVE monitoring sites were in 20% worst case days of the
year 2001. For the sites that were 20% worst case days, the average contribution of fine
soil to PM2.5 is ~60% (with a standard deviation of 13%), and dusts (fine soil and coarse
mass) contributed in average ~ 46% (S.D. 13%) to the aerosol light extinction.
The average contribution of fine soil to PM2.5 is ~54%, and to aerosol light extinction is
~41% for all WRAP sites on April 16, 2001.
Average Contributions of Major Checm ial Com ponnents to Light
Extinction for 68 WRAP Sites (1997-2001 April Average)
Average Contributions of Major Checm ial Com ponnents to Light
Extinction for 68 WRAP Sites (April 16, 2001)
CM
15%
CM
20%
Sulfate
28%
Sulfate
36%
FS
7%
EC
7%
FS
21%
OC
20%
Nitrate
15%
Nitrate
14%
EC
3%
OC
14%
31
Asian Dust Signature
• Asian dust may cause haze in a large area and last
several days depending on the regional and local
weather conditions in the United States.
• Usually, dust elements dominate the aerosol light
extinction in the whole western United States during
the Asian dust episode. The dust cloud may also
move to the Eastern U.S. and influence some of the
eastern sites, although the influence is usually much
smaller in both spatial scale and loading.
• Most of the Asian dust episodes happen in the
spring during the Month of March to May.
32
Asian Source Attribution Evidence
• The desert regions in Mongolia and China,
especially Gobi desert in Northwest China, are
important sources of mineral aerosols. Given
suitable weather conditions, dust can be lifted
from the dry surface of the Asian Gobi desert
region and transported to the United States in
about 7-10 days. Extremely high aerosol loadings
dominated by dust components are observed in
Northern China and Korea during the episode.
33
Origin of the Asian Dust
Strong low pressure system sitting in northeast Mongolia caused surface wind speeds to be as high as ~30 m/s
34
The August 2001 Western US Wildfire Episode
Most of August
experienced
heavy OC in the
west.
44 sites in the
WRAP region
experienced the
worst 20% day
on August 17.
3 sites had this day as worst Bext:
CABI1, Cabinet Mountains, MT
NOCA1, North Cascades, WA
PASA1, Pasayten, WA
35
The August 2001 Western US Wildfire Episode: August 11-23
Fires started early
in August and lasted
all month. 2001
was not a
particularly bad fire
year in terms of
number of acres
burned
36
Databases available
• IMPROVE data
• Text, photos and maps on fires from NIFC,
USFS, newspapers
• Text on meteorology from NWS, NCDC
• Weather maps from CDC, NCEP, WXP
• MODIS satellite images from UWisc, USFS
• Coarse fire locations fron MODIS, AVHRR
37
Large Fire Locations
Typical of late
season fires, most
of the fires are in
the northwest and
northern Great
Basin
38
August 15 Terra MODIS at 18:38 UTC
shows transport of regional smoke plumes
39
August 17 Terra
MODIS
at 18:38 UTC
Can see thick
smoke in NW
and coastal areas
The Moose Incident was one of the largest fires of the
2001 season. This lightning caused fire occurred
August16th on the Flathead National Forest. The
Moose fire ultimately burned approximately 71,000
acres before it was controlled.
40
Conditions
According to NIFC National Fire News the national level of
preparedness increased to the highest point on the 16th, as more
than one half million acres are burning in 42 large fires across the
United States. Nearly 21,000 firefighters are working on the fire
lines. Record high temperatures in Oregon, Washington, and
Idaho may increase large fire activity. Predicted strong winds will
challenge firefighters on the 17th. Media reports on the 16th
indicate federal troops will join the 21,000 firefighters. Fire
activity as of mid-August was near to or slightly above the 10year average.
41
Drought
http://lwf.ncdc.noaa.gov/img/climate/research/2001/wildfire/08-07Statewideprank_pg.gif
42
Morning surface weather 17 August (12:00 UTC)
Notice radar echos
(see end of presentation)
43
5 day backtrajectories from morning of 17 Aug
NM sites point toward regional fires
in Northwest and upper Great Basin
Fires
Possible Canadian
influence on Montana
sites
Fires
44
Arizona Local (Regional) Dust Episode
45
Arizona Regional Dust Episode
46
Surface Weather Map Before the Sampling Day
47
Surface Weather Map
Midnight October 16, 2001
48
Annual Percentile of Bext, OC, FS and CM on
10/16/2001
Location
BALD1
CHIR1
IKBA1
PHOE1
QUVA1
SAGU1
SIAN1
TONT1
Bext %
0.99
0.99
0.97
0.96
1.00
1.00
1.00
1.00
OC%
0.97
0.98
0.96
0.90
0.95
1.00
0.99
0.94
FS %
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
CM %
1.00
0.98
0.97
1.00
1.00
1.00
1.00
0.99
49
Transport of the Dust
50
Percentile of Bext, OC, FS and CM on 10/16/2001
Location
BALD1
CHIR1
IKBA1
PHOE1
QUVA1
SAGU1
SIAN1
TONT1
Bext %
0.989473684
0.991596639
0.973451327
0.95890411
1
1
1
1
OC%
0.968421053
0.983193277
0.955752212
0.904109589
0.948717949
1
0.99047619
0.938596491
FS %
CM %
1
1
1 0.983193277
1 0.973451327
1
1
1
1
1
1
1
1
1 0.99122807
51
Case study for September 3, 1997-Eastern US
sulfate transport to Colorado Plateau
12 spatially coherent
sites in TX, NM, AZ,
CO 20% worst
haze,with sulfate
dominant
52
Table 1.
SITE
GRSA
MEVE
MOZI
ROMO
WEMI
BAND
GICL
BIBE
GUMO
TONT
CHIR
GRCA
S percentile
(1997)
98
100
98
94
100
100
99
98
98
100
100
100
S conc μg/m3
(adj to STP)
0.69
0.74
0.47
Highest S for all 1997 at
0.42
Arizona, New Mexico sites
0.62
and Mesa Verde, 94-99
0.75
1.18
percentile other sites
2.09
1.71
1.11
1.26
0.74
53
2 days prior to event, flow was light from the east to southeast
upwind of sites. Haze reported at many eastern US NWS
stations
54
Weak cold front passes- stronger flow accelerates westward
movement of hazy airmass? Flow in response to high pressure
system along Canadian border
55
Bandalier, Big Bend 315 hour backtrajectories from noon
local time Sep 3
BAND
BIBE
56
Tonto, Grand Canyon 315 hour
backtrajectories from noon local time Sep 3
TONT
57
Mt. Zirkel, Great Sand Dunes, 315 hour backtrajectories
from noon local time Sep 3
MOZI
GRSA
58
Conceptual model of event
• Highest S day for 1998 for Arizona and New Mexico sites
long-range transport from eastern US
• Specific source areas in eastern US, unknown, could be large
hazy blob and many source areas included from southeast to
possibly Ohio River Valley
• Initial inspection of aerosol maps indicates rare to get this far
west- but not unusual to reach Big Bend (from BRAVO study
analyses)
• Movement of large early autumn high pressure west to east
along Canadian border associated with this pattern and other
cases of eastern US transport to Big Bend
• Additional analysis could look at airport visibility data, other
aerosol data, additional cases (if found)
59