Title: Reducing Air Pollution In Los Angeles
1Review of WRAP Regional Modeling Center (RMC)
Deliverables Related to the Technical Support
System (TSS)
Ralph Morris and Gerry Mansell ENVIRON
Corporation Gail Tonnesen and Zion
Wang University of California, Riverside
September 14-15, 2005 Attribution of Haze
Workgroup Meeting San Francisco, California
2Overview
- 2002 Base A Base Case CMAQ/CAMx Modeling and
Model Evaluation - 2002 CAMx PSAT Source Apportionment Modeling
- PSAT/TSSA Comparisons
- RMC BART Modeling Plans
- 2018 Simulations and Visibility Projections
- Modeling Elements of the Visibility SIP Weight of
Evidence (WOE) Reasonable Progress Goal (RPG)
Demonstration
32002 Base A Modeling
- CMAQ emissions ready September 12, 2005
- Start annual 2002 36 km CMAQ run September 19,
2005 - CAMx emissions ready September 19, 2005
- Compare Jan/Jul 2002 CMAQ/CAMx October 3, 2005
- Make decisions on model for 12 km modeling and
control strategy evaluation - Finish annual 2002 36 km CAMx run October 10,
2005 - Perform PSAT PM Source Apportionment using CAMx
October 31, 2005 - 2018 Emission Inventories October 31, 2005
42002 Base A Modeling
- Example of Model Performance Evaluation (MPE)
displays of use to the TSS - UCR MPE Tool
- Scatter Time Series Plots by subregion
- allsite_allday (SO4 example for WRAP States)
- allday_onesite (SO4 example for Canyonlands)
- onesite_allday
- Monthly Bias/Error plots
- By subregion (Bias example for SO4 in WRAP
States) - Stacked 24-hr average extinction plots
- Observed vs. Model (Canyonlands example)
- Comparisons for Worst/Best 20 Days
5Example UCR Tool MPE Plots, CMAQ vs. CAMx for
January July 2002 allsite_allday for WRAP States
6MPE Plots for SO4 at Canyonlands and July 2002 ?
CMAQ vs. CAMx Scatter Plot and Stats
Observed, CMAQ, and CAMx ? Time Series Plot
7- SO4 IMPROVE in WRAP States
- Monthly Fractional Bias
- CAMx
- CMAQ
8Observed vs. Modeled Daily Extinction _at_
Canyonlands
Observed
Observed
CMAQ
CAMx
9Source Apportionment Approaches
- CALPUFF Lagrangian non-steady-state puff model
- Chemistry highly simplified, incorrect and
over 20 years old (1983) - Fails to adequately account for wind shear
- SCICHEM Lagrangian model with full chemistry
- Needs 3-D concentrations fields
- Currently computationally demanding
- Photochemical Grid Models CMAQ/CAMx
- Zero-Out Runs (actually sensitivity approach)
- Reactive Tracer PSAT/TSSA approaches
10PM Source Apportionment Technology (PSAT) in CAMx
- Reactive tracer approach that operates in
parallel to the host model to track PM precursor
emissions and formation - Set up to operate with families of tracers that
can operate separately or together
- Sulfate (SO4)
- Nitrate (NO3)
- Ammonium (NH4)
- Secondary Organic Aerosols (SOA)
- Mercury (Hg)
- Primary PM (EC, OC, Soil, CM)
11PSAT Conceptual Approach
- Modify CAMx to include families of tracers
(tagged species) for user selected source
groups - Source group source category and/or geographic
area - Build on CAMx ozone apportionment schemes (OSAT,
APCA) - Tag primary species as they enter the model
- SO2i , NOi , VOCi , primary PM (crustal, EC,
etc.) - When secondary species form, tag them according
to their parent primary species - SO4i , NO3i , SOAi
12Zero-Out Comparisons for Sulfate
- Use Eastern US/Canada modeling domain
- Add four hypothetical point sources to base
emissions - Test large and small emission rates to
investigate signal/noise - Large SOx 850 TPD
- Small SOx 0.85 TPD
X
X
X
X
13MRPO Large Source Episode Maximum SO4 PSAT
versus Zero Out
PSAT
Zero-Out
14Oxidant Limiting Sulfate Example
PSAT
Zero-Out
- PSAT attributes 50 of SO4 to source A (and 50
to B) - Zero-out attributes zero SO4 to source A (no
source is culpable) - Zero-out result (sensitivity) is not a reasonable
apportionment for this example
15PSAT Sulfate Evaluation
- Good agreement for extent and magnitude of
sulfate impacts between PSAT and zero-out - Comparing the outer plume edge is a stringent
test - Zero-out impacts can be smaller or larger due to
oxidant limited sulfate formation and changes in
oxidant levels. - Run times look very good
- Two tracers per source group for sulfate
- PSAT obtains 50 SO4 source contributions in time
needed for 1 zero-out assessment
16PSAT Chemical Scheme for NOy Gasses
- PSAT tracks 4 groups of NOy gasses
- RGN
- TPN
- HN3
- NTR
- Conversion of RGN to HN3 and NTR is slowly
reversible - Conversion of RGN to TPN is reversible rapidly
or slowly
17PSAT for SOA
- CAMx SOA scheme
- VOC -- OH, O3, NO3 --gt Condensable Gas (CG) ltgt
SOA - CGs partition to an SOA solution phase
- PSAT implementation straightforward, but many
terms - Three types of VOC precursor
- alkanes, aromatics, terpenes
- Five pairs of CG/SOA
- four anthropogenic, one biogenic
- low/high volatility products
- PSAT tracers for VOC, CG and SOA species
- 14 tracers per source group
18PSAT Evaluation for NO3 and SOA
- Independent check against SOME
- Source Oriented External Mixture (Kleiman et al
at UC David) - SOME uses explicit species for each source group
that are integrated in the model - Highly computationally demanding
- Zero-Out comparisons not appropriate for VOC/NOx
due to nonlinear chemistry - Good agreement between PSAT and SOEM for NO3 and
SOA - http//pah.cert.ucr.edu/aqm/308/meetings/March_200
5/03-08_09-05.SF_CA/Alternative_Model_Mar8-9_2005_
MF_Meeting.ppt
19CAMx/PSAT and CMAQ/TSSA Comparisons Feb/Jul 2002
- PSAT Configuration
- 15 source regions
- 5 Source Categories (1) Biogenic (2) On-Road
Mobile (3) Points (4) Fires and (5)
AreaNon-Road - Initial and Boundary Concentrations
- 77 Source Groups (7715 x 5 2)
- SO4, NO3 and NH4 families of tracers
- Did not run SOA, Hg and Primary PM tracers
- TSSA Configuration
- Differences in source group source categories
(e.g., mv on-road non-road, fires?, BC??) - Other category in TSSA for unattributable PM
20PSAT/TSSA Source Region Map CA, NV, OR, WA, ID,
UT, AZ, NM, CO, WY, MT, ND, SD, Eastern States
and Mex/Can/Ocean
21Grand Canyon, Arizona Day 182 (07/01/02) 2nd
Worst Visibility Day in 2002 NV Points
Highest AZ Points (5xsmall) Mex Points TSSA
Units??? TSSA Other???
22Grand Canyon, Arizona Day 188 (07/07/02) 15th
Worst Visibility Day in 2002 Some differences
TSSA and PSAT Pts_Mex, Other, BC
23Grand Canyon, Arizona Day 32 (02/01/02) 8th Best
Visibility Day in 2002 PSAT UT_Points BC
AZ_Points UT_NonRoad NM_Points TSSA UT_Points
Other OR_Points WA_Points ID_Points
24Rocky Mtn. NP, Colorado Day 182 (07/01/05) Worst
Day of 2002 PSAT UT_Fires CO_Pts NV_Pts
CO_Fires UT_Pts. TSSA Other CO_Pts UT_Pts
NV_Pts If Fires in Other then fairly good
agreement
25Conclusions PM Source Apportionment
- PSAT results mostly consistent with TSSA
- Some differences, TSSA Other category makes it
hard to interpret - Version of CMAQ with TSSA has known mass
conservation problems - Powerful diagnostic tool that can be used for
source culpability (e.g., BART) and to design
optimally effective control PM/visibility control
strategies - PSAT explains 100 of the PM, doesnt suffer
Other unexplained portion of PM like TSSA - TSSA being implemented in latest versions of CMAQ
26PSAT Plans for WRAP
- 2002 Base A Emissions
- Source Regions
- WRAP States plus others and IC/BC
- Source Categories
- Anthropogenic versus Natural emissions
- SO4, NO3 and NH4 initially, test SOA and primary
PM - 2018 Base Case emissions
- Source regions and categories TBD
2722 Pre-Merged Emission Files
- Argts Area sources except dust sources
- Arfgts Area fires from CENRAP
- Awfgts3d WRAP wild, prescribed and agricultural
fires - Bsfgts3d Canadian Wild fires/Blue Sky algorithm
- fdgts_RPO Fugitive dust (Ag construction) for
entire domain - mbgts_WRAP On road mobile sources for WRAP RPO
- mbgts_CANDA_MEX On road mobile sources for
Can/Mex - mbvgts_CENRAP36 On-road mobile sources for
CENRAP states - mbvgts_RPO_US36 On road mobile sources for MW,
VISTAS, MAINE-VU - nh3gts_RPO36 Ammonia from agricultural sources
for CENRAP/MW states - nh3gts_WRAP36 Ammonia emissions ag sources for
WRAP GIS model - Nrygts Off road mobile with annual IDA files
- Nrmgts Off road mobile with monthly or seasonal
IDA files - Nwfgts3d Point sources fires from non WRAP
states (CENRAP and VESTAS)
2822 Pre-Merged Emission Files
- Ofsgts3d Off shore point sources in the Gulf of
Mexico - Ofsmagts Off shore Marines shipping in the
Pacific Ocean - Ofsargts Off shore area sources in the Gulf of
Mexico - ptgts3d_RPO_US36 Point sources emissions for all
RPOs, Can Mex - rdgts_RPO Road dust for the entire domain
- B3gts_RPO Biogenc emissions from BIES3 for the
entire domain - wb_dus Wind blown dust for entire domain
- Oggts3d Oil and gas for WRAP states (except CA)
- 2002 PSAT run need to define natural emissions
- Arfgts Area fires from CENRAP
- Awfgts3d WRAP wild, prescribed and agricultural
fires (will need to process wildfires separately) - Bsfgts3d Canadian Wild fires/Blue Sky algorithm
- Nwfgts3d Point sources fires from non WRAP
states (CENRAP and VESTAS)? - B3gts_RPO Biogenc emissions from BIES3 for the
entire domain - wb_dus Wind blown dust for entire domain
29WRAP RMC BART Modeling
- RMC will perform regional photochemical grid
model of alternative regional strategies using
CMAQ and/or CAMx with PSAT - RMC will assist States who desire to perform
source-specific CALPUFF modeling - Provide States with 3-tears of CALMET ready MM5
fields (2001, 2002 and 2003) - May perform source-specific modeling using PSAT
for 2002
30Example of BART Modeling using Grid Models
- Midwest RPO (MRPO)
- Use combination of photochemical grid and CALPUFF
modeling in the BART analysis - Comprehensive Air-quality Model with extensions
(CAMx) PM Source Apportionment Technology (PSAT)
31CALPUFF estimates higher visibility impacts than
CAMx/PSAT and consequently generally more days
and larger spatial extent of dV gt 0.5 deciview
PSAT
CALPUFF
32July 19, 2002 24-Hour SO4 Concentrations IN
Source (isgburn) CALPUFF much higher
concentrations away from source. Why secondary
CALPUFF SO4 peak over Cape Cod?
CAMx PSAT
CALPUFF
33CALPUFF More Conservative than Grid Models
- CALPUFF chemistry overstates NO3 and SO4 in
winter - CALPUFF understates dispersion because it fails
to adequately account for wind shear and wind
variations across the puff - Uses just one wind to advect entire column of
puff - IWAQM found CALPUFF overestimation bias of a
factor of 3-4 at distances beyond 200-300 km - When encountering stagnant conditions, puffs pile
up on each other and stop dispersing - Violates 2nd Law of Thermodynamics
34CALPUFF puff column advected north by winds at
300 m AGL even though surface winds from east and
north
Surface Winds 0600
Surface Winds 1200
300 AGL Winds 0600
352018 Modeling/Visibility Projections
- Visibility projections use 2018 and 2002 modeling
results in relative sense to scale observed
2000-2004 visibility to 2018 - Draft EPA Guidance (2001)
- 2018 Visibility Goal based on Glide Path from
current (2000-2004) observed visibility to
Natural Conditions in 2064 - EPA Guidance for default Natural Conditions (2003)
36Baseline Conditions 28.9 dv Natural Conditions
11.4 dv 2018 Visibility Goal 24.9 dv
2018 Reduction Goal 4.1 dv 2018 Modeled
Reduction 5.2 dv GRSM achieves 2018 Vis Goal
37Great Smoky Mountains Obs vs. Model Extinction
W20
gt 80 extinction due to SO4
38Modeled Visibility Goal Test will be Difficult
for WRAP Class I Areas
- Worst days not always dominated by SO4 -- OMC,
NO3 and/or CM can be more important than SO4 at
many sites - California NO3 issue
- Southwestern Desert dust (CM)
- Fires, Fires, Fires, Fires
- Posses unique and special conditions for modeling
visibility projections - May be more difficult to model achievement of
visibility goal - Many sites dominated by fires for Worst 20 days
and assumed to remain unchanged from 2002 to 2018 - Dont CAIR states
- Point source SO2 and NOx controls much less
effective at reducing visibility in west compared
to east
39Five examples of WRAP visibility
projections WHIT, NM GRCA, AZ CRLA, OR SAGO,
CA DENA, AK
40 Dust ?
41White Mountain, NM Worst 20 Days in 2002
Observations vs. Predictions
? Fires
Obs Dust ?
? Nitrate
42(No Transcript)
43Grand Canyon, AZ Worst 20 Days in
2002 Observations vs. Predictions
? Fires in model
? Dust in obs
44(No Transcript)
45(No Transcript)
46(No Transcript)
47Denali Glide Path to Natural Conditions, Baseline
for Current Worst Days (10 dv) gt 2064 Natural
Conditions for many eastern Class I areas (e.g.,
GRSM _at_ 11 dv) Denali 2018 RPG Reduction 0.61
dv
48Denali National Park Best 20 Days (B20) Current
5-Year Average for B20 Days (1.91 dv) lower than
EPA default natural conditions for best days
(2.30 dv)
49Conclusions WRAP Vis Projections (1)
- Much more diverse PM mixture in western US on
Worst 20 days than in the east - Fires and wind blown dust much more important
little opportunity to control - Focus reasonable progress on days with high
anthropogenic contributions? - Incorporate fires and dust in Natural Conditions
endpoint? - Mexico, Canada and global transport can have
large influence at some Class I areas - Modeled visibility goal test will likely not be
achieved at many WRAP Class I areas
50Conclusions WRAP Vis Projections (2)
- Need to start developing strategy for
demonstrating reasonable progress for WRAP - Weight of Evidence (WOE) RPG demo needed
- Enforceable emission reductions
- Treatment of extreme events (fires/dust/internatio
nal) - Visibility improvements on days due to US anthro
sources - Examine extinction improvements by species?
- Smoke management plan
- Modeled visibility changes are just one element
of WOE RPG demonstration
51Modeled WOE RPG Elements
- Glide paths and modeled RPG test (EPA)
- Eliminate days dominated by natural events in
modeled RPG test (e.g., fires, dust) - 2018 projections for species dominated by
anthropogenic emissions (e.g., SO4, NO3) - 2018 projections for modeled worst visibility
days, worst sulfate days, etc. - Other???
52RMC 2018 Modeling Schedule
- 2018 SMOKE Emissions Modeling Oct05
- 2018 36 km CMAQ/CAMx Modeling Nov05
- Preliminary 2018 visibility projections Dec05
- 2018 12 km modeling Nov-Dec05
- 2018 Source Apportionment Modeling Jan06
- 2018 Control Strategy Modeling 2006