P1246341516VOdgb

1
Oklahoma 8-Hour Ozone Early Action Compact
Update on 2007 Control Strategy Modeling Ralph
Morris, Michele Jimenez, Yiqin Jia, Cuong Tran,
and Edward Tai ENVIRON International
Corporation (rmorris_at_environcorp.com)
Oklahoma EAC Technical Advisory Committee and
Public Meeting Tuesday, March 16, 2004 Tulsas
Chambers 2nd Floor Conference Room Two West
Second St. Tulsa, Oklahoma
2
Todays Topics

• Overview of OK EAC Modeling To Date
• Revised 2007 Base Case and Initial 2007 Control
  Strategy Runs
• Additional Planned Runs
• Clean Air Action Plan (CAAP) and Weight of
  Evidence (WOE) Attainment Demonstration
• Source Apportionment and Emissions Sensitivity
  Tests (presented January 22, 2004)
• Denver EAC WOE Analysis
• UCI Oil and Gas Emissions Sensitivity Tests

3
Oklahoma EAC Modeling ? Current (03/16/04) Status

• Dec 2002 Modeling Protocol
• Apr 2003 MM5 Meteorological Modeling Report
• August 13-25, 1999 Oklahoma ozone episode
• Sep 2003 MM5 Meteorological Modeling Report
• August 13 September 1, 1999 Oklahoma ozone
  episode
• Oct 2, 2003 Oklahoma Ozone Modeling Meeting
• Summary Modeling Protocol, MM5 Met Model
  Evaluation, and Emissions Modeling Summary
• 1999 Base Case Modeling and Model Performance
  Evaluation
• Run1 OTAG Clean Boundary
  Conditions (BCs)
• Run2 Spatially varying
  BCs
• Run3 Corrected emissions,
  spatially varying BCs
• EPA notes availability of August 1999 CMAQ
  results for BCs
• Desire to perform VOC/NOx emission reduction
  sensitivity tests to determine whether modeled
  ozone is sensitive to changes in inventory

4
Oklahoma EAC Modeling ? Current (03/16/04) Status

• Dec 22, 2003 1999 Base Case Model Performance
  Evaluation and Sensitivity Report
• Run5 Base Case simulation for August 1999
  Oklahoma ozone episode
• Model Performance Evaluation
• Meets EPAs 8-hr ozone performance
  goal of within 20 at most monitor
  days (85 and 95 of time using
  Maximum and Best Fit comparisons)
• Two recommendations
• Base Case simulation with CMAQ BCs
• Sensitivity simulation using UCI
  (Katzenstein, 2003) Oil and Gas VOC
  emissions
• 30 VOC/NOx Emission Reduction Sensitivity Tests

5
Oklahoma EAC Modeling ? Current (03/16/04) Status

• January 22, 2004 EAC Meeting in OKC
• Emission updates to 1999 Base Case (Run6)
• Update BCs using EPA CMAQ results (Run7)
• 2002 Base Case modeling (CMAQ BCs)
• 2007 Base Case modeling (CMAQ BCs)
• Projection of 8-Hour Ozone Design Values to 2007
• Begin development of 2007 Control Strategy
  Scenarios
• Base Case Modeling and MPE (Run6 Run7)
• 2007 8-Hour Ozone Design Value Projections
• 1999 Ozone Source Apportionment Modeling
• 1999 VOC/NOx Sensitivity Simulations (12/22/03
  Report)
• Planned 2007 Control Strategies

6
Oklahoma EAC Modeling ? Current (03/16/04) Status

• Decisions at January 22, 2004 EAC Meeting in OKC
• Concerns were expressed regarding the EPA 2007
  inventory based on 1996 NET
• Generate new 2007 Base Case from 1999 NET
  inventory
• Develop a set of control strategy runs to perform
• UCI OilGas Sensitivity
• 5 Anthropogenic VOC, NOx and VOXNOx in Tulsa
• Remove expiring permitted sources
• 7.8 RVP
• Stage I
• NOx Controls on EGUs
• Stage II
• Basic I/M

7
Oklahoma 8-Hour Ozone Modeling

• August 13 September 1, 1999 Modeling Episode
• 36/12/4 km Grid Structure
• CAMx Photochemical Grid Model
• MM5 Meteorological Model
• EPS2x Emissions Model
• GLOBEIS Biogenic emissions
• MOBILE6 and NONROAD mobile sources
• Local Traffic Demand Model (TDM) for Tulsa and
  Oklahoma City (link based data)
• Day-specific point source emissions using CEM
  data
• Other Local Data as Available

8
36/12/4 km Oklahoma Modeling Domain
9
Future-Year Emission Scenarios

• 1999 and 2002 Base Case
• Based on 1999 NEI Inventory
• 2002 Summer Season Average EGU Emissions (CEM)
• Tulsa/OKC on-road mobile source emissions using
  link-based VMT and MOBILE6
• August 1999 Biogenics
• Original 2007 Base Case
• Based on EPAs Heavy Duty Diesel Rulemaking 2007
  emissions inventory for Area and Point Sources
• MOBILE6 for On-Road NONROAD for Off-Road Mobile
• Area and Non-EGU Point based on 1996 NET
• Revised 2007 Base Case
• Area and Non-EGU Point from 1999 NEI

10
Projected 8-Hr Ozone Design Values (DVs)Using
1999-2003 Observed DVsOriginal 2007 Base Case
EPA HDD Rule
Attainment demonstrated all sites using 1999,
2002, 2003 DVs Skiatook projected to still
violate 8-hour ozone standard using 2000 and 2001
DVs (88.0 and 85.1 ppb) Tulsa close to violating
using 2000 DVs (84.7 ppb)
11
Project 8-Hour Ozone Design Values (DVs)Revised
2007 Base Case (1999 NEI)
Attainment demonstrated all sites using 1999,
2001, 2002, 2003 DVs Using 2000 DVs, Skiatook
(87.5 ppb) and Tulsa (85.2 ppb) still
violate Using 2001 DVs, Skiatook close (84.7 ppb)
12
6. actually MSA
11. TTMA link-based network
13
Weight of Evidence Attainment Demonstration

• EPA Draft 8-hr Ozone Guidance (EPA, 1999)
• Modeled max 8-hr ozone Design Value
• Air Quality Modeling Analysis
• Change in grid-hours with ozone 84 ppb
• Number grid cells 84 ppb
• Change in ppb-hr with ozone 84 ppb
• Air Quality and Emission Trends
• Extrapolation of DV to attainment year
• Emission Trends
• Observation Based Models (OBM)
• Not quantitative, support VOC/NOx selection

14
Weight of Evidence Attainment Demonstration

• Other corroborative analysis
• Quantifying model uncertainties
• Other year Design Values
• Have done 1999 to 2003
• Examine basis for excluding days
• Use higher base year base case cutoff
  concentration than 70 ppb (80 ppb)
• Additional data collection
• More monitors in more current years
• Other?

15
Weight of Evidence (WOE) Attainment Demonstration
Modeling Component

• EPA Draft 8-hr Ozone Guidance (EPA, 1999)
• Modeled maximum 8-hr ozone Design Value must be
  less than 90 ppb
• Tulsa EAC 2007 Base Case projected maximum 8-hour
  ozone Design Value 87 ppb
• Air Quality Modeling Analysis
• Change in grid-hours with ozone 84 ppb
• Change in Number grid cells 84 ppb
• Change in ppb-hr with ozone 84 ppb
• Also other data analysis (meteorology, emissions,
  trends, etc.)

16
Modeled WOE Attainment Demonstration Tests

• EPA Guidance WOE Metrics
• Grid-Hours 84 ppb
• Integrated exposure metric of modeled exceedances
  over time and area
• Grid-Cells 84 ppb
• Integrated exposure metrics over area
• Relative Reduction (RD) Measure of amount of
  time and area ozone 84 ppb (ppb-hr)
• Integrated dosage metric
• EPA Guidance States that large reductions in
  these metrics are desirable
• EPA defines large as 80

17
Modeled WOE Attainment TestsEPA Guidance Desires
80
Across OK 4 km grid 80 large reduction target
not met for Cntl5 (w/o expiring permits) and
Cntl11 (7.8 RVP in TTMA w/ 85 penetration) Need
to isolate calculations to Tulsa area
18
Next Steps -- Near Term by 03/31/04

• Clean Air Action Plan
• TCMs in Tulsa and OKC
• WOE Attainment Demonstration
• Modeled Metrics
• Other Year DVs
• DV Trends
• Emission Trends
• Anomalous conditions of 1999/2000???
• Other???

19
Next Steps -- After March 31, 2004

• Justification for modifying BCs
• BCs based on August 1999 CMAQ run
• SIP Call, Tier 2, HDDR, NONROAD, etc. will lower
  future-year ozone and precursors
• No future-year CMAQ run for August 1999
• Develop Procedures for Modifying BCs based on
  other regional runs?
• CAMx EPA IAQR 1995, 2001, 2010
• CAMx HDDR 1995, 2007
• Run CMAQ August 1999 episode for 2007?
• Unlikely to get all files

20
Next Steps -- After March 31, 2004

• Justify use of more recent observed DVs (e.g.,
  2001-2003)
• EPA Guidance (EPA, 1999, pg.37)
• States should review monitored data from (a) the
  3-year period straddling the year represented
  by the most recent available emissions inventory
  (e.g., 1995-1997 for 1996 inventory), and (b) the
  3-year period used to designate an area
  attainment. (or more recent 3-year period)
• The current design value used in the modeled
  attainment and screening tests is the higher of
  the two estimated obtained from (a) and (b).

21
Next Steps -- After March 31, 2004

• Justify use of more recent observed DVs (e.g.,
  2001-2003)
• (a) 1999 NEI most recent 1998-2000 DVs
• (b) Attainment designation 2001-2003 DVs
• Next emission year update is 2002 NEI
• Initial State submissions (CERR) to EPA due June
  2004
• If update OK emissions using 2002 Emissions
• Justifies use of 2001-2003 observed DVs
• Pass modeled ozone attainment demonstration test
  without WOE (still recommend do WOE procedures)

22
Ozone Source Apportionment Modeling

• Ozone Source Apportionment Technology (OSAT)
  Anthropogenic Precursor Culpability Assessment
  (APCA)
• Uses reactive tracers to track VOC and NOx
  emissions from source regions and categories
  (source groups)
• Ozone formed is allocated to source groups based
  on the relative contribution of the appropriate
  precursor (VOC or NOx) present
• Mass balance accounting system where all ozone is
  accounted for

23
Ozone Source Apportionment Modeling

• OSAT/APCA Source Apportionment
• Modeling tool that can be used to help identify
  sources that contribute the most to ozone
  concentrations
• Because ozone formation is nonlinear, source
  apportionment is not unique
• Changes in emissions or BCs (e.g., controls) will
  change ozone source apportionment
• Used in EPA NOx SIP Call Rulemaking to identify
  which States contribute significantly to
  nonattainment

24
OK Ozone Source Apportionment Modeling

• 1999 Base Case Emissions Scenario
• 11 Source Regions
• Tulsa MSA 2. OKC MSA
• 3. 6. SW, SE, NE NW OK 4-km Grid
• 7. 8. N S Texas
• 9. KS 10. LA/AR/MO/IL 11. Southeast
• 4 Source Categories
• On-Road Mobile
• Other Low-Level Anthropogenic
• Elevated Points
• Biogenic
• 11 x 4 2 (IC/BC) 46 Source Groups

25
Ozone Source Apportionment Modeling
26
Ozone Source Apportionment Modeling
27
OSAT/APCA Skiatook on August 16, 1999
28
OSAT/APCA Skiatook on August 17, 1999
29
OSAT/APCA Skiatook on August 21, 1999
30
OSAT/APCA Skiatook on August 23, 1999
31
Daily Max 8-Hr Ozone August 23, 1999 Run7
32
OSAT/APCA Skiatook on August 28, 1999
33
Daily Max 8-Hr Ozone August 28, 1999 Run7
34
OSAT/APCA Skiatook on August 30, 1999
35
OSAT/APCA Skiatook on September 1, 1999
36
OK Ozone Source Apportionment Preliminary
Conclusions

• Boundary Conditions and Tulsa Emissions
  consistently highest contributors to elevated
  8-hr ozone concentrations at Skiatook
• Next most important contributor varies day-to-day
  (NE OK SE OK, OKC Central US Southeast US N
  S Texas)
• Using APCA approach results in very little ozone
  attributed to biogenic emissions in Oklahoma due
  to little biogenic NOx in the area

37
OK Ozone Source Apportionment Preliminary
Conclusions

• Use of 1999 CMAQ BCs for Future-Year (2002 and
  2007) modeling very conservative as regional
  control strategies not accounted for
• Tier 2/Low Sulfur On-Road Mobile
• Heavy Duty Diesel
• Non-Road Engines
• Clear Skies Initiative (CSI) ???
• PM Transport Rule ???

38
OK Ozone Source Apportionment Preliminary
Conclusions

• Sources that contribute the most to daily maximum
  8-hr ozone at Skiatook during August 1999 episode
• Tulsa MSA On-road Mobile
• Tulsa MSA Other Low-Level Anthropogenic
  (non-road, area low-level point)
• Tulsa MSA Elevated Point
• NE OK (elevated point)
• Central and Southeast US
• N S Texas
• OKC

39
VOC/NOx Emission Reduction Sensitivity Runs

• 1999 Base Case Emissions Scenario
• Reduce Anthropogenic VOC and/or NOx Emissions by
  30 in the Oklahoma 4 km Grid plus extra 30
  reduction in OK Point Source NOx
• Effects of estimated daily max 8-hr ozone across
  Oklahoma
• Effects on projected 8-hr ozone Design Values
• Documented in December 22, 2003 Report (Morris,
  Tai and Jia, 2003)

40
VOC/NOx Emission Reduction Sensitivity Runs
41
30 NOx Reduction Aug 17 (lt) and Aug 30 (rt)
42
30 VOC Reduction Aug 17 (lt) and Aug 30 (rt)
43
Effects of Emission Reductions on 8-Hr O3 DVs
44
Emission Reduction Sensitivity -- Conclusions

• Regional NOx controls in Oklahoma most
  beneficial, although local NOx disbenefits occurs
  in the Tulsa and OKC areas
• Regional VOC controls in Oklahoma are only
  effective at reducing ozone in and immediately
  downwind of Tulsa and OKC and mitigate some (but
  not all) of the NOx disbenefits
• Effectiveness of Oklahoma emission controls is
  dampened by contributions of ozone transport
• Oklahoma air quality will (and has?) likely see
  benefits from numerous regional ozone/PM
  transport rules

45
Denver/New Mexico EACCommon Modeling Tools

• MM5 Meteorological Model
• Common 36/12 km grid
• Separate local 4/1.33 km (Denver) and 4 km (New
  Mexico) grids
• EPS2x Emissions Model
• NEI 1999 base inventory
• Colorado emissions provided by CDPHE
• Local Oil and Gas Emissions (COGA, NMOGA)
• CAMx Photochemical Grid Model

46
Denver/New Mexico EAC -- Common Episode and
Regional 36/12 km Grid

• June July 2002 Regional Episode
• Embedded High 8-hr Ozone Episodes

Denver (4/1.33 km) June 8 - 12, 2002 June 25
July 1, 2002 July 18 21, 2002
New Mexico (4 km) June 4 8, 2002 June 16 - 19,
2002 June 30 July 2, 2002 July 17 - 18, 2002
47
Denver 36/12/4/1.33 km Modeling Domain (1.33 km
grid not used)
48
Differences in Denver and New Mexico EACs

• Both regions were attaining the 8-hour ozone
  standard at the initiation of the studies in
  early 2003
• During 2003, ozone remained relatively low in the
  Four Corners region maximum 8-hour ozone Design
  Value during 2001-2003 is 74 ppb (attainment ppb)
• During 2003, Denver had the worse summer for
  ozone in many years three monitors now violate
  the 8-hour ozone standard based on 2001-2003
  data
• 87 ppb _at_ Rocky Flats North
• 85 ppb _at_ Chatfield
• 85 ppb _at_ South Boulder

49
Denver EAC Modeling ? Timeline

• May 2003 Modeling Protocol (includes episode
  selection)
• July 2003 MM5 Meteorological Modeling Report
• ? During Summer 2003 8-hr ozone standard
  violated
• Sep 2003 2002 Emissions Inventory Report and
  Addendum
• Oct 10, 2003 Modeling Review Panel (MRP) Meeting
• Nov 19, 2003 2002 Base Case/Evaluation Report
• Dec 10, 2003 Modeling Review Panel (MRP) Meeting
• Dec 2003 Preliminary 2007 Modeling
• Jan-Feb 2004 Revised 2007 Base Case and Control
  Strategy Modeling
• Feb 4, 2004 Report on 2007 Emission Reduction
  Sensitivity Modeling
• Feb 27, 2004 Report on 2007 Control Strategy
  Modeling
• Feb 27, 2004 Draft EAC Ozone Action Plan
• Mar 2004 Preliminary Ozone Source Apportionment
  Modeling

50
CAMx Base Case Evaluation

• Follow EPA Draft 8-Hour Ozone Modeling Guidance
  Evaluation Procedures
• Big Picture Graphical Performance
• Spatial Maps of Predictions and Observations
• Scatter and Q-Q Plots
• Time Series Plots
• Ozone Metrics
• New 8-Hour Ozone Performance Metrics
• Performance Goals
• Initial Evaluation for Ozone Only

51
Denver Ozone Monitoring Network
52
2002 Base Case 8-Hour Ozone on July 1, 2002

• Modeled ozone near monitor matches observations
• Spatial displacement of modeled ozone away from
  DMA
• DMA ozone suppression overstated

53
EPA Draft 8-Hour Ozone Guidance Performance Goals
(EPA, 1999)
54
EPA 8-Hour Ozone Performance Metrics

• bias pred/obs mean 8-hr ( 1-hr) daily maxima
  near each monitor (EPA, 1999)
• 20 most monitors (8-hr comparisons only)
  (EPA, 1999)
• How to define near? Use same NX by NY array
  of grid cells centered on monitor with 15 km
  radius as used in the attainment test (e.g., 9 by
  9 for 4 km grid)
• What predicted ozone to select for comparison
  with observed maxima? several approaches,
  including
• Maximum ozone near monitor (same as used in the
  attainment test)
• Nearest ozone near monitor (closest to
  observation)

55
Max Predicted 8-Hr Ozone Near Monitor June
2002 Episode and 2002 Base Case 4 km

• EPA Performance Goal of within 20 at most
  monitors
• 96 of Max Pred near monitor value
• 3 pred/obs pairs do not meet Cnty 6/26 CO Springs 6/28 and Boulder on 7/1)

56
Nearest Predicted 8-Hr Ozone Near Monitor June
2002 Episode and 2002 Base Case 4 km

• EPA Goal
• 98 of pred/obs pairs meet
• Weld Cnty 6/26 pred/obs of 57/81 ppb
• CO Springs 6/28 pred/obs of 57/74 ppb
• Meets goal with underprediction tendency

57
Summary of 2002 Base Case Ozone Performance for
Denver June 2002 Episode

• Meets EPA performance goals
• 95 of monitor/days meet EPAs performance goal
  
• EPA Bias and Gross Error performance goals (15
  35) met using maximum and nearest predicted 8-hr
  ozone near the monitor
• Underestimation bias sometimes exceeds performance goal but gross error always meets
  
  pred/obs 8-hr ozone at the monitor
• Some ozone spatial alignment issues
• Underestimation bias (but within performance
  goals)
• Understated ozone transport?
• Understated emissions?
• Model performance for July 2002 episode was not
  as good so dropped for initial control strategy
  evaluations

58
2007 Emission Reduction Sensitivity Analysis

• Across-the-Board 10 reduction in anthropogenic
  emissions in DMA Weld County (plus on-road and
  off-road reductions).
• Key Findings
• Modeled ozone stiff response to local emission
  reductions (i.e., ozone not very responsive to
  local emission controls)
• VOC control in DMA is more effective than NOx
  control in DMA
• 10 VOC control in DMA results in 0.3-0.4 ppb
  ozone reduction at Rocky Flats North monitor
• 10 NOx control in DMA results in 0.4 ppb ozone
  increase at Rocky Flats North monitor

59
2007 Emission Scenarios

• 2007 Base Case CO Data Provided by CDPHE
• 9 psi RVP gasoline w/ 25 Ethanol penetration
• 2007 Control Package w/ 8.1 RVP
• 2007 Control Package w/ 7.8 RVP
• Control Measures Modeled (DMA Weld County)
• 37.5 control on Flash VOC emissions
• Control of RICE natural gas units and Dehydrators
• 8.1 or 7.8 psi RVP gasoline on-road mobile
  sources in DMA with 25 Ethanol market penetration

60
2007 Emissions Scenarios DMAWeldVOC Emissions
(tons per day)
DMAWeld Adams, Arapahoe, Broomfield, Boulder,
Denver, Douglas, Jefferson and Weld Counties.
On-road mobile includes extra areas in link
based network and should be 65 for DMAWeld
counties
61
2007 Emissions Scenarios DMAWeldNOx Emissions
(tons per day)
DMAWeld Adams, Arapahoe, Broomfield, Boulder,
Denver, Douglas, Jefferson and Weld Counties.
On-road mobile includes extra areas in link
based network and should be 65 for DMAWeld
counties
62
Daily Max 8-Hour Ozone on July 1, 2002
2007 Base 2007
Cntl Pkg
w/ 8.1 RVP - Base
63
How Are Modeling Results Used to Demonstrate
Attainment of the 8-Hour Ozone NAAQS

• EPA Draft Guidance uses model in a relative sense
  to scale current-year observed 8-hour ozone
  Design Values (DVC) to the future-year (DVF)
• This is done using monitor specific Relative
  Reduction Factor (RRFi) that is the ratio of the
  future-year to current-year 8-hour ozone model
  estimates near the monitor
• DVFi RRFi x DVCi

64
How Are Modeling Results Used to Demonstrate
Attainment of the 8-Hour Ozone NAAQS

• The RRFi for monitor i is the ratio of the
  average daily maximum 8-hour ozone concentrations
  near the monitor for the future-year (O3F) to the
  current-year (O3C) scenario for all episode days
  when the current-year ozone is greater than 70
  ppb (O3Cij 70 ppb)

65
How Are Modeling Results Used to Demonstrate
Attainment of the 8-Hour Ozone NAAQS

• Selecting maximum estimated ozone near the
  monitor
• Near defined by NX x NY array of cells centered
  on monitor that encompasses 15 km radius
• 5 km 7 x 7 4 km 9 x 9 etc.
• Exclude days in which 2002 Base Case estimated
  8-hr ozone near monitor is
• Eliminates low ozone (background) days with low
  emissions contributions

66
2007 Projected 8-Hour Ozone Design
Values Attainment Demonstrated when DVF 67
Denver EAC Modeled Attainment Test

• Projected 2007 8-hour ozone Design Value at Rocky
  Flats is 86.5 ppb for 2007 Base Case and 86.0 ppb
  for 2007 Control Strategies
• All other monitors demonstrate attainment (ppb)
• Why is modeled ozone so stiff?
• Contributions of ozone transport
• 2002 episode not as adverse as conditions that
  produced observed Design Values that includes
  summer 2003
• Model underestimation bias

68
Ozone Source Apportionment10 Source Regions in
36 km West US Grid

• DMA
• Weld
• Elbert
• Morgan
• Larimer
• El Paso
• North CO
• South CO
• Central US
• Western US

69
Denver Ozone Source Apportionment

• Source Apportionment results currently under
  review by State and Denver RAQC, report not yet
  available
• Preliminary results suggest the following
• A majority (75 to 85) of the peak 8-hour ozone
  concentrations at the Rocky Flats monitor come
  from outside of the Denver Metropolitan Area
  (DMA)
• For sources in the DMA, on-road mobile sources
  are most important followed by non-road mobile
  sources
• Helps explain why modeling results are so stiff
  in response to local controls

70
Weight of Evidence (WOE) Attainment Demonstration
Modeling Component

• EPA Draft 8-hr Ozone Guidance (EPA, 1999)
• Modeled maximum 8-hr ozone Design Value must be
  less than 90 ppb
• Denver EAC 2007 Control Strategy maximum
  projected 8-hour ozone Design Value 86 ppb
• Air Quality Modeling Analysis
• Change in grid-hours with ozone 84 ppb
• Change in Number grid cells 84 ppb
• Change in ppb-hr with ozone 84 ppb
• Also other data analysis (meteorology, emissions,
  trends, etc.)

71
WOE Attainment Demonstration Issues

• Model is Stiff in Response to Local Controls
• Model meets EPA performance goals (within 20)
  but has underprediction bias
• Using some days with ozone estimates of 70-75 ppb
  to scale 87 ppb Design Value at Rocky Flats
• June 2002 episode not as adverse as observed
  2001-2003 8-hr ozone Design Values that include
  the Summer of 2003
• Local emissions not contributing as much to the
  ozone maximums in the model as likely occurred in
  the observed ozone Design Values

72
Modeled WOE Attainment Demonstration Tests

• Project 2007 8-hour ozone Design Values (DVs)
  using modeled ozone 80 ppb
• Modeled ozone more representative of ozone DVs
• 2007 Projected 8-hour ozone at Rocky Flats is
• 85.4 Control Pkg. w/ 8.1 RVP
• 85.2 Control Pkg. w/ 7.8 RVP
• Project 2007 DVs using observed 2000-2002 DVs
• Conditions more consistent with June 2002 episode

73
Modeled WOE Attainment Demonstration Tests

• EPA Guidance WOE Metrics
• Grid-Hours 84 ppb
• Integrated exposure metric of modeled exceedances
  over time and area
• Grid-Cells 84 ppb
• Integrated exposure metrics over area
• Relative Reduction (RD) Measure of amount of
  time and area ozone 84 ppb (ppb-hr)
• Integrated dosage metric
• EPA Guidance States that large reductions in
  these metrics are desirable
• EPA defines large as 80

74
Modeled WOE Attainment TestsEPA Guidance Desires
80
75
Denver EAC WOE Attainment Tests

• Anomalous Meteorological Conditions of 2003
  produced unusually high ozone
• Unusually high temperatures during Summer 2003
• July 2003 4rth highest mean temperature on record
• Unusually low mixing depths trapped local
  pollutants causing high ozone
• Much higher contribution of local emissions to
  ozone than in 2002 episodes

76
Denver EAC WOE Attainment Tests

• Zurbenko-Rao filter applied to ozone at Rocky
  Flats to decompose ozone trends into long-term
  and short-term components accounting for
  temperature
• Over 1993-2003 downward trend in 4rth highest
  8-hr ozone of 1.2 per year

77
Denver EAC WOE Attainment Tests

• Downward trend in emissions in the DMA
• From 2002, VOC emissions are projected to be
  reduced by 10 (2007) and 13 (2012)
• From 2002, NOx emissions are projected to be
  reduced by 15 (2007) and 19 (2012)
• Model Uncertainty and Representativeness
• Model underestimation bias
• Use of days 80 ppb reduces DV from 86 ppb to 85
  ppb
• Contributions of local emissions understated
• 2003 observed DV of 87 has higher local
  contribution than 2002 episodes

78
National Academy of Sciences (NAS) Report

• Extensive Regional Atmosphere Hydrocarbon
  Pollution in the Southwestern US
• Aaron Katzenstein, Lambert Doezema, Isobel
  Simpson, Donald Blake and F. Sherwood Rowland,
  University of California at Irvine (UCI) (October
  14, 2003)
• Collected many hydrocarbon samples in OK/KS/TX
• Found elevated alkane (HC) concentrations
• Attributed to Oil and Gas production in region
• Estimated regional-wide emission rates

79
UCI Ethane Observations(Source Katzenstein et
al., 2003)
80
UCI OG Emissions Sensitivity Test-- Approach to
Integrate UCI Emissions with Photochemical Model

• Contacted UCI who (Lambert Doezema) provided a
  low and high estimate region-wide hydrocarbon
  emission estimates from their measurement grid
• Methane, ethane, propane, ni butane, ni pentane
• Determine county equivalent to UCI measurement
  grid
• Allocate UCI OG hydrocarbon emissions to
  counties based on average OilGas production
• Methane considered non-reactive VOC so is dropped

81
UCI Measurement Grid
82
(No Transcript)
83
(No Transcript)
84
Comparison of OilGas Production to NEI99 OilGas
VOC Emissions
Average Oil and Gas Production
1999 NEI OilGas VOC Emissions
85
UCI OG Emissions Sensitivity Test

• Replace existing county-level OilGas VOC
  emissions from NEI99 with UCI estimates allocated
  to OK/KS/TX/CO counties based on OilGas
  production
• Process with EPS2x emissions to speciate, grid
  and temporally allocate emissions for CAMx
  modeling
• Run CAMx model for high and low estimates of UCI
  OilGas HC emissions
• Compare ozone estimates

86
Comparison of NEI99 and UCI OilGas Production
VOC Emissions (tons per day)
87
Ozone Results using UCI OG HC Emissions

• Very small increases in ozone concentrations,
  primarily downwind of NOx sources and near high
  OilGas production
• Maximum increase in 8-hour ozone concentrations
  are 0.5-1.0 ppb
• Small increases elsewhere
• No effect on model performance
• Normalized bias of 10.1 reduced to 10.0
• Does not solve underprediction tendency

88
Daily Max 8-Hour Ozone PerformanceNEI99 OG VOC
UCI OG VOC
89
Difference in 8-Hour Ozone Concentrations (ppb)
from UCI OilGas HC Emissions
90
UCI OilGas HC Emissions Conclusions

• Preliminary results still being examined
• Use of UCI OG VOC estimates results in small
  increases in ozone (always
• Just included potentially missing VOC emissions,
  results may be different if potentially missing
  NOx emissions included
• Does not explain western U.S. ozone
  underprediction tendency seen in Denver, San Juan
  and Oklahoma EACs and with multiple models (CAMx
  and CMAQ)
• Return to EAC Home Page