stlouis-8hr-o3-tsd

              ST. LOUIS 8-HOUR OZONE
TECHNICAL SUPPORT DOCUMENT
Illinois Environmental Protection Agency
Bureau of Air
1021 N. Grand Ave. East
P.O. Box 19276
Springfield, Illinois 62794-9276
March 26, 2007
i
TABLE OF CONTENTS
1.0 EXECUTIVE SUMMARY...............................................................................................................................1
2.0 INTRODUCTION.............................................................................................................................................3
2.1 BACKGROUND AND PURPOSE.................................................................................................................3
2.2 STATE AGENCY ORGANIZATIONS AND WORK GROUPS...................................................................4
2.3 OVERVIEW OF APPROACH.......................................................................................................................5
2.3.1 Modeling Protocol................................................................................................................................5
2.3.2 Model Selection.....................................................................................................................................5
2.3.3 Modeling Domains................................................................................................................................7
2.3.4 Vertical Structure of Modeling Domain...............................................................................................11
2.3.4.1 Air Quality Data.............................................................................................................................................13
2.3.4.2 Ozone Column Data.......................................................................................................................................14
2.3.4.3 Initial and Boundary Conditions Data.............................................................................................................14
2.3.5 Episode Selection................................................................................................................................14
2.3.6 Conceptual Model...............................................................................................................................20
2.3.7 Emissions Input Preparation and QA/QC............................................................................................25
2.3.8 Meteorological Input Preparation and QA/QC...................................................................................26
2.3.9 Air Quality Model Input Preparation and QA/QC...............................................................................28
2.3.10 Base Case Modeling and Model Performance Evaluation..................................................................29
2.3.11 Future-Year Modeling and Modeled Attainment Demonstration.........................................................31
2.3.12 Weight of Evidence (WOE) Analysis....................................................................................................32
3.0 EMISSIONS MODELING.............................................................................................................................32
3.1 2002 BASE 4 MODEL VALIDATION EMISSIONS INVENTORY...........................................................35
3.1.1 2002 Base 4 Model Validation Inventory Data Sources......................................................................35
3.1.1.1 All Point Sources Except EGUs in Midwest RPO and Minnesota..................................................................35
3.1.1.2 EGU Point Sources in the Midwest RPO and Minnesota................................................................................36
3.1.1.3 Area Sources..................................................................................................................................................36
3.1.1.4 Offroad Mobile Sources.................................................................................................................................37
3.1.1.5 Onroad Mobile Sources..................................................................................................................................38
3.1.1.6 Biogenic Sources............................................................................................................................................39
3.1.2 2002 Model Validation Inventory Emissions Summaries.....................................................................39
3.2 2002 BASE 4 TYPICAL EMISSIONS INVENTORY..................................................................................47
3.3 2009 BASE 4 ON-THE-BOOKS EMISSIONS INVENTORY.....................................................................49
3.3.1 2009 Base 4 On-the-Books Inventory Data Sources............................................................................49
3.3.1.1 All Point Sources Except EGUs in Midwest RPO and Minnesota..................................................................49
3.3.1.2 EGU Point Sources in the Midwest RPO and Minnesota................................................................................51
3.3.1.3 Area Sources..................................................................................................................................................51
3.3.1.4 Offroad Mobile Sources.................................................................................................................................52
3.3.1.5 Onroad Mobile Sources..................................................................................................................................52
3.3.1.6 Biogenic Sources............................................................................................................................................56
3.3.2 2009 Base 4 On-the-Books Inventory Emissions Summaries...............................................................56
3.4 QUALITY ASSURANCE AND QUALITY CONTROL..............................................................................61
4.0 MODEL PERFORMANCE EVALUATION................................................................................................65
4.1 MODEL PERFORMANCE EVALUATION APPROACH..........................................................................65
4.2 MODEL PERFORMANCE METRICS AND GOALS.................................................................................67
4.3 OZONE MODEL PERFORMANCE STATISTICS......................................................................................67
4.3.1 Performance Evaluation for Episode 1: June 10-24, 2002.................................................................68
4.3.1.1 Episode 1 Ozone Performance Metrics...........................................................................................................68
4.3.1.2 Episode 1 Scatter Plots...................................................................................................................................71
4.3.1.3 Episode 1 Spatial Plots and Conceptual Model Comparison (June 19-23).........................................................72
4.3.2 Performance Evaluation for Episode 2: July 2-16, 2002.....................................................................75
4.3.2.1 Episode 1 Ozone Performance Metrics...........................................................................................................75
4.3.2.2 Episode 2 Scatter Plots...................................................................................................................................76
4.3.2.3 Episode 2 Spatial Plots and Conceptual Model Comparison..........................................................................76
4.3.3 Performance Evaluation for Episode 3: July 29 – August 5, 2002......................................................80 ii
4.3.3.1 Episode 3 Ozone Performance Metrics...........................................................................................................80
4.3.3.2 Episode 3 Scatter Plots......................................................................................................................................81
4.3.3.3 Episode 3 Spatial Plots...................................................................................................................................81
4.4 EVALUATION AT KEY MONITORS FOR ATTAINMENT DEMONSTRATION DAYS.......................83
4.5 CONCLUSIONS..........................................................................................................................................88
5.0 ATTAINMENT DEMONSTRATION MODELING ANALYSES..............................................................89
5.1 FUTURE-YEAR MODELING INPUTS.......................................................................................................90
5.2 PROJECTION OF 2009 8-HOUR OZONE DESIGN VALUES...................................................................90
5.3 SCREENING ATTAINMENT DEMONSTRATION TEST FOR UNMONITORED AREAS....................95
5.4 SUMMARY OF MODELED ATTAINMENT DEMONSTRATION...........................................................97
6.0 WEIGHT OF EVIDENCE ANALYSIS.........................................................................................................98
6.1 OVERVIEW OF WOE ANALYSIS..............................................................................................................98
6.2 MODELED ATTAINMENT DEMONSTRATION USING CAMX............................................................98
6.3 ADDITIONAL MODELING METRICS......................................................................................................99
6.4 ATTAINMENT TEST WITH ALTERNATIVE CUT-OFFS......................................................................100
6.5 INDEPENDENT CORROBORATIVE MODELING ANALYSIS.............................................................103
6.5.1 EPA Interstate Air Quality Rule CAMx Modeling.............................................................................103
6.5.2 Lake Michigan Air Directors Consortium CAMx Modeling..............................................................104
6.6 COMPARISONS OF 2002/2009 EMISSION REDUCTIONS WITH OTHER STUDIES..........................105
6.7 OZONE SOURCE APPORTIONMENT MODELING...............................................................................108
6.8 TRENDS IN AMBIENT AIR QUALITY....................................................................................................119
6.9 TRENDS IN EMISSIONS..........................................................................................................................120
6.10 CONCLUSIONS OF ST. LOUIS WOE.......................................................................................................121
7.0 REFERENCES..............................................................................................................................................123
Appendix A. Episode Selection and Conceptual Model
Appendix B. MM5 Evaluation
Appendix C. St. Louis Base 4 Emissions
Appendix D. Model Performance Evaluation iii
LIST OF TABLES
TABLE 2-1. RPO UNIFIED GRID PROJECTION DEFINITION...........................................................................................11
TABLE 2-2. GRID DEFINITIONS FOR MM5, SMOKE/EMS, AND CMAQ/CAMX.........................................................11
TABLE 2-3. VERTICAL LAYER DEFINITION FOR MM5 SIMULATIONS (LEFT-MOST COLUMNS) AND APPROACH FOR REDUCING CMAQ/CAMX LAYERS BY COLLAPSING MULTIPLE MM5 LAYERS (RIGHT COLUMNS)....................12
TABLE 2-4. 2002 8-HOUR OZONE EXCEEDANCE DAYS IN THE ST. LOUIS AREA..........................................................18
TABLE 3-1. SUMMARY OF WEEKDAY NOX EMISSIONS FROM THE 2002 BASE 4 TYPICAL AND 2009 ON-THE-BOOKS INVENTORIES FOR ST. LOUIS NONATTAINMENT COUNTIES.................................................................................34
TABLE 3-2. SUMMARY OF WEEKDAY VOC EMISSIONS FROM THE 2002 BASE 4 TYPICAL AND 2009 ON-THE-BOOKS INVENTORIES FOR ST. LOUIS NONATTAINMENT COUNTIES.................................................................................34
TABLE 3-3. COUNTIES WITH LINK-BASED VMT AND SPEED DATA FROM EAST WEST GATEWAY................................38
TABLE 3-4. 2002 BASE 4 MODEL VALIDATION INVENTORY – WEEKDAY, SATURDAY, SUNDAY NOX EMISSIONS FOR ST. LOUIS NONATTAINMENT AREA COUNTIES BY SOURCE TYPE........................................................................41
TABLE 3-5. 2002 BASE 4 MODEL VALIDATION INVENTORY – WEEKDAY, SATURDAY, SUNDAY VOC EMISSIONS FOR ST. LOUIS NONATTAINMENT AREA COUNTIES BY SOURCE TYPE........................................................................41
TABLE 3-6. 2002 BASE 4 MODEL VALIDATION INVENTORY – WEEKDAY, SATURDAY, SUNDAY CO EMISSIONS FOR ST. LOUIS NONATTAINMENT AREA COUNTIES BY SOURCE TYPE..............................................................................42
TABLE 3-7. 2002 BASE 4 MODEL VALIDATION INVENTORY – DAILY ONROAD MOBILE NOX EMISSIONS FOR ST. LOUIS NONATTAINMENT AREA COUNTIES....................................................................................................................43
TABLE 3-8. 2002 BASE 4 MODEL VALIDATION INVENTORY – DAILY ONROAD MOBILE VOC EMISSIONS FOR ST. LOUIS NONATTAINMENT AREA COUNTIES....................................................................................................................44
TABLE 3-9. 2002 BASE 4 MODEL VALIDATION INVENTORY – DAILY ONROAD MOBILE CO EMISSIONS FOR ST. LOUIS NONATTAINMENT AREA COUNTIES....................................................................................................................45
TABLE 3-10. 2002 BASE 4 MODEL VALIDATION INVENTORY – DAILY BIOGENIC VOC EMISSIONS FOR ST. LOUIS NONATTAINMENT AREA COUNTIES....................................................................................................................46
TABLE 3-11. 2002 BASE 4 MODEL VALIDATION INVENTORY – EGU NOX EMISSIONS BY DAY...................................47
TABLE 3-12. 2002 BASE 4 TYPICAL INVENTORY – WEEKDAY, SATURDAY, SUNDAY NOX EMISSIONS FOR ST. LOUIS NONATTAINMENT AREA COUNTIES BY SOURCE TYPE........................................................................................48
TABLE 3-13. 2002 BASE 4 TYPICAL INVENTORY – WEEKDAY, SATURDAY, SUNDAY VOC EMISSIONS FOR ST. LOUIS NONATTAINMENT AREA COUNTIES BY SOURCE TYPE........................................................................................48
TABLE 3-14. 2002 BASE 4 TYPICAL INVENTORY – WEEKDAY, SATURDAY, SUNDAY CO EMISSIONS FOR ST. LOUIS NONATTAINMENT AREA COUNTIES BY SOURCE TYPE........................................................................................49
TABLE 3-15. COMPARISON OF MISSOURI-SIDE 2002 AND 2009 VMT (MILES/ANNUAL AVERAGE WEEKDAY)..............54
TABLE 3-16. COMPARISON OF ILLINOIS-SIDE 2002 AND 2009 VMT (MILES/ANNUAL AVERAGE WEEKDAY)................55
TABLE 3-17 .VMT GROWTH FACTORS FOR ST. LOUIS NON-ATTAINMENT AREA COUNTIES........................................55
TABLE 3-18. MAJOR DIFFERENCES BETWEEN 2002 AND 2009 MOBILE6 SETTINGS IN NONATTAINMENT AREA COUNTIES..........................................................................................................................................................56
TABLE 3-19. 2009 BASE 4 ON-THE-BOOKS INVENTORY – WEEKDAY, SATURDAY, SUNDAY NOX EMISSIONS FOR ST. LOUIS NONATTAINMENT AREA COUNTIES BY SOURCE TYPE..............................................................................57
TABLE 3-20. 2009 BASE 4 ON-THE-BOOKS INVENTORY – WEEKDAY, SATURDAY, SUNDAY VOC EMISSIONS FOR ST. LOUIS NONATTAINMENT AREA COUNTIES BY SOURCE TYPE..............................................................................57
TABLE 3-21. 2009 BASE 4 ON-THE-BOOKS INVENTORY – WEEKDAY, SATURDAY, SUNDAY CO EMISSIONS FOR ST. LOUIS NONATTAINMENT AREA COUNTIES BY SOURCE TYPE..............................................................................58
TABLE 3-22. 2009 BASE 4 ON-THE-BOOKS INVENTORY – DAILY ONROAD MOBILE NOX EMISSIONS FOR ST. LOUIS NONATTAINMENT AREA COUNTIES....................................................................................................................59
TABLE 3-23. 2009 BASE 4 ON-THE-BOOKS INVENTORY – DAILY ONROAD MOBILE VOC EMISSIONS FOR ST. LOUIS NONATTAINMENT AREA COUNTIES....................................................................................................................60
TABLE 3-24. 2009 BASE 4 ON-THE-BOOKS INVENTORY – DAILY ONROAD MOBILE CO EMISSIONS FOR ST. LOUIS NONATTAINMENT AREA COUNTIES....................................................................................................................61
TABLE 3-25. 2002 BASE 4 TYPICAL INVENTORY – COMPARISON OF RAW ANNUAL AND GRIDDED (36 KM) ANNUAL POINT SOURCE EMISSIONS.................................................................................................................................62
TABLE 3-26. 2002 BASE 4 TYPICAL INVENTORY – COMPARISON OF TEMPORALIZED (JUNE WEEKDAY) AND TEMPORALIZED (JUNE WEEKDAY) AND GRIDDED (36 KM) POINT SOURCE EMISSIONS.......................................63
TABLE 3-27. 2002 BASE 4 TYPICAL INVENTORY – COMPARISON OF RAW ANNUAL AND GRIDDED (36 KM) ANNUAL AREA SOURCE EMISSIONS..................................................................................................................................63
TABLE 3-28. 2002 BASE 4 TYPICAL INVENTORY – COMPARISON OF TEMPORALIZED (JUNE WEEKDAY) AND TEMPORALIZED (JUNE WEEKDAY) AND GRIDDED (36 KM) AREA SOURCE EMISSIONS........................................64
TABLE 3-29. 2002 BASE 4 TYPICAL INVENTORY – COMPARISON OF RAW ANNUAL AND GRIDDED (36 KM) ANNUAL OFFROAD MOBILE SOURCE EMISSIONS...............................................................................................................64 iv
TABLE 3-30. 2002 BASE 4 TYPICAL INVENTORY – COMPARISON OF TEMPORALIZED (JUNE WEEKDAY) AND TEMPORALIZED (JUNE WEEKDAY) AND GRIDDED (36 KM) OFFROAD MOBILE SOURCE EMISSIONS....................64
TABLE 3-31. 2002 BASE 4 TYPICAL INVENTORY – COMPARISON OF RAW ANNUAL AND GRIDDED (36 KM) ANNUAL VMT INPUTS FOR ONROAD MOBILE SOURCES....................................................................................................64
TABLE 3-32. COMPARISON OF TEMPORALIZED (JUNE WEEKDAY) AND TEMPORALIZED (JUNE WEEKDAY) AND GRIDDED (36 KM) ONROAD MOBILE SOURCE EMISSIONS...................................................................................65
TABLE 4-1. EPA PERFORMANCE BENCHMARKS (EPA, 1991)......................................................................................67
TABLE 4-2. PERFORMANCE METRICS FOR EPISODE 1. YELLOW HIGHLIGHTING INDICATES THAT A METRIC DOES NOT FALL WITHIN THE PERFORMANCE BENCHMARK...................................................................................................71
TABLE 4-3. PERFORMANCE METRICS FOR EPISODE 2...................................................................................................76
TABLE 4-4. PERFORMANCE METRICS FOR EPISODE 3...................................................................................................81
TABLE 4-5. MODELED AND OBSERVED PEAK 1-HOUR OZONE CONCENTRATIONS AT THE SUNSET HILLS MONITOR DURING EPISODE 2 DAYS USED IN THE ATTAINMENT TEST................................................................................85
TABLE 4-6. MODELED AND OBSERVED PEAK 1-HOUR OZONE CONCENTRATIONS AT THE WEST ALTON MONITOR DURING EPISODE 1 DAYS USED IN THE ATTAINMENT TEST................................................................................86
TABLE 4-7. MODELED AND OBSERVED PEAK 1-HOUR OZONE CONCENTRATIONS AT THE WEST ALTON MONITOR DURING EPISODE 2 DAYS USED IN THE ATTAINMENT TEST................................................................................87
TABLE 4-8. MODELED AND OBSERVED PEAK 1-HOUR OZONE CONCENTRATIONS AT THE MARGARETTA MONITOR DURING EPISODE 1 DAYS USED IN THE ATTAINMENT TEST................................................................................87
TABLE 4-9. MODELED AND OBSERVED PEAK 1-HOUR OZONE CONCENTRATIONS AT THE MARGARETTA MONITOR DURING EPISODE 2 DAYS USED IN THE ATTAINMENT TEST................................................................................88
TABLE 4-10. MODELED AND OBSERVED PEAK 1-HOUR OZONE CONCENTRATIONS AT THE MARGARETTA MONITOR DURING EPISODE 3 DAYS USED IN THE ATTAINMENT TEST................................................................................88
TABLE 5-1. MAXIMUM OBSERVED DESIGN VALUE CONSISTENT WITH 2009 ATTAINMENT SHOWN FOR EACH RRF CUTPOINT SHOWN ON THE SCALES IN FIGURE 5-4..............................................................................................97
TABLE 6-1. NUMBER OF GRID HOURS WITH 8-HOUR DAILY MAXIMUM OZONE > 85 PPB..........................................100
TABLE 6-2. NUMBER OF GRID CELLS WITH 8-HOUR DAILY MAXIMUM OZONE > 85 PPB...........................................100
TABLE 6-3. RELATIVE DIFFERENCE (RD) IN 8-HOUR OZONE CONCENTRATIONS > 85 PPB...........................................100
TABLE 6-4. MODELED DESIGN VALUES FROM CAIR FOR ST. LOUIS AREA MONITORS FROM THE TECHNICAL SUPPORT DOCUMENT FOR THE INTERSTATE AIR QUALITY RULE AIR MODELING ANALYSES (EPA 2004)......................104
TABLE 6-5. ST. LOUIS 2002 AND 2009 TOTAL ANTHROPOGENIC EMISSIONS FOR A TYPICAL SUMMER WEEKDAY (TONS PER DAY)..........................................................................................................................................................105
TABLE 6-6. SUMMARY OF TOTAL ANTHROPOGENIC EMISSIONS FOR A TYPICAL SUMMER WEEKDAY IN THE TULSA AREA FOR 2002, 2007 AND 2012 (TONS PER DAY).............................................................................................106
TABLE 6-7. SUMMARY OF TOTAL ANTHROPOGENIC EMISSIONS FOR A TYPICAL SUMMER WEEKDAY IN THE OKLAHOMA CITY AREA FOR 2002, 2007 AND 2012 (TONS PER DAY)................................................................106
TABLE 6-8. 2002 AND 2007 BASE CASE VOC AND NOX EMISSIONS ON THE DENVER METROPOLITAN AREA AND WELD COUNTY REGIONS (TYPICAL SUMMER WEEKDAY AND COUNTY SPECIFIC EMISSIONS IN TONS PER DAY) FROM MORRIS ET AL., (2004D).........................................................................................................................106
TABLE 6-9. 1999 AND 2009 ANTHROPOGENIC NOX EMISSIONS IN THE DALLAS-FORT WORTH AREA (TYPICAL SUMMER WEEKDAY EMISSIONS IN TONS PER DAY) FROM TAI AND YARWOOD, (2006).......................................107
TABLE 6-10. 1999 AND 2009 ANTHROPOGENIC NOX EMISSIONS IN THE DALLAS-FORT WORTH AREA (TYPICAL SUMMER WEEKDAY EMISSIONS IN TONS PER DAY) FROM TAI AND YARWOOD, (2006).......................................107
TABLE 6-11. 2002 AND 2009 ONROAD MOBILE SOURCE EMISSIONS IN THE STATE OF MISSOURI FROM THE ASIP BASE G EMISSION INVENTORY (TONS PER DAY) (ENVIRON AND ALPINE, 2006)......................................................108
TABLE 6-12. MONITOR-SPECIFIC DESIGN VALUE TRENDS.........................................................................................120 v
LIST OF FIGURES
FIGURE 2-1. NESTED 36/12/4 KM ST. LOUIS MODELING DOMAINS FOR PHOTOCHEMICAL (TOP) AND EMISSIONS (BOTTOM) MODELING...........................................................................................................................................9
FIGURE 2-2. REVISED NESTED 36/12/4 KM ST. LOUIS MODELING DOMAINS FOR PHOTOCHEMICAL MODELING............10
FIGURE 2-3. OZONE MONITORING SITES IN THE ST. LOUIS AREA.................................................................................13
FIGURE 4-1. 8-HOUR OZONE MODEL PERFORMANCE STATISTICS FOR THE THREE ST. LOUIS EPISODES. 2002 BASELINE EMISSIONS SCENARIO.........................................................................................................................................69
FIGURE 4-2. 1-HOUR OZONE MODEL PERFORMANCE STATISTICS FOR THE THREE ST. LOUIS EPISODES. 2002 BASELINE EMISSIONS SCENARIO.........................................................................................................................................70
FIGURE 5-1. DESIGN VALUE PROJECTION FOR THE FUTURE YEAR 2009 (OTB CONTROLS SCENARIO) FOR THE ST. LOUIS 4-KM DOMAIN MONITORS...................................................................................................................................93
FIGURE 5-2. NUMBER OF DAYS USED IN DETERMINING THE PROJECTED FUTURE YEAR (2009) DESIGN VALUE FOR EACH MONITOR IN THE ST. LOUIS 4-KM DOMAIN..........................................................................................................94
FIGURE 5-3. NORMALIZED BIAS PERFORMANCE STATISTICS FOR DAYS USED IN THE MODELED ATTAINMENT TEST FOR THE ORCHARD FARM MONITOR.........................................................................................................................95
FIGURE 5-4. EPISODE 1-3 AVERAGE 2009/2002 MODELED RELATIVE REDUCTION FACTORS FOR THE ST. LOUIS NAA. BLACK NUMBERS WITHIN DOMAIN INDICATE 2002 MONITOR DESIGN VALUES. UPPER (LOWER) PANEL SHOWS RRFS CALCULATED WITH A DAILY MAXIMUM 8-HOUR OZONE THRESHOLD OF 70 PPB (85 PPB)...........................96
FIGURE 6-1. NUMBER OF DAYS USED IN THE 8-HOUR OZONE FUTURE YEAR DESIGN VALUES PROJECTIONS................101
FIGURE 6-2. 2009 FUTURE YEAR DESIGN VALUE PROJECTIONS FOR MONITORS IN THE 4-KM ST. LOUIS DOMAIN........102
FIGURE 6-3. FUTURE YEAR DESIGN VALUES BASED ON THE MAXIMUM DESIGN VALUES FOR THE 2000-2002, 2001-2003, 2002-2004 TIME PERIODS CALCULATED SEPARATELY.............................................................................103
FIGURE 6-4. PROJECTED 2009 8-HOUR OZONE DESIGN VALUES AS PRESENTED BY MIKE KOERBER, LAKE MICHIGAN AIR DIRECTORS CONSORTIUM (LADCO) OCTOBER 31, 2005 “REGIONAL AIR QUALITY PLANNING FOR THE UPPER MIDWEST: ATTAINMENT STRATEGY OPTIONS”......................................................................................104
FIGURE 6-5. SOURCE REGIONS FOR OZONE SOURCE APPORTIONMENT IN THE ST. LOUIS MODELING STUDY..............111
FIGURE 6-6. AVERAGE CONTRIBUTION FROM EACH SOURCE REGION TO ST. LOUIS 8-HOUR OZONE ≥ 85 PPB FOR EPISODE 3........................................................................................................................................................112
FIGURE 6-7: AVERAGE CONTRIBUTION FROM EACH SOURCE REGION TO ST. LOUIS 8-HOUR OZONE ≥ 85 PPB FOR EPISODE 3. DARK RED PORTION OF BAR REPRESENTS CONTRIBUTION OZONE FORMED UNDER VOC-LIMITED CONDITIONS, AND LIGHT BLUE PORTION OF BAR REPRESENTS THE CONTRIBUTION FROM OZONE FORMED UNDER NOX-LIMITED CONDITIONS..............................................................................................................................112
FIGURE 6-8. AVERAGE CONTRIBUTION FROM EACH SOURCE REGION TO ST. LOUIS 8-HOUR OZONE ≥ 85 PPB FOR AUGUST 2........................................................................................................................................................113
FIGURE 6-9. AVERAGE CONTRIBUTION FROM EACH SOURCE REGION TO ST. LOUIS 8-HOUR OZONE ≥ 85 PPB FOR AUGUST 2. DARK RED PORTION OF BAR REPRESENTS CONTRIBUTION OZONE FORMED UNDER VOC LIMITED CONDITIONS, AND LIGHT BLUE PORTION OF BAR REPRESENTS THE CONTRIBUTION FROM OZONE FORMED UNDER NOX-LIMITED CONDITIONS..............................................................................................................................114
FIGURE 6-10. AVERAGE CONTRIBUTION FROM EACH SOURCE REGION TO ST. LOUIS 8-HOUR OZONE ≥ 85 PPB FOR AUGUST 3........................................................................................................................................................114
FIGURE 6-11. AVERAGE CONTRIBUTION FROM EACH SOURCE REGION TO ST. LOUIS 8-HOUR OZONE ≥ 85 PPB FOR AUGUST 3. DARK RED PORTION OF BAR REPRESENTS CONTRIBUTION OZONE FORMED UNDER VOC LIMITED CONDITIONS, AND LIGHT BLUE PORTION OF BAR REPRESENTS THE CONTRIBUTION FROM OZONE FORMED UNDER NOX-LIMITED CONDITIONS..............................................................................................................................115
FIGURE 6-12. AVERAGE CONTRIBUTION FROM EACH SOURCE REGION TO ST. LOUIS 8-HOUR OZONE ≥ 85 PPB FOR AUGUST 4........................................................................................................................................................115
FIGURE 6-13. AVERAGE CONTRIBUTION FROM EACH SOURCE REGION TO ST. LOUIS 8-HOUR OZONE ≥ 85 PPB FOR AUGUST 4. DARK RED PORTION OF BAR REPRESENTS CONTRIBUTION OZONE FORMED UNDER VOC LIMITED CONDITIONS, AND LIGHT BLUE PORTION OF BAR REPRESENTS THE CONTRIBUTION FROM OZONE FORMED UNDER NOX-LIMITED CONDITIONS..............................................................................................................................116
FIGURE 6-14. AVERAGE CONTRIBUTION FROM EACH SOURCE REGION TO ST. LOUIS 8-HOUR OZONE ≥ 85 PPB. AUGUST 5......................................................................................................................................................................116
FIGURE 6-15. AVERAGE CONTRIBUTION FROM EACH SOURCE REGION TO ST. LOUIS 8-HOUR OZONE ≥ 85 PPB FOR AUGUST 5. DARK RED PORTION OF BAR REPRESENTS CONTRIBUTION OZONE FORMED UNDER VOC LIMITED CONDITIONS, AND LIGHT BLUE PORTION OF BAR REPRESENTS THE CONTRIBUTION FROM OZONE FORMED UNDER NOX-LIMITED CONDITIONS..............................................................................................................................117
FIGURE 6-16. SOURCE REGIONS FOR OZONE SOURCE APPORTIONMENT USED IN THE 5-STATE STAKEHOLDER STUDY OSAT ANALYSIS..............................................................................................................................................117
FIGURE 6-17. 5 STATE STAKEHOLDER OSAT STUDY ST. LOUIS JUNE-AUGUST AVERAGE CONTRIBUTION TO 8-HOUR vi
OZONE > 85 PPB................................................................................................................................................118
FIGURE 6-18. ST. LOUIS 8-HOUR OZONE DESIGN VALUE AND NUMBER OF OZONE CONDUCIVE DAYS.....................119
FIGURE 6-19. 1990 – 2009 TREND IN ANTHROPOGENIC NOX AND VOC EMISSIONS IN THE ST. LOUIS NONATTAINMENT AREA................................................................................................................................................................121 1
1.0 EXECUTIVE SUMMARY
On April 15, 2004, U.S. EPA designated portions of the St. Louis metropolitan area, including counties in both Missouri and Illinois, as nonattainment for the 8-hour ozone NAAQS. These designations became effective on June 15, 2004. Nine counties in the St. Louis area are designated as “moderate” nonattainment area for this new 8-hour standard (based on 2001-2003 observed ozone data). In Missouri, they are St. Louis City, Franklin, Jefferson, St. Charles, and St. Louis Counties. In Illinois, the nonattainment counties are Jersey, Madison, Monroe, and
St. Clair.
One of the primary goals of the St. Louis 8-hour ozone modeling study was to develop photochemical modeling databases and allied analysis tools necessary to reliably simulate the processes responsible for 8-hour ozone exceedances in the region. This is done to assist the States of Missouri and Illinois in their development of realistic emissions reduction strategies for inclusion in the St. Louis ozone State Implementation Plan (SIP) due by June 2007. The St. Louis modeling study included episodic emissions, meteorological, and ozone simulations using a nested 36/12/4 km grid covering the central U.S. and centered on St. Louis. The modeling effort used SMOKE and supplemental EMS emissions, MM5 meteorological, and the CAMx and CMAQ air quality modeling systems for estimating ozone on the nested 36/12/4 km St. Louis grid during three 8-hour ozone episodes from the summer of 2002.
The 2002 Baseline CAMx and CMAQ modeling databases were evaluated against monitored ozone data from the St. Louis area in order to evaluate the fitness of the databases for use in the modeled attainment test. Initial simulations illustrated that the CMAQ modeling system exhibited a larger under-prediction ozone bias than CAMx. Given this large under-prediction bias, the higher computational efficiently of CAMx over CMAQ and the resource constraints of the study, the MDNR and IEPA elected to proceed with CAMx as the lead model and CMAQ as a corroborative model.
After several iterations of modeling inventories, meteorology, and modeling set-up, the modeling team reached a consensus regarding the appropriate inputs and model for the best and most accurate base case. On most episode days, the model achieved EPA’s model performance evaluation goals for surface layer 8-hour and 1-hour ozone concentrations. Many of the days that did not meet these goals exhibited low ozone concentrations. These days were included in the modeling because they were bounded by two periods of high ozone concentrations or were needed as “ramp-up” days for the study. In general, the 1-hour and 8-hour ozone performance statistics suggest a systematic underestimation of ozone that is related to the over-estimation of ozone suppression by oxides of nitrogen in the St. Louis urban core, and the model’s tendency to delay ozone formation in the St. Louis urban plume relative to observations. However, the St. Louis 2002 baseline model simulation exhibited sufficient skill in meeting most performance goals (especially on key days). Therefore, the modeling team decided that it may be used to project future-year ozone air quality and 8-hour ozone attainment, recognizing the inherent uncertainties in the atmospheric modeling process.
After detailed performance testing of the 2002 basecase simulation, the CAMx modeling system was exercised with a 2009 On-the-Books (OTB) emissions control scenario aimed at assessing the effects of future year emission control strategies on ozone in the St. Louis Nonattainment Area (NAA). The projected 8-hour ozone design values (using observed 2000-2004 5-year baseline 8-hour ozone design values) in the St. Louis NAA for the 2009 OTB emission scenario were all below 85 ppb, thereby demonstrating attainment. However, the projected 2009 design 2
value for one St. Louis NAA monitor (Orchard Farm) was very nearly 82 ppb and therefore, a weight of evidence determination was completed to provide additional confidence in the study results. Note, the CMAQ modeling system never was able to meet the model performance evaluation goals using the final basecase inventory and was discarded from further consideration due to lack of acceptable performance.
Based on the model’s response to sensitivity analyses, the final attainment demonstration, and an Ozone Source Apportionment Technology (OSAT) scenario, elevated ozone concentrations in
St. Louis are responsive to NOx emission control. Upwind and local NOx emission control are beneficial to reduce ozone in the area and necessary to demonstrate attainment in St. Louis.
The weight of evidence analyses lead to a determination that the St. Louis area will be in attainment of the NAAQS by 2010. Every one of the supplemental analyses performed was consistent in predicting attainment for St. Louis; not a single study suggested that the St. Louis area will not reach attainment by 2010. Therefore, the evidence for attainment was overwhelming and conclusive. 3
2.0 INTRODUCTION
2.1 BACKGROUND AND PURPOSE
On April 15, 2004, U.S. EPA designated portions of the St. Louis metropolitan area, including counties in both Missouri and Illinois, as nonattainment for the 8-hour ozone National Ambient Air Quality Standard (NAAQS). These designations became effective on June 15, 2004. Nine counties in the St. Louis area are designated as “moderate” nonattainment for this new 8-hour standard (based on 2001-2003 observed ozone data). In Missouri, they are St. Louis City, Franklin, Jefferson, St. Charles, and St. Louis Counties. In Illinois, the nonattainment counties are Jersey, Madison, Monroe, and St. Clair.
For “moderate” nonattainment areas, U.S. EPA established a deadline of June 15, 2007, for states to develop and adopt SIPs, and June 15, 2010, for areas to attain the 8-hour ozone standard. The June 2007 8-hour ozone SIP must include a demonstration that the St. Louis nonattainment area (NAA) will achieve the 8-hour ozone standard by 2010. An important component of this attainment demonstration is the use of photochemical grid models to project future-year ozone air quality. On April 15, 2004, U.S. EPA issued Phase I of its implementation rule for the 8-hour ozone NAAQS. This rule provides for classification of nonattainment areas for the 8-hour ozone standard, and describes U.S. EPA’s policy regarding revocation of the 1-hour ozone NAAQS, attainment dates, and timing of emissions reductions necessary to demonstrate attainment.
Phase II of the Implementation Rule was released in late 2005 and addressed mandatory control measures, interstate transport, attainment demonstrations, reasonable further progress, conformity, reasonable available control measures, NOx exemptions, and new source review
(70 FR 71612-71705, Nov. 29, 2005).
One of the primary goals of the St. Louis 8-hour ozone and PM2.5 modeling study was to develop photochemical modeling data bases and allied analysis tools necessary to reliably simulate the processes responsible for 8-hour ozone exceedances in the region. This was done to develop realistic emissions reduction strategies for inclusion in the St. Louis ozone SIP due by June 2007. This Technical Support Document (TSD) describes the modeling activities performed by the Missouri Department of Natural Resources (MDNR), Illinois Environmental Protection Agency (IEPA) and the St. Louis Modeling and Data Analysis Workgroup (MDAW) as well as the contractors for the study (ENVIRON/Alpine Geophysics) for the 8-hr ozone attainment demonstration for the St. Louis NAA. The MDAW consists of experienced air quality modelers at four (4) ‘modeling hubs’: MDNR, IEPA, EPA Region VII and Ameren that performed much of the St. Louis ozone modeling, with assistance from ENVIRON/Alpine. Collectively, the MDAW modeling hubs conducted the episodic 8-hour ozone modeling for St. Louis. Both the Missouri Department of Natural Resources, Air Pollution Control Program and the Illinois Environmental Protection Agency, Bureau of Air expressed a strong desire to work cooperatively with affected parties in the development and implementation of reliable, effective and equitable 8-hour ozone control strategies for the St. Louis metropolitan area. Both agencies have maintained the authority and flexibility to promulgate plans and necessary rules, given the dictates of the rulemaking process in each state. 4
2.2 STATE AGENCY ORGANIZATIONS AND WORK GROUPS
The states of Missouri and Illinois determined the committee structure described below that was used to manage the development and evaluation of control strategies, research, modeling, and other activities:
• State Air Agencies: Responsible for providing policy direction and guidance, selecting achievable emissions strategies, and resolving disputes as they arose. The state air agencies met as appropriate to oversee the progress of the effort. The Missouri Air Conservation Commission has final authority to adopt Missouri’s control plan. Similarly, the Illinois Pollution Control Board has the final authority to adopt control requirements in Illinois.
Participants: Air Directors from Missouri DNR and Illinois EPA.
• Modeling and Data Analysis Workgroup (MDAW): Responsible for the planning and management of the technical work necessary to demonstrate attainment, including emissions, meteorological, and photochemical modeling. The Modeling Workgroup contained four (4) modeling hubs (MDNR, IEPA, EPA Region VII and Ameren) that each assumed primary responsibility for the treatment of one meteorological episode for ozone. The Workgroup was also responsible for contractor selection, data analysis, source apportionment, coordination and communication of model results to AQAC, the Control Strategy Development Workgroup, and the state agency air directors. The Modeling and Data Analysis Workgroup met on a regular basis to coordinate the development and performance of technical activities. Meetings were open to stakeholders and representatives from local agencies having the technical expertise to contribute to work activities.
Participants: IEPA, MDNR, U.S. EPA Region VII, U.S. EPA Region V, and East-West Gateway. Local organizations, stakeholders, and academics that were able to contribute technical capabilities or resources were also invited to participate.
• Air Quality Advisory Committee (AQAC): Served as a forum for communication and outreach between local governmental agencies, stakeholders, the Modeling and Data Analysis Workgroup, Control Strategy Development Workgroup, and the state agency air directors. The AQAC met on a regular basis, and was also responsible for identifying emissions control options for evaluation by the Control Strategy Development Workgroup, for developing conformity budgets, and preparing conformity demonstrations that are consistent with the 8-hour ozone SIPs. The Modeling and Data Analysis Workgroup, the Control Strategy Development Workgroup, and, when possible, the state agency air directors, were present at the meetings to report on activities, and to solicit input on control strategy recommendations.
Participants: East West Gateway, MDNR, IEPA, U.S. EPA Regions 5 and 7, St. Louis County, St. Louis City, Federal Highway Administration, Missouri Department of Highway and Transportation (MDHT), Illinois Department of Transportation (IDOT), Federal Highway Administration (FHWA), Environmental Groups, Industry, and other local representatives.
5
• Control Strategy Development Workgroup (CSDW): Responsible for the identification and technical evaluation of control strategies needed to demonstrate attainment of the 8-hour ozone standards, and meet other regulatory requirements (e.g. contingency measure identification). The Control Strategy Development Workgroup was also responsible for coordination and communication of strategies and technical information to AQAC, the Modeling and Data Analysis Workgroup, and the State Agency Air Directors. The Control Strategy Development Workgroup met on a regular basis to coordinate the performance of technical activities. Meetings were open to stakeholders and representatives from local agencies having the technical expertise to contribute to work activities.
Participants: IEPA, MDNR, U.S. EPA Region VII, U.S. EPA Region V, East-West Gateway. Local organizations, stakeholders, and academics that were able to contribute technical capabilities or resources were also invited to participate.
2.3 OVERVIEW OF APPROACH
2.3.1 Modeling Protocol
The St. Louis 8-Hour Ozone Study meteorological, emissions and air quality modeling followed the procedures outlined in the Modeling Protocol (ENVIRON and Alpine Geophysics, 2005). The Modeling Protocol describes the overall modeling activities performed by all the participants in the project. Its main function was to serve as a means for planning and communicating how the modeled attainment demonstration would be performed. The protocol guided the technical details of the modeling study and provided a formal framework within which the scientific assumptions, operational details, commitments and expectations of the various participants were communicated explicitly. The modeling protocol also set forth means for resolution of potential differences of technical and policy opinion to be worked out openly and within prescribed time and budget constraints.
2.3.2 Model Selection
The model selection methodology for the St. Louis ozone modeling rigorously adhered to EPA’s guidance for regulatory modeling in support of ozone and fine particulate attainment demonstrations (EPA, 1991; 1999; 2005; 2006). Unlike previous ozone modeling guidance, the agency now recommends that models be selected for SIP studies on a ‘case-by-case’ basis with appropriate consideration being given to the candidate model’s:
> Technical formulation, capabilities and features,
> Pertinent peer-review and performance evaluation history,
> Public availability, and
> Demonstrated success in similar regulatory applications.
Detailed discussion of the selection process for each model component may be found in the Modeling Protocol. Here follows a brief summary of each of the model components and a description of how it fits into the St. Louis 8-hour ozone modeling.
• MM5: The Mesoscale Meteorological Model (MM5) is a nonhydrostatic, prognostic meteorological model routinely used for urban- and regional-scale photochemical, fine particulate, and regional haze regulatory modeling studies (Dudhia, 1993; Seaman, 2000). Developed in the 1970s, the MM5 modeling system maintains its status as a state-of-the-6
science model through enhancements provided by a broad user community worldwide (Stauffer and Seaman, 1990; Xiu and Pleim, 2000; Byun et al., 2005a,b). MM5 is used nearly exclusively for regulatory air quality applications in the U.S. In recent years, the modeling system has been successfully applied in continental-scale annual simulations.
• SMOKE: The Sparse Matrix Operator Kernel Emissions (SMOKE) modeling system is an emissions modeling system that generates hourly, gridded, speciated emission inputs of mobile, nonroad, area, point, fire, and biogenic emission sources for photochemical grid models (Coats, 1995; Houyoux et al., 2000). As with most ‘emissions models’, SMOKE is principally an emission processing system and not a true emissions modeling system in which emissions estimates are simulated from ‘first principles’. This means that, with the exception of mobile and biogenic sources, its purpose is to provide an efficient, modern tool for converting emissions inventory data into the formatted emission files required by an air quality simulation model. For mobile sources, SMOKE actually simulates emissions rates based on input mobile-source activity data, emission factors and outputs from transportation travel-demand models.
• EMS: The Emissions Modeling System-2003 (EMS-2003) is an emissions processing and modeling system with core functionality---spatial allocation, temporal allocation, and speciation of emissions---effectively the same as the SMOKE modeling system. Emissions inventory data representing point, area, fire, nonroad, mobile, and biogenic emissions are processed to produce inputs that are properly formatted for acceptance by an air quality simulation model. Only mobile and biogenic emissions are obtained from ‘fundamental’ calculations or ‘first principles’, the remaining emissions categories are input as pre-determined estimates that are ‘reduced’ through processing to the required level of resolution. The software was primarily used to create Electrical Generating Unit (EGU) emission estimates, and to provide supporting quality assurance/quality control checks.
• CAMx: The Comprehensive Air Quality Model with Extensions (CAMx) modeling system is a state-of-science ‘One-Atmosphere’ photochemical grid model capable of addressing ozone, particulate matter (PM), visibility and acid deposition at regional scale for periods up to one year (ENVIRON, 2006). CAMx is a publicly available open-source computer modeling system for the integrated assessment of gaseous and particulate air pollution. Built on today’s understanding that air quality issues are complex, interrelated, and reach beyond the urban scale, CAMx is designed to (a) simulate air quality over many geographic scales, (b) treat a wide variety of inert and chemically active pollutants including ozone, inorganic and organic PM2.5 and PM10 and mercury and toxics, (c) provide source-receptor, sensitivity, and process analyses and (d) be computationally efficient and easy to use. The U.S. EPA has approved the use of CAMx for numerous ozone and PM State Implementation Plans throughout the U.S. and has used this model to evaluate regional mitigation strategies.
• CMAQ: EPA’s Models-3/Community Multiscale Air Quality (CMAQ) modeling system is also ‘One-Atmosphere’ photochemical grid model capable of addressing ozone, particulate matter (PM), visibility and acid deposition at regional scale for periods up to one year (Byun and Ching, 1999). The CMAQ modeling system was designed to approach air quality as a whole by including state-of-the-science capabilities for modeling multiple air quality issues, including tropospheric ozone, fine particles, toxics, acid deposition, and visibility degradation. CMAQ was also designed to have multi-scale capabilities so that separate models were not needed for urban and regional scale air quality modeling. The CMAQ modeling system contains three types of modeling components: (a) a meteorological module 7
for the description of atmospheric states and motions, (b) an emission models for man-made and natural emissions that are injected into the atmosphere, and (c) a chemistry-transport modeling system for simulation of the chemical transformation and fate.
The MM5 meteorological model was applied to generate the meteorological fields used with the SMOKE emissions and CMAQ/CAMx air quality models. The MM5 meteorological modeling was conducted in a similar fashion as was done for the Central Regional Air Planning Association (CENRAP) visibility modeling (Johnson, 2004). These simulations used the Pleim-Xiu PBL scheme (Xiu and Pleim, 2000), the Kain-Fritsch II cumulus parameterization for the 36 and 12 km domains (Kain and Fritsch, 1993), the RRTM radiation scheme (Mlawer et al. 1997) and the Reisner I mixed phase moist physics parameterization (Reisner et al., 1998). Model-ready emissions inputs were generated by processing emissions inventories developed by CENRAP (Strait, Roe and Vuckovich, 2004; Reid et al., 2004a,b) and the Midwest RPO (MRPO) using the SMOKE emissions modeling system. In the first phase of the St. Louis Modeling, EPA’s Models-3 Community Multiscale Air Quality (CMAQ; Byun and Ching, 1999) modeling system and the Comprehensive Air-quality Model with extensions (CAMx; ENVIRON, 2006) air quality models were both applied. The application of the CMAQ/CAMx air quality models benefited from the extensive testing and evaluation conducted by CENRAP (Morris et al., 2005c), Visibility Improvement State and Tribal Association of the Southeast (VISTAS) (ENVIRON et al., 2003b,c,d; Morris et al., 2004a,b,c; 2005a,b) and MRPO (Baker, 2004). The CMAQ/CAMx model application followed the relevant guidance documents (EPA, 1991; 1999; 2001; 2003a, b; 2005; 2006). Note: As with all long-term modeling projects conducted for St. Louis, there was a consistent effort to use the most up-to-date scientific algorithms in each modeling system. For example, several different versions of CAMx were used in the base-case evaluation process (v4.11, v4.20, and v4.30). It was the intention of the modeling hubs to use the most technically defensible tools for the model performance and attainment demonstration exercises.
2.3.3 Modeling Domains
The 36 km continental U.S. horizontal domain for each of the models was identical to those used by Western Regional Air Partnership (WRAP), CENRAP, and VISTAS Regional Planning Organizations (RPOs). The CMAQ/CAMx air quality modeling domain is nested within the MM5 domain. The selection of the MM5 domain is described by Johnson (2004). Figure 2-1 displays the nested 36/12/4 km domains established by the MDNR for photochemical modeling and emissions modeling of the three summer 2002 8-hour ozone episodes.
During the course of the photochemical analyses, the modeling team decided to utilize a smaller 4km grid than the one described above for sensitivity testing purposes. This was done to maximize the amount of modeling work that could be accomplished given the computing resources available during the analyses. This smaller 4km grid was more narrowly focused around the St. Louis area and provided much shorter run times in CAMx to allow for more efficient processing of the sensitivity analyses. When evaluating the final model performance and proceeding with the future year analyses, the larger of the 4km grids was used to minimize affects from the 12km to 4km grid transition in St. Louis. In order to use the MM5 outputs for the smaller 4km domain, MM5CAMx had to be re-run for each episode and the SMOKE emission output had to be “windowed” out to allow for input to CAMx. Figure 2-2 displays the revised domains utilized in the study.
Both MM5 and CMAQ/CAMx employed the RPO unified grid definition for the 36 km 8
continental domain for the ozone modeling. The RPO unified grid consists of a Lambert-Conformal map projection using the projection parameters listed in Table 2-1.
The MM5 36 km grid includes 164 cells in the east-west direction and by 128 cells in the north-south direction. The CMAQ/CAMx 36 km grid includes 148 cells in the east-west direction and 112 cells in the north-south direction. Because the MM5 model is also nested within the Eta model, there is a possibility of boundary effects near the MM5 boundary that occur as the Eta meteorological variables are simulated by MM5 and are forced into dynamic balance with MM5’s meteorological fields. Thus, a larger MM5 domain was selected to provide a buffer of 6 grid cells around each boundary of the CMAQ/CAMx 36 km domain. This was designed to eliminate any errors in the meteorology from boundary effects in the MM5 simulation at the interface of the MM5 and Eta models. The buffer region used here complies with the EPA suggestion of a buffer of at least 3-6 grid cells at each boundary (EPA, 2006).
Table 2-2 lists the number of rows and columns and the definition of the X and Y origin (i.e., the southwest corner) for the 36/12/4 km domains used by MM5, SMOKE and CMAQ/CAMx. In Table 2-2, “Dot” refers to the grid mesh defined at the vertices of the grid cells while “Cross” refers to the grid mesh defined by the grid cell centers. Thus, the dimension of the dot mesh is equal to the dimension of the cross mesh plus one.
9
-2000-1500-1000-50005001000150020002500-2000-1500-1000-500050010001500Coarse: -2736.0,-2088.0, 36x36, 148x112Fine1 : -660.0,-1344.0, 12x12, 203x200Fine2 : 68.0, -580.0, 04x04, 254x218
-2000-1500-1000-50005001000150020002500-2000-1500-1000-50005001000150012 km Domain12 km DomainSW corner: -1092, -1524NX, NY: 245, 22436 km Domain36 km DomainSW corner: -2736, -2088NX, NY: 148, 112
4 km DomainSW corner: 68, -580NX, NY: 254, 2184 km Domain
Figure 2-1. Nested 36/12/4 km St. Louis modeling domains for photochemical (top) and emissions (bottom) modeling.
10
Figure 2-2. Revised Nested 36/12/4 km St. Louis modeling domains for photochemical modeling.11
Table 2-1. RPO Unified Grid Projection Definition
Parameter
Value
Projection
Lambert-Conformal
Alpha
33 degrees
Beta
45 degrees
x center
-97 degrees
y center
40 degrees
Table 2-2. Grid Definitions for MM5, SMOKE/EMS, and CMAQ/CAMx
Model
Columns
dot(cross)
Rows
dot(cross)
Xorigin
(meters)
Yorigin
(meters)
2.3.4 Vertical Structure of Modeling Domain
The CMAQ/ CAMx model vertical structure is primarily defined by the vertical grid used in the MM5 modeling. The MM5 model employed a terrain-following coordinate system defined by pressure, and had 34 vertical layers that extend from the surface upward to 100 mb. CAMx and CMAQ were applied with exactly the same vertical layer structure. A layer averaging scheme was adopted for CMAQ/CAMx to reduce the computational burden of the CMAQ and CAMx simulations. The effects of layer averaging were evaluated by WRAP and VISTAS and were found to have a relatively minor effect on the model performance metrics when both the 34 layer (no layer averaging) and 19 layer (layers averaged) CMAQ model simulations were compared to ambient monitoring data (Morris et al., 2004a). For the St. Louis ozone modeling, 16 vertical layers were used. Table 2-3 details the mapping from the 34 vertical layers used by MM5 to the 16 vertical layers used by CMAQ and CAMx in the St. Louis study.
MM5
36 km grid
12 km grid
4 km grid
165 (164)
265 (264)
271 (270)
129 (128)
241 (240)
235 (234)
-2952000
-1188000
24000
-2304000
-1620000
-600000
SMOKE/EMS
36 km grid
12 km grid
4 km grid
(148)
(245)
(254)
(112)
(224)
(218)
-2736000
-1092000
68000
-2088000
-1524000
-580000
CMAQ/CAMx
36 km grid
12 km grid
4 km grid
small 4km grid
(148)
(203)
(254)
(92)
(112)
(200)
(218)
(92)
-2736000
-660000
68000
320000
-2088000
-1344000
-580000
-32800012
Table 2-3. Vertical Layer Definition for MM5 Simulations (left-most columns) and Approach for Reducing CMAQ/CAMx Layers by Collapsing Multiple MM5 Layers (right columns)
MM5
CMAQ/CAMx
Layer
Sigma
Pres (mb)
Height (m)
Depth (m)
Layer
Pres (mb)
Height (m)
Depth (m)
34 (top)
0.000
100
18123
2856
16
100
18123
7987
33
0.050
145
15267
2097
32
0.100
190
13170
1659
31
0.150
235
11510
1374
30
0.200
280
10136
1173
15
280
10136
3106
39
0.250
325
8963
1024
28
0.300
370
7938
909
27
0.350
415
7030
817
14
415
7030
2866
26
0.400
460
6213
742
25
0.450
505
5471
680
24
0.500
550
4791
627
23
0.550
595
4163
582
13
595
4163
1635
22
0.600
640
3581
543
21
0.650
685
3038
509
20
0.700
730
2528
386
12
730
2528
664
19
0.740
766
2142
278
18
0.770
793
1864
269
11
793
1864
443
17
0.800
820
1596
174
16
0.820
838
1421
171
10
838
1421
338
15
0.840
856
1251
167
14
0.860
874
1083
164
9
874
1083
324
13
0.880
892
920
161
12
0.900
910
759
79
8
910
759
158
11
0.910
919
680
78
10
0.920
928
601
78
7
928
601
155
9
0.930
937
524
77
8
0.940
946
447
76
6
946
447
152
7
0.950
955
371
75
6
0.960
964
295
75
5
964
295
149
5
0.970
973
220
74
4
0.980
982
146
37
4
982
146
37
3
0.985
987
109
37
3
987
109
37
2
0.990
991
73
36
2
991
73
36
1
0.995
996
36
36
1
996
36
36
0 (ground)
1.000
1000
0
0
0
0
0
0
13
2.3.4.1 Air Quality Data
Data from ambient monitoring networks for both gas and aerosol species were used in the model performance evaluation. In the model performance evaluation presented in this TSD, the focus is on the evaluation of modeled surface layer ozone within the St. Louis NAA. Figure 2-3 displays the locations of monitoring sites in the St. Louis area including monitors outside the current NAA.
Figure 2-3. Ozone monitoring sites in the St. Louis Area
14
2.3.4.2 Ozone Column Data
Additional data used in the air quality modeling include the Total Ozone Mapping Spectrometer (TOMS) data. TOMS data are available for 24-hour average time periods, and are obtained from http://toms.gsfc.nasa.gov/eptoms/ep.html. The TOMS data are used in the CMAQ (JPROC) and CAMx (TUV) radiation models to calculate photolysis rates. The TOMS data were completely missing for the period of 3 August through 12 August, 2002, as well as on 10 June, 2002. In addition, 2 August and 18-19 November, 2002 had partially missing data.
The CAMx TUV processor allows for the use of monthly average data, so that option was used and the missing data ignored. The CMAQ JPROC processor does not allow for the use of monthly average data so the data from 1 August was used for 2 August through 7 August, and the data from 13 August was used for 8 August through 12 August. Data from 9 June was used for 10 June. Data from 17 November was used for 18 November, and data from 20 November was used for 19 November.
2.3.4.3 Initial and Boundary Conditions Data
For the episodic ozone simulations, the MDAW modeling hubs utilized a nominal 48-72 hour spin up period to initialize the simulations. The CENRAP annual CMAQ results were used for Initial Concentrations (ICs) for the CMAQ and CAMx episodic simulations. The CMAQ and CAMx boundary conditions were based on results from a 2002 GEOS-CHEM global climate model simulation (Jacob, 1999). The 2002 GEOS-CHEM model output has been processed to define day-specific high time resolved (i.e., 3-hourly) CMAQ and CAMx boundary conditions for 2002.
2.3.5 Episode Selection
The methodology for episode selection for the St. Louis 8-hour ozone modeling adhered to the criteria set forth by EPA in their guidance document for regulatory modeling in support of ozone, PM2.5 and regional haze analyses (EPA, 1991; 1999; 2005; 2006). In general, EPA recommends the following main criteria for selecting time periods to model for 8-hour ozone:
• The time periods selected should represent a variety of meteorological conditions. 8-hour ozone should exceed 85 ppb at multiple monitors.
• Model episodes with observed ozone close to the area design value (~90 ppb)
• Model time periods with robust observational databases
• Model a sufficient number of days to ensure a robust Relative Response Factor (RRF) for each monitor (minimum 5 days, 10-16 days preferred).
EPA recognized that some of these criteria may be in conflict with each other and acknowledged that some secondary criteria may help resolve the issue. Some additional considerations for selection are: choosing time periods that have been previously modeled successfully in other demonstrations, selecting time periods drawn from the period upon which the baseline design value was calculated (i.e., 2000-2004), choosing episodes as close to the NAAQS on as many days are possible, and including weekend days if the area commonly has violations then.
The St. Louis Ozone Technical Group accounted for these criteria when choosing the appropriate 15
time periods to model. The group also recognized other key state-specific considerations in the selection process, such as constraints due to limited regulatory timeframe, human resources, and computing capacity.
In accordance with the guidance, episode selection focused on the key years surrounding the calculation of the area’s baseline design value. Particularly, the years 2001-2003 were examined in order to determine appropriate representative time periods to model. Below is a brief summary of the elevated ozone time period specifics for 2001 – 2003 as it recorded by the 17 urban area monitors in and around St. Louis.
2001 Ozone Season: 14 8-hour ozone exceedance days, 12 single-day events, one 2-day event, weak multiple-monitor days (no days > 4 monitors exceeding).
2002 Ozone Season: 32 exceedance days, 5 single days, but significant multi-day events, wide breadth of meteorology, many multiple-monitor days (17 days > 4 monitors exceeding).
2003 Ozone Season: 11 exceedance days, four single-day events, two 2-day events, one 3-day event ( 6 days > 4 monitors exceeding)
In order to affirm that the episode selection represented a variety of meteorological conditions associated with high ozone events, MDNR’s July 2003 Technical Support Document for Determination of Nonattainment Boundaries in Missouri for the 8-hr Ozone National Ambient Air Quality Standard was reviewed (Bennett, Froning, Mefrakis 2003). A summary of the different meteorological regimes identified for St. Louis is below.
Meteorological Regime #1
Synoptic features
Regime #1 occurs as a high pressure area develops over the Ohio River Valley forcing any lingering frontal boundaries to be pushed out of the region. As the day wears on, the center of the high pressure system migrates to the northeast and establishes itself over the New England states. Frontal boundaries typically remain to the northwest with their area of influence limited to the High Plains.
Surface features
The presence of the high pressure center over the Ohio River valley during the morning hours often leads to calm, potentially haze conditions. As the high pressure center migrates eastward, the surface wind speeds increase slightly, but remain below ten knots. In most instances the predominant wind direction is from the southeastern quadrant. Slight variations in the position of the high pressure center determine if the winds are from the east-southeast, southeast, or south-southeast.
Meteorological Regime #2
Synoptic Features
Regime #2 occurs as a high pressure area over the New England states retreats southward over the Mid-Atlantic states. The frontal boundary positioned over the High Plains in Regime #1 continues to move toward the Midwest as the afternoon high pressure center drifts off the eastern seaboard. Depending on the strength of the area of high pressure, the frontal boundary may continue its southeasterly path, or it may become stationary along the Missouri/Iowa border
Surface features 16
The surface conditions occuring during the 2nd regime are not as consistent as those associated with the first. The largest contributor to this variation in wind direction is often due to the proximity of the frontal boundary to the St. Louis metropolitan area. The predominant wind direction is often from the southwest with wind speeds less than ten knots. Again, a.m. calms are common. As frontal boundaries approach, the winds may shift to the southeast or north. With few exceptions, the winds remain at speed less than ten knots.
Meteorological Regime #3
Synoptic Features
Regime #3 occurs as the stationary front positioned along the Missouri/Iowa border, as seen in Regime #2, becomes mobile and continues its southerly advance though the State of Missouri. As the front approaches the St. Louis and Kansas City regions, early morning precursor emissions and/or ozone are forced southward causing higher concentrations of ozone to the south of each metropolitan area. The timing and intensity of the frontal boundary determines which sites report elevated concentrations.
Surface Features
The surface conditions occurring during this regime do not follow a consistent pattern due to the proximity of the frontal boundary to the St. Louis metropolitan area. Hazy conditions are often reported prior to the passage of a cold front with calm, variable winds common. As frontal boundaries approach, the winds may shift to the southeast or north. With few exceptions, the winds remain at speeds less than ten knots.
Meteorological Regime #4
Synoptic Features
Regime #4 occurs as a high pressure area develops over the State of Iowa and migrates southward over Missouri. Further tracking of the high pressure center indicates that it will continue to move eastward over Illinois and Indiana. No predominate frontal systems are present within the region.
Surface Features
The presence of the high pressure center over the midsection of the United States during the morning hours often leads to calm, potentially hazy conditions. As the high pressure center migrates eastward into Illinois and Indiana, the surface wind speeds increase slightly, but remain below ten knots. In most instances the predominant wind direction is from the northeast quadrant. Slight variations in the position of the high pressure center determine the pattern of the surface flow.
Meteorological Regime #5
Synoptic Features
Regime #5 occurs less frequently than previous regimes as a high pressure areas develop over Canada and the Northern New England states. A frontal boundary will approach and pass through the State of Missouri and will remain to the east over the Ohio River Valley as a second boundary approaches from the West.
Surface Features
The presence of multiple frontal boundaries in the region typically leads to little or no formation of ozone. However, on the days with reported ozone exceedances, the frontal systems were in close proximity to one another and often trapped pollutants between their boundaries. With little or no precipitation 17
reported and sunny skies, the ozone precursors had little chance for dilution and were available for ozone production.
Meteorological Regime #6
Synoptic Features
Regime #6 resulted in a high pressure buildup over West Virginia as a stationary front remained in an east/west configuration along the I-70 corridor. The frontal boundary advanced and retreated across the immediate area causing ozone episodes with significant differences in ozone maximums from day to day depending on what air mass was over each metropolitan area.
Surface Features
The presence of the frontal boundary to the north or the south of the city caused the wind speeds and directions to vary from day to day depending upon the air mass over the region.
Meteorological Regime #7
Synoptic Features
Regime #7 occurs when an area of strong high pressure develops over the Eastern United States. Depending on the strength of the high pressure region, centers may develop over Missouri and Illinois. The strongest subsidence regions remain over the East Coast. The St. Louis region was the only area within the State of Missouri that reported ozone exceedances during this meteorological regime.
Surface Features
The presence of the high pressure centers throughout the region leads to calm conditions during the morning hours allowing precursor emissions to remain in the urban core. As the high pressure centers migrate and/or weaken as the day continues, the ozone plume will begin to migrate in the direction of the surface flow. The wind directions vary under this regime and are extremely dependent upon the development and position of individual high pressure centers.
The meteorological conditions associated with Regimes #2, #4, and #7 resulted in the most severe 8-hour ozone concentrations within the St. Louis area. Each of these meteorological regimes resulted in days exceeding 110 parts per billion based upon the 8-hour average. Regime’s #1 and #3 were the next most severe, with concentrations exceeding 100 parts per billion at several ambient air quality sites. Both regimes #6 and #7 remained below 100 parts per billion.
In addition to reviewing the severity of ozone concentrations under certain meteorological conditions, the likelihood that ozone concentrations in excess of the 8-hour ozone standard would occur was also evaluated. Regime’s #1 and #2 occurred most frequently and often were associated with the same episode. Regimes #3, #4, and #7 also occurred on a regular basis, with Regime #3 ending ozone episodes with the passage of a frontal system that ushered in new, cleaner air masses.
Balancing the existing capabilities of the technical group and the necessity to develop a robust modeling demonstration, the 2002 time period offered the best candidate ozone events. By limiting the modeling to a particular year, it allowed for a significantly streamlined database acquisition process without compromising episode quality or quantity per guidance recommendations. Listed in Table 2-4 are the 8-hour exceedance days from the summer 2002 ozone season in the St. Louis area. Highlighted in shaded, bolded text are the exceedance days 18
selected to model.
Table 2-4. 2002 8-Hour Ozone Exceedance Days in the St. Louis Area
06/08/2002
07/16/2002
06/19/2002
07/20/2002
06/20/2002
07/25/2002
06/21/2002
07/30/2002
06/22/2002
08/01/2002
06/23/2002
08/02/2002
07/02/2002
08/03/2002
07/03/2002
08/04/2002
07/04/2002
08/09/2002
07/05/2002
08/10/2002
07/07/2002
09/01/2002
07/08/2002
09/06/2002
07/09/2002
09/07/2002
07/13/2002
09/08/2002
07/14/2002
09/09/2002
07/15/2002
09/14/2002
Note: Shaded, bolded text indicates days chosen for modeling
For this study, 21 of the 32 exceedance days (and necessary ramp-up days) in 2002 were modeled. Of the 5 multi-day episodes of three or more consecutive days, four of them are captured in the modeling. The meteorology during the September 6-9, 2002 timeframe bears a striking resemblance to the weather conditions prevalent during the high ozone stretch in early July, thus it was removed from consideration to avoid duplication or overweighting from a particular meteorological regime.
The meteorological regimes identified previously are summarized as follows:
Regime #1 June 19-21, August 9, September 1
Regime #2 June 22-23, July 8-9, July 25, August 1-2, August 10, September 14
Regime #3 July 5, July 20
Regime #4 July 13-16
Regime #5 July 30
Regime #6 August 2-4
Regime #7 July 2-4, September 7-9
All the meteorological regimes are contained in the episodes selected with the most frequent regimes represented by more than one episode. The selection of these episodes assures that we have a variety of meteorological conditions that are conducive to elevated 8-hour ozone formation in the St. Louis area.
Below is a brief description of the air quality and meteorology from the episode days selected in the attainment demonstration modeling. See Appendix A for more detailed air quality and meteorological information for the chosen modeled days.
19
6/19/02-6/23/02: A high-pressure center at the 500-millibar level persisted over the State of Missouri for the entirety of the review period with surface high pressure evident over the Ohio River Valley and New England states at the onset of the episode. The presence of the surface high over the New England states resulted in calm, hazy conditions across the region during the morning hours, with light southeasterly flow apparent by the mid-afternoon hours on June 19 and 20, 2002. Elevated ozone concentrations were reported at West Alton (93 ppb – 8 hour), Jerseyville (91 ppb), Orchard Farm (86 ppb) and Alton (87 ppb) on June 19th. The highest 8-hour concentrations on June 20th were again north of the downtown area and the maximum was monitored at Jerseyville (100 ppb). The back trajectory analysis indicates that on June 19, 2002 little transport occurred from areas outside the non-attainment area. As the 500-millibar high pressure center intensified and the air mass became more stagnant by June 21, 2002, the ozone concentrations across the region increased with a southerly push still evident as maximum ozone concentrations of 110 ppb and 100 respectively were reported at Jerseyville and West Alton. The back trajectory analysis indicates that transport from the Gulf Coast states is occurring and corresponds to the southerly push that was noted on the meteorological charts. On June 22, 2002 a weak frontal boundary was located over the Great Lakes region with high pressure continuing to dominate the East Coast with the center over the mid-Atlantic states. Early morning conditions continued to be calm, with haze reported at several National Weather Service sites. Higher ozone concentrations continued on June 22nd continued to occur with maximum concentrations reported at Orchard Farm (111 ppb), Jerseyville (109 ppb), and West Alton (111 ppb). Transport from the Tennessee and Ohio River Valley’s is evident based upon the back trajectory analysis that was conducted for the 22nd and 23rd of June. Widespread rainfall associated with a low pressure center over Florida and Georgia brought the June episode to an end.
7/2/02-7/16/02: Long-term elevated ozone period, essentially 3 different episodes with one to two days break for air-mass change. July 2- July 5 was dominated by low wind speeds, variable wind directions in St. Louis, moderate stagnation, and upper-level winds from south to southeast. Weak high pressure drifted from KY to northern AL during the 4-day stretch. Daily peak 8-hour ozone ranged from 92 ppb on July 2 (1 site exceeding) to 109 ppb on July 5 (6 sites exceeding). A front in the area on July 6 marked the transition to the 2nd elevated period. A fast moving high pressure migrated from Michigan to South Carolina over the 3 days of July 7-July 9. For the period, back trajectories indicated transport of ozone into the St. Louis area from the east/northeast. Winds backed to southwesterly by July 9 as a cold front approached from the northwest and monitors on the Illinois side were high. On July 7, transport from the northeast was evident southwesterly winds. Daily 8-hour ozone peaks were 93 ppb on July 7 (4 sites exceeding), 119 ppb on July 8 (13 sites) , and 90 ppb on July 9 (1 site over). From July 10-12, a low pressure system and cold front cleaned out the air mass, but a new high pressure system migrating out of Canada moved slowly from northern MN on July 13 to southern Alabama on July 16, resulted in a significant ozone episode in St. Louis. Transport indicated incoming ozone from the IN, OH, MI area. Daily peaks ranged from 89 ppb (3 sites exceeding) of July 13 to 114 ppb (145 sites exceeding) on July 15. Though high pressure was still in place in the deep-south, daily localized rainfall in St. Louis air-shed on July 17th and 18th effectively suppressed ozone formation.
7/30/02-8/4/02: A 500 mb high pressure centered itself over the central plains during this period, with a weak surface high residing in the southeast early on, then another surface high pressure area strengthening from MI to VA as the episode progressed. A weak frontal boundary near St. Louis during the August 2-4 timeframe likely muted the intensity of the exceedances, but did result in highly variable wind directions day to day. Daily 8-hour ozone peaks were 93 ppb on 20
July 30 (1 site over), then a 1 day break on July 31, followed by 96 ppb on August 1 (3 sites), 85 ppb on August 2 (1 site), 99 ppb on August 3 (7 sites), and 98 ppb on August 4 (6 sites exceeding). Backward trajectories indicate long term transport from the southwest until August 3rd, then an air-mass origination from the Ohio River Valley on August 4th. Ozone levels remained high but just below 85 ppb on August 5th. A strong cold frontal passage ended the episode on August 6.
The technical group determined that modeling this selection of episodes provides a full range of the typical or historical ozone conducive meteorology in St. Louis, including modeling full synoptic cycles of some of the longer term elevated ozone event from the summer of 2002. The quantity of days modeled also provided the necessary robustness (for each monitor) to draw conclusions from the EPA recommended RRF test with reasonable confidence.
2.3.6 Conceptual Model
The conceptual model is designed to provide an explanation of events that transpired to cause high ozone during these modeling time periods. Typically, it includes a discussion of meteorology, emissions, and transported ozone and precursors into the metropolitan area. As discussed previously in Episode Selection, there are several types of synoptic weather patterns associate with high ozone in St. Louis. Most of the local surface weather patterns are calm or light winds in the morning hours and continued calm or a “push” to the suburban areas in the afternoon resulting in high 8-hour concentrations.
The following is a description of the conceptual model for 8-hour ozone exceedance days within all three ozone episodes evaluated in the attainment demonstration analyses:
June 19-23, 2002
As discussed previously in Section 2.3.5, there was a high pressure center at the 500-millibar level that remained over the state of Missouri throughout this episode. This in conjunction with surface high pressure over the Ohio River Valley and Great Lakes Region translate to Meteorological Regime #1 in St. Louis for June 19-21.
On June 19th, the surface winds were light and predominantly from the south-southeast and south over the course of the day. The 72-hour back trajectory for June 19th entering St. Louis demonstrates low-wind speed conditions and limited transport of ozone and precursors from eastern Tennessee and Kentucky along with southern Illinois and Missouri. As expected the highest 8-hour ozone concentrations were found north of the metropolitan area at Jerseyville (91 ppb) along with exceedances at Orchard Farm, West Alton, and Alton. In addition, there was an exceedance at the Houston, IL (86 ppb) monitor to the south and east of the metropolitan area. These high concentrations are likely due to large proximate NOx source impacts on the monitor and/or near-field transport from the south or south-southeast. The other upwind monitors for this day (Bonne Terre, Arnold) were 70-75 ppb (max 8-hour average).
Surface wind conditions on June 20th were again from the south and south-southeast. The 72-hour back trajectories indicate potential transport from the south and south-southeast (eastern Tennessee, Kentucky, Mississippi). Again, the highest 8-hour concentrations were downwind of the urban core at Jerseyville (100 ppb) with exceedances at West Alton, Orchard Farm, Alton, Breckenridge Hills, and Nilwood. The wind directions were consistent and the highest 1-hour value was monitored at Jerseyville (115 ppb) at 5:00 PM with the highest 1-hour concentrations at West Alton, Orchard Farm, and Alton earlier (2:00-3:00 PM). This is indicative of persistent 21
flow within the region and transport of the morning urban emission plume into the suburban and rural areas in the early/late afternoon hours. Maximum 8-hour upwind monitor concentrations on this day were between 58 and 69 ppb.
The 500-millibar high pressure center over Missouri intensified and the air mass became more stagnant on June 21st. As expected, the early morning surface winds slowed down further on this day and were variable, but remained from the south as the day progressed. In addition, a weak front was located over Iowa and the Great Lakes. The back trajectories again illustrated transport from the south and south-east (more easterly on this day). The highest 1-hour concentrations were downwind of the area at Jerseyville (119 ppb at 4:00 PM), but the sites closer to the downtown experienced 1-hour concentrations over 100 ppb at 11:00 AM and continued into the early afternoon. There were many concentrations that exceeded the 8-hour standard on this day. The maximum was observed at Jerseyville 110 ppb, with West Alton (100 ppb), Orchard Farm and Breckenridge Hills (96 ppb), Ferguson (95 ppb), Alton (94 ppb), Sunset Hills (90 ppb), Nilwood (89 ppb), Maryville (88 ppb), Arnold and Margaretta (85 ppb) also experiencing concentrations over the standard. Maximum upwind concentrations were 75-80 ppb. The more stagnant air mass around the metropolitan area contributed to more sites and overall higher concentrations on this day than previous days in this episode.
June 22nd and 23rd were characterized in Section 2.3.5 as being from meteorological regime #2 with surface high pressure centered over the Mid-Atlantic states with the frontal boundary remaining over the Great Lakes region. June 22nd was predominated by hazy conditions over much of the Midwest and East. The surface winds were very light and variable during the morning hours with calms reported. Winds were more easterly than previous days, but still remained from the southeast throughout the afternoon hours. The back trajectories for June 22nd were more easterly from Kentucky and southern Illinois. All sites within the St. Louis area monitored concentrations over the 8-hour standard with the maximums occurring north and west of the urban core (West Alton and Orchard Farm – 111 ppb). The maximum 1-hour concentration at Orchard Farm was 125 ppb at 2:00 PM. This day was the highest monitored day in the June 2002 episode. This day can be characterized as a high regional event with concentrations over the 8-hour ozone standard in most of the Midwestern United States along with the very light and variable winds in the morning hours contributing allowing a buildup of precursors that were later slowly pushed to the north and west resulting in very high concentrations throughout the entire area. The maximum 8-hour “upwind” concentrations on this day were above 90 ppb.
Surface winds were again light and variable in the morning hours on June 23rd. Late morning and early afternoon winds were similar to June 21st (southerly). Back trajectories illustrated transport from the Ohio River Valley. The maximum 8-hour average concentration was 101 ppb at Orchard Farm. Jerseyville, West Alton, Queeny Park, Arnold, Alton, Sunset Hills, Ferguson, Ladue, Maryville, Edwardsville, Wood River, and Margaretta monitored concentrations over the standard. Two separate 1-hour ozone peaks were observed at the Orchard Farm and Jerseyville monitors on this day (early peak at 11:00 AM-12:00 PM and late peak at 4:00 PM-5:00 PM). This episode ends on June 24th due to late afternoon and evening rain showers in and around the St. Louis area.
July 2-5
The first portion of the July 2-5, 2002 episode was characterized as Meteorological Regime #7 in Section 2.3.5. This regime develops with strong high pressure over the eastern United States. 22
Strong subsidence remains over the east coast and high pressure centers can develop over Missouri and Illinois. Surface wind conditions are dependent on the location of the high pressure centers and are typically light and variable in the morning hours.
Surface high pressure was located over southern Illinois on July 2nd with light and variable surface winds in the morning with a south-southeasterly flow in the afternoon. The maximum 8-hour concentration was 88 ppb at the Orchard Farm monitor with elevated 1-hour concentrations also observed at West Alton and Jerseyville. The 72-hour back trajectories for July 2nd illustrated transport from the south-southwest (southern Missouri and western Arkansas). Cloud cover on this day may have led to somewhat reduced ozone concentrations. Maximum 8-hour upwind concentrations were below 60 ppb.
The high pressure over the central/eastern United States remained on July 3rd with light and variable surface winds throughout the day. Back trajectories indicated stagnation around the St. Louis area. The maximum 8-hour concentration was observed at West Alton (90 ppb) with the maximum 1-hour concentration at the Ferguson monitor (107 ppb). It appears that the ozone plume did not extend much beyond the suburban area during this day. Morning clouds in the St. Louis area may have somewhat limited ozone formation on this day. Maximum 8-hour upwind concentrations were between 60-70 ppb.
Surface high pressure was located directly over Missouri on July 4th with light and variable winds in the morning and a westerly/northerly push by the afternoon. Back trajectories were indicative of stagnation around the St. Louis area. Maximum 8-hour concentrations were observed at Ladue (103 ppb) with higher concentrations measured in the southern half of the monitoring network. Bonne Terre monitored a maximum 8-hour concentration of 90 ppb. Maximum 8-hour upwind concentrations in the northern part of the area were around 65 ppb. Also, a frontal boundary was present over Iowa in the afternoon.
The front approached from the north on July 5th and light/variable winds continued. Back trajectories still demonstrated stagnation with influence from the east (Illinois, Indiana). Again, the maximum 1-hour and 8-hour concentrations were seen in the south-western portions of the St. Louis area (Sunset Hills – 128 ppb 1-hour and 109 ppb 8-hour). Concentrations over the 8-hour NAAQS were observed at Queeny Park, Ladue, West Alton, S. Broadway, Margaretta, and Arnold. Upwind, maximum 8-hour concentrations were near 70 ppb.
The frontal passage on July 6th caused reduced ozone concentrations, but the maximum 8-hour concentration was still 84 ppb at Sunset Hills.
July 7-9
A surface wind shift is evident on July 7th with early morning winds light and from the south with a switch to northeast in the late morning and early afternoon hours. A surface high was located over the Great Lakes region on this day. Back trajectories illustrate transport around that high pressure center from the northern Ohio River Valley. Maximum 1-hour concentrations were observed early (10:00 AM) at West Alton -- 111 ppb with a substantive decrease in concentration over the next few hours. 1-hour and 8-hour concentrations increase back over the urban core and the highest 8-hour concentration was monitored at the Sunset Hills monitor (93 23
ppb). Several sites exceeded the 8-hour standard on this day including: Ladue, Queeny Park, and Margaretta. Maximum 8-hour upwind concentrations (north of the area) were near 70 ppb.
The surface high migrated to the south and east over the Mid-Atlantic states and a frontal boundary was observed over the High Plains on July 8th (Meteorological Regime #2). Light and variable surface winds (southerly component) switched to southwesterly and westerly after noon and continued until the early evening hours. Back trajectories on this day illustrate transport from the north and east (Illinois, Indiana). As with July 7th, the suburban sites to the north (Orchard Farm, West Alton, and Alton) all monitored higher 1-hour concentrations at 10:00-11:00 AM than the typical mid-to-late afternoon. The highest 1-hour and 8-hour concentrations were monitored at Maryville (135 ppb and 119 ppb). This day recorded the highest monitored concentrations in the July 2002 episode. Several other sites monitored exceedances of the 8-hour standard: Margaretta (111 ppb), Ferguson (110 ppb), Edwardsville (104 ppb), E. St. Louis (102 ppb), West Alton (99 ppb), Breckenridge Hills (94 ppb), South Broadway (93 ppb), Maryville (92 ppb), Orchard Farm (91 ppb), Ladue (87 ppb), Clark (86 ppb), and Sunset Hills (85 ppb). Maximum 8-hour upwind concentrations (south of the area) were 76-77 ppb.
The front over Iowa progressed south toward St. Louis on July 9th and some rain showers developed during the day mitigating ozone production. Light and variable winds during the early morning give way to westerly surface flow in the late morning that continues with increased speed into the evening. Back trajectories reflect limited transport into the area. Only one monitor recorded an 8-hour exceedance on this day (Maryville - 90 ppb). After the front passed, “clean air” arrived from the north and ended this portion of the episode. It is interesting to note that Bonne Terre observed two 1-hour concentrations over 100 ppb on the morning of July 10th (carry-over from the previous episode days).
July 13-16
This portion of the episode was identified as Meteorological Regime #4 in Section 2.3.5. This regime has a surface high pressure center over Iowa with light surface winds from the northeast. Back trajectories for July 13th show influence from the high pressure center and illustrate transport from the northern Ohio River Valley. High 1-hour and 8-hour concentrations were observed over the southern portion of the monitoring network with the maximum at Sunset Hills (8-hour 89 ppb). Exceedances were also monitored at Arnold (88 ppb), Ladue (86 ppb), and Bonne Terre (85 ppb). As expected, the Bonne Terre concentrations become elevated after the near-metro sites. Maximum 8-hour upwind concentrations (north) were 76-77 ppb.
The high pressure center remained over Iowa and surface winds were variable in the morning hours and transitioned to the northeast in the late morning and early afternoon. Back trajectories again illustrated transport from Illinois, Indiana, and Ohio. Highest 1-hour and 8-hour concentrations were again in the southern portion of the monitoring network. Sunset Hills monitored the maximum 8-hour ozone concentration at 97 ppb. The Arnold and Bonne Terre monitors also monitored 1-hour concentrations over 100 ppb. 8-hour exceedances were monitored at Arnold, Bonne Terre, Sunset Hills, Ladue, S. Broadway, Margaretta, West Alton, E. St. Louis, and Alton. Maximum 8-hour upwind concentrations were near 80 ppb.
The high pressure center moved south over northeast Missouri and surface winds were very light and variable on July 15th (the entire day). Back trajectories still exhibited transport from Illinois and Indiana. This day was dominated by the stagnant air near the high pressure center and ozone built-up around the entire metropolitan area. Sunset Hills and Margaretta both monitored one-24
hour concentrations above 120 ppb at 2:00 PM. Nearly all urban and suburban sites monitored 8-hour exceedances on this day (Wood River, Bonne Terre, Houston, Jerseyville, and Nilwood were the only sites not to monitor over 85 ppb). Maximum 8-hour background concentrations were 75-83 ppb around the metropolitan area. On this day, emissions were not pushed out of the urban center and resulted in high concentrations throughout the area.
The high pressure center migrated over into central Illinois on the 16th and surface winds were again light, but with a late morning/early afternoon push from the south. Limited transport from the east due to the migration of the surface high was prevalent in the back trajectories. The highest concentrations were monitored at the West Alton and Alton locations (93 and 90 ppb – 8 hour). There were also exceedances at Orchard Farm, Jerseyville, and Arnold. The higher values at Arnold are likely due to precursor carry-over to the south of the area from the previous day being blown back over the monitor on the 16th. Upwind concentrations around 70 ppb were observed. This is a “normal” high ozone day for St. Louis (light winds in the morning with a southerly push around noon.
Localized rainfall in St. Louis on July 17th and 18th suppressed ozone formation and ended this extended period of high ozone.
July 30 and August 1-4
A 500 mb high pressure system centered itself over the central plains during this episode. July 30th was a mild ozone exceedance day with only one 8-hour exceedance at Queeny Park (90 ppb). Transport was indicated by the back trajectories from the southwest (southwestern Missouri, Oklahoma and Texas). Winds were variable in the morning and shifted to the northeast around noon (contributing to the exceedance at Queeny Park). Widely scattered rain showers were prevalent on this day around the St. Louis area.
July 31st had sustained winds higher than the previous day and following days from the south and was not an exceedance day in this episode. The highest 8-hour concentration was 78 ppb at Orchard Farm.
A weak frontal boundary approached from the north on August 1st and a surface high pressure center developed over the West Virgina. Winds were from the south, southeast, and southwest during the day. The highest 8-hour concentrations were observed at Wood River (92 ppb), West Alton (88 ppb), and Alton (85 ppb) on the 1st. Back trajectories again illustrated transport from the southwest. Background concentrations were observed near 50 ppb for 8-hour averages.
High pressure was still located over West Virginia on August 2nd, with the frontal boundary near the St. Louis area pushed past the metropolitan area (Meteorological Regime #6). This caused a wind shift to the north and northwest in the late afternoon. Winds were highly variable on this day due to the frontal passage. Back trajectories show transport from southwest Missouri, Oklahoma, and Texas. The maximum concentrations happened as the front passed and precursors were pushed to the south. Arnold was the only site that observed an 8-hour exceedance (85 ppb), but Arnold, Sunset Hills, and S. Broadway all monitored 1-hour concentrations over 100 ppb. Background concentrations were still generally low (50-60 ppb) on August 2nd.
The frontal boundary retreated back to the north of St. Louis on August 3rd with high pressure remaining over the eastern seaboard. The “old”, dirty airmass returned to the area and caused a 25
sharp rise in monitored concentrations. Light and variable winds in the morning were predominant with the south-southeasterly flow during the late morning and early afternoon. Back trajectories again show transport from the southwest. The highest 1-hour and 8-hour concentrations were monitored at West Alton (99 ppb – 8 hour), Alton, and Orchard Farm. Background concentrations were monitored aorund 60 ppb in the area.
The front remains to the north on August 4th with high pressure remaining over West Virginia. Variable winds on this day contributed to higher concentrations in the urban core than on previous days. The highest observed concentrations were detected at Margaretta (116 ppb – 1 hour and 98 ppb – 8 hour). Other exceedances were monitored at Edwardsville, Maryville, Alton, and Ladue. Background concentrations near 70 ppb were again observed.
2.3.7 Emissions Input Preparation and QA/QC
The purpose of the emissions processing is to format the emission inventory for the photochemical model. Specifically, the emission inventory is allocated:
• Temporally – to account for seasonal, day of week and hour of day variability,
• Spatially – to reflect the geographic distributions of emissions, and
• Chemically – to account for the chemical composition of VOC and NOx emissions in terms of the Carbon Bond 4 (CB4) chemical mechanism.
Three sets of emissions inputs were prepared for the St. Louis 8-hour ozone modeling study—the 2002 model validation inventory, the 2002 typical emissions inventory, and the 2009 on-the-books inventory. The 2002 model validation inventory was used in the model performance evaluation. It includes day- and hour-specific continuous emissions monitoring (CEM) data for Electric Generating Units (EGUs). The 2002 typical emissions inventory is similar to the validation inventory except that it does not include day- and hour-specific CEM data. The typical emissions inventory was used as the basis for applying growth and control factors and calculating RRFs. Lastly, the 2009 on-the-books inventory accounts for emissions growth and incorporates federal, state, and local controls implemented between 2002 and 2009.
The process of producing a model-ready inventory is iterative, with data corrections or improvements invariably leading to a succession of more refined modeling inventories. The St. Louis modeling commenced with Base 1 and concluded with Base 4, and the latter reflects processing of Midwest RPO inventory updates through the Base K inventory as well as CENRAP inventory updates through the Base B inventory. Section 3 of this document describes the preparation of the Base 4 model-ready emissions inputs in greater detail. Below is a summary of the most significant changes from the Base 1 through the final Base 4 versions of the modeling inventories.
Base 1 Modeling Emissions Inventory
• CENRAP Base A inventory for all source categories plus CEM data for all CENRAP EGUs
• MRPO Base I inventory for all categories
• Initial draft link-based VMT from EW Gateway
Base 2 Modeling Emissions Inventory
26
• CENRAP Base B inventory for all source categories (no changes to CEM data)
• Corrected MO statewide recreational marine emissions in the offroad inventory
• MRPO Base J inventory for all categories except onroad mobile
• Draft link-based VMT from EW Gateway received Aug 2005 (~84 million DVMT total in counties covered by network)
Base 3 Modeling Emissions Inventory
• Draft MRPO Base K emissions for aircraft, commercial marine, locomotives for MRPO states (categories were missing from MRPO Base J)
• "Final" link-based VMT from EW Gateway received Oct 2005 (~70 million DVMT total)
Base 3b Modeling Emissions Inventory
• Corrected point source stack parameters in MRPO states
• Refined criteria for elevated and plume-in-grid (PiG) sources and set PiG sources consistently across all episodes
• Reran biogenics and onroad mobile for July episode using "Base 2" meteorological inputs
Base 4 Modeling Emissions Inventory
• MRPO Base K inventory for all categories except onroad mobile
• Added portable fuel containers (gas cans) to MO area inventory statewide
• Corrected aircraft refueling emissions at Lambert International Airport (applied a 90% control factor)
• Corrected VOC emissions at a point source in Arkansas -- 52,000 tons/yr changed to 52 tons/yr
An important part of this process is the quality control checks integral to the emissions modeling software that generate warning or error messages on suspect or incorrect records because of data that is missing, “out-of-bounds,” duplicative, lacking matching cross-reference data, or is otherwise deficient. In addition, programs external to the emissions modeling systems for generating data summaries, graphical depictions of emissions data, and other data-probing or analyses were relied upon extensively in the processing of inventories to assure the highest quality emission inputs are being used for photochemical modeling. Illinois and Missouri, both separately and jointly, conducted quality assurance checks on their respective state inventories. Additionally, ENVIRON conducted independent assessments of the inputs and outputs to the emissions model.
2.3.8 Meteorological Input Preparation and QA/QC
Meteorological data were generated using the MM5 prognostic meteorological model. Episodic MM5 runs at 5-day increments on the 36/12/4 km domains were performed by the four MDAW modeling hubs, with a minimum 12 hour spin up period used for each episode. The following table illustrates the physics options selected for the 36/12/4km MM5 analyses. An example of the configure.user files and mm5.deck files are included in Appendix B.
Physics Options
Selection
Configure.user file 27
Moisture
Mixed Phase (Reisner 1)
IMPHYS = 5,5,5 (MPHYSTBL=0)
Cumulus Parameterization
Kain-Fritsch 2 (36/12km), None (4km)
ICUPA=8,8,1
Planetary Boundary Layer
Pleim-Xiu
IBLTYP=7,7,7
Radiation
RRTM
FRAD=4,4,4
Land Surface Model
Pleim-Xiu
ISOIL=3
Shallow Convection
No
ISHALLO=0,0,0
Nudging
Sfc = Varies
Analysis = Yes
FDDAGD=1
The MDAW Modeling Hubs processed the MM5 data using the MCIP and MM5CAMx processors to generate meteorological inputs for the CMAQ and CAMx models, respectively. The outputs from this processing were used by the Modeling Hubs to perform base case modeling using both the CAMx and CMAQ models.
The MDNR, IEPA, and Ameren modeling staff reviewed the performance of the meteorological output using METSTAT. As with the inventory discussion, the process of producing model-ready meteorological can be iterative. In this study after several rounds of sensitivity analyses, the modeling team discovered the July 2002 episode was dramatically underpredicting ozone. Based on the METSTAT outputs for the Base 1 meteorological files and a substantive ozone underprediction, a strong temperature bias in the 4km grid (model underprediction) for this episode was discovered that was not present in the other two episodes.
Based on recommendations from ENVIRON, a sensitivity analysis was completed on the MM5 4-kilometer domain that evaluated the use of:
1) 3-D analysis nudging (FDDA) to the 4 km grid for MXRATIO, temperature, and winds.
2) No nudging in planetary boundary layer for temperature or MXRATIO, but included for winds (i.e surface analysis nudging).
The results of this sensitivity were evaluated along with the original performance using METSTAT and the resulting photochemical analysis. The changes in the mm5.deck file are included in Appendix B along with the other MM5 input files. As can be seen in the METSTAT results, this sensitivity provided enhanced performance for the 4-km temperature field and, ultimately, improved photochemical model underprediction for this episode. Therefore, the final basecase modeling analyses (emissions and photochemical) were performed with this Base 2 meteorological dataset for the July 2002 episode. Both sets of METSTAT results (Base 1 and Base 2) are included in Appendix B along with the other two episodes’ results.
In addition to the METSTAT analyses detailed in Appendix B, ENVIRON performed some additional quality assurance/quality control measures for this study:
• Analyses of the MM5 data to assure that it had been transferred correctly.
• Evaluation of upper-air MM5 meteorological estimates by comparison them to upper-air observations and satellite images.
• Comparison of the MDAW modeling hub’s 2002 36 km MM5 simulation with the 28
ones generated by WRAP and VISTAS.
2.3.9 Air Quality Model Input Preparation and QA/QC
Key aspects of QA for the CMAQ and CAMx input and output data included the following:
• Verification that correct configuration and science options were used in compiling and running each module in the CMAQ and CAMx modeling systems, where these included (for CMAQ) the MCIP, JPROC, ICON, BCON and the CCTM.
• Verification that the correct configuration and science options were used in running each model in the CAMx modeling system where these included MM5CAMx, TUV, land use, CAMx, and the CMAQ-to-CAMx emissions and IC/BC processors.
• Verification that correct input data sets were used when running each model.
• Evaluation of CMAQ and CAMx results to verify that model output was reasonable and consistent with general expectations.
• Processing of ambient monitoring data for use in the model performance evaluation.
• Evaluation of the CMAQ and CAMx results against concurrent observations and each other.
• Backup and archiving of critical model input data.
During the processing of the MM5 data for use into CAMx, there are two different options for computing vertical turbulent diffusivity when using the MM5 options chosen for the St. Louis study: (1) the O’Brien scheme (OB70) and (2) the CMAQ scheme. Each of these schemes was utilized in the early photochemical analyses to determine the better fit for this exercise. The minimum Kv value (Kz_min) was set at 0.1 m2/s for OB70 and 1.0m2/s for CMAQ. The modeling group performed all the Base 1, 2, and 3 runs with both sets of meteorological inputs. After establishing that the CMAQ scheme was slightly superior to the OB70 scheme based on model performance in the small 4km domain, the remaining photochemical analyses were developed with the CMAQ-processed dataset.
In addition to this choice and based on recommendations from ENVIRON, a program called kvpatch was utilized to better represent mixing over the urbanized area. This program applies minimum Kv values to layers below a user-defined height based on input landuse fields and maximum Kv within that depth. The kvpatch program was utilized for both the OB70 and CMAQ diffusitivity schemes with improved results from both when compared to the original scenarios. However, once the decision was made to pursue the CMAQ scheme, two versions of this program were run: kvpatch and super-kvpatch for one of the episodes (June 2002). The super-kvpatch version utilized a value of 2.0 m2/s over urban land use areas, while the kvpatch version utilized a value of 1.0 m2/s. The user-defined height was chosen as the default (100 meters) for all these sensitivities. This last sensitivity illustrated that the kvpatch program and not the super-kvpatch program should be used for the final basecase analyses.
The most critical element for CMAQ and CAMx simulations was the QA/QC of the meteorological and emissions input files, which is discussed above. The major QA issue specifically associated with the air quality model simulations was verification that the correct science options were specified in the model itself and that the correct input files were used when running the model. For CMAQ modeling, MDNR employed a system of naming conventions using environment variables in the compile and run scripts that guarantee that correct inputs and science options are used. Similar procedures were used in CAMx modeling using file and 29
directory naming conventions. A redundant naming system was employed so that the names of key science options or inputs are included in the name of the CMAQ and CAMx executable program, in the name of the CMAQ and CAMx output files, and in the name of the directory in which the files were located. This was accomplished by using the environment variables in the scripts to specify the names and locations of key input files.
A second key QA procedure was to avoid “recycling” run scripts, i.e., the original run scripts and directory structures that were used in performing a model simulation were preserved.
The MDAW modeling hubs and ENVIRON also performed a post-processing QA of the CMAQ and CAMx output files similar to that described for the emissions processing. Animated graphic files were generated using PAVE, and were viewed to search for unexpected patterns in the CMAQ and CAMx output files. In the case of model sensitivity studies, the animated graphic files were prepared as difference plots for the sensitivity case minus the base case. This was done to screen for errors in the emissions inputs. Finally, 24-hour average plots were produced for each day of the CMAQ and CAMx simulations. This provided a summary that was useful for quickly comparing various model simulations. A table detailing all the scenarios completed is included in Appendix D along with example run scripts for the photochemical models.
2.3.10 Base Case Modeling and Model Performance Evaluation
The St. Louis 8-hr Ozone Modeling Study simulated three high 8-hour ozone episodes from the summer of 2002. The three episodes were: June 10-24, July 2-16, and July 29-August 5, 2002. During Phase I of the St. Louis 8-hour ozone modeling, both the CAMx and CMAQ models were used to simulate these three episodes, and model performance was evaluated following EPA Guidance (EPA 2006) augmented by other recommendations (e.g., Boylan, 2004; McNally and Tesche, 1994; Pun, Chen and Seigneur, 2004; and Morris et al., 2005a, b).
Initially, the four MDAW Modeling Hubs performed ozone modeling on the 36/12/4 km grid using both the CAMx and CMAQ models with the Base 1 2002 base case emissions for the June, July and July/August ozone episodes. CMAQ and CAMx were applied using the exact same horizontal and vertical structure using the Carbon Bond IV chemical mechanism (Gery et al., 1989). The ozone performance of both models during the Phase I modeling was characterized by a general underestimation bias. The underestimation bias in CMAQ was more severe, so, given the time constraints imposed by the SIP deadline, the primary focus was placed on the CAMx model, in order to diagnose the causes of its underestimation bias to improve performance for the 8-hour ozone attainment demonstration. NOTE: the final basecase evaluation results using CMAQ are included in Appendix D. As can be seen, the CMAQ results illustrate dramatic underprediction when compared to the CAMx results. Therefore, the CMAQ modeling analyses were not conducted for the attainment demonstration analyses due to poor model performance.
The modeling group performed several iterations of photochemical analyses including the various emissions scenarios (Base 1, 2, 3, 3b, and 4), CAMx model versions (v4.11s, v4.20, and v4.30), vertical diffusivity schemes (CMAQ, OB70, with and without kvpatch), 4 km domain sizes, photochemical mechanisms (CAMx Mech3, Mech4_CF, and Mech4_None), different meteorological modeling inputs, and Plume-in-grid treatment of large point sources. Many of these sensitivity combinations were performed on all three ozone episodes, but several were attempted only on one or two episodes. Each analysis provided useful information with respect to the air quality model’s response to emission changes and the model’s performance was tracked to identify the set of options that provided the best performance. The results of these 30
analyses are included in tables for each episode in Appendix D.
In addition to the myriad sensitivity analyses discussed above, ENVIRON was tasked with performing other separate sensitivity analyses. These included investigating the possibility that the model’s exaggerated ozone suppression in the St. Louis urban core may be due to insufficient vertical mixing. A summary of their analysis and findings is provided below.
ENVIRON - Analysis of the first round of CAMx modeling of the three episodes showed that part of the ozone underprediction could be attributed to the model’s tendency to overstate the ozone suppression in the St. Louis urban core. In addition, the modeled ozone formation occurred too slowly, so that peaks occur further downwind from the St. Louis urban core in the model than observed. The magnitudes and the timing of the modeled peaks, however, were comparable to observations.
Several potential causes of these ozone performance issues were evaluated. Ozone formation that occurs too slowly may be caused by an insufficient free radical supply. This may due to several factors. One possibility is insufficient sunlight to generate radicals vital for ozone formation. Incoming solar radiation may be overly attenuated if the meteorological model has a bias toward excessive cloudiness. The MM5 cloud fields were compared to satellite imagery during the June and July episodes, and the observed and modeled cloud fields were found to agree reasonably well; this suggests that the MM5 cloud field is not the cause of the delayed ozone formation.
NOx is emitted into the model layers near the surface, and if the vertical mixing in the model is more subdued than in the real world, NOx concentrations in the near surface layers will be unrealistically high in the model. This can cause excessive titration of ozone, leading to ozone suppression. It is also possible that the reactivity of emitted VOCs in the emissions inventory is understated or that the VOC inventory (mobile sources in particular) is simply underestimated. A series of sensitivity tests was performed to evaluate these hypotheses.
First, the chemical mechanism used in the initial basecase I modeling, CAMx Mechanism 3, was replaced with CAMx Mechanism 4, and Episode 2 was rerun; the two Episode 2 simulations were identical except for the choice of chemical mechanism. CAMx Mechanism 3 uses the OTAG (Ozone Transport Assessment Group) version of the CB4 (Carbon Bond 4) mechanism with updated PAN chemistry and revised radical and isoprene reactions. CAMx Mechanism 4 gas-phase chemistry is based on an updated version of the CB4 Mechanism that was developed by extending the inorganic reactions in the OTAG version of the CB4 mechanism (Yarwood et al. 2005). The use of CAMx Mechanism 4 improved the normalized bias by 2-5 percentage points. This is likely due to renoxification through HNO3, which is the reaction of gaseous NO with HNO3 on surfaces to form NO2 and HONO. Removal of HNO3 via deposition is therefore not a permanent sink, but represents a pathway through which NOx can be cycled back into the atmosphere, and possibly increase ozone production. Reactions for renoxification through HNO3 are present in Mechanism 4, but not Mechanism 3.
A second set of sensitivity studies examined the effect of increasing the vertical diffusivity (Kv). Kv is not calculated within MM5, but is determined within the meteorological preprocessor MM5CAMx. Kv values near the ground were modified because the MM5 tends to underestimate real world near-surface mixing of trace gases. In the original St. Louis CAMx application, land-use dependant minimum Kv values were applied in the surface layer. Then, in each vertical column, the maximum diffusivity encountered in the lowest 100 meters was applied to all layers 31
below 100 meters. In the sensitivity test, the minimum Kv value was increased from 1.0 m2 s-1 to 2.0 m2 s-1. This change improved model performance slightly, reducing the mean normalized bias during Episode 2 by ~2 percentage points.
Next, the model’s sensitivity to changes in VOC and NOx emissions was explored, with the intent of determining the cause of the exaggerated ozone suppression in the urban core. In the first emissions sensitivity test, biogenic isoprene emissions were increased throughout the 4-km domain by a factor of 1.5 in order to determine whether underestimation of biogenic VOCs might be the cause of the excessive ozone suppression near the downtown St. Louis. There was some improvement (a 3-5 percentage point increase in the normalized bias) in the June episode. During periods of southerly winds during the June episode (e.g. June 21), air rich in biogenic emissions was advected toward the downtown St. Louis, supplying additional VOCs to the St. Louis NAA. However, increasing the biogenic isoprene had little effect during the July episode.
As a result of this sensitivity test, the biogenic and mobile emissions were recalculated using MM5 temperatures adjusted to correct a known bias (see Section 2.3.8) in the modeled temperature fields.
Several other emissions tests were carried out that did not have a significant effect on model performance. Anthropogenic NOx emissions were reduced by 25%. This had a small effect on the normalized bias (<+2% change), with urban ozone increases offset by rural ozone reductions. A 50% increase in non-isoprene (i.e., anthropogenic) VOC emissions resulted in only a minor improvement in performance (<1% change in normalized bias). Finally, a test in which NOx was reduced by 25% and non-isoprene VOCs increased by 50% did not have a significant impact on performance.
As a result of all these sensitivity analyses and model performance evaluations, the final 2002 Base Case episodic simulations were carried out using CAMx version 4.30, Mechanism 4 with no PM chemistry, the minimum Kv set to 1.0 m2/s using the CMAQ kv scheme with the kvpatch adjustment, Plume-in-Grid treatment for large sources, and Base 4 emissions.
An operational evaluation of model performance of the final 2002 episodic CAMx simulation was carried out according to EPA Guidance (EPA 2006). Although on many days model performance was characterized by an under-prediction bias, the model performance was found to exhibit sufficient skill in meeting most performance goals that it could be used to project future-year ozone air quality and 8-hour ozone attainment for St. Louis, recognizing the inherent uncertainties in atmospheric modeling process. The model performed best during the periods of high ozone that are critical to the attainment demonstration. The method of performing the attainment test (EPA 2006) effectively removed from consideration the low ozone days on which the model’s performance issues were most in evidence.
2.3.11 Future-Year Modeling and Modeled Attainment Demonstration
Future-year modeling for ozone was performed with CAMx for 2009. The 2002 emissions were projected to 2009 assuming growth and currently on-the-book (OTB) controls. These growth and control factors were developed by the contracting team. Regional growth and control factors developed by EPA for the Clean Air Interstate Rule (CAIR) and by the various RPOs were used, and were enhanced with information specific to the St. Louis area. The control factors reflect federally promulgated Maximum Achievable Control Technology (MACT) standards, New Source Performance Standards, implementation of the NOx SIP Call, and court settlements 32
(consent decrees) reached with refineries and ethanol producers.
The St. Louis future-year modeling used the 2002 MM5 meteorological conditions. That is, the meteorological conditions for the 2009 future-year were assumed to the same as for 2002. This allowed for the comparison of the changes in 8-hr ozone concentrations in the study area from the current (2002) to future-year due to changes in emissions. This means that the effects of climate change, land use variations and climatic variations were not be accounted for in the future-year meteorological inputs.
The St. Louis modeling results were used to demonstrate attainment of the 8-hour ozone standards. The procedures for performing a modeled ozone attainment demonstration are outlined in EPA’s 8-hour ozone modeling guidance (EPA, 2006) and are discussed in Section 4. These procedures involve the use of the model in a relative sense to scale the observed 8-hour ozone design value based on the relative changes in the modeled 8-hour ozone concentration between the current-year (2002) and future-year (2009).
In the St. Louis Nonattainment Area, there are 8 monitoring sites for which the current year (2002) design values exceeded the 8-hour ozone NAAQS (85.0 ppb or higher). The projected design values for 2009 using the modeling results show that no monitor exceeds 85 ppb, so that all St. Louis monitors are projected to attain the 8-hour standard. The attainment test using the St. Louis modeling results suggested that on-the-books controls are sufficient for the St. Louis area to pass the modeled attainment demonstration test.
2.3.12 Weight of Evidence (WOE) Analysis
EPA guidance states that if there is a future design value between 82-87 ppb at one or more sites/grid cells, then “a weight of evidence demonstration should be conducted to determine if aggregate supplemental analyses support the modeled attainment test” (EPA 2006). In fact, EPA suggests that a weight of evidence (WOE) always be performed to corroborate the modeled attainment demonstration test. In a WOE determination, results from several types of air quality analyses are considered and the results reviewed for consistency with the conclusion of the modeled attainment test regarding the likelihood that the proposed control strategy will result in a NAA meeting the NAAQS. The credibility of each type of analysis used in the WOE determination must be assessed and finally, a conclusion reached regarding the likelihood of attainment.
The trends in air quality and emissions, corroborative modeling analyses, and additional modeling metrics all support the conclusion of the CAMx modeled attainment demonstration that the St. Louis NAA will be attainment by 2010. Every one of the analyses presented was consistent in predicting attainment for St. Louis; not a single component of the weight of evidence determination suggested that the St. Louis area will be out of attainment in 2010. Thus, the evidence that the St. Louis NAA will reach attainment of the ozone NAAQS by 2010 is overwhelming and conclusive.
3.0 EMISSIONS MODELING
This section describes the preparation of the 36/12/4 km emissions inputs for the June 10-24, July 2-16 and July 29-August 5, 2002 episodes. Three separate sets of CAMx-ready Base 4 emissions files were prepared—the 2002 model validation inventory, the 2002 typical emissions inventory, and the 2009 on-the-books inventory. The emissions modeling projection and grid 33
structure are described in Figure 2-1 and Tables 2-1 and 2-2.
In building the Base 4 inventory for St. Louis ozone modeling, the best available emissions datasets were obtained from RPOs that are developing comprehensive inventories for their member states and tribes for regional haze purposes. Most notably, the St. Louis Base 4 inventory incorporated the Base B inventory generated by CENRAP and the Base K inventory from the MRPO.
Emissions for each major source group (e.g., mobile, off-road mobile, area, point, and biogenic) were processed separately and merged together to create model-ready emissions files. Emissions for all source categories were processed for each day in the three ozone episodes.
MDNR, IEPA, and ENVIRON/Alpine collaborated on preparing the emissions. IEPA used EMS-2003 to process the emissions for electric generating units (EGUs) for the MRPO states plus Minnesota.1 ENVIRON/Alpine generated biogenic and onroad mobile emissions, and MDNR prepared emissions for the remainder of the source categories. Both ENVIRON/Alpine and MDNR used SMOKE v. 2.1 as their emissions processing tool. MDNR, IEPA, and ENVIRON/Alpine were all involved in quality-assuring the raw emissions inputs as well as the gridded, model-ready emissions files.
Tables 3-1 and 3-2 present summaries of anthropogenic NOx and VOC emissions in the St. Louis nonattainment counties for 2002 and 2009. The emissions in these tables are based on the 2002 typical and 2009 on-the-books inventories for a weekday in the June episode (Wednesday, June 19). Additional emissions summaries and details of the emissions preparation are included in Sections 3.1 through 3.4. NOTE: due to rounding, the sum of the individual entries may differ slightly from the totals shown in the emission summary tables.
1 Minnesota, though a CENRAP member state, was actively involved with the MRPO in reviewing and revising its inventory beyond CENRAP’s contractor-supported base year inventory. Minnesota gave approval to the MRPO for using these updates in their emissions processing, and this lead was generally followed for Minnesota’s emissions for the St. Louis study. 34
Table 3-1. Summary of Weekday NOx Emissions from the 2002 Base 4 Typical and 2009 On-the-Books Inventories for St. Louis Nonattainment Counties
Area
NOx (tpd)
Offroad Mobile
NOx (tpd)
Onroad Mobile
NOx (tpd)
Non-EGU Point
NOx (tpd)
EGU Point
NOx (tpd)
Total: All Source Categories
NOx (tpd)
County
2002
2009
2002
2009
2002
2009
2002
2009
2002
2009
2002
2009
Missouri
Franklin
1.7
1.8
4.8
3.8
15.1
7.7
0.1
0.1
24.2
28.1
46.0
41.6
Jefferson
1.4
1.5
5.3
5.4
17.7
9.6
15.4
18.3
10.6
15.2
50.4
50.0
St Charles
2.4
2.5
7.1
7.7
23.5
12.8
1.2
1.2
44.9
21.9
79.1
46.1
St Louis
9.9
10.2
31.5
31.6
97.4
52.7
2.3
2.3
23.1
17.9
164.1
114.8
St Louis City
4.0
4.4
12.0
10.5
26.4
14.4
5.4
5.1
0.0
0.0
47.9
34.5
MO NAA Subtotal
19.4
20.4
60.8
59.1
180.1
97.2
24.4
27.1
102.8
83.1
387.4
286.9
Illinois
Jersey
0.1
0.1
2.6
2.4
1.7
1.0
0.0
0.0
0.0
0.0
4.4
3.5
Madison
0.7
0.8
15.6
13.1
24.2
11.7
29.7
25.7
13.9
9.5
84.1
60.8
Monroe
0.1
0.1
5.5
4.2
3.8
1.9
0.1
0.1
1.7
0.0
11.2
6.4
St Clair
0.5
0.6
10.8
8.8
23.6
11.4
3.3
4.7
0.4
0.0
38.6
25.5
IL NAA Subtotal
1.4
1.5
34.5
28.5
53.3
26.1
33.1
30.5
15.9
9.6
138.3
96.2
NAA Totals
20.8
21.9
95.3
87.6
233.4
123.3
57.4
57.6
118.7
92.7
525.7
383.1
Table 3-2. Summary of Weekday VOC Emissions from the 2002 Base 4 Typical and 2009 On-the-Books Inventories for St. Louis Nonattainment Counties
Area
VOC (tpd)
Offroad Mobile
VOC (tpd)
Onroad Mobile
VOC (tpd)
Non-EGU Point
VOC (tpd)
EGU Point
VOC (tpd)
Total: All Source Categories
VOC (tpd)
County
2002
2009
2002
2009
2002
2009
2002
2009
2002
2009
2002
2009
Missouri
Franklin
3.9
4.2
2.7
3.3
5.3
3.7
1.9
1.9
0.8
0.8
14.5
14.0
Jefferson
8.7
9.0
4.0
4.5
8.3
4.7
1.7
1.7
0.4
0.5
23.1
20.4
St Charles
9.1
10.1
7.2
7.2
11.4
6.5
3.0
3.1
0.6
0.6
31.3
27.4
St Louis
36.2
39.2
27.5
20.3
49.3
28.0
13.2
13.4
0.2
0.3
126.4
101.3
St Louis City
13.4
14.1
5.6
4.1
14.5
8.3
11.1
10.8
0.0
0.0
44.6
37.2
MO NAA Subtotal
71.4
76.6
46.9
39.4
88.9
51.2
30.8
30.9
1.9
2.3
239.9
200.3
Illinois
Jersey
2.2
2.2
0.5
0.5
1.1
0.5
0.1
0.0
0.0
0.0
3.8
3.2
Madison
13.8
13.3
4.4
3.3
11.3
5.5
11.0
7.9
0.3
0.2
40.8
30.1
Monroe
2.3
2.3
0.6
0.5
2.0
1.0
0.1
0.0
0.1
0.0
5.1
3.9
St Clair
11.6
11.0
3.4
2.4
11.7
5.7
3.7
3.9
0.0
0.0
30.5
22.9
IL NAA Subtotal
29.9
28.8
9.0
6.5
26.1
12.7
14.9
11.9
0.4
0.2
80.2
60.1
NAA Totals
101.3
105.4
55.9
45.9
115.0
63.9
45.7
42.7
2.3
2.5
320.1
260.4
35
3.1 2002 BASE 4 MODEL VALIDATION EMISSIONS INVENTORY
The 2002 Base 4 model validation inventory was used for the CAMx model performance evaluation. Because this inventory is used to calibrate the CAMx model, the intent is to include episode-speci