1. 1 Exceptional Event Rule Summary NESCAUM MAC Meeting May 16 and 17, 2006 Newport, RI

2. 2 Exceptional Events Rulemaking Proposal General Overview March 1, 2006 US EPA

3. 3 Key Requirements of SAFETEA-LU EPA must publish a proposed rule on exceptional events in the Federal Register by March 1, 2006, and must issue a final rule within 1 year of proposal. (Comments were due May 9, 2006) When a State wishes to have air quality data associated with an exceptional event excluded from regulatory determinations, the exceptional event must be demonstrated by reliable and accurate data. The State must show that there is a “clear causal relationship” between exceedance(s) of a National Ambient Air Quality Standard (NAAQS) and the event. There must be a public review process related to an exceptional event determination. The rule must set criteria and procedures for States to petition EPA to exclude data affected by exceptional events. Until the rule is finalized, certain existing EPA guidance and regulations will stay in place.

4. 4 Principles to Be Followed, as Established by SAFETEA-LU Protection of public health is the highest priority. Air quality data should be carefully screened to ensure that events not likely to recur are represented accurately in all monitoring data and analyses. Timely information should be provided to the public in any case in which the air quality is unhealthy. All ambient air quality data should be included in a timely manner in an appropriate Federal air quality database that is accessible to the public. Each State must take necessary measures to safeguard public health, regardless of the source of the air pollution.

5. 5 Outline of the Proposed Rule Definitions and examples– what is an exceptional event?  Rule follows SAFETEA-LU definition & gives examples  New section 319 defines exceptional event as an event that: Affects air quality; Is not reasonably controllable or preventable; Is an event that is caused by human activity that is unlikely to recur at a particular location, or is a natural event; and Is determined by the Administrator through the process established in the rule to be an exceptional event.  The rule recognizes natural events as a subset of exceptional events “in which human activity has no substantial or direct causal connection,” and also recognizes that natural events are likely to recur.

6. 6 Examples of Exceptional Events Chemical Spills and Industrial Accidents Structural Fires Exceedances due to Transported Pollution Exceedances due to a Terrorist Attack Natural Events:  Volcanic & Seismic Activities  Natural Disasters & Associated Clean-up Activities  High Wind Events  Unwanted Fires  Stratospheric Ozone Intrusions

7. 7 Process for Identifying “Exceptional” Events Proposed rule outlines the basic steps through which an event is determined to be “exceptional”:  State must flag the data in the Air Quality System (AQS) database as being influenced by an exceptional event.  State must submit documentation and demonstration to EPA for concurrence related to the event.  EPA must concur on the flag for the data to be excluded from regulatory decisions. First major question: Which events are eligible to be treated as “exceptional”? What does it mean for an event to “affect” air quality, as required by the statutory definition?

8. 8 When Does an Event “Affect” Air Quality? EPA is taking comment on 3 options:  OPTION 1: To be eligible for concurrence, flagged values must be above 95 th percentile of non-event days for the calendar quarter, based on previous 3-5 years. Appropriate documentation would still be required, but it should be easier to show these days are “exceptional” because they fall 2 standard deviations above the mean (simple statistical test) This would include approximately 85% of the days that have earned EPA concurrence in the past.  OPTION 2: 95 th /75 th percentile tiered approach Days between the 75 th and 95 th percentile would also be eligible for exclusion, pending more substantial demonstration.  OPTION 3: General case-by-case evaluation without threshold criteria. In addition, daily value is only eligible for exclusion if State shows that an exceedance of the applicable air quality standard would not have occurred “but for” the influence of exceptional events.

9. 9 Timelines for Flagging Data & Submitting Demonstrations Rule proposes 3 options for comment:  OPTION 1: Early Flagging by States (within 90 days after end of calendar quarter in which event occurred) & Demonstration Submission (within 90 days after flagging)  OPTION 2: Early Flagging (90 days) & Delayed Demonstration Submission (3 years)  OPTION 3: Delayed Flagging and Demonstration Submission (no later than 6 months prior to regulatory determination) EPA must concur or not concur within 30 days after submission of documentation, with possibility of 30-day extension for more complex demonstrations

10. 10 State Demonstrations In Support of a Flag The demonstration must show a “clear causal relationship” between the affected data and the event. However, rule proposes no specific requirements regarding the contents of the demonstration– this is a case-by-case submittal with an emphasis on weight-of-evidence. (If it is exceptional you will know it!) Such demonstrations generally include documentation showing:  The event occurred and emissions related to the event were transported in the direction of violating monitor  Size of the affected area  Relationship in time between event, transport of emissions, and recorded concentrations  Pollutant species-specific information supporting causal relationship, if appropriate Types of data that states could employ:  Satellite images  Surface measurement data  Monitoring data from one or more downwind monitors  Chemical composition analysis from speciation monitors or laboratory analysis of filters  Meteorological data and historical record

11. 11 Public Health Protection Rule also takes comment on what, if anything, States should be required to do to protect public health during or following an exceptional event Preferred Option: (How about AQI and AirNow?)  Prompt public notification that an event is occurring or is expected to occur  Public education on how to reduce individual exposures to air pollution due to an event  Implementation of reasonable measures to protect public health (This could include mitigation of significant contributing anthropogenic sources, if present, or otherwise minimizing or abating public health impacts.) Other Options for Comment:  Mitigation Plan with more specific control requirements (e.g. RACM on contributing anthropogenic sources) developed in advance as part of section 110 SIP.  Mitigation Plan developed after the occurrence of a natural event that we expect will recur, but not submitted as part of SIP. This may also include RACM requirements on contributing anthropogenic sources.  No specific requirements for public health protection or mitigation of events– let States design the approach they think best

12. 12 Fireworks Fireworks are not explicitly covered by section 319 or this rulemaking. However, EPA proposes as a policy matter to treat certain types of fireworks events in a manner similar to exceptional events. Specifically, where States can show that the use of fireworks displays is integral to significant traditional national, ethnic, or other cultural events (e.g., 4 th of July celebrations, Chinese New Year), EPA is proposing that air quality data associated with such events could be excluded from regulatory determinations. For such events, public health protection efforts may be appropriate. EPA requests comment on the treatment of fireworks and any requirements that should apply. (NY does not support the exclusion of data influenced by fireworks.)

13. 13 Around 2002 the EPA transitioned from the old “AIRS” database of pollutant concentrations to the new “AQS” database and, as part of this process, revised the list of data qualifiers. Valid data can now be flagged as being an “exceptional event” or having a “quality assurance” issue, while invalid (or “null”) data are excluded from the database. There are 20 such qualifiers:

14. 14 02-05 PM-2.5 Data Valid data pointsPercent valid CT9144830990.868 DE5617509790.742 ME5264495194.054 MA146721205382.15 NH5106489895.926 NJ108591072598.766 NY233502116290.63 PA366093548596.93 RI5219453686.913 VT3492330394.588 PR8215690083.993 VI50337173.757

15. 15 Structural fires (F) appeared 15 times, including once each at sites in AZ, GA, KS, KY, NV, NJ, NC, and TX. One site outside of Charleston, WV reported this flag on three consecutive days (1/7, 1/10, and 1/13/2005). In OH, two FRM sites (one collocated with an STN) in Canton reported this flag on 6/27/2005), while an STN site in Cincinnati reported this flag on 8/19/2004. Sandblasting (D) appeared 17 times; 11 of these occurred at a monitor in Ste. Genevieve, MO (various times 2003-2005), while the other six were for the STN monitor at Queens College (August-September 2005). Sanding/salting of streets (N) appeared 18 times – once at a site in Reno, NV; twice at a site in Portsmouth, NH; 10 times at a site in Cleveland, OH; and five times at a site in Kingsport, TN. Sixteen of the 18 appearances occurred during the winter of 2004/05.

16. 16 Agricultural tilling (K) appeared 21 times. Eleven of these occurred at three sites in CA; two occurred at a site in Washington, DC; two occurred at a site in Kinston, NC; two occurred at a location outside of Phoenix, AZ; two occurred at two different sites in WI; and one each at Covington, KY and Harrison Co., TX. Rerouting of traffic (M) appeared 27 times, all at a single location in San Juan, PR in June 2003. Prescribed burning (Q) appeared 29 times at various times and locations: CA (6); TX and DC (4 each); NV and MI (3 each); MN, MT, and NC (2 each); and FL, GA, and MO (1 each). Infrequent large gatherings appeared 35 times. Sixteen of these occurred at four sites in HI on New Year’s eves and New Year’s days. Six of these occurred at a site in West Orange, TX (five during the mid-April time). Five of these occurred at three sites in UT on July 4 th s. Four of these occurred at a site in Charleston, SC (one of which was a July 4 th ). Single occurrences at sites in KY (a July 4 th ), MI, MN, and WI corresponded to summertime dates

17. 17 Roofing operations (P) occurred 94 times, and 74 of these occurred at various locations in CA. There were three occurrences each in MI and AK; two occurrences each in AZ, MT, SC, TN, and WA; and one each in GA, ND, OH, and WI. High winds (A) occurred 133 times, and 56 of these occurred at various sites in CA. Other states with more than five occurrences were AL (14), UT (12), OH (9), TX (8), and AZ (6). Four or fewer occurrences were from sites in DC, FL, KY, MA, MI, MN, MO, MT, NC, ND, PA, SC, TN, and WI. Highway construction (L) occurred 206 times, and 159 of these occurred at a site in San Juan, PR from 2002-2003. Twenty-six of these occurred at several sites in CA. The balance of occurrences were from sites in AL (2), AZ (4), FL (1), GA (3), IN (5), IA (1), NJ (1), OH (1), SC (1), and WV (2). Volcanic eruptions (C) occurred 214 times, all of which occurred at numerous sites in PR and VI from 2002-2005.

18. 18 Sahara dust (U) occurred 704 times, and 696 of these occurred at various locations in PR and VI. Five occurred at two sites in TX (July 2005) and the other three occurred at three sites in FL (two days in July 05, one in April 04). Construction/demolition (J) occurred 888 times, and 675 of these occurred at various locations in CA. Other states with more than 20 occurrences were NJ (37), TX (28), KY (26), SC (23), and PA (21). The next highest total was from DC (10). Forest fires (E) occurred 941 times. In the northeastern states from DE to NH, this flag occurred 237 times, and 236 of these corresponded to the Jul 6-9, 02 period. Other states with substantial wildfires were SC (192), MT (149), OR (85), TN (80), CA (64), AK (44), KY & NV (16 each), CO (12), and AZ & TX (10 each). The states of GA, NC, UT, and WA had fewer than 10 occ. of this flag. Clean up after a major disaster (R) occurred 1235 times, the most frequently occurring exceptional event flag. A large majority of these (1181) occurred in LA, MS, and TX during the fourth quarter of 05, presumably in the wake of the hurricane season. In addition to these there were 24 occurrences in AL during the Sept-Nov 04 time period, and 30 occurrences in NC during the Dec 7 & 10, 02 time period.

19. 19 Seismic activity (S) appeared one time at a site in Akron, OH on 3/29/2002. Unusual traffic congestion (I) appeared three times; twice at a site in Hazard, KY (6/1/2002 & 8/7/2004) and once at a site in Onamia, MN (8/22/2004). Chemical spills & industrial accidents (H) appeared seven times, once each at a site in AR, FL, GA, KY, MT, NC, and VA. Stratospheric ozone intrusion (B) and High pollen count (G) – did not appear even once.

20. 20