Presentation on theme: "An Analysis of Ozone Monitoring Seasons in the U.S. Louise Camalier (not attending) (Presented by David Mintz) National Air Quality Conference Portland,"— Presentation transcript:
An Analysis of Ozone Monitoring Seasons in the U.S. Louise Camalier (not attending) (Presented by David Mintz) National Air Quality Conference Portland, Oregon April 8, 2008
Purpose The ozone NAAQS level is now 0.075 ppm Are states current official monitoring seasons adequate to protect against the adverse health effects protected by the future primary air quality standard?
Analytical Plan Use most recently certified 3-yr period 2004-2006 Look at ambient data, examine actual exceedences Predict ozone to examine potential exceedences (across time) Useful for monitors without year round data
Official Monitoring Seasons Current CFR vs. AQS Current CFR (mandated season) Generally, seasons are consistent within a state, excluding Texas and Louisiana (defined by AQCR) AQS (mandated & modified season) Seasons can be modified on a site-by-site basis, based on judgment of regional administrator Examples: California, Nevada, Arizona
May-Sep Mar-Oct Mar-Sep Mar-Nov Jun-Sep Apr-Nov Monitoring Season Apr-Oct Official Ozone Monitoring Seasons Where is the year round monitoring? Good spatial representation Year round monitoring Official seasons in AQS Monitoring Data in AQS Jan-Dec Apr-Sep May-Oct Wisconsin is April 15 - Sept 15
Using Ambient Data What do we see? Ambient, year round data from 531 sites (~45% of total) Examine number of observed exceedances (8hr daily max) using as much data as possible Full year Partial year Only within monitoring season With the data available, are we seeing exceedances occurring outside of a states official season?
Exceedances: in or out of season? Common (core) monitoring season across all states is June-Sept Months displayed on the following maps are the fringe months Feb-May (4 months before) Nov-Jan (4 months after) Are there out of season exceedances when the concentration threshold is lowered? Scenarios: 0.075 ppm 0.060 ppm* *indicator for the yellow AQI level
Ozone AQI Summary CategoryAQI Value Previous 8-Hour Ozone AQI (ppm) New 8-Hour Ozone AQI (ppm) 1-Hour Ozone AQI (ppm) Good0-500.000-0.0640.000-0.059 Moderate51-1000.065-0.0840.060-0.075 Unhealthy of Sensitive Groups 101-1500.085-0.1040.076-0.0950.125-0.164 Unhealthy151-2000.105-0.1240.096-0.1150.165-0.204 Very Unhealthy201-3000.125-0.3740.116-0.3740.205-0.404 Hazardous 301-4000.405-0.504 401-5000.505-0.604 Changes are in red Used in conjunction with one another
Using the new standard (0.075 ppm) to locate areas that may need seasonal modifications
Standard scenario: 0.075 ppm April March February in season exceedances in blue out of season exceedances in red May 4 months before common O 3 season 2/1-2/27 3/30-3/31 This situation doesnt in itself justify expanding the season for the entire month of March
OctoberNovember December Standard scenario: 0.075 ppm in season exceedances in blue out of season exceedances in red January 4 months after common O 3 season 1/24-1/31
Using the AQI yellow level (0.060 ppm) to locate areas that may need seasonal modifications
February March Scenario: 0.060 ppm April May in season exceedances in blue out of season exceedances in red 4 months before common O 3 season
Scenario: 0.060 ppm OctoberNovember DecemberJanuary in season exceedances in blue out of season exceedances in red 4 months after common O 3 season
Estimating ozone at existing sites when data is not available Predicting during the off-season months
Why Statistically predict ozone? We would like to fill temporal gaps where little or no ozone data are available Statistical model is used and tailored to accurately predict exceedance rates during off- season months* Off season month ranges are area specific but typically include months such as February, March, October and November *Assumes that relationship between ozone and meteorology during other months is similar to data used in fitting (do not use core months) Example for South Carolina Red dots are data in the predicted months
Statistical Prediction of Ozone (in non-monitored months) Case Study: Columbia, South Carolina South Carolinas official monitoring season: April - September We want to predict ozone during months outside of the official season Focusing on predicting for: February, March, and October Core months for ozone season: June, July, and August Ozone and meteorological relationships are different during core months, therefore we only use the surrounding (cooler) months (March, April, May, September, and October) in the model Using the cooler months is best as this better represents the kind of meteorology and ozone response that occurs during the months which we are trying to predict
About the model Urban area ozone data is combined with meteorological data (1997- 2006) Relationship is developed between maximum 8-hr ozone values and meteorology - Maximum 8-hr ozone is modeled as a function of daily meteorological variables (max temperature, humidity, etc.) Best predictions obtained when excluding summer months during the fitting process (June, July, Aug) Summer relationship is different from spring/fall/winter relationship Columbia has observed data for months which we are trying to predict (e.g., February, March, October, November) Use these data to validate our model predictions Results are shown for Columbia For more details: Camalier, L., Cox, B., and Dolwick, P., 2007. The effects of meteorology on ozone in urban areas and their use in assessing ozone trends. Atmospheric Environment 41, 7127-7137.
Example: Columbia, South Carolina Scatter plot of observed and predicted values for only the data used in the fitting process (March, April, May, Sept, Oct) Scatter plot of observed and predicted number of exceedances for all month/year combinations (2004-2006) with observed data Values in red not used in fitting process (February, November) Model Validation
exceedances > 75 ppb occur in months outside of current monitoring season (red bars) Jan 0 Feb 0 Mar 0.6 April 3.4 May 4.8 June 3.9 July 3.7 Aug 1.2 Sep 0.7 Oct 0.1 Nov 0 Dec 0 Columbia, South Carolina June, July, and August are not used in the fitting process, however they are behaving the way we expect
Using ambient & predicted data Case Example: South Carolina Season: April-September Used urban area with highest expected exceedences Ambient Data (2004-2006) On average, for 60 ppb ~10 exceedences/year between 2/15-3/31 We are predicting ~8 exceedences/year between February and March Feb: 1.2 March: 6.4 Predicted Exceedences, days above 60 ppb predicted months
Conclusions One can use ambient, existing data along with statistically predicted data to guide informed decisions Any modifications of the official season will be based on monitoring judgment and the results from this analysis
Other Questions? Contact: Louise Camalier Camalier.Louise@epa.gov (919) 541-0200 EPA,OAR,OAQPS,AQAD, Air Quality Analysis Group (RTP, NC)