EPA is considering a revision of the ambient daily maximum ozone standard to better protect public health. It is proposed to increase the averaging time from the current 1-hour to an 8-hour averaging time. The revised 8-hour standard would be in the range of 60-90 ppb, compared to the current 120 ppb for 1-hour standards.
USEPA has established a FACA subcommittee to advise the agency on certain areas of transition and implementation of the new standard(s). Workgroups are being established, several with relevance to the OTAG process, including transport, nonattainment designation, and local control strategies.
The proposed changes may occur during the OTAG process and it would be prudent to examine its impact in the OTAG process. Any change to the ozone standard, including the form of the standard (e.g. 8 vs. 1-hour averaging time) can affect the importance of the degree of transport and, therefore OTAG G policy.
In a Policy Paper, the OTAG Modeling and Assessment Subgroup has recommended that the evaluation of observational results and, perhaps, model projections should take into account the potential future ozone standards as well as the current standard.
The Air Quality Analysis Workgroup has identified the review of air quality data in comparison to the current standard and potential future standards, as an activity. This report, prepared for the OTAG Air Quality Analysis Workgroup, is in response to the expressed need.
The air quality analysis below, makes extensive use of the concepts of pattern and pattern analysis. Pattern analysis is a structured approach to the organization and presentation of air quality data. The ozone pattern analysis over the OTAG region is also based on the following physical considerations..
Back to Contents
The mission of OTAG is to identify control strategies and implementation options for the reduction of regional ozone over the eastern US The operational goals of OTAG are stated as (1) A general reduction in ozone and ozone precursors aloft throughout the OTAG region and (2) a reduction of ozone and ozone precursors at the boundaries of nonattainment areas.
The policy-relevant section of this report addresses the spatial pattern of nonattainment for 8-hour and 1-hour. It does not deal with the purpose of the 8 and 1 hour standards per se. Rather, it is intended to address the question:
Should the OTAG policy development process be concerned about the differences between the 8-hour and 1-hour ambient standard for nonattainment, in particular as it pertains to transport?.
The results of ozone pattern analysis indicates that there are discernible differences in the exceedence pattern based on the 8 and 1 hour standard. In particular, the regional (non-urban) ozone exceedences over the industrial states (from Illinois to Pennsylvania) are higher for the 8-hour standard. On the other hand, the 8-hour exceedences in the vicinity of New York City and Houston, TX. are reduced compared to the 1-hour standard. However, the general pattern of exceedences is similar for both measures. Hence, the significance of the 8 and 1 hour standard difference may depend on the particular policy issue at hand.
Back to Contents
The ozone data used in this report were collected from multiple
Data from each network were extracted and combined into a single integrated data set. The details of the data sources and quality control procedures are discussed in the "Preparation of Ozone Files for Data Analysis."
The first examination of average daily maximum ozone maps has revealed anomalous ozone "holes" and peaks at unexpected locations. For those sites the hourly and daily maximum ozone values were re-examined for possible inconsistencies. Sudden systematic changes in the ozone concentrations, as well as major deviation from neighboring sites were the main clues for anomalous behavior. As a result of this quality control process, 6 out of 709 monitoring sites were discarded. The remaining data were used in all the subsequent computations exactly as submitted by the networks.
The data processing has occurred in the following major steps:
The daily maximum 8-hour ozone concentrations were derived from the hourly data, as described presented in the section on data sources The 8-hour daily average was computed using the recommended EPA procedures. Running 8-hour averages are computed from the hourly ozone concentration data for each hour of the year and the result is stored in the first, or start, hour of the 8-hour period. An 8-hour average is considered valid if at least 75 percent of the hourly averages for the 8-hour period are available. In the event that only six (or seven) hourly averages are available, the 8-hour average is computed on the basis of the hours available using six (or seven) as the divisor.
There are 24 possible running 8-hour average ozone concentrations for each calendar day during the ozone monitoring season. The daily maximum 8-hour concentration for a given calendar day is the highest of the 24 possible 8-hour average concentrations computed for that day. This process is repeated yielding a daily maximum 8-hour average ozone concentration for each day with ambient ozone monitoring data. Because the 8-hour averages are recorded in the start hour, it is possible that the daily maximum 8-hour concentrations from two consecutive days may have some hourly concentrations in common. Generally, overlapping daily maximum 8-hour averages are not likely, except in those non-urban monitoring locations with less pronounced diurnal variation in hourly concentrations. An ozone monitoring day is counted as a valid day if valid 8-hour averages are available for at least 75 percent of possible hours in the day (i.e., at least 18 of the 24 averages).
Back to Contents
The resulting pattern of 8-hour and 1-hour average ozone is illustrated in Figures 1a, b, and c for an industrial site Hammond, IN, a near-metropolitan site in Greenwich, CT, and for a high elevation site, Whiteface Mountain, NY. It is evident, that both the 8-hour and the 1-hour moving average show the diurnal ozone cycle. However, the 8-hour moving average is smoother and void of multiple daily peaks. Also, the daily ozone amplitude is reduced by the 8-hour averaging: the daily maximum for the 8-hour average is less than the 1-hour average and the daily minimum is higher than the 1-hour average.
Another feature of the hourly comparisons is that the 8-hour average pattern is phase shifted by 4 hours back in time. Hence, the 8-hour daily average peak always occurs prior to the 1-hour peak. The reason for this phase shift is that the current EPA guidelines suggest to place the value of the 8-hour average into the first of the eight averaged hours, rather then into the center of the 8-hour window at the fourth or fifth hour.
Another consequence of placing the 8-hour average in the first hour is that in some instances the daily maximum 8-hour average may exceed the daily maximum for 1-hour average. This is counter-intuitive, since averaging should always reduce the peak values of a time-varying signal. Such anomalous behavior occurs generally at high elevation sites during rising ozone conditions at night and early hours of the next day. The phenomenon is illustrated in Figure 1c for Whiteface Mountain, NY, on August 1, 1995.
The daily maximum ozone is detected by examining the values for each 24 hourly concentrations. For reasons discussed above, the daily maximum derived from the 8-hour moving average is generally lower than the 1-hour daily maximum. A more detailed comparison of the two daily maximums is illustrated in Figures 2, 3, and 4. The daily time series are plotted for June, July, and August, 1995 at Hammond, IN, Greenwich, CT, and Whiteface Mountain, NY. The time series for each location is accompanied by a corresponding scatter-chart that compares the 8-hour and 1-hour values for each day of the season.
For Hammond, IN the ratio of the average 8-hour to the average 1-hour concentration is 0.85 and the slope of the linear regression is 0.86 (Figure 2a and b). Hence, for Hammond, IN, on the average the 8-hour daily maximum ozone is about 85% of the 1-hour daily maximum. On all 92 days in 1995, the 8-hour daily maximum ozone was less than the 1-hour maximum.
The Greenwich, CT site is located adjacent to New York City. The data show significant day-to-day variation (Figure 3a and b), with the 1-hour daily maximum exceeding the 8-hour average by a substantial margin (25% or more). This relationship is particularly evident in the scatter chart. During high ozone concentrations, exceeding 100 ppb the 8-hour average daily maximum is only 75% of the 1-hour daily maximum. Consequently, the slope of the regression line is 0.75, while the ratio of the averages is 0.82
The Whiteface Mountain, NY (Figure 4a and b) is a high elevation site at 1480 meters, that is exposed to ozone carried within the 1-2 km high planetary boundary layer. The site is like a sampling probe that sticks into the bulk of the planetary boundary layer, uninfluenced by surface effects. The main feature of this site is that the day-to-day variation is less significant. Also, the deviation between the 8- and 1-hour daily maximum values is less pronounced. The scatter chart further reveals that a substantial fraction of the daily values exhibit higher 8-hour than 1-hour maximum values. The reason for this behavior has been discussed above and it is related to the diminishing of the diurnal cycle at the higher elevation site. At Whiteface Mountain, NY the ratio of the average 8-hour and 1-hour values is 0.92 and the slope of the regression line is 0.82.
In summary, the above illustrative examples show that the 8-hour daily maximum ozone may range from 75% to 95% of the 1-hour daily maximum, sometimes exceeding the 1-hour values. The lower values can occur in the vicinity of ozone source areas where the hourly concentrations vary significantly and the 8-hour filter tends to diminish the hourly peak values. On the other hand, at remote high elevation sites, which exhibit little variation from one hour to another, the 8-hour filter does not change he signal substantially. Furthermore, at the high elevation sites there is an anomalous behavior, such as that the 8-hour daily maximum is frequently higher than the undampened 1-hour signal. This is attributable to the placement of 8-hour average values into the first, rather than in the fourth or fifth hour.
This section will examine the statistical relationship between 8-hour and 1-hour ozone values for different subregions of the OTAG domain.
The overall relationship between the two daily measures of ozone throughout the OTAG region is shown in Figure 5. The scatter chart contains all available daily data for June, July, and August, between 1991 and 1995. Thus, it represents the average 8-hour - 1-hour relationship for the entire OTAG region. The average 1-hour ozone is 59 ppb, while the average 8-hour daily maximum ozone is 51 ppb, i.e. 86% of the 1-hour daily maximum. The correlation coefficient for the linear regression line is 0.96, with the slope of 0.83 and offset of 1.4 ppb.
In order to detect the regional differences in 8 and 1-hour measures, the same correlation plots were produced for four sub-regions of OTAG, denoted as Northeast (NE), Northwest (NW), Southeast (SE), and Southwest (SW) (Figure 6a, b, c, and d). The Northeast and Southeast regions exhibit similar correlations. The ratio of 8-hour and 1-hour averages is about 0.85 and the slope of the regression line is also 0.85. The Northeast scatter-chart also indicates that a substantial fraction of the 8-hour daily maximum values exceed the 1-hour daily maximum ozone.
The northwestern region covering the Midwest and Upper Midwest shows somewhat higher ratio of 8-hour to 1-hour averages (0.87), and slope of the regression line is 0.85. It is not clear whether these deviations are significant. However, it can be clearly stated that the southwestern OTAG region differs significantly from the other regions. The average 1-hour (54 ppb) and 8-hour (45 ppb) concentrations are lower than the rest of the OTAG region. The ratio of 8-hour to 1-hour averages is only 0.83. Furthermore, the slope of the best fit regression line is only 0.79 compared to 0.85 in the entire regions.
In summary, throughout the OTAG region, on the average, the 8-hour daily maximum is about 85% of the 1-hour daily maximum values. Regionally, the Northeast and Southeast sub-region of OTAG are virtually identical to the overall average, without distinguishing characteristics. However, over the Southwest the 8-hour ozone is only 80-83% of the 1-hour values, while over the Northwest is 86-87% of the 1-hour values. Hence, using 8-hour rather than 1-hour daily maximum ozone metric, the Southwest would show lower values, while the Northwest higher daily maximum values.
The purpose of this section is to examine the spatial pattern of exceedences as measured by the current 1-hour daily maximum standards and by the proposed 8-hour standard. In particular, the spatial differences in the spatial exceedence pattern between the two ozone metric will be evaluated.
The spatial pattern of exceedences was obtained by calculating the average number of daily maximum values that exceeded a threshold value. For the 1-hour daily maximum the current standard is 120 ppb. For the new 8-hour standard, a threshold range of 70-90 ppb is being considered. In the following analysis, we have taken 80 ppb as the working value for the 8-hour daily maximum threshold.
The calculation of the number of exceedences is sensitive to the number of missing observations. For this reason, it was required that every monitoring station has valid daily maximum values for at least 75% of the observations. Furthermore, for the accepted stations the exceedences were scaled upward in proportion to the missing data.
A spatial comparison of 1-hour and 8-hour exceedences is made difficult by differing absolute magnitudes. The 1-hour exceedences with 120 ppb threshold yields a range of 0-4 exceedences per year. On the other hand, the exceedences for 8-hour daily maximum ozone with 80 ppb threshold ranges up to 12 exceedences per year. In order to make the two metric comparable, the threshold levels were adjusted to yield roughly comparable average exceedences. In doing so, the results of the statistical analysis were utilized which showed that on the average the 8-hour maximum ozone is 85% of the 1-hour values.
In the analysis that follows, the 1-hour daily maximum ozone exceedences with 120 ppb threshold were compared to the 8-hour daily maximum values with 102 ppb (120*0.85) threshold. Conversely, the 8-hour daily maximum ozone with 80 ppb threshold was compared to the 1-hour values with 94 ppb (80/0.85) threshold.
The spatial pattern of 1-hour exceedences (>120 ppb) is shown in Figure 7a. The corresponding 8-hour exceedences (>102 ppb) is shown in Figure 7b. The maps clearly indicate that the exceedences of 1 or more days per year are confined to the vicinity of major metropolitan areas. The most pronounced exceedence region is the Washington, DC - Boston, MA corridor in the Northeast. Additional exceedence hot-spots are evident downwind of Chicago, Atlanta, Houston, and Dallas. Further exceedence regions are present for St. Louis and the cities in the Ohio River Valley.
The gross features of the 8-hour, (>102 ppb) exceedences (Figure 7b) are similar to the corresponding 1-hour exceedences, in that they occur over the same set of metropolitan areas. A more detailed comparison is given below.
The spatial pattern of 1-hour (>94 ppb) and 8-hour (>80ppb) exceedences is shown in Figure 8a and b. The number of exceedences at this threshold level is more than 12 days per year at urban industrial areas. The largest number of the exceedences occur at the Washington-Boston corridor. Additional hot-spots are visible for Atlanta, Houston, Dallas, St. Louis, and Chicago. A broad area of high ozone exceedences are also evident over the industrial states stretching from Illinois to Pennsylvania, just north of the Ohio River Valley.
A comparison of the exceedence pattern for 1-hour and 8-hour daily maximum ozone shows very similar spatial distribution. In fact, differences between the 1-hour and 8-hour exceedence pattern can only be discerned after numerical manipulation of the data. Figure 9 shows the map of the arithmetic difference between the 8-hour and 1-hour exceedence measures. Areas in red mark the geographic regions where the 8-hour measure produces higher exceedences than the corresponding 1-hour metric. Conversely, the blue areas indicate the regions where the 8-hour metric yields less exceedences. The exceedence difference map (Figure 9) shows that throughout the industrial Midwest from Illinois through Pennsylvania the 8-hour exceedences will be higher than the corresponding 1-hour values. On the other hand, in the areas of higher hourly variability, in the vicinity of Houston and adjacent to New York City, the 8-hour exceedences will be reduced compared to the 1-hour standard.
In summary, the spatial pattern of 1-hour and 8-hour exceedences clearly points to large metropolitan areas and the industrial Midwestern states as the predominant exceedence zones. Switching from the 1-hour to the 8-hour exceedence metric yields an increase of exceedences over the industrial Midwestern states from Illinois to Pennsylvania and a decrease of exceedences around Houston and New York City. However, the overall exceedence pattern for 8-hour and 1-hour metric is rather similar.
Ozone exhibits strong day to day variation throughout the summer. For this reason, the time-averaged concentration is an inadequate measure of spatial pattern. Since the shape of the ozone distribution function of concentrations is also variable from one region to another, it is beneficial to examine the ozone distribution using non-parametric statistics, i.e. percentiles.
The ozone concentration percentiles for each monitoring station were calculated using June, July, August data during 1991-1995. Data completeness condition of 25% was applied, such that during the 5 summers (5*90=450 days) at least 112 days had valid daily maximum ozone concentration data. The spatial maps were obtained by computing the percentile for each station and contouring the values using standard procedures. Such maps were generated for 10, 50, and 90 percentiles. The three percentile maps are shown in Figures 10a, b, and c.
The 10th percentile map (Figure 10a) represents the 8-hour daily maximum ozone concentration during the cleanest 10% of the days. At the edges of the OTAG region, 10th percentile ozone daily maximum concentration is below 20-30 ppb. Elevated 10th percentile values, in excess of 35 ppb extend through the industrial Midwest, from Missouri through Illinois, Indiana, Ohio, Kentucky and Virginias. Throughout this region the 10th percentile ozone concentration varies by about 5 ppb, 35-40 ppb. In other words, there is a persistent and spatially uniform area that has about 15 ppb excess over the tropospheric background even during the cleanest days. Conspicuously absent are higher concentrations in the 10th percentiles for major metropolitan areas. There is virtually no concentration difference between an urban area and its surrounding. This implies that during ozone-free periods both urban and rural areas are well ventilated, preventing the occurrence of local urban ozone buildup.
The 90th percentile map given in Figure 10c is substantially different from the 10th percentile. The ozone concentration during the highest 10% of the days is elevated mostly in the vicinity of major metropolitan centers. In particular, the entire eastern seaboard from Virginia to Rhode Island is covered with high 8-hour daily maximum ozone levels (>80 ppb) at least 10% of the days. Additional high 90th percentiles in excess of 80 ppb can be observed for Dallas-Ft. Worth, Atlanta, as well as downwind of Chicago, on eastern shores of Lake Michigan. Southern Indiana and Ohio also indicate elevated 90th percentile ozone. Hence, the dramatic difference between the 10th and 90th percentiles is that the 90th percentile is highest near urban metropolitan centers, while the 10th percentile is highest over the industrial Ohio River Valley. It is remarkable that the southern Appalachian Mountains have low 90th percentile ozone. In fact, the entire spine of the Appalachian Mountains has lower 90th percentiles than their surrounding. Evidently, high ozone concentration that accumulate in lower lying areas do not reach the higher elevation Appalachian sites. A further complication arises in examining the high elevation sites, such as Whiteface Mountain, NY. It shows higher concentrations than their surrounding low lying areas. This pattern is opposite to the high elevation pattern over the southern Appalachians. A full understanding of the high-low elevation difference can only be reached if the vertical ozone profiles are better understood.
The 50th percentiles daily maximum ozone (Figure 10b) shows a pattern that is a in-between of the 10th and the 90th percentiles. The elevated 8-hour daily maximum ozone concentrations over industrial Midwest, just north of the Ohio River coexist with the high 50th percentile concentrations over the eastern megalopolis. Virtually all major metropolitan areas show higher values than their regional surrounding, but the urban excess is much less than for the 90th percentile.
In summary, the 10th percentile daily 8-hour maximum ozone is low (20 -30ppb). Throughout the OTAG region except over the industrial Midwest from Illinois to Virginia. The 90th percentile shows the highest values surrounding the major metropolitan areas.
The magnitude of the day to day ozone variation is reflected in the width of the concentration frequency distribution functions as measured by the 90 - 10 percentile difference. Monitoring sites that have highly variable ozone levels from one degree to another have a broad distribution function with large differences between the upper and lower percentiles.
Ozone variability can be viewed as an indication of nearby ozone/precursor sources and sinks. Monitoring sites that show great day to day variability tend to be in the vicinity of fresh ozone sources. Where urban/industrial plants perturb their surrounding areas. Sites with small day to day variability tend to be background sites exposed to spatially more homogeneous and temporarily uniform background concentrations.
The map of the difference between the 90th and 10th percentiles is shown in Figure 11. The highest variability (90-10th percentile difference) of 8-hour daily maximum ozone concentration occurs in the vicinity of metropolitan areas. In particular, in Washington-New York megalopolis, Houston, Dallas-Ft. Worth and Atlanta exhibit the highest range with the 90th percentile exceeding the 10th percentile by 50 ppb. Virtually all major metropolitan areas tend to cause high ozone variations within their neighborhoods. A further, broad moderately variable ozone region stretches throughout the industrial Midwest from Illinois to Pennsylvania.
It is interesting to observe that the corners of the OTAG region covering southern Florida, Texas Gulf Coast, Upper Midwest, and Maine show the least variance in ozone concentration. This is consistent with the notion that background monitoring sites, sampling aged tropospheric ozone, do not have high day to day variation. It is also interesting to observe that over the spine of the Appalachian Mountains the concentrations are much less variable than over their surrounding low lying areas.
In summary, the spatial pattern of 8-hour daily maximum summertime ozone over the OTAG region is highly variable, particularly in the vicinity of populated metropolitan centers. The urban impact is virtually undetectable during clean days (10 percentile of the ozone concentration). However, the urban influence during high ozone levels (90th percentiles) is very pronounced, but confined largely to the vicinity (few hundred miles) of metropolitan areas. It is therefore evident, that in establishing and quantifying the urban influence on near surface ozone concentrations, the upper percentiles of the ozone concentration are more sensitive indicators than lower percentiles. A puzzling feature of the 10th percentile maps is the elevated ozone region in the Ohio River Valley region. It is evident that it is not tropospheric background, nor urban ozone. Could it be due to major point sources in the Ohio River Valley?
Back to Contents
|Submit your comments, feedback, questions, and ideas pertaining this page. Your input will be automatically added to the existing annotations. In order to add a new comment, you must be registered with the OTAG/AQA Peoples Page.|