Historical Perspective on the Climatological Potential for "Local" Pollution Episodes

Submitted to the OTAG Air Quality Analysis Workgroup

by R. Poirot and P. Wishinski, VT DEC

February 3, 1997


A necessary precondition for the transport of "high" concentrations of ozone or ozone precursors from a "source area" to a "receptor area" is a degree of "local" pollution accumulation in or near the source area. Key factors which influence local pollution accumulation include local emissions density and local meteorology (especially mixing depth and wind speed). During the 1960's and early 1970's, there was considerable interest in evaluating the long-term "climatological" potential for local pollution episodes. Many of these historical evaluations were conducted strictly from a meteorological perspective - i.e. without regard to emissions densities. As such, they can provide a useful complement to recent OTAG modeling and Air Quality Analysis results. In particular, they may provide useful insights into the following general OTAG AQA observations:

  1. Importance of local stagnation in Southern and Midwestern sections of the OTAG domain,
  2. Absence of high peak 1-hour ozone levels in Midwest (in apparent conflict with #1 above),
  3. Importance of the Midwestern source region as contributor to multiple OTAG receptor areas.



The airmass over a given "source" location does not reside indefinitely over that location (fortunately), but moves eventually over other "downwind" locations. "Airmass transport" does not necessarily result in a substantial contribution to downwind ozone concentrations, as physical and chemical processes may diminish ozone and/or precursor (O&P) concentrations en route between source and receptor. However, with the exception of ozone formed en-route from source area precursors, there are no other mechanisms by which concentrations of "transported" ozone concentrations can increase during transport. If the initial concentrations of O&P are "low" at the time the airmass leaves the source region, the net effect of transport would be to dilute the concentration at the receptor. Consequently, some degree of build-up or accumulation of O&P in or near a source region is a necessary pre-condition for any "substantial" degree of transported ozone at a downwind receptor. Thus "substantial" ozone transport can be conceptually viewed as a two-stage process: 1. local accumulation, followed by 2. airmass transport.

Back to Contents

Climatology vs. Episodes

Many of the recent OTAG Air Quality Analysis studies are based on long-term (5-10 year) "climatological" analyses 1-5. Over time, the seasonal patterns of ozone and meteorological conditions exhibit reasonably consistent and predictable patterns from year to year; however, short-term variations in ozone and meteorology are statistically random with respect to time, and are unpredictable beyond more than a few days in the future1. Specific meteorological conditions associated with a single historical episode will never occur again. Consequently, episodic modeling results provide a limited basis for prediction of meteorologically-dependent aspects of future conditions (for example: the duration of local stagnation/accumulation in specific source regions, or the distance, direction or persistence of subsequent downwind transport). Relationships derived from long-term data sets provide a more robust basis for probabilistic predictions of future conditions. In the absence of substantial long-term changes in climate or emissions, the meteorology (and ozone) over a 5-10 year period in the future may be probabalistically anticipated to exhibit similar characteristics to a 5-10 year period in the past.

Back to Contents

Consistencies and Inconsistencies

The historical "wisdom" on the nature of (and required control strategies for reducing) ozone pollution has presumed that ozone is a local pollutant, responsive primarily to changes in ground-level VOC emissions in urban areas. A recent review of (episodic) ozone modeling studies suggested that NOx emissions controls in general, and stationary NOx controls in particular are relatively ineffective at reducing ozone concentrations13. In contrast, recent OTAG model results suggest that NOx reductions are relatively much more effective than VOC reductions over broad areas of the Eastern US for selected episodes. Other evaluations suggest that the OTAG model results may overstate NOx control benefits - for example through overestimates of biogenic VOC concentrations14 and/or artificially enhanced mixing, resulting from use of excessively large (12 km) grid squares.15

One recurring pattern observed in several recent OTAG climatological analyses 2,3,4 is that the Midwest (a region characterized by relatively high emissions of NOx from stationary sources) appears to be an important ozone source region - or is at least frequently and persistently upwind - for high ozone concentrations in a variety of surrounding receptor areas. In apparent paradox, the Midwest itself does not experience a high incidence of exceedances of the current, 1-hour ozone standard.2

Back to Contents

Why the Midwest?

That the Midwest is frequently upwind of high ozone concentrations at a variety of downwind receptors is clear. However, specific reasons or mechanisms by which emissions from this source region might contribute to high ozone levels in other regions remain unclear. In addition to high NOx emissions densities, other possible explanations suggested in OTAG AQA discussions include:

  1. Central Location - air masses at locations throughout OTAG have a higher "geometric"probability of having previously resided toward the center of, than at the periphery of the OTAG domain.
  2. Large Size - the Midwest (like any individual urban area) appears to act as a coherent source region, but is spatially larger than any urban area, and so has a higher probability of experiencing forward or backward "trajectory hits". As with flows passing over the elongated East Coast urban corridor, these Midwestern trajectory hits are likely to traverse a relatively long regional cross-section.
  3. Unique Source Configuration - The Midwestern region includes a number of moderately large urban centers, but also includes uniquely high densities of stationary NOx sources. The tall stacks physically separate this NOx from (ground-level) VOC emissions - slowing the rates of ozone formation (and subsequent destruction). Elevated releases are also subject to higher, more directionally persistent wind speeds, and to a lower potential for physical and chemical destruction of O&P at the earth's surface.
  4. Continuous Emissions - Utility emissions exhibit less of a diurnal and day-of-week cycle than do mobile and area source emissions (consistent with high 10th percentile concentrations, high 8-hour concentrations and lack of weekly cycle in Midwestern ozone levels). Consequently the Midwest is always, chronically "there" as a potential source region; whereas urban centers are periodically cleaned out (low 10th percentile) and have much higher weekday/weekend ozone ratios.2
  5. Aerosol Enhancement - Recent measurements/ analyses from the NARSTO-NE field study suggest that sulfate aerosols (which Midwestern emissions affect) can significantly accelerate ozone production.16
  6. Meteorology - The highest Midwestern (and Southern) ozone concentrations are typically associated with variable, meandering wind directions and low wind speeds (local stagnation). Lowest Midwestern ozone concentrations are higher than in other sections of OTAG and are associated with ventilating flows which subsequently pass through other OTAG regions.3

Back to Contents

Historical Climatological Evaluations

During the 1960's and early 70's - when national air pollution control efforts were just getting under way - there was also a considerable interest in the evaluation of climatological conditions associated with high pollution concentrations 6-12. Many of these earlier climatological studies focused exclusively on the space and time patterns of meteorological conditions conducive to the formation of high pollutant concentrations - without regard to actual areal emissions densities or reaction chemistry. They were all based on long-term (5 to 30 year) meteorological data sets, and evaluated spatial and temporal patterns of variables like: frequency of low-level inversions and wind speeds 6, diffusion conditions 7, mixing depth 8, stagnating anticyclones 9, and various combinations of the above.10 Many of these "pollution-conducive" meteorological factors were originally observed as common conditions in a number of severe, localized (or regional) pollution episodes in the 1940s, '50s and '60s (Donora, PA, 1948; London, 1952, '56, '57, '62; NY City, 1953, '63, '66; etc.). In turn, the subsequent identification of key meteorological factors led to development of a routine, national-scale air pollution episode forecast program in the 1960s.11

Back to Contents

Selected Results from Holzworth (1972) 12

Among the most comprehensive of the historical climatological analyses was an EPA publication entitled "Mixing heights, wind speeds, and potential for urban air pollution throughout the contiguous United States" (Holzworth, 1972).12 The focus was on identification of spatial and seasonal patterns of meteorological conditions conducive to the occurance of local urban or community-scale pollution episodes. Actual urban locations, emissions densities and chemical reactions and inter-regional transport were not considered. The estimated episode potential was for non-reactive ("slowly reacting") pollutants. The meteorological calculations were based primarily on a record of morning and afternoon mixing heights and wind speeds from a national network of 62 upper air NWS sites for the 5 year period of 1960 - 1964. Holzworth also presented results of "number of forecast days of high meteorological potential for air pollution in a five year period" - which for the eastern US, was based on a 10 year (1960-1970) record of predictions from the National Air Pollution Potential Forecast Program.

Although the meteorological data sets are dated - in excess of 30 years old, there is no reason to assume that current or future climatic conditions will differ substantially from the 1960s. Because of the focus on potential for episodes of non-reactive pollutants, the relevance to ozone episodes relates only to the initial - local accumulation stage - of ozone and precursors, and not to any subsequent influences of transport or removal processes. Selected results are presented below in Figures 1 through 6. Holzworth's original graphics were electronically scanned, cropped to focus on the OTAG domain, and colored to add clarity.

Mean Summer morning mixing heights for the OTAG domain are displayed in Figure 1. The most distinct feature of this plot is the relatively high mixing heights (500 to > 1,000 meters) along the Atlantic and Gulf coastal areas. There is a relatively steep gradient of decreasing morning mixing heights moving inland from the moderating influences of warm nocturnal ocean temperatures.

Lower morning mixing heights (< 500 meters) prevail throughout the interior of the OTAG domain, with the lowest mixing heights evident in the Northwest. The most easterly extent of relatively low morning mixing heights (< 400 meters) includes areas of the Southeast (GA, SC) and Midwest (KY, OH, WV).

By mid-afternoon, substantially higher mixing heights prevail throughout the East (Figure 2), and are greatest - in excess of 2,000 meters - along the western edge of the domain. A broad region of high mixing heights - 1,800 to 2,000 meters - is also evident throughout much of the South and in a long South-to-Northeast corridor aligned approximately with the Appalachian Mountains. Lower mixing heights - less than 1,600 meters are evident in the Northwest, and especially along the Atlantic and Gulf coasts.

These coastal areas (with relatively high morning mixing heights) experience a limited (2X) growth in mixing height from morning to afternoon; whereas some interior sections are characterized by a greater than 5-fold expansion of mixing height as the day progresses.

Along with mixing depth (volume), wind speeds are also exert an important influence on the potential for local pollutant accumulation. For a given emission rate (grams/second) and mixing depth (meters), a lower wind speed (meters/second) increases the quantity of pollutant released into a given volume of air. Figure 3 shows mean summer morning wind speeds, averaged through the morning mixed layer. The highest morning wind speeds - from 5 to > 7 meters/second - are typically found in western (plains) sections of the OTAG domain. Lowest Summer morning wind speeds - less than 4 meters/second - occur in a broad band just west of the Appalachians from the central Gulf coast up through OH and PA. An area of minimum morning wind speed (< 3 m/s) is located near the intersection of the OH/KY//WV/PA borders.

Mean Summer afternoon wind speeds (Figure 4) are typically 2 to 3 meters per second faster than morning conditions. Lowest afternoon speeds (< 5 m/s) occur in a broad region extending from the Gulf Coast up through the southeastern Midwest. Highest afternoon wind speeds (> 7 m/s) tend to occur in Western and New England sections of the OTAG region.

Various combinations of low mixing heights, low wind speeds and their persistence, in the absence of precipitation, over 2 or more days were observed during many of the most severe historical pollution episodes. Projected occurrences of these conditions formed the basis for forecasts issued by the National Air Pollution Potential Forecast Program.11

Figure 5 displays a count of the number of forecast days of high meteorological potential for air pollution in a five year period from the National Air Pollution Potential Forecast Program. Within the OTAG domain, these forecasted episode frequencies range from less than 10 (< 2 days/year) to greater than 40 (> 8 days/year). The highest projected episode frequency includes an area just west of the Appalachian Mountains, extending northward from central Georgia and Alabama up through West Virginia and Ohio. The maximum forecasted episode frequency occurs in an area near the intersection of the Ohio/Kentucky/West Virginia borders.

The forecasts in Figure 5 are based on annual data, and include a number of forecast days in the Fall, Winter and Spring which are not relevant to ozone episodes. Figures 1, 2, 3, 4 and 6 are based on Summer-only data.

Holzworth also developed estimates of potential pollutant concentration ( ) as a function of a normalized emission rate (Q) for hypothetical urban areas with a sizes of 10 and 100 km. These calculations were derived through application of a simple dispersion model with wind flows perpendicular to a series of line sources for a distance of 10 or 100 km. Figure 6 displays upper quartile Summer morning X/Q values for a 100 km city size. Also displayed (but not color shaded) are similar (dashed) isopleths for a smaller hypothetical city size of 10 km.

A common feature of Figure 5 and 6 is an area with high meteorological potential for local pollution accumulation along the western edge of the Appalachian Mountains, with the highest values in both cases in an area near the intersection of the OH/KY/WV/PA borders.

Back to Contents


The Midwestern area identified in Figures 5 and 6 as having a high (1960's) meteorological potential for local pollution episodes is also characterized by very high (1990's) stationary source NOx emissions densities (Figure 7). The same general region is also identified by (1990's) backward trajectory calculations4 as being persistently upwind of high ozone concentrations at a variety of receptor sites throughout the OTAG domain (Figure 8). However, the Midwest itself does not currently experience frequent exceedances of the 1-hour ozone standard (Figure 9). How can this be?

The Holzworth (and other 1960's) climatological estimates of local episode potential were developed for "slowly-reacting" pollutants - which typically build to maximum concentration (under low mixing heights and low wind speeds) in the early morning hours. In the above Figures 1-6 from Holzworth, the Midwest is identified as having a high potential for pollution episodes primarily in the "morning" plots (Figures 1, 3 and 6). Ozone and most ozone precursors are "quickly-reacting" pollutants. In the absence of sunlight, precursors may build to high local concentration during the early morning hours, but ozone itself is predominantly an "afternoon" pollutant. Holzworth's plot of upper quartile Summer morning X/Q concentrations (Figure 6) indicates a high pollution potential in the Midwest. However, a similar calculation of Summer afternoon X/Q values (not shown here) projects afternoon X/Q concentrations which are an order of magnitude lower than the morning values, and which exhibit almost no spatial variation throughout the eastern US. Note also from Figures 1 and 2 that while morning mixing heights in the Midwest are typically quite low (300-400 meters), they increase to roughly 5 times this height (< 1,800 meters) by mid afternoon (when they are typically higher than in surrounding regions).

So while morning precursor concentrations may build to high concentration in a shallow mixed layer, afternoon ozone production takes place within a substantially expanded volume. This "ballooning" of the Midwestern mixing depth may help explain the relative absence of high 1-hour ozone concentrations, despite high precursor emissions and low wind speeds. These factors may also help explain the persistence of relatively high chronic concentrations (high 10th percentile and high 8-hour averages), despite the absence of high 1-hour concentrations. As the Midwestern mixing height descends rapidly in the evening hours, a high fraction of the preceding day's accumulated ozone is preserved aloft, protected from chemical and physical destruction at/near the earth's surface. This deep layer or "reservoir" of moderately, but chronically high nocturnal ozone aloft can be subsequently subject to higher wind speeds which can transport the "substantially" high ozone concentrations to other regions.

The subsequent transport from this centrally-located region can and does occur in many directions, as wind directions in the central eastern US can be quite variable from one episode to another (but may tend to be more uni-directional during a specific episode). Figure 11 from Schichtel and Husar (1997,3c shows transport vectors and residence times derived from 1991-95 NGM meteorological data. The met data have been sorted at each location on the map to reflect the average conditions during high ozone events (the highest 10% of ozone concentrations at each location). The color shaded "residence times" are approximately equal to the inverse wind speed. The vectors show the average wind direction, while the vector length reflects the persistence of the wind direction (during high ozone events). The very short vector length in the Midwest indicates varied wind directions during high ozone events. Locations around the periphery of the Midwest show vectors pointing outward from this central location. Note also that the residence times show a spatial pattern (from 1990's data during high ozone conditions) which is quite similar to the (Figures 3 and 4, 1960's-based) wind speed plots from Holzworth.12

Thus, from a climatological perspective, wind speeds derived from a 5-year 1960's data set appear similar to (and relevant to) wind speeds during high ozone events during a 5-year period in the 1990's. Summer wind speeds are low in the South and Midwest. Wind directions from the Midwest during high ozone events are variable from episode to episode, with high ozone episodes in areas surrounding the Midwest in all directions frequently associated with flows from the Midwest.

Figure 12 shows similar transport vectors and residence times based on NGM data during the July, 1991 OTAG model episode. Unlike Figure 11, it includes periods and locations of relatively low ozone during model "ramp-up" days, and on "off-peak" days during the model period. Note, however, that residence times in the Midwest are generally shorter for this episode than in the Figure 11 plot of "climatologically defined episodes". Note also that the transport vectors from the Midwest are longer (more uni-directionally persistent) for the individual episode. Thus any directional implications of ozone transport based on any individual episode may reflect only part of the story, and should not be viewed as "predictive" of future conditions.

Back to Contents


  1. Long-term, climatological evaluations of meteorological conditions provide a more robust basis than individual episodes for prediction of long-term meteorological conditions in the future.
  2. Transport of "substantial" concentrations of ozone pollution requires an initial buildup or accumulation of ozone and/or precursors in/near a "source region".
  3. Historical climatological evaluations of upper air data indicate distinct spatial variations in the meteorological potential for local pollution accumulation during the Summer in the Eastern US.
  4. The Southern and Midwestern sections of the Eastern US typically experience combinations of low wind speeds and low morning mixing heights which are conducive to local pollution accumulation.
  5. The highest meteorological potential for accumulation of high morning pollutant concentrations occurs in a Midwestern region centered near the convergence of the OH/WV/PA/KY borders.
  6. The Midwest experiences a relatively extreme (5 fold) diurnal variation in Summer mixing depth, which - in combination with low morning and afternoon wind speeds and low morning mixing depth - can lead to the formation of a deep reservoir of moderately but chronically high ozone concentrations.
  7. During/following periods of local stagnation in the Midwest, wind directions from the Midwest are highly variable from episode to episode.
  8. Regardless of emissions, these unique meteorological conditions in the Midwest can help explain:
    1. Absence of extreme peak 1-hour ozone concentrations,
    2. Presence of relatively high 8-hour average ozone concentrations,
    3. Importance of the Midwest as a large, centrally located, chronic source region in which ozone can accumulate to high concentrations, and from which "substantial" contributions to ozone transport can occur in a variety of downwind locations.

Back to Contents


1. Porter, P. S., S.T. Rao, I. Zurbenko, E. Zalewsky, R.F. Henry and J.Y. Ku (1996) at


2. Husar, R. B. (1996a,b,c,d) at





3. Schichtel, B. and R. B. Husar (1996a,b,c) at




4. Poirot, R. and P. Wishinski (1996a,b,c,d) at





5. Rao, S. T. and E. Brankov (1997) Role of atmospheric transport on ozone concentrations at Whiteface Mt., presentation to OTAG AQA Workgroup, Washington, DC.

6. Hosler, C. R. (1961) "Low level inversion frequency in the contiguous United States", Mon. Weather Rev. 89: 319-339.

7. Hosler, C. R. (1964) "Climatological estimates of diffusion conditions in the United States", Nuclear Safety 5: 184-192.

8. Holzworth, G. C. (1964) "Estimates of mean maximum mixing depths in the contiguous United States". Mon. Weather Rev. 92: 235-242.

9. Korshover, J. (1967) "Climatology of stagnating anticyclones east of the Rocky Mountains, 1936- 1965", Public Health Service Publication No. 999-AP-34, Cincinnati, OH.

10. Holzworth, G.C. (1967) "Mixing depths, wind speeds, and air pollution potential for selected locations in the United States", J. Appl. Meteor. 6:1039-1044.

11. Gross, E. (1970) "The national air pollution potential forecast program", EESA Tech. Memo. WBTM NMC 47, National Meteorological Center, Suiteland, MD.

12. Holzworth, G.C. (1972) "Mixing heights, wind speeds, and potential for urban air pollution throughout the contiguous United States", Office of Air Prog. pub. AP-101,USEPA, RTP, NC.

13. Morris, R.E. (1995) "Review of Recent Ozone Measurement and Modeling Studies in the Eastern United States Nov 30 1995 Draft", ENVIRON Corp., Navato, CA 949455010.

14. Edgerton, E. S., (1996) Presentation to OTAG AQS Workgroup on: Comparison of isoprene concentrations from UAM-V and ambient measurements at selected SOS and NARSTO-NE sites.

15. Imhoff, R. E. (1996) "Preliminary proposal: evaluation of the effect of the use of a 12 kilometer grid on otag modeling results", at http://capita.wustl.edu/otag/reports/mod12km/ufgrid12.html

16. Dickerson, R., S. Kondragunta, G. Stenchikov and W. Ryan (1996) "The impact of aerosols on photochemical smog formation", Presented at the First NARSTO-NE Data Analysis Symposium and Workshop, Norfolk, VA.

Back to Contents


Figure 1 Mean Summer Morning Mixing Height (m) 12

Figure 2 Mean Summer Afternoon Mixing Height. (m) 12

Figure 3 Mean Summer Morning Wind Speed (m/s)

Averaged through the Morning Mixed Layer 12

Figure 4 Mean Summer Afternoon Wind Speed (m/s)

Averaged through the Summer Afternoon Mixed Layer 12

Figure 5 Number of Forecast Days per 5 Years with

High Meteorological Potenential for Air Pollution 12

Figure 6 Upper Quartile Summer Morning X/Q Values

(normalized concentration) for a 100 km City Size 12

Figure 7 1991 Point Source NOx Emissions from EPA Interim Inventory

Figure 8 Areas with Deviations > 0.3 GSD Above Mean,

Averaged for 23 Trajectory Sites 4

Figure 9 Areas with High 1-Hour Ozone 2

Figure 10 Areas with High 8-Hour Ozone 2

Figure 11 Transport Vectors and Residence Times for

Upper 10% Ozone Days, Summers: 1991-95 3

Figure 12 Transport Vectors and Residence Times for

July 1991 OTAG Modeled Episode 3

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.