Jump to main content.


Trend Analyses

Note: EPA no longer updates this information, but it may be useful as a reference or resource.


Table of Contents

Objectives and Trend Analysis Procedure
Selecting an Indicator
Adjusting for Meteorological Variability
Meteorological Variables Potentially Associated with Ozone Formation
Classification Methods
EPA Meteorological Trend Adjustment Method
SUNYA Trend Analysis Method
Detecting Trends
Summary
References

 [Workbook Table of Contents] [Top of Trend Analyses] [Previous Section] [Next Section]


TREND ANALYSES

One of the PAMS objectives is to provide data to assess trends in ozone and ozone precursor concentrations over time and reconcile these data with trends in precursor emissions. 

  • Trend Analysis Procedure
  • Select indicator (summary statistic).
  • Adjust indicator to remove meteorological influences.
  • Apply statistical procedures for detecting trends.
  • Evaluate the trend for direction, rate of change, etc.

 [Workbook Table of Contents] [Top of Trend Analyses] [Previous Section] [Next Section]


SELECTING AN INDICATOR

Examples for annual summary statistics include the following:

Ozone

  • Percentile of daily maximum 1-hr concentration (i.e., 99th, 95th, 90th)
  • Second highest daily 1-hr maximum
  • Seasonal mean or median of daily maxima
  • Number of annual exceedance days or hours
  • Average of highest 30 maximum daily concentrations

Hydrocarbons

  • Mean 6-9 a.m. total NMHC concentrations
  • Mean 6-9 a.m. individual chemical species concentrations or species ratios

Useful Initial Information

  • Determine number, location, magnitude, time of year, and time of day of ozone exceedances in a region

 [Workbook Table of Contents] [Top of Trend Analyses] [Previous Section] [Next Section]


Figure 1

Number of Exceedance Days in the OTR


Figure 2

Days above Ozone Standard - 1981 to 1995


Figure 3

Average Maximum Ozone Concentration in the OTR during a NAAQS Exceedance - 1981 to 1995


Figure 4

Hours above Ozone Standard in the OTR - 1981 to 1995


Summary of the ranking for 1995 Air Quality Statistics


Figure 5

Number of Ozone Exceedance days for the past 15 years.


Figure 6

Frequency of Exceedances for the period from 1981 to 1995


ADJUSTING FOR METEOROLOGICAL VARIABILITY

  • Meteorological conditions may have a confounding influence on air quality, which could obscure underlying trends. For example, techniques to adjust ozone statistics for the effects of meteorological variability include:
  • Classification of days on the basis of meteorological conditions into categories that define their ozone formation potential, examine classified data for trends.
  • Development of mathematical relationships between ozone concentration and meteorological factors, use the relationship to predict ozone concentrations expected to occur under standardized meteorological conditions, examine "adjusted" concentrations for trends.
  • Development of statistical methods of filtering out effects of meteorology from ozone concentration data, examine modified ozone data for trends.

  [Workbook Table of Contents] [Top of Trend Analyses] [Previous Section] [Next Section]


METEOROLOGICAL VARIABLES POTENTIALLY ASSOCIATED WITH OZONE FORMATION

Insolation

     Sky cover, integrated clear or cloudy sky photolysis rate, ceiling height, number of daylight hours

 Ventilation

    Vector and scalar average wind speeds, wind fluctuation ratio (ratio of scalar to vector average wind speed), mixing height, ventilation coefficient (mixing height times wind speed)

Transport

     Wind direction, recirculation index

Indirect Measures

    Previous day's daily maximum ozone concentration, temperature (daily maximum, range), humidity indicators, surface pressure and range, precipitation, temperature at 500 and 850 mb, synoptic weather pattern type

  [Workbook Table of Contents] [Top of Trend Analyses] [Previous Section] [Next Section]


CLASSIFICATION METHODS

High ozone is likely to occur with wind patterns that bring, keep, or return high background concentrations to the region; elevated temperature; intense solar radiation (i.e., no cloud cover); and shallow mixing depth. VOC concentrations are most affected by wind speed, wind direction when there is inhomogeneity in the spatial distribution of sources, and temperature (e.g., for isoprene).

Examples

  • Number of days ozone exceedances/number of ozone-conducive days (e.g., ozone-conducive days defined as temperature in excess of 90°F).
  • Classify days by high, moderate, low ozone formation potential and calculate year-to-year trend on number of days in each category.
  • Group years according to frequency of occurrence of ozone-conducive days and calculate trend in annual summary statistics only for years with similar numbers.


Additional approaches are discussed in reports listed in the reference section.


Figure 7

Number of days above the Ozone Standard and days above 90 degrees F in the OTR


Figure 8

Number of days above the Ozone Standard and days above 90 degrees F in the Lake Michigan Region


Figure 9

Actual and Normalized Number of days above the Ozone Standard in the Lake Michigan Region 

 [Workbook Table of Contents] [Top of Trend Analyses] [Previous Section] [Next Section]


EPA METEOROLOGICAL TREND ADJUSTMENT METHOD
(Cox and Chu Trend Analysis Method)

The Cox and Chu Method is a regression model that can be used to predict the probability distribution of daily maximum ozone concentrations in a metropolitan area given the values of several meteorological parameters including:

Surface temperature

  • Average 0700-0900 wind speed
  • Average 1300-1600 wind speed
  • Average morning wind direction
  • Average afternoon wind direction
  • Total opaque cloud cover
  • Relative humidity

The model includes a term to account for the long-term trends in ozone air quality.


Figure 10

Plot of the actual and Cox and Chu model-predicted 95th percentile of the ozone daily maximum concentration in the New York City CMSA

      Plot of the actual and Cox and Chu model-predicted 95th percentile of the ozone daily maximum
      concentration in the New York City CMSA (Cox and Chu, 1996).


Figure 11

Actual and adjusted trends in the number of days on which ozone concentrations exceed 0.12 ppm in the Chicago area.


Figure 12

Cumulative Reduction in the Meteorologically-Adjusted Dailay Maximum Ozone Concentration from 1984-1994

 [Workbook Table of Contents] [Top of Trend Analyses] [Previous Section] [Next Section]


SUNYA TREND ANALYSIS METHOD
(Rao and Zurbenko Trend Analysis Method)

The Rao and Zurbenko method is a statistical method of moderating the influence of meteorology on surface ozone concentrations. Surface temperature is used as a surrogate for all meteorological parameters assumed to affect ozone.

Time series of the log of the daily maximum ozone (O) and daily maximum temperature (T) are represented by

 X(t) = e(t) + S(t) + W(t) = Baseline(t) + W(t)

e(t) = Long-term trend component

S(t) = Seasonal variation

W(t) = Short-term variation (white noise)

The baseline is separated from the short-term variation using the Kolmogorov-Zurbenko (KZm,p) filter.

Filtered Ozone-Temperature Correlation

Om,p(t) = aTm,p(t+lag) +b + e (t)

e (t) reveal changes in ozone concentrations which cannot be attributed to temperature fluctuation.

e (t) = e m,p(t) + d (t)

e m,p(t) = Long-term emission effects unexplained by temperature

d (t) = Ozone seasonal variation induced by meteorological variables other than temperature


Figure 13

Components of the log of the daily maximum ozone concentrations.


Figure 14

Variation in the of the log of the daily maximum when the temperature effect is removed from the ozone concentrations


Figure 15

Geographical map depicting trends

      Geographical map depicting trends, significant at the 95% confidence level, in the temperature-independent
      ozone (percent per year) during the period 1980-1992 (Zurbenko et al., 1995).

 [Workbook Table of Contents] [Top of Trend Analyses] [Previous Section] [Next Section]


DETECTING TRENDS

Linear Model

Use simple linear regression on annual summary statistics or to logged statistics (if lognormal); perform analysis of variance.

Nonparametric Methods

To test for and estimate a trend without making distributional assumptions (e.g., Spearman's rho test of trend, Kendall's tau test of trend).

Time Series Models

Statistical modeling of ozone concentrations taking into account their serial dependence (e.g., auto-regressive integrated moving average - ARIMA).

Extreme Value Theory

To estimate distributions of annual maximum hourly concentrations, estimate distributions of the number of days exceeding the standard (e.g., Chi-square test, Poisson process approximation).


Figure 16

Detecting Trends - Illustration using confidence intervals to determine statistically significant changes.


Figure 17

Ambient Air Quality change related to Fuel RVP decrease.

 [Workbook Table of Contents] [Top of Trend Analyses] [Previous Section] [Next Section]


PRECURSOR TREND ANALYSES

Summary of average expected changes in emissions of several hydrocarbons and ambient data values using species weight fractions at Los Angeles North Main (0600-0900 PDT, summer 1992 and 1993, screened data). Indicator usefulness is expressed as Yes, No, or Possible. A "Possible" indicates the estimated emission change is equal to about one standard deviation or at the edge of the interquartile range. (Stoeckenius et al., 1995)

 Species

Average Expected Change (Years)

Daily Value Statistics (weight percent)

Estimated Changec (weight percent)

 Useful Indicator?

   

Mean

Median

STDa

25thb

Mean

Median

Meand

Mediane

 1,3-Butadiene

 -17% (95-96)

 0.29

 0.31

 0.1

 0.1

 -0.05

 -0.05

 N

 N

 Benzene

 -30% (94-96)
-35% (94-96)f

 3.0

 2.8

 0.5

 0.3

 -0.9
-1.0

 -0.9
-1.1

Y
Y

Y
Y

C7-C8 Aromatic Hydrocarbons

 -10% (94-95)

 17

 17

 2.1

 1.6

 -1.7

 -1.7

N

P

 Total Xylenes

 -10% (94-95)

 6.7

 6.8

 1.1

 0.7

 -0.7

 -0.7

N

P

 I-Butene

 +200% (94-95)

 0.8

 0.8

 0.3

0.1g

 +1.7

 +1.6

Y

Y

 n-Butane

 -25% (94-95)

 3.7

 3.7

 0.7

 0.5

 -1.0

 -1.0

Y

Y

C9-C10 Aromatic
Hydrocarbons

 -38% (95-96)
-45% (94-96)f

 5.3

5.3

 0.9

 0.6

 -2.0
-2.4

 -2.0
-2.4

Y
Y

Y
Y

a Standard deviation of the daily mean values in the 1992 and 1993 CARB 3-hr hydrocarbon data.
b Median - 25th percentile for the 1992 and 1993 CARB 3-hr hydrocarbon data.
c Estimated change to the mean and median values if fuel changes are implemented.
d For the indicator to be useful, the estimated change must be greater than one standard deviation.
e For the indicator to be useful, the estimated change must be outside the interquartile range.
f Estimated change of
£ 10 percent in 1995-1996 probably too small to be observed in the data.
g 75th percentile - median for the 1992 and 1993 CARB 3-hr hydrocarbon data.


Figure 18

Changes in the California ambient benzene concentrations.


Figure 19

NOx and VOC Emissions Trends in the Northeast ozone transport region.


Figure 20

NOx and VOC Emission Trends - Baltimore MSA 1985 to 1991


Figure 21

VOC and NOx Emissions vs. Ozone Exceedance Days for the whole OTR


Figure 22

Ozone and Ozone Precursor Concentration trends in the South Coast Air Basin

 [Workbook Table of Contents] [Top of Trend Analyses] [Previous Section] [Next Section]


SUMMARY

  
Analysis


Tool(s)


Special Data Requirements

Met-adjustments using statistical methods, classification methods Statistical software, programming, spreadsheets, databases Non-PAMS meteorological measurements
Trend Analyses Statistical software Non-PAMS meteorological measurements
Ozone Precursor Emission Trend Analyses Statistical software,
Spreadsheet, programming, databases
E.I. data,
Non-PAMS meteorological measurements

  [Workbook Table of Contents] [Top of Trend Analyses] [Previous Section] [Next Section]


 REFERENCES

Bloomfield P., Royle J.A., Steinberg L.J., and Yang Q. (1996) Accounting for meteorological effects in measuring urban ozone levels and trends. Atmos. Environ. 30, 3067-3077.

Chinkin L.R., Reiss R., Eisinger D.S., Dye T.S., Jones C.M. (1996a) Ozone exceedance data analysis: representativeness of 1995, Phase I. Final report prepared for the American Petroleum Institute by Sonoma Technology, Inc., Santa Rosa, CA. STI-996031-1574-FR, August.

Chinkin L.R., Reiss R., and Eisinger D.S. (1996b) Ozone exceedance data analysis: representativeness of 1995, Phase II. Final report prepared for the American Petroleum Institute by Sonoma Technology, Inc., Santa Rosa, CA. STI-996032-1586-FR, October.

Cox W.M. and Chu S.H. (1992) Meteorologically adjusted ozone trends in urban areas: a probability approach. In Transactions of the Tropospheric Ozone and the Environment II International Specialty Conference, Air & Waste Management Association, Pittsburgh, PA, pp. 342-353.

Cox W.M. and Chu S.H. (1993) Meteorological adjusted ozone trends in urban areas: a probabilistic approach. Atmos. Environ., 27B, 425-434.

Cox W.M. and Chu S.H. (1996) Assessment of interannual ozone variations in urban areas from a climatological perspective, submitted to Atmos. Environ.

Fairley D. and Blanchard C.L. (1991) Rethinking the ozone standard. J. Air & Waste Manag. Assoc. 41, 928-936.

Flaum J.B., Rao S.T., and Zurbenko I.G. (1996) Moderating the influence of meteorological conditions on ambient ozone concentrations. J. Air & Waste Manage. Assoc., 46, 35-46.

Hammond D. (1996) Ambient trends of benzene in California from 1990 through 1995. Paper presented at the U.S. Environmental Protection Agency and Air & Waste Management Association International Symposium on Measurement of Toxic and Related Air Pollutants, Research Triangle Park, NC, May 7-9.

Kolaz D.J. and Swinford R.L. (1990) How to remove the influence of meteorology from the Chicago area ozone trend. Presented at the Air & Waste Management Association 83rd Annual Meeting, Pittsburgh, PA, June 24-29.

LADCO (1995) Lake Michigan Ozone Study. 1994 data analysis report, version 1.1. Report prepared by Lake Michigan Air Directors Consortium, Des Plaines, IL, May.

LADCO (1996) Lake Michigan Ozone Study: 1995 data analysis report, version 1.1. Report prepared by Lake Michigan Air Directors Consortium, Des Plaines, MI, April.

Porter P.S., Rao S.T., Zurbenko I., Zalewsky E., Henry R.F., and Ku J.Y. (1996) Statistical characteristics of spectrally-decomposed ambient ozone time series data. Final report prepared for the Ozone Transport Assessment Group by the University of Idaho, the State University of New York at Albany and the New York Department of Environmental Conservation, August.

Rao S.T. and Zurbenko I.G. (1994) Detecting and tracking changes in ozone air quality. J. Air & Waste Manage. Assoc., 44, 1089-1092.

Rao, S.T., Zalewsky E., and Zurbenko I.G. (1995) Determining spatial and temporal variations in ozone air quality, J. Air & Waste Manage. Assoc., 45, 57-61.

Roberts P.T., Main H.H., and Yocke M.A. (1996) Operating plan for ozone modeling data collection in El Paso-Ciudad Juarez-Sunland Park. Draft final report prepared for U.S. Environmental Protection Agency, Research Triangle Park, NC under subcontract to SAIC, McLean, VA by Sonoma Technology, Inc., Santa Rosa, CA, STI-95370-1573-DFR, EPA Contract No. 68-D3-0030, Option Year II, Work Assignment No. 11-77, SAIC Project No. 01-1030-07-3823-xxx, March.

Sabo E.J. and Hawes J.T. (1990) User's guide and program documentation for the transported ozone design value model. Report prepared for U.S. Environmental Protection Agency, Research Triangle Park, NC, EPA Contract No. 68-02-4393.

Schmidt, M. (1997) Personal communication.

Smith B.E. and Adamski W.J. (1996) Long term ozone trends in the Lake Michigan airshed. Paper No. 97-A112 presented at the Air & Waste Management Association 90th Annual Meeting & Exhibition, Toronto, Ontario, Canada, June 8-13.

Stoeckenius T.E., Ligocki M.P., Cohen B.L., Rosenbaum A.S., and Douglas S.G. (1994b) Recommendations for analysis of PAMS data. Final report prepared by Systems Applications International, San Rafael, CA, SYSAPP94-94/011r1, February.

Stoeckenius T.E., Yarwood G., Ligocki M.P., Cohen J.P., Shepard S.B., Looker R.E.,Fujita E.M., Main H.H., and Roberts P.T. (1995) Feasibility study for a 1995-1996 Southern California air quality monitoring program. Final report prepared for Coordinating Research Council, Atlanta, GA by Systems Applications International, San Rafael, CA, Desert Research Institute, Reno, NV, and Sonoma Technology, Inc., Santa Rosa, CA, SYSAPP94-94/065, January.

U.S. Environmental Protection Agency (1994a) Clean air act ozone design value study: a report to Congress. Final report prepared by OAQPS, U.S. Environmental Protection Agency, Research Triangle Park, NC, EPA-454/R-94-035, December.

U.S. Environmental Protection Agency (1994b) Air pollutant emission trends: 1900-1994. Prepared by the Office of Air and Radiation , U.S. Environmental Protection Agency, Research Triangle Park, NC.

Zurbenko I.G., Rao S.T., and Henry R.F. (1995) Mapping ozone in the eastern United States, Environmental Manager, 1, 24-30.

 [Workbook Table of Contents] [Top of Trend Analyses] [Previous Section] [Next Section]


Local Navigation


Jump to main content.