Introduction to Environmental Statistics: Module 1: Interpreting your monitoring data
Developed for USEPA by:
Dr. Peter Scheff Salvatore Cali & Justin Ford
University of Illinois at Chicago Environmental and Occupational Health Sciences Division School of Public Health, 2121 West Taylor St. Chicago, IL 60612-7260
UIC The University of Illinois at Chicago
Program Objectives
• Provide context and background for air pollution sampling and sampling design • Provide an introduction to statistical assessment of air pollution data
UIC The University of Illinois at Chicago
Differences between Criteria Pollutant and Air Toxics Monitoring Programs: Criteria Pollutants: CO, Pb, PM, SO2, NO2, O3 • Human health effects well defined by epidemiologic studies • NAAQS standards based on these health studies • Monitoring aimed at measuring attainment of NAAQS
UIC The University of Illinois at Chicago
1
Differences between Criteria Pollutant and Air Toxics Monitoring Programs:
Air Toxics: (e.g. benzene, Cr, formaldehyde, PAHs)
• 188 Air Toxics listed in ’90 Clean Air Act Amendments • Health effects not as well documented: human health risks generally derived from animal studies • There are no air quality standards for air toxics • Monitoring data aimed at estimating human health risks, either alone or in concert with modeling efforts
UIC The University of Illinois at Chicago
Table 1: National Ambient Air Quality Standards
Pollutant
Carbon monoxide Lead Nitrogen dioxide Particulate matter (PM10) Particulate matter (PM2.5) Ozone Sulfur oxides
Criteria Pollutants
Primary Standards
9 ppm (10 mg/m3) 35 ppm (40 mg/m3) 1.5 g/m3 g/m3)
Averaging Times
8-hour 1-hour Quarterly avg. Annual (Arithmetic mean) Annual (Arith mean) 24-hour Annual (Arith mean) 24-hour 8-hour 1-hour Annual (Arith mean) 24-hour 3-hour None None
Secondary Standards
0.053 ppm (100 50 g/m3 150 g/m3 15 65 g/m3 g/m3
Same as primary Same as primary Same as primary ---------0.5 ppm (1300
0.08 ppm 0.12 ppm 0.03 ppm 0.14 ppm -----
g/m3)
UIC The University of Illinois at Chicago
Primary and Secondary Pollutants
Primary:
• Found in the same chemical form in which it was emitted into the atmosphere For example: Coarse PM, CO, benzene, metals • Many travel far, some may settle out or react quickly
Secondary:
• Formed by chemical reactions in the atmosphere For example: formaldehyde, ozone, much of PM2.5 • May be formed far from source; may be reactive
UIC The University of Illinois at Chicago
2
Sampling and Analytical Methods
Criteria Pollutants may be particles (PM10, PM2.5, Lead) or gases (CO, NO2, SO2, O3). Required methods are described in the NAAQS Air Toxics may be particles (various metals), gases (VOCs), or combination (semivolatiles, incl. PAHs, PCBs) For recommendations on sampling and analytical methods, see Air Toxics Monitoring Concept Paper: http://www.epa.gov/ttn/amtic/files/ambient/airtox/ cncpsab.pdf
UIC The University of Illinois at Chicago
Purposes for Air Quality Monitoring
UIC The University of Illinois at Chicago
Purposes for Air Quality Monitoring
1) To demonstrate compliance or non-compliance with regulations
UIC The University of Illinois at Chicago
3
Purposes for Air Quality Monitoring
2) Establish a baseline for future reference
UIC The University of Illinois at Chicago
Purposes for Air Quality Monitoring
3) Health and exposure studies
UIC The University of Illinois at Chicago
Purposes for Air Quality Monitoring
4) Provide information on spatial variability
Peak AQI-Ozone Green = Good Red = Unhealthy
UIC The University of Illinois at Chicago
4
Air Quality Index for Ozone and PM2.5
Index Values 0-50 51-100 51101-150 101AQI>100 “Unhealthy Days” Category Good Moderate Unhealthy for Sensitive Groups Unhealthy Very Unhealthy PM2.5 (ug/m3) 0-15.4 15.5-40.4 15.540.5-65.4 40.5Ozone 8-hour (ppm) 0-.064 .064-.084 .064.085-.104 .085-
151-200 151201-300 201-
65.5-150.4 65.5150.5-250.4 150.5-
.105-.124 .105.125-.374 .125-
UIC The University of Illinois at Chicago
UIC The University of Illinois at Chicago
Purposes for Air Quality Monitoring
5) Provide information on temporal variability
UIC The University of Illinois at Chicago
Grey area = Sample results between 90 and 10 percentile 0 Dotted line = # of days greater than 90 F White line = mean of sample results
th
th
5
Purposes for Air Quality Monitoring
6) Establish trends and evaluate the effectiveness of HAP reduction strategies
UIC The University of Illinois at Chicago
Decline in lead in gasoline and BLLs in the population 1976-1980
Air Quality Criteria for Lead – U.S. EPA; EPA 600/8-83-0288, September -1984, Volume 1. Figure 1-16. Parallel decreases in blood lead values observed in the NHANES 11 Study and amounts of lead used in gasoline during 1976-1980. Source: Annest (1983).
UIC The University of Illinois at Chicago
Purposes for Air Quality Monitoring
7) Compare monitoring results with computer models
UIC The University of Illinois at Chicago
6
Steps in design and analysis process
UIC The University of Illinois at Chicago
Steps in design and analysis process
1. Define purpose of study 2. Define the population of interest (time and space) 3. Collect information on the physical environment (site history, weather, direction of flow, ...) 4. Define types of samples to be collected (location, averaging time, media, personal vs area...) 5. Develop a quality assurance plan (measurement objectives, sampling procedures, laboratory procedures, data handling, data manipulation, statistical analysis, …)
UIC The University of Illinois at Chicago
Steps in design and analysis process
6. Examine data from prior studies, baseline & trends 7. Develop a field sampling program including number of samples, locations, days, replicates 8. Determine required statistical procedures, data plots, summaries, ... 9. Conduct study according to written plans 10. Evaluate results 11. Evaluate whether study objectives have been met
UIC The University of Illinois at Chicago
7
Example of steps in compliance monitoring
• Ozone monitoring • PM 2.5 monitoring
UIC The University of Illinois at Chicago
Step 1: Define purpose of study
• Determine if the area of concern is in compliance with NAAQS Standards for Ozone • Determine the diurnal pattern in the concentration of toxic air pollutants
UIC The University of Illinois at Chicago
Ozone: a secondary pollutant
VOC + NOx + heat + sunlight
ozone
• Ozone is formed by complex chemical reactions between volatile organic compounds (VOC) and oxides of nitrogen (NOx) in the presence of sunlight. • Ozone “precursors” (NOx and VOCs), as well as ozone itself, can be carried hundreds of miles from their origins, causing air pollution over wide regions. • Ozone monitors can represent a large geographic area.
UIC The University of Illinois at Chicago
8
Reactions
Sources
Exposures
UIC The University of Illinois at Chicago
National Research Council. Rethinking the Ozone Problem in Urban and Regional Air Pollution, National Academy Press, 1991.
Concentration (ppb)
NO, NO2, and O3 Concentrations, Lake Michigan Ozone Study Sites, June 25, 1991
O3
NO2 NO
UIC The University of Illinois at Chicago
Hour of the Day
Benzene: Afternoon decline due to mixing and ozone formation
UIC The University of Illinois at Chicago
Credit: Battelle Memorial Institute
9
NYC Benzene
2 1.8 1.6 1.4 1.2 ppb 1 0.8 0.6 0.4 0.2 0 0 24 Time from 13 July 1995 48 72
UIC The University of Illinois at Chicago
Adapted from Seigneur et al, Comparison of Simulated Urban and Rural Diesel Concentrations, Atmospheric and Environmental Research, Inc. www.cmascenter.org/2003_workshop/session2/seigneur_presentation.ppt
Step 2: Define the population of interest (time and space)
• Temporal scales • Spatial scales
UIC The University of Illinois at Chicago
Spatial Scales
• Air quality measurement at single site is based on a relatively small sample It is useful to know how large an area this value represents • Spatial scale of representativeness: Physical dimensions of the area of a monitoring station, located so pollutant concentrations within 25% of area mean See:
NETWORK DESIGN AND SITE EXPOSURE CRITERIA FOR SELECTED NONCRITERIA AIR POLLUTANTS EPA Number: 450484022 ; NTIS number PB95-232419
UIC The University of Illinois at Chicago
10
Spatial Scales
As defined in the CFR (app D of part 58, title 40):
• Urban Scale –overall citywide conditions (4 to 50 km) • Neighborhood Scale – Extended area with uniform land use (0.5 – 4km) • Middle Scale- Areas up to several blocks (100 to 500 meters) • Microscale – Represents localized areas (several to 100 meters)
UIC The University of Illinois at Chicago
Exposure Assessment at Different Scales
UIC The University of Illinois at Chicago
Step 3: Collect information on the physical environment
• Site history • Location of major emitters • Landscape elevation; elevation barriers and low elevation areas • Seasonal average daily temperature and humidity • Seasonal predominant wind directions
UIC The University of Illinois at Chicago
11
PAMS Network measures VOCs. Monitor locations picked based on the geography of O3 formation.
UIC The University of Illinois at Chicago
Where to get meteorological info?
• Why is it useful? It is important to note the location of pollutant sources, populated areas, and monitoring sites relative to one another. • Work with your agency’s meteorologist or air modeler to get site specific meteorological data For Example: • What are the prevailing winds? • Windroses show prevailing wind direction and speeds over a year or several years • Visit http://home.pes.com/windroses for a general sense of your study area
UIC The University of Illinois at Chicago
San Francisco Strong prevailing wind from the ocean.
St. Louis Gentler winds, coming from all directions over time
UIC The University of Illinois at Chicago
12
Consider wind speed vs. concentration
UIC The University of Illinois at Chicago
Consider temperature vs. concentration
UIC The University of Illinois at Chicago
Consider wind direction versus concentration
Pocatello Wind Hours at Each Direction sorted by PM10 Concentration Feb 12 -- 15, 1996
25 20 15 10 5 0 >150 ug/m3 >100, <150 ug/m3 >50, <100 ug/m3 <50 ug/m3
UIC The University of Illinois at Chicago
13
Pollution Rose of H2S at Paul, Idaho
UIC The University of Illinois at Chicago
UIC The University of Illinois at Chicago
Consider temperature vs. concentration
UIC The University of Illinois at Chicago
14
Consider RH vs. concentration
UIC The University of Illinois at Chicago
Where to get topographic info?
Why? To get an idea of local air movements: open plains? deep valleys? Get USGS Topo maps from your agency mapping staff or local planning dept. Look online at www.topozone.com
UIC The University of Illinois at Chicago
Inversion in Donora, PA
• Oct 1948 • 20 dead and left hundreds seriously injured and dying • Caused by fluoride emissions from Donora Zinc Works & steel plants owned by US Steel Corporation
UIC The University of Illinois at Chicago
15
Background Graphic
NASA6-703-044 Po River Valley Smog, Italy Fall/Winter 1997. The snow capped higher elevations of the Italian, Swiss, and French Alps mountains provide a natural barrier to the outward migration of smog shown here filling the Po River Valley of northern Italy. This southwest-looking view shows a dense sea of air pollution (extensive, lighter colored areacenter of image) that is trapped from time to time by air stagnation in the heavily industrialized Po River Valley. The picture shows that air pollution even infiltrates the numerous mountain valleys that surround this river basin.
UIC The University of Illinois at Chicago
Step 4: Define types of samples to be collected (location, averaging time, media, personal vs area...)
• Long-term vs short-term vs personal • Location: Chicago metropolitan area • Averaging time: 3-year average of annual 4th highest maximum 8-hour average ozone concentrations • Media: Reference or equivalent method (40 CFR, Pt 53.9 EPA-600/4-79-057)
UIC The University of Illinois at Chicago
Long-term ambient air monitoring may be used to determine:
• Highest concentrations expected to occur in the area covered by the monitoring network • Representative concentrations in areas of high population density • Impact on ambient pollution levels of significant sources or source categories • General background concentration levels • Extent of regional pollutant transport among populated areas
UIC The University of Illinois at Chicago
16
Short-term monitoring can be useful to:
• Facilitate proper assessments of geographic variability • Characterize environmental justice concerns • Assess ambient concentrations representative of small geographic areas like schools, which may be potentially impacted by specific sources (“hot spots”).
Personal exposure monitors are:
• Used primarily by non-EPA organizations; limited data • Most inhalation exposure characterizations rely on model predictions using ambient monitoring data
UIC The University of Illinois at Chicago
Step 5: Develop a quality assurance plan (measurement objectives, sampling procedures, laboratory procedures, data handling, data manipulation, statistical analysis, …)
• Covered in Module 3: QA/QC for Air Pollution Monitoring
UIC The University of Illinois at Chicago
Step 6: Examine data from prior studies, baseline & trends
B ackground concentrations assum ed in the N ATA -N ational Scale A ssessm ent and C EP:
Pollutant Background Conc. (ug/m 3)
B enzene C arbon tetrachloride C hloroform E thylene dibrom ide E thylene dichloride Form aldehyde H exachlorobenzene M ercury com pounds M ethylene chloride P olychlorinated biphenyls P erchloroethylene (Tetrachloroethylene) Trichloroethylene
UIC The University of Illinois at Chicago
0.48 0.88 0.083 0.0077 0.061 0.25 .000093 0.0015 0.15 0.00038 0.14 0.081
17
NATA estimate of background contribution to ambient air
Formaldehyde
0.9% 20.0% 11.0%
Benzene
2.7% 6.3%
34.5%
30.0% 39.8%
Major Area,Other Onroad Mobile Nonroad Mobile Estimated Background
38.1% 16.8%
PERC
7.1%
Carbon Tetrachloride
0.1%0.1%
43.3%
49.5%
UIC The University of Illinois at Chicago
99.9%
Step 7: Develop a field sampling program (# of samples, locations, etc.)
1. Daily continuous monitoring at 10 local sites, 9 for ozone and 8 for PM 2. Collect one field blank for every 20 samples 3. Set up replicate sampler at 10% of monitoring sites 4. Background samples for one full year
UIC The University of Illinois at Chicago
Step 8: Determine required statistical procedures, data plots, summaries, ... • • • • • •
UIC The University of Illinois at Chicago
Trends Diurnal patterns Averaging time Percentiles Isopleths maps Wind and Pollution roses
Pocatello Wind Hours at Each Direction sorted by PM10 Concentration Feb 12 -- 15, 1996
25 20 15 10 5 0 >150 ug/m3 >100, <150 ug/m3 >50, <100 ug/m3 <50 ug/m3
18
Step 9: Conduct study according to written plans
Monitoring PM2.5: • 3 fixed-site IMPROVE samplers located to represent rural area • PTFE membrane filters for gravimetric analysis • 24-hr average samples once every 3 days • 120 readings/ year total • 3 year study period
UIC The University of Illinois at Chicago
Things to Consider Before Trying to Interpret an Existing Dataset
• What was the purpose of the monitoring study? • What is the site’s spatial representativeness? • What is the nature of monitoring site in question? • Where are the pollutant sources? • What is the meteorology topography? • What is the nature of the pollutant in question? • Will it travel far from its sources? • What is its reactivity?
UIC The University of Illinois at Chicago
Step 10: Evaluate results
The following procedures are detailed in USEPA Publication #: EPA-454/R-98017 “Guideline on Data Handling Conventions for the 8-hour Ozone NAAQS”
UIC The University of Illinois at Chicago
19
Interpretation
• Compile the data used for 8-hour ozone monitoring……..
UIC The University of Illinois at Chicago
OZONE SAMPLING RESULTS
Calculating 8 hour average
• Within any 24 hr day calculate all possible 8 hr averages • If data points are missing, average over the period using the new denominator • Select the highest value as the 8 hr average for the day
Time of Hourly average 8-hour average Day (ppm) (ppm) 12:00 AM 0.030 1:00 AM 0.020 2:00 AM 0.030 3:00 AM 0.020 4:00 AM 0.030 5:00 AM 0.040 6:00 AM 0.030 7:00 AM 0.030 8:00 AM 0.040 0.030 9:00 AM 0.050 0.032 10:00 AM 0.070 0.038 11:00 AM 0.070 0.042 12:00 PM 0.090 0.050 1:00 PM 0.060 0.053 2:00 PM 0.070 0.057 3:00 PM 0.080 0.062 4:00 PM 0.070 0.067 5:00 PM 0.060 0.069 6:00 PM 0.060 0.070 7:00 PM 0.070 0.070 8:00 PM 0.050 0.068 9:00 PM 0.030 0.061 10:00 PM 0.030 0.058 11:00 PM 0.020 0.052 12:00 AM 0.020 0.046 Maximum Concentration: 0.070
Hourly average (ppm) with missing data 8-hour average points (ppm) 0.030 0.020 0.030 0.020 0.040 0.030 0.030 0.040 0.050 0.070 0.070 0.090 0.070 0.080 0.070 0.060 0.060 0.070 0.050 0.030 0.020 0.020
0.030 0.033 0.039 0.044 0.053 0.053 0.056 0.063 0.068 0.070 0.071 0.071 0.069 0.061 0.061 0.055 0.048 0.071
UIC The University of Illinois at Chicago
Highest 8-hour average
Highest 8-hour average of data set with missing samples
More on Missing Data
• If there are fewer than 6 values in a given 8 hour period, that period is invalid unless the concentration clearly exceeds the level of the standard. • There are requirements for data completeness: For ozone, 75% of the days in a year with an average of 90% over three years of the ozone season as defined for the site
UIC The University of Illinois at Chicago
20
Calculating Ozone Standard Compliance
8hr maximum concentrations (ppm)
Percent Year valid days
2001 2002 2003
98% 93% 100%
1st 0.094 0.091 0.087
2nd 0.092 0.091 0.086
3rd 0.089 0.087 0.083
4th 0.088 0.084 0.082
• By ordering the daily maximum concentrations you can obtain the absolute maximums for each year. • Take the 4th highest for each year and average them to determine if you are in compliance
0.088 + 0.084 + 0.082 = 0.08466 3
UIC The University of Illinois at Chicago
0.08466 rounds to 0.08; results are below the 8-hour Primary Standard of 0.08 ppm (If 0.08500 or greater, round to 0.09)
Example 2: PM 2.5
• Data interpretation for compliance purposes
UIC The University of Illinois at Chicago
Secondary PM 2.5 formed from primary SO2, NO2, and Primary/Secondary VOCs
A substantial fraction of PM2.5 is secondary sulfate and nitrate, formed as a result of atmospheric reactions. Such reactions involve: • Gas phase conversion of SO2 to H2SO4 by OH radicals; • Aqueous-phase reactions of SO2with H2O2, O3, or O2 (catalyzed by Fe and Mn) Atmospheric reactions involving volatile organic compounds such as alkanes, alkenes, aromatics, cyclic olefins, and terpenes (or any reactive organic gas that contains at least seven carbon atoms) yield organic compounds.
UIC The University of Illinois at Chicago
21
PM 2.5 Chemical Composition
UIC The University of Illinois at Chicago
PM 2.5 Standard
• Annual PM2.5 Standard is met when 3-year average of spatially averaged annual mean concentrations (among designated monitors) are 15.0 µg/m3
– Designated monitors are sites designated for spatial averaging in a State PM Monitoring Network Description. This applies only if opting to do spatial averaging in the area as discussed in 40 CFR Part 58. If not opting to do spatial averaging, use annual average of the single site
• The 24-hour PM2.5 Standard is based on 3-year average of the annual 98th percentile PM2.5 concentration and the values determined for PM2.5 at each monitoring site are less than or equal to 65 µg/m3
UIC The University of Illinois at Chicago
How do we compute the 3-year average annual mean for PM2.5 ?
• Start with averaging the quarterly data. • Take the sum of all the samples and divide it by the total number of samples
Quarter Jan-Mar Apr-Jun Jul-Sept Oct-Dec Average (µg/m3) 20.15 13.94 15.44 13.89
Sum = (63.42 µg/m3)/4 = 15.85 µg/m3
Data From EPA Site 4850 WILSON AVE Cook County IL 2003
UIC The University of Illinois at Chicago
22
From Quarters to Years
• Calculate the yearly average for each site by averaging the Quarterly data
20.15 + 13.94 + 15.44 + 13.89 = 15.85µ g 4
Note that 15.85 rounds to 15.9 and may exceed standard of 15.0 µg/m3
UIC The University of Illinois at Chicago
You Can Also Calculate a Spatial Average
• Assume spatial averaging with three sites • Find the mean and the 98th percentile Annual means in µg/m3 Year 1 2 3
UIC The University of Illinois at Chicago
Site 1 15.9 17.2 16.9
Site 2 11.5 13.2 12.7
Site 3 NA 19.1 17.6
Calculating the Mean
• The mean is just the simple average of the sites each year • These averages are then averaged to determine if there is an exceedance
Annual means in µg/m3 Year 1 2 3 Site 1 Site 2 Site 3 mean 15.9 17.2 16.9 11.5 13.2 12.7 19.1 17.6 13.7 16.5 15.7 15.3
Average of the 3 yearly averages
UIC The University of Illinois at Chicago
15.3 > 15.0 µg/m3 so these sites may exceed the standard
23
order
conc 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 2.6 3.2 3.3 3.9 4.2 4.3 4.4 4.7 4.8 5.1 5.2 5.2 5.3 5.5 5.5 5.6 5.7 5.8 5.8 5.9 6 6.1 6.2 6.2 6.3 6.3 6.3 6.4 6.6 6.7 6.9 6.9 7.1 7.3 7.6 7.6 7.7 7.7 7.7 7.8 8 8 8.3 8.4 8.5 8.6 8.6
98th Percentile
• The average of the 98th percentile is found by first getting the 98th percentile for each year at each site. • The numbers than can be averaged to determine compliance • 1st sort the data from lowest to highest and assign an order to each sample starting with 1 • In our example we have 118 samples
order conc µg/m3
1 2 3 4 5
2.6 3.2 3.3 3.9 4.2
UIC The University of Illinois at Chicago
113 114 115 116 117 118
25.5 27.4 29.5 31.9 35.2 44
Finding the 98th percentile value
1) Multiply the total number of samples by 98% or 0.98 118 x 0.98 = 115.64 2) Take the integer of your answer and add one 115 + 1 = 116 3) This is you sample number for the 98th percentile 113 Order 116 = 31.9 µg/m3 114
115 116 117 118
UIC The University of Illinois at Chicago
25.5 27.4 29.5 31.9 35.2 44
Average the 98th percentile
• Once you have the 98th percentile for each site you can average the data over 3 years to determine compliance • Each site stands alone for this computation so you should end up with something like this
Year 2003 2002 2001 Average
UIC The University of Illinois at Chicago
98% Concentration (ug/m3) 31.9 29.2 36.3 32.5
32.5 < 65 µg/m3 This data is below the 98th percentile standard
24
Step 11: Evaluate whether study objectives have been met
• Was the data sufficient to meet the requirements of the compliance standard? • Does the data make sense? • Does it meet the Data Quality Objectives (DQO)…. To be presented in modules 2 and 3
UIC The University of Illinois at Chicago
What you should have learned…
• Context, background, and purpose of the EPA Air Pollution Monitoring Program • 11 steps in the design and analysis process • Basic data interpretation for the Ozone and PM2.5 Standards
UIC The University of Illinois at Chicago
25