VISIBILITY DEGRADATION AND METEOROLOGY IN THE BOULDER VALLEY
by George M. Mathews, Jr.
AKTHim LAKES LÏB1A1Y COlOmDO gC’JOOL oi MINES nOLDSN. COLORADO 80*01 ProQuest Number: 10783590
All rights reserved
INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted.
In the unlikely event that the author did not send a com plete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion. uest
ProQuest 10783590
Published by ProQuest LLO (2018). Copyright of the Dissertation is held by the Author.
All rights reserved. This work is protected against unauthorized copying under Title 17, United States C ode Microform Edition © ProQuest LLO.
ProQuest LLO. 789 East Eisenhower Parkway P.Q. Box 1346 Ann Arbor, Ml 48106- 1346 T-3870
À thesis submitted to the Faculty and the Board of
Trustees of the Colorado School of Mines in partial fulfillment of the requirements for the degree of Master of
Science, Mineral Resources Ecology.
Golden, Colorado
Date ^I 11^0 Signed: ITI eorge M. Mathews, Jr.
Approved: :dward A. Howard, Ph.D. Thesis Advisor
Golden, Color^o
Date
JohnVAC Cordes Department Head Dq^rtment of Environmental Sciences and Engineering Ecology
11 T-3870
ABSTRACT
This study demonstrated significant differences in the micrometeorologic conditions associated with different types of poor visual air quality events and that these differences were linked with mesoscale and synoptic weather features.
For the study period December 1988 through January 1989, 51 identified poor visual air quality events in Boulder,
Colorado were categorized into two groups, those caused by transported pollution and those caused by locally generated pollution. The classification procedure was based on the ratio of the peak particle light scattering coefficient during an event to the mean coefficient of the 4 preceding hours. Periods of acceptably clear visibility, 10 for each month, were randomly selected from the remaining time spans.
The classification procedure was more applicable to the
December 1988 group than to the January 1989 group.
Nonparametric and parametric statistical tests were applied to seven météorologie variables for the three visibility groups. Significant differences were found among the wind speeds, wind directions, relative humidities, inversion heights, and event timings in relation to a diurnal wind shift associated with the three groups. No significant differences were detected for the net radiation and snow
iii T-3870
cover associated with the two types of poor visual air quality events.
A brief comparison of poor visibility events and high carbon monoxide periods was conducted. This project suggested that meteorology was a major factor governing the levels of fine particulates, but not of carbon monoxide, in instances of pollution transport. In contrast, during local pollution events meteorology played a minor role in the different temporal patterns of the two pollutants.
Three case study days, one each to represent acceptable visibility, transport pollution, and local pollution were examined in detail. Regional changes in the pressure gradient along the lee side of the mountains, because of movement of a synoptic low pressure trough, were seen as the driving force behind transport of a polluted air mass.
IV T-3870
TABLE OF CONTENTS
Page ABSTRACT iii
LIST OF FIGURES vii
LIST OF TABLES ix
ACKNOWLEDGMENTS x
INTRODUCTION 1 Visibility Theory 1 State of Colorado Visibility Standard 5 Previous Visibility Studies in the Denver Metro Area 6 Project Objectives 11
AVAILABLE DATA SET 12
METHODS 17
ANALYSIS AND RESULTS 21 Interpretation of a Poor Visibility Event 21 Determination of a Threshold b 21 Frequency Distributions of Nephelometer Data 29 Effect of Relative Humidity 29 Definition of a Poor Visibility Event 38 Identification of Poor Visibility Events 39 Classification of Poor Visibility Events 44 Statistical Methods 49 Météorologie Variables 49 Wind speed 49 Wind direction 51 Relative humidity 51 Inversion height 53 Time of event start in relation to a diurnal wind shift 56 Net radiation 63 Snow cover 63 Discussion 66 Relationships Between Poor VAQ Events and High Carbon Monoxide Periods 71 T-3870
TABLE OF CONTENTS (continued)
Page
Phenomenological Descriptions of Selected Case Days 74 Acceptable Visibility Period: December 03, 1988 74 Transport Poor VAQ Event: December 29, 1988 80 Locally Generated Poor VAQ Event: December 30, 1988 89
CONCLUSIONS 95
REFERENCES CITED 98
Appendix A. Visual Air Quality Events of December 1988 104
Appendix B. Visual Air Quality Events of January 1989 110
VI T-3870
LIST OF FIGURES
No. Title Page
1. Boulder Air Quality Study, Winter of 1988-1989, Monitoring Sites. 13
2. PROFS Mesonet Monitoring Sites. 16
3. Heated Nephelometer, SCT. December 1988. 27
4-a. Frequency Distribution of Unheated Nephelometer, SCT. Dec 1988. 3 0
4-b. Frequency Distribution of Heated Nephelometer, SCT. Dec 1988. 31
4-c. Frequency Distribution of Unheated Nephelometer, SCT. Jan 1989. 32
4-d. Frequency Distribution of Heated Nephelometer, SCT. Jan 1989. 33
5. Field and Calculated Unheated Nephelometers, SCT. Dec 1988. 3 5
6. Field and Calculated Nephelometers, SCT. Dec 15, 1988. 36
7. Field and Calculated Nephelometers, SCT. Dec 24-25, 1988. 37
8. Heated Nephelometer, SCT. Dec 6-7, 1988. 41
9. Heated Nephelometer, SCT. Dec 15, 1988. 42
10. Heated Nephelometer, SCT. Dec 27, 1988. 43
11. Frequency Distribution of Peak/Mean Ratios of 28 Poor VAQ Events, Dec 1988. 47
12. Frequency Distribution of Diurnal Wind Shift Times, 47 Observations, Dec 1988-Jan 1989. 57
Vll T-3870
LIST OF FIGURES (continued)
No. Title Page
13. Frequency Distribution of the Times of Event Peaks, 51 Poor VAQ Events, Dec 1988-Jan 1989. 58
14. Frequency Distribution of the Times of Event Starts, 51 Poor VAQ Events, Dec 1988-Jan. 1989. 59
15. Heated Nephelometer, SCT vs CO, XRD. Dec. 1988. 72
16. Dec 03, 1988. Wind Direction and Speed. SCT. 76
17. Dec 03, 1988. Pressure Differential: PROFS Erie - Boulder. 77
18. Dec 03, 1988. Heated Nephelometer, SCT. CO, XRD. 79
19-a.Dec 29, 1988, llOOMST. Surface Pressures. 82
19-b.Dec 29, 1988, 1400MST. Surface Pressures. 83
19-c.Dec 29, 1988, 1700MST. Surface Pressures. 84
20-a.Dec 29, 1988. Heated Nephelometer, SCT. Wind Direction, XRD. 85
20-b.Dec 29, 1988. CO and Wind Direction, XRD. 85
21. Dec 29, 1988, 1300hr. Vertical Temperature Profile, XRD. 86
22. Dec 29, 1988. Pressure Differential: PROFS Erie - Boulder. 88
23. Dec 30, 1988. Wind Direction, XRD. Wind Speed, SCT. 90
24. Dec 30, 1988. Heated Nephelometer, SCT. Carbon Monoxide, XRD. 91
25. Dec. 3, 1988. Pressure Differential: PROFS Erie - Bouler. 94
viii T-3870
LIST OF TABLES
No. Title Page
1. Statistics and Test Results for Wind Speed. 50
2. Statistics and Test Results for Wind Direction. 52
3. Statistics and Test Results for Relative Humidity. 54
4. Statistics and Test Results for Inversion Height. 55
5. Statistics and Test Results for Time of Event Start in Relation to a Diurnal Wind Shift. 61
6. Statistics and Test Results for Net Radiation. 64
7. Statistics and Test Results for Snow Cover. 65
8. Summary of Statistical Test Results for Météorologie Variables. 68
IX T-3870
ACKNOWLEDGMENTS
I am grateful to my advisor Dr. Edward A. (Al) Howard for the optimism and encouragement that he extended to me from my initial inquiries about the environmental science programs at the Colorado School of Mines through the research and writing of this thesis project. I also want to recognize the other faculty members of my advisory committee. Dr. Ronald R. Cohen and Dr. Bruce Van Haveren, for their thoughtful suggestions, advice, and enthusiastic participation. To the secretary of the Department of
Environmental Sciences and Engineering Ecology, Juanita
Chuven, I extend a heartfelt thanks for her energy and dedication to the students of the department.
I am deeply indebted to the staff of the National
Oceanic and Atmospheric Administration, Wave Propagation
Laboratory, in Boulder, Colorado. I am especially thankful to John E. Gaynor for giving me day to day guidance and teaching on this project, and for sharing his office for many months. Without his active interest and participation in these endeavors, this project could not have been accomplished. I am also grateful to David C. Welsh, Leslie
K. Lewis, and Vivian LeFebvre for their patient instruction T-3870
in computer science, and to Daniel E. Wolfe, Richard B.
Fritz, Clark W. King, Herman Sievering, Barry A. Bodhaine,
and Amy S. Wyngaard for their technical support.
Finally, I want to thank the Boulder County Health
Department staff, especially Tom Douville, Environmental
Health Program Manager, for allowing me the scheduling
flexibility to pursue my educational goals.
XI T-3870
INTRODUCTION
Like many communities along the Front Range of Colora do, the City of Boulder endures days and longer periods of visibility degrading pollution during the winter months.
While the influence of meteorology on visual air quality in nearby Denver and the South Platte River basin has been investigated in several studies, research into relationships between meteorology and air pollution of the Boulder Valley has only recently commenced. To date, examination into these relationships has addressed primarily carbon monoxide and only incidentally visibility degrading particulates. A detailed analysis of micrometeorologies, with related meso scale and synoptic weather patterns, that lead to episodes of poor visual air quality in the upper Boulder Valley has not yet been presented. This project used statistical analysis and descriptive phenomenology to investigate the relationships among micrometeorologic variables to two types of poor visual air quality (VAQ) events, those caused by pollution transport and those produced by local sources.
Visibilitv Theory
Visibility has historically been defined in terms of the greatest distance at which an observer can just distin guish a black object contrasted against the horizon sky T-3870
(Malm, 1983). Because visibility is a psychophysical human evaluation considering other VAQ components besides con trast, such as discoloration, haze layering, and plume blight, extensive research has gone into the methodologies of human observer judgments of VAQ (Stewart, Middleton, and
Ely, 1983; Stewart et al., 1984). Yet, because reliance on teams of human observers to systematically monitor visibil ity conditions can be costly and time consuming, standard ized equipment is typically used to measure visibility conditions in both urban and rural settings.
Visual range is reduced by atmospheric particles and gases scattering and absorbing light. The total reduction or extinction of light is measured by the term extinction coefficient, b^^^, which is expressed in units of inverse length, m”\ 3cm‘\ or Mm"'.
The most responsible agents of visual range reduction are particles nearly the same size as the wavelength of visible light, 0.4-0.7 micrometers, which efficiently scat ter light, primarily in a forward direction. The light scattering ability of aerosol particles is dependent on the relative humidity of the atmosphere because hygroscopic growth can affect the optical properties of the aerosols.
These effects are small below 50-70% relative humidity
(Lindgren and Cooper, 1969; Covert et al., 1980; Pilinis, T-3870
Seinfeld, and Grosjean, 1989). Atmospheric particles, prin cipally carbonaceous, absorb less light than they scatter, absorption effects being on the order of 10% of scattering effects in clean rural areas and perhaps equal in urban areas. The scattering of light by atmospheric gases, known as Rayleigh scattering, constitutes a baseline level of light extinction in pristine atmospheres against which all other factors are additive. Rayleigh scattering depends on air density and is thus inversely proportional to altitude.
The only significant light absorption due to gases is at tributable to nitrogen dioxide which is usually of less importance in urban atmospheres than in combustion source emission plumes (Waggoner et al., 1981).
An instrument frequently used to evaluate visibility is the integrating nephelometer (Ruby, 1985), which directly measures the light scattered by particles and gases in a sample volume integrated over essentially all scattering angles over a weighted range of visible wavelengths. The sample volume is illuminated by a side lamp behind a diffus ing window. Light scattered by the gases and suspended particulate matter in the sample volume is measured by a photomultiplier tube. Since the light scattered out of the sight path is the same as the reduction of light along the sight path due to scattering, the integrating nephelometer T-3870
gives a direct measure of the scattering extinction coeffi cient, . The term b^^^^ is comprised of a scattering ex tinction coefficients due to the particulate fraction, b^^, and the gaseous fraction, b^g, or Rayleigh's coefficient.
The term can be related to meteorological visual range by the Koschmieder equation:
A = 3 -
(Horvath and Noll, 1969) but reporting results as has not been recommended because of significant assumptions that must be made in such a conversion (Ruby and Waggoner, 1981).
The term can also be related to mass concentration of fine particulate matter less than 2.5 um diameter:
Mass (ug/m^) = 3 . 8*10^ (b^^^^)
(CharIson et al., 1969). For optical purposes, an unheated nephelometer is recommended to approximate the conditions observed by the human eye. For mass concentration purposes, a nephelometer with a heated sample line to drive off water associated with hygroscopic particles is recommended (Ruby
1985). During the calibration procedure, the instrument is purged with a freon gas to remove the Rayleigh scattering T-3870
effect (Ruby and Waggoner, 1981).
State of Colorado Visibilitv Standard
Degradation of visual air quality, characterized by atmospheric discoloration, haze layering, plume blight, and particularly brightness contrast, has been of increasing concern to environmental groups, the general public, regu lators, and elected officials since the 1970*s. Formalized into the 1977 Amendments of the Clean Air Act, this aware ness was primarily directed toward national parks and pris tine areas. Public interest in visual air quality in urban areas has grown since the passage of the amendments to the
Act. On December 21, 1989 the Colorado Air Quality Control
Commission adopted a visibility standard for the metropoli tan Denver area, well known for its winter brown cloud
(Colorado Air Quality Control Commission, 1989).
Both the Governor and the General Assembly directed the
AQCC to establish such a standard, and it is based on the public's opinion regarding the level of acceptable visibil ity. The visibility standard is a total atmospheric ex tinction of 0.76 km‘\ equivalent to a standard visual range of 32 miles. The averaging time is four contiguous daylight hours when relative humidity is less than 70%. The standard will be effective beginning 1993. T-3870
Previous Visibilitv Studies in the Denver Metropolitan Area
The role of météorologie conditions in air pollution episodes has been recognized for some time. Stable air masses of prolonged anticyclonic high pressure systems with secondary inversions, accompanied by the presence of fog and occurring in autumn or early winter, characterized the air pollution disasters in the Meuse Valley, Belgium, 193 0, in
Donora, Pennsylvania, 1948, and in London, 1952 and 1962
(Cassell, 1969).
In Colorado, the meteorology, as well as the constitu ents and sources associated with the Denver brown cloud, have been examined repeatedly in increasingly complex and detailed studies. Crow's report (1976), as part of a 1973
Denver air pollution study, sponsored by the U.S. Environ mental Protection Agency, is notable for several reasons.
The 1973 study encompassed the Denver basin from Rocky Flats on the west to Buckley Air National Guard Base on the east.
The Boulder Valley was not included. Crow discussed the typical diurnal windshift pattern, from downslope to upslope in the forenoon and then back to downslope in the late afternoon, and he emphasized the importance of the stability of the air profile in determining the transport, vertically and horizontally, of pollutants. Rather than météorologie conditions which set up air pollution episodes, he focused T-3870
on describing two météorologie mechanisms for cleansing the city of air pollutant accumulations; strong synoptic crosswinds and a chimney effect produced by the urban heat island. He related three pollution episodes in November and how météorologie events interrupted them.
In November and December 1978, the General Motors
Research Laboratories and the Motor Vehicle Manufacturers
Association organized an extensive study to determine the causes of Denver's brown cloud. Groblicki, Wolff, and
Countess (1981) and Wolff et al. (1981) described in joint articles the contributions of various chemical species and emissions sources to the fine particulate mass loading, light extinction, and visual range reduction in Denver. As
Crow did, Ferman, Wolff, and Kelly (1981) discussed the average diurnal pattern and indicated that the cause for this pattern is the formation and destruction of the noc turnal radiation inversion. During the 1978 study, carbon monoxide readings were highest in the morning. One extended pollution episode in mid-December was discussed with focus on the strength and persistence of the ground-based inver sion.
A subsequent study of Denver's winter haze, sponsored by the U.S. ERA, was conducted in 1982. Lewis et al. (1986) reported that research, and it also concentrated on source T-3870
apportionments of the fine particulate mass and light ex tinction, not on meteorology.
The 1987-88 Metro Denver Brown Cloud Study was multi purpose in its dimensions. It attempted to interlock the disciplines of meteorology and atmospheric chemistry to characterize quantitatively the relationships between visi bility reduction and pollution source apportionment. The descriptions of the météorologie phenomena that affect the formation, duration, and dispersion of high pollution epi sodes were expanded significantly over previous reports; drainage flow, return flow, stagnation, upslope snow storm, upslope circulation, and downslope westerlies. The study compared and contrasted the meteorologies of two extended high pollution episodes and narrated the effects of several weather features on visibility and pollutant transport and transformation (Watson et al., 1988). Again the Boulder
Valley was not specifically examined.
Growing out of the 1987-88 Metro Denver Brown Cloud
Study, other research into the météorologie and pollution relationships in the Denver basin has continued. In review ing four years of data. Summers, Neff, and King (1989) reported that a decrease in surface barometric pressures along the lee side of the Colorado Rocky Mountains accompa nies most episodes of declining air quality in Denver. The T-3870
presence of a 500 mb ridge on the windward side of the
Rockies was also noted in most cases. Wolfe and Gaynor (1989) examined a particular high pollution day, December
29, 1987, for météorologie factors contributing to the episode and concluded that not only the presence, but also the strength and timing of synoptic patterns are important.
King and Russell (1989) investigated the interrelationships of temperature structures in comparisons of urban Denver and outlying rural locations. They also examined the vertical visibility distribution and vertical temperature gradient in a downtown Denver setting. Ruffieux (1989) modeled the micrometeorology on the shadow side of the tall downtown
Denver buildings.
Although Boulder is often grouped with the Denver metroplex for social and political purposes, the air quality and meteorology of the Boulder Valley are sufficiently distinct to warrant a particular examination of this local ized regime. Gaynor and Wittenmeier (1989) reviewed nine teen occurrences of elevated carbon monoxide levels over almost seven years in the City of Boulder and correlated these with wind speed and direction. They noted frequencies of low wind speed and east northeast direction and discussed the weakened influences of the lee side surface low pressure trough and the windward 500 mb pressure ridge, described by T-3870 10
Summers, Neff, and King (1989). Gaynor (1989) selected four case days in December 1987 that exhibited high levels of carbon monoxide for Boulder and observed commonalities of lack of the normal westerly flows and presence of low inver sions. He analyzed evidence for a three-layered system above the Boulder Valley: a shallow, cold northeast flow under a layer of light, cold mountain drainage, both of which were beneath strong, warm, synoptic westerlies.
During the 1988-89 winter pollution season, the meteo rology of the Boulder Valley was investigated in association with levels of gaseous and particulate pollutants. More limited in scope than the Denver research efforts, the
1988-89 Boulder Air Quality Study (BAQS) focused on the influence of meteorology on carbon monoxide pollution and, to some extent, visibility degradation, and on source appor tionment of visibility degrading constituents. Concerning the météorologie questions, the BAQS concluded that regional weather patterns and the City of Boulder's topographic situation reduce the opportunities for long-range transport of pollutants into the upper Boulder Valley but rather promote the buildup of locally generated pollutants. Specif ic occasions of pollutant transport down as well as up the
Boulder Valley were recognized (Gaynor et al., 1989). T-3870 11
Project Objectives
By describing typical météorologie conditions that prompt high pollution episodes in Front Range Colorado settings, these prior investigations provided an excellent foundation for the project of this thesis, which examined specifically the causal relationships of the Boulder Valley meteorology to periods of high concentrations of visibility degrading air pollutants.
This research hypothesized that two kinds of wintertime poor visual air quality (VAQ) events in the City of Boulder can be identified: those caused by advection of pollution into the city from outside locations and those produced by immediate local pollution sources. The micrometeorologic conditions that induce these two distinct categories of poor
VAQ events are definable and distinctly different, from each other and from periods of acceptable visibility. Both statistical methods and phenomenological approaches were enlisted to investigate the météorologie differences between the two categories of poor VAQ events. T-3870 12
AVAILABLE DATA SET
The Boulder Air Quality Study, Winter of 1988-89, provided a core data base for relating visibility degrading pollution levels to météorologie conditions.
Sponsored by the City of Boulder, the BAQS was a coop erative effort involving the National Oceanic and Atmospher ic Administration (NCAA), the University of Colorado at
Denver, the National Center for Atmospheric Research (NCAR), the Colorado Department of Health, and the Boulder County
Health Department. The main body of data is stored at the
NCAA Wave Propagation Laboratory (NOAA-WPL) in Boulder.
These data were gathered from nine sites, eight of which were more or less aligned along the Boulder Valley drainage. Figure 1 maps the monitoring sites of the BAQS.
While not all parameters were sampled at each location, the following presents a list of the meteorological and pollu tion parameters, with the instrumentation manufacturer when information was available, measured during the study:
1. ambient air temperature, Campbell Scientific, Inc.,
2. dew point temperature, EG&G cooled mirror,
model 110, T-3870 13
W Q) •H-p CO S' •H p O P> •H c o s '•1,
CT> a\CO rH I • } | ': - - 4 CO CO ! Jf. G\ grmt! (W o u o -p c •H IS
, J k F ^ 73> 1 O 4J CO >1 4J •H rH <0 a p ■H < u Q) 73 rH 0 o CQ >. :. _ (Ü r / .K1 tJ£U f-V) «5U
0) p 3 U» •H (H T-3870 14
3. barometric pressure, variable instrumentation with site,
4. solar radiation, Campbell Scientific, Inc.,
LI-COR 200,
5. net radiation, Campbell Scientific, Inc., Radia
tion Energy Balance Systems,
6. wind speed. Met One Instruments,
7. wind direction, Met One Instruments,
8. carbon monoxide, Beckman model 866,
9. intensity of light scatter at ambient relative
humidity, MRI integrating nephelometer, model
1550,
10. intensity of light scatter at low relative
humidity, MRI integrating nephelometer, model
1550, with sample line preheater (Ruby and
Waggoner, 1981),
11. condensation nuclei. Thermal Systems, Inc. (TSI),
Model 5000,
12. aerosol particle chemical constituent analysis,
proton induced x-ray emission analysis by Elemen
tal Analysis, Inc.,
13. crosswinds, NOAA-WPL designed sensor, and
14. atmospheric turbulence or vertical temperature
structure, NOAA-WPL designed acoustic mini-sodar. T-3870 15
Data for all parameters, with two exceptions, were averaged over 60-minute and 5-minute time periods. Chemical con stituent analyses were collected in 8-hour intervals, and sodar soundings were available on continuous strip charts.
As a supplement to the BAQS monitoring sites, a reposi tory of continuous météorologie data collected at the PROFS mesonet stations (Pratt and Clark, 1983) resides at NCAA.
These twenty-two stations, shown in Figure 2, are situated throughout northeast Colorado from the Front Range mountains out onto the high plains. Data measured are:
1. ambient air temperature,
2. dew point temperature,
3. barometric pressure,
4. wind speed,
5. wind direction,
6. horizontal radiation,
7. 40 deg angle radiation, and
8. precipitat ion.
Those PROFS sites located in or near the Boulder Creek, St.
Vrain Creek, and South Platte River drainages were of prima ry interest to this study.
The National Weather Service provided synoptic scale and historical weather information. T-3870 16
A NUN
Nunn A BCD FOR Bnggsdale Fort Collins
LVE Estes Park Loveland GLY Greeley EPK FTM A PTL Longmont Fort Morgan Platteviile LGM^m Keenesburg WRO Boulder ERI BRI Ward
Roiltnsviile SOU Brighton
Fritz Peak Arvada ARVA AUR BYE Idaho Spring) Lakewood] ISG Aurora Byers Squaw LAK Mountain Littleton LTN
ELB Elbert
30 kilometers A Station with forward scatter 30 miles meter Contour interval 2500 feet
Figure 2. PROFS Mesonet Monitoring Sites T-3870 17
METHODS
This project was divided into four main activities; defining, identifying and classifying poor VAQ events from the available BAQS data set; preparing a data base of se lected météorologie variables and applying statistical analyses to ascertain the differences between types of poor
VAQ events; briefly studying the differences in the meteorologies that affect levels of fine particulate and carbon monoxide pollution; and examining in detail selected representative case study days in order to integrate the separate météorologie variables into a comprehensive whole.
The Colorado standard will be measured by transmis- someter instrumentation, which provides total atmospheric extinction values (Malm, 1983). The BAQS, however, collect ed data by nephelometer instrumentation, which provides light extinction values due only to scattering by gases, aerosol particles, and physically associated water (Charlson et al., 1969; Ruby and Waggoner, 1981; Ruby 1985). In order to utilize the BAQS data, it was necessary to compute a particle scattering coefficient, termed a threshold b^^, that was a surrogate for the total extinction coefficient standard. This was accomplished by calculation from the equivalent visual range standard, after accounting for and T-3870 18
subtracting the particle absorption coefficient, the gas absorption coefficient, and the Rayleigh scattering coeffi cient for air at ambient elevation. To assist in the selec tion of a threshold b^^, frequency distributions of the 60- minute averages of unheated and heated nephelometer values for the months of December 1988 and January 1989 were plot ted. Because relative humidity affects the particle scat tering coefficient, this variable was considered when se lecting a threshold bsp. To objectively identify periods of visibility degradation from the extensive BAQS data base of hourly particle scattering coefficients throughout December
1988 and January 1989, a very specific extraction definition of a poor visibility event was formulated and a computer program written to perform the extractions. The actual start of a poor visibility event was pinpointed by examining the 5-minute data traces of the initial hours of an event and the several hours preceding an event. A classification procedure for dividing the identified poor VAQ events into the categories of transport and local events was devised and applied.
A data base of seven meteorological variables of the identified poor VAQ events and randomly selected acceptably clear periods was constructed in preparation for performing the statistical analyses. The data base was divided into T-3870 19
two separate months and the months combined and into trans port, local, and clear events. That portion of the data base which included wind speed, wind direction, relative humidity, and net radiation was compiled from the BAQS data set into a data base file in the commercial MATLAB software format. That portion which included inversion height, timing of the event in relation to a diurnal wind shift, and snow cover was manually tabulated. Data collected at the
Scott Carpenter Park site were utilized unless flagged, in which case data from the Crossroads site were used. All sodar information for determination of inversion heights was collected at Crossroads. National Weather Service records provided information about presence of snow cover on the ground for the City of Boulder.
Initially, nonparametric tests were applied because the number of events in each sample was usually less than 30, because the population distribution may not have been nor mal, and because the data were not quantitative for certain factors. All statistical methods that were utilized in this project are described in Walpole and Myers (1989). The nonparametric Kruskal-Wallis H test was applied to evaluate the presence of a significant difference among three data sets: transport pollution events, local pollution events, and acceptably clear events. The nonparametric Mann-Whitney T-3870 20
test, also known as the Wilcoxon rank-sum test, was applied to evaluate the presence of a significant difference between the two data sets, transport and local pollution events.
The nonparametric sign test was applied to single data sets as noted. When possible, a t-test was performed in order to quantitatively bracket the difference between the météo rologie variables of the groups. In all tests a 0.05 level of significance was chosen.
A preliminary examination into the météorologie factors affecting fine particulate and carbon monoxide pollution was conducted by plotting a scatter diagram of b^^ values versus carbon monoxide values and by comparing pollutant levels on specific case days.
Three case study days, one each to represent acceptable visibility, transport pollution, and local pollution, were examined in detail. T-3870 21
ANALYSIS AND RESULTS
Interpretation of a Poor Visibility Event
Determination of a Threshold b^^
The computation of a particle scattering coefficient, threshold b^p, that would substitute for the total extinc tion coefficient standard began with the visibility standard for the Denver metropolitan area. It is an atmospheric extinction, b^^^, of 0.076 km’^, or 76*10'^ m‘\ equivalent to a standard visual range, L^, of 32 miles (Colorado Air
Quality Control Commission, 1989). As discussed above
(Appel et al., 1985), five components are additively ac countable for total atmospheric extinction:
^ext “ ^sp-dry ^sw ^ ^sg ^ap ^ag where
^sp-dry “ the scattering coefficient for aerosols with out associated liquid water,
bg^ = the scattering coefficient for liquid-phase
water associated with hygroscopic aerosol
constituents,
bgg = the Rayleigh scattering coefficient for air,
bgp = the absorption coefficient for particles,
and
bgg = the absorption coefficient for gases. T-3870 22
In the discussion below, ^^p-dry ^sw termed b^^. The light extinction components other than the particle scattering coefficient were estimated from the research conducted in 1978, 1982, and 1988-89. Groblicki, Wolff, and
Countess (1981) summarized the mean extinction measurements of the 1978 Denver Winter Haze Study:
Parameter Value (10’^ m*^) Percent
bap 66 28.9
^sp-dry 92 40.4
bsw 55 24.1 (^sp-dry"^^sw 147 64.5)
bag 15 6.6
b'axt 228 100.0
The total of the extinction coefficients, b ' , does not include the Rayleigh scattering coefficient.
In the 1982 investigation of Lewis, Baumgardner, and
Stevens (1986), light extinction values averaged over day time and nighttime findings were reported: T-3870 23
Parameter Value (10'^ m*'’) Percent
bap 22.4 17.5
^sp-dry 66.2 51.7
bsw 28.6 22.3 (^sp-dry"^bsi, 94.8 73.0)
bag 10.8 8.4
b*ext 128.0 100.0
In the 1988-89 Denver Brown Cloud Study, Watson et al.
(1989) determined the average extinction values for periods when only gas or only coal was burned at all area power plants. For the needsI of this research. only those values for the coal burning periods were of interest because the
BAQS was conducted only when coal was burned throughout the
Denver-Boulder area. The Watson investigation team summa- rized their findings:
Parameter Value (10'^ m’^) Percent
b.p 41.0 33.8 ^sp-dry^^sw 67.8 55.8
bag 12.6 10.4
b'ext 121.4 100.0 T-3870 24
From the information reported in these three studies, the average ratio of particle absorption to the combined particle with associated water scattering (b^p.y^y+bg^) ] was calculated. The same was done for the ratios of gas absorption to the combined particle with associated water scattering [b,/ (b,p.^,y+b,J ] •
Average of Groblicki (1981) Lewis (1986) Watson (1989) 3 studies
(^a/^sp-dry*^sw) 66/147=0.449 22.44/94.8=0.236 41.0/67.8=0.605 0.430
(^ag/^sp-dry^^sw) 15/147=0.102 10.8/94.8=0.114 12.6/67.8=0.186 0.134
Then,
b'ext = [ (b,p dry+b,J + + b„]
= (b,p dry+b.w) +0 • 430 (b,p.^^+b„) +0 . 1 3 4 (b,p.^,^+b,J and
b'ext = l-564(b,p_j^y+b,J.
Since
bsp bgp-jjry + bgy,
b ’ext = 1-564 b,p.
Standard visual range, L^, is related to light extinc tion by the Koschmieder equation, = 3.S/b^^^ (Horvath and
Noll, 1969). The format of the eequation utilized by the
State of Colorado to account for the elevation effect on the T-3870 25
Rayleigh coefficient and the conversion from miles to kilo meters is:
= [3.912/(b„, - b^g + 0.01 km'’)] * 0.6214
Since
bext " ^sg “ ^ ext and
b'ext = 1-564 b,p then,
= 32 mi = [3.912/(1.564 b^p +0.01 km'^) ] * 0.6214
bgp = 0.042 km"^ = 42 * 10’* m'\
This, then, was the particle scattering coefficient, mea sured by the unheated nephelometer during the BAQS, substi tuted for the total extinction coefficient of the Colorado visibility standard, that was not measured in the BAQS.
Ruby (1985) advised that the data set from the unheated nephelometer should be used because this set represented light scattering values under ambient conditions where both the fine particulate mass and the atmospheric water associ ated with the particulates reduce visual range as observed by the human eye. Unfortunately, it did not prove practical to rely on the unheated nephelometer data for two reasons. T-3870 26
First, the data recovery was insufficient: 84.9% in Decem ber 1988, but only 25.6% in January 1989. Data recovery for the heated nephelometer over the same time periods was
97.3% and 97.6%, respectively. Secondly, the unheated nephelometer was calibrated for only a three week period beginning December 19, 1988 and ending January 8, 1989.
Since the correlations between the two instruments were high, 0.9432 for December and 0.9725 for January, substitu tion of heated nephelometer results for unheated nephelome ter results, with appropriate adjustments, was justified.
The heated nephelometer values for December 1988 (Fig ure 3) indicated that there was an apparent baseline below which the heated nephelometer did not record. At the time of the research, it was known that the nephelometer instru ments had been zeroed with freon gas to eliminate the Ray leigh scattering effect. By this calibration procedure, nephelometers would register zero during times of exception ally clean air. The apparent baseline was then assumed to be an unidentified equipment limitation, and an attempt was made to estimate its value. Baseline periods were chosen and the average heated nephelometer readings calculated.
The instrument baseline for the heated nephelometer was estimated to be 24 * 10'^ m"**. T-3870 27
o CD
m o o cn C\J 00
I—Ii CD
m o 0) U •Hg-
o o oo o oo o o o in o in o in in m m C\J CM
a59J3A9 a:^nuTuj-og -ujg-goT *Ajp-dsq T-3870 28
Date/Hour Average b^p.^^y (10'*
12/12/88,0000-12/14/88,2300 28.78
12/22/88,0000-2300 23.83
01/03/89,0000-2000 22.97
01/05/89,0000-20000 24.40
01/06/89,0100-01/07/89,1800 23.85
01/09/89,0000-01/10/89,1600 24.57
01/21/89,0300-2300 24.29
However, it was later learned that the apparent base line was indeed the Rayleigh scattering coefficient which had been re-entered into the computerized data set. At
Boulder's elevation, the Rayleigh scattering coefficient is
17-18*10"^ m"^ (Ruby and Waggoner, 1981). It was retrospec tively decided that the difference of 6-7 units would not have significantly altered the results of the poor VAQ event extraction process described below, and the estimated equip ment baseline was used for the baseline Rayleigh scattering coefficient.
The estimated threshold bgp_y^y, 42*10"^ m"\ was added to this apparent instrument baseline, 24*10 1-'^ m’\ to give
66*10'^ m'% an adjusted threshold for determining a poor visibility event.
AlTmra LAKSS LÎ3B.AEY C O L O m m O flCSOOl. d MINES GOLDEN, COLORADO 80401 T-3870 29
Frequency Distributions of Nephelometer Data
The frequency distributions of the 60-minute averages of unheated and heated nephelometer values for the months of
December 1988 and January 1989 are shown in Figures 4 a-d.
The distributions indicated a breakpoint in the 60-70*10'^ m'^ range. They also demonstrated the high number of non functional unheated nephelometer values.
Effect of Relative Humidity
If the heated nephelometer data set was to be used, the significance of associated water in visibility deterioration had to be remembered. It was necessary to add an adjustment factor to the heated nephelometer results to infer the scattering coefficient of the liquid water associated with the aerosol particles, b^^, that was removed by the heating process of the instrument. The b^^ value needed to be added to the bgp_y^y value of the heated nephelometer only when the ambient relative humidity exceeded 70%, since below this level the effects of associated water on visual range were negligible (Covert et al. 1980). Gaynor et al. (1989) in the BAQS determined that 37% of the visibility reduction in
Boulder for the study period was due to particle associated water. T-3870 30
CO CO in o r—I ÜQ) Q 54 o U o CO
Q) ti i cn r4 OJ o oi 0) fO m c_ 0) §■ m> z OJ 4D -1 C (0 Q) o OI € CD D 4-1 O I E C CD o I •H OLU +j o o •H CL JD (0 •H Q >1 Ü § g- 0) k k
<0 I o o o o o o ru o
co co
o U 0) U k A I UQ) san^GA ; o jaqiunN & T-3870 32 o o 0) o ■p 0 en e QJ rH (Uen (U c_ CD 1 (13> 2 Q) D -P C (0 d) E o I jG o O c tû D o C ID o I -H o LUO 4J •H CL P m P> D Cfl •H o Q > 1 ü c Q) o 0) D P (P ü I o O o o o o o O o o o LO D m ru 0) P G saniBA io jaquunN T-3870 33 a\ OJ 00 p H c (0 h) o o k d) cn cnOJ m r4I c_ CD Q) > I Z o 3 T3 C 0) E t (0 O Q) UD S (N I o E LO c in I o LU •H O 4-> >> •H c_ ■Q 4J I U) a jQcn & § o 2 T3 I o o o o o o o Q> U & santeA ^o jaqujriN •H T-3870 34 To summarize, when ambient relative humidity was less than or equal to 70%, the heated nephelometer value, was substituted directly for the unheated nephelometer value, bgp = bgp_j^y+bg^. When ambient relative humidity was more than 70%, the heated nephelometer value was multiplied by 1.37 to provide an estimated, or calculated, unheated nephelometer value. The validity of this procedure was verified by examin ing graphic comparisons of direct field data from the un heated nephelometer with unheated nephelometer numbers theoretically calculated. Figures 5-7 show these plots for the entire month of December 1988 and for two selected days. December 15, 1988 was a day when the ambient relative humidity exceeded 70% all day, and the correlation between the field and calculated unheated nephelometer values was 0.9957. December 24, 0800, - December 25, 0700, 1988 was a time period when the ambient relative humidity exceeded 70% during the last six hours. T-3870 35 o o CO CO 00 cn U Q) a cn u 0 ) +i J rH Q) c_ I z 03 I: 1 (U <0 0 ) P 13 CO 0 ) CO CJl fH( 0 c_ : 3 sz sz QJ a C L O Ü QJ cu e rH c c QJ <0 U U n "O QJ QJ QJ O -U J-» ? fü fD (0 fZQJ JC QJ 3c 3 c CD T3 13 •H r - l u QJ"-I r—(fO u t in o o o o o o o o o o o 2 o in o in o in o in o in in m m rvj OJ & •H 969J3A9 a;nuiLU-09 -ujg-gg; ’dsq T-3870 36 in m OJ CO CO rH in <—I Ü0) Q 54 U w w M 0) c (Ü D § a fH c VO 2 o o o o o o o o g- o in o m •H m n CM OJ sSejaAg a;nuTuu-og ’t-ujg-goT ‘Ajp-dsq s dsq T-3870 37 00 CO 0 > «H lO 03 I r^o OJ c_- OJ Œ Ü O 0 ) o a OJ c 3 o JC in u 0 ) OJ -P E 0 ) I— § rH o 0 ) o o 0 ) z 00 T3 CL CL \ (U \n CL J 3 d OJ \ (0 n OJ 1-4 JZ 0 CL o o r—I CL o (Q oCD U CO T3 CO C <Ü OJ 'Ü \ r—I OJ (U TD •H CO CL OJ. Q) P o o o o o o o o o o o CO UO OJ o 00 OJ OJ & •H aôejSAe a:^nutuj-09 'T-wg-30T ‘Ajp-dsq s dsq T-3870 38 Definition of a Poor Visibility Event Periods of visibility degradation were extracted from the extensive BAQS data base of hourly particle scattering coefficients throughout December 1988 and January 1989 by formulating a narrow definition of a poor visibility event and writing a computer program to perform the conforming extractions. A poor visibility event was defined to occur when at least two consecutive 60-minute bsp_dry values (or 1.37*bgp_ypy when the relative humidity exceeded 70%) exceeded both the threshold b^p, 66*10’^ m“^, and the average of the previous four b^p.^^^ values, and to end when a sequential bgp.dry value fell below the threshold. It should be noted that this definition did not coin cide with the State of Colorado's definition of a violation of the visibility standard, which requires an averaging time of four consecutive daytime hours, when ambient relative humidity is below 70%. This definition was constructed for extraction purposes only. Both its strength and weakness lay in the emphasis given to the commencement of a poor visibility event. While strict criteria disregarded modest elevations in b^p and hence highlighted serious degradations in visual air quality, the actual beginnings of the rise in concentrations of light scattering fine particles may have been missed. It then became necessary during later analysis T-3870 39 to return to the data base to verify the actual start and end of the extracted event and to assure that the event was discrete. Spikes in the data were avoided, which allowed a focus on more significant occurrences in elevations. On the other hand, a singular circumstance may have escaped identification or have been blurred into a more massive occurrence. On the basis of this definition, 28 poor visibility events were extracted from the December 1988 data set and 24 from the January 1989 set. Identification of Poor Visibility Events from the Data Set The extraction procedure highlighted, on the basis of the imposed definition of a poor visibility event, those contiguous hours during which the 60-minute averages of the particle scattering coefficients exceeded a calculated threshold value. In order to pinpoint the actual start of a poor visibility event, it was necessary to examine the 5-minute data traces of the initial hours of an event and the several hours preceding an event. Even then, the specific start of a poor visibility event was usually not immediately evident. Three examples will illustrate the difficulties encountered in this process. T-3870 40 Figure 8 shows the heated nephelometer results for the hours preceding the extracted poor visibility event, 12/6/88,2300 - 12/7/88,0100. The initiation of the event could have beeen identified as occurring at 12/6/88, 122 5, 1505, 1810, 2125 or 2320. Figure 9 presents the gradual increase for the hours preceding the 12/15/88,1200-1300 extracted event. The question here was to determine when a non-event evolved into a poor VAQ event. Figure 10 shows that the extracted event for 12/27/88, 1500-2200 commenced quite suddenly. It was important not to advance too far ahead of the beginning of this event, or any event, so that the météorologie factors of this abrupt change could be directly studied. To resolve these ambiguities and impose a uniform standard of identifying the start of an extracted event, the following rule was applied; The start of a poor visibility event commenced at the time of the minimum 5-minute value within the hour recording a b^p.^^^ 60-minute average greater than 66 units, or within the preceding hour. In the examples cited above, the event on 12/6/88 started at 2320, the event on 12/15/88 at 1100, and the event on 12/27/88 at 1505. T-3870 41 in C\J CO o c\00 o I—( I VO Ü g o u Ü3 u Q) i rH o Q) t z OJ 73 Q) o (0 o Q) 53 o 00 2 & •H o ooo o o o CM 00 VO CM X BuiuuiBaq [eAja;ui aqnutuj-g aq; jo ‘t-ujg-aoT ‘Ajp-dsq ueaw T-3870 42 o C\J 00 GO CT> r4 ir> o o ü g S en o E 00 1 u in 0) 4->i rH ru o Q) OJ k •Hg. O o o o o o o 00 uo in m OJ X BuiuuiBaq leAjaqui aqnuiuu-g aqq q ‘t- uj9-30î ‘Ajp-dsq ueaw T-3870 43 o o OJ ru o en -p& 00 o 00 CD ro -Pd- > c_ in U 0) -P * [ -GO O Q) LD g r H 0) un un O ru » I z o o 73 o o o o o o o o o un o un o un o un m m ru ru X Buiuurôaq iSAja^ur aqnuruj-g aq; ‘î-mg-goî ‘Ajp-dsq ueaw T-3870 44 The end of the beginning of a poor visibility event was identified as the peak value recorded before the 5-minute trace showed a decrease in bg^.y^y. To identify the actual end of an entire poor visibility event was not necessary in this project, since its focus was on the météorologie factors contributing to the establish ment of poor visibility events not the breakup of these periods. As a result of carefully examining the heated nephelometer values, for each of the 52 extracted events, 1 event in December was recognized as 2 distinct occurrences, and 2 events, 1 in each month, contained too much flagged data to be studied. The identification total became 51, with 28 in December 1988 and 23 in January 1989. This set of 51 identified poor VAQ events was the subject of the statistical analyses described below. Classification of Poor Visibilitv Events The 51 identified poor visibility events were classi fied into two groups: events hypothetically caused by transport of pollutants into the city and events hypotheti cally produced by immediate local sources of pollution. Several methods of classification were explored but rejected. The first depended upon the time of day of the T-3870 45 start of the poor visibility event. However, this tech nique was abandoned when it was realized that time of day was actually a reflection of the concurrent meteorology. Since the purpose of the project was to determine the sig nificance of météorologie variables, any classification on the basis of time would have been a confounding factor. The classification scheme had to be based solely on the charac teristics of the bgp_y^y values of the event. The second technique attempted to calculate the rates of rise in ^sp-dry values during the beginning of events. A frequency distribution of the mean db^p.^^^/5-minutes values did not produce any definitive breakdown of the 51 events into the two groups. A corollary procedure that was examined was a classification based on the first maximum, or on the total of the first and second maximum, 5-minute rise in b^p.^^y value during the beginning of an event. Although more successful in producing a frequency distribution with clear breakpoints, this technique was also not as definitive as required. The fourth unsuccessful approach was the calcu lation of the cumulative sum of the b.^sp-apy values from the identified start to the peak of each event, a close approx imation of the area under the curve. The selected method was a comparison of the peak event ^sp-dry value with particle scattering coefficient values of T-3870 46 hours preceeding the event. Munn described the use of a "widely adopted statistic in air pollution studies," the peak-to-mean ratio (Munn, 1970). This P/M ratio is the maximum concentration averaged over a short time interval divided by the mean concentration for some longer interval. The P/M ratio is used mainly in dispersion studies involving stationary sources. Its application for this project was a modification to study the area-source problem of a city. This project hypothesized that the maximum b^^^y^y values attained during transport poor VAQ events in relation to the values preceding the events would be higher than those reached in local events. The ratio of the peak bgp.y^y value of an event divided by the mean of the four hours preceding the identified start of that event was calculated for each identified event. On the basis of the frequency distribu tion of the P/M ratios for the poor VAQ events of December 1988, presented in Figure 11, a breakpoint ratio of 4 was selected. Events which exceeded the P/M ratio of 4 were classified as transport incidents, while those with P/M ratios below 4 were categorized as locally generated events. For this breakpoint decision, the project relied on the frequency distribution pattern for December 1988 events T-3870 47 in V s s I u n o C- o D cu o c 00 c_ OJ o c o a in o •H 4J OL iS tn a XD § c m OJ E \ (0 01Q s nTD 0) k s;uaA3 ^o jaquunN & •H pH T-3870 48 alone. Several clearly displayed a trace that sudden ly rose dramatically, an indication of transport of fine particlulate pollution to the monitoring site. When the method was extended to the January 1989 data set, it ap peared that this classification technique would probably be of questionable utility because many of the January 1989 events so categorized as transport did not display the similar pattern of sudden rise in bgp.y^y values. Neverthe less, the P/M ratio classification scheme was utilized on the January 1989 group of poor VAQ events, and the decision was made to apply any statistical methods to each month separately as well as to the two months combined. The P/M ratio classification system divided the 28 identified poor VAQ events of December 1988 into a group of 8 hypothetically transport events and a group of 20 hypo thetically local events. The system broke the 23 identified poor VAQ events of January 1989 into a group of 8 hypothet ically transport events and a group of 15 hypothetically local events. Two samples of acceptably clear visibility days, 10 each for December 1988 and for January 1989, were selected at random from those a.m. and p.m. times of days that had not been extracted as poor visibility events, as defined. T-3870 49 Appendices A and B present the December 1988 and Janu ary 1989 extracted events, times of event starts and peaks, and the numbers used in the various classification ap proaches. Statistical Methods Because the categorization procedure was judged more reliant for the December 1988 events than for the January 1989 events, the statistical tests on the seven météorologie variables were performed on the data for each month sepa rately and on the combined data for both months. Météorologie Variables Wind Speed. The 5-minute wind speed averages for the first 60 minutes of the identified start of a poor VAQ event were averaged. Data and test results for wind speed are summarized in Table 1. The mean wind speeds during the first 60 minutes of the periods of acceptable visibility were significantly higher than those during the first 60 minutes of poor VAQ events, whether caused locally or by transported pollution. The mean wind speeds during the transport poor VAQ events were higher than those during the locally caused events; however, the difference was low, less than 2 m/s. T-3870 50 i— * 5 ir> w a - s OC»— .A-w ^ 5M « £ 'e « S im 73 A 0) g = #5 Q> Wind direction. The 5-minute wind direction averages for the first 60 minutes of the identified start of a poor VAQ event were vectorially averaged. Data and test results for wind direction are summarized in Table 2. In the case of wind direction, a t-test was not applied. Instead, the sign test with binomial probability of 0.25 assigned to the northeast quadrant was performed on each of the three groups to estimate the importance of a northeast component to the wind direction. The mean wind directions during the first 60 minutes of the poor VAQ events, whether caused locally or by trans ported pollution, were significantly different from each other and from those during the first 60 minutes of the periods of acceptable visibility. There was a high proba bility that poor VAQ events caused by transported pollution were initiated by winds from the northeast quadrant. This could not be said of locally generated poor VAQ events, with which more variable winds were associated. Relative humiditv. Relative humidities were calculated from the 60-minute averages of the ambient temperature, dew point temperature, and pressure data collected at Scott Carpenter Park. The 60-minute average for the hour during which a poor VAQ event or acceptably clear event started T-3870 52 I ra C4 ro «» II 3 wo wo g 5: CO % T?S 5 C o •H ro 2 II •H 2 £ o A 2 3 c 0» •H wo S ^ p i i i «n o O' 4# s 50 9 S ^ g( <» — «» <> 51 0 - Ï is? s 2M 2 V c ? % 5 u 1 j i ill o ll§g§ 10 p rH o 0) iiill « 4J (0 cm wo.C g* £ 0) =#w OH M» ##“ « o E4 "I o_ '— CJ ■s I issi ” (d to Ü •H P to .§ J g T •H 3 S * 2 -s 4J > » I— w« at (tJ P S 09 ^ CA IS -s II -S « -K ^ 8 ^ ^ was selected for the relative humidity data set. Data and test results for relative humidity are summarized in Table 3 . The mean relative humidities during the first hour of poor VAQ events were higher than the mean relative humidi ties during the first hour of acceptably clear periods, at least 10% higher for locally generated poor VAQ events but not more than 5% higher for transport pollution events. A significant difference between the two groups of poor VAQ events themselves was, however, not discernible. Inversion height. Inversion height information was obtained by visually examining the sodar sounding strip charts and manually estimating the height of the lowest inversion during the first 60 minutes of the initiation of a poor VAQ event. Cases in which no inversion layer was observed were assigned a height of 500 m, the maximum ob servable height on the strip charts. Cases of indefinite sodar trace were deleted from the data sets. Data and test results for inversion height are summarized in Table 4. The mean inversion heights observed for transport pollution events were at least 50 m highe^than the mean inversion heights observed for the locally generated poor VAQ events. T-3870 54 esi •.r oi % 3 R 2 s K 3: 2 % Z (rt 5 3 VO 1 2 M « II w -o z e « c e a* : g - 5 3 s 5 i 8 3 5 8 S 8 3 " 5 o < > % c > ? 1 g 1 2 1^ 1 1 r M w o 2 S <-> •H 7 3 a = •H I i s 2 jg ? : g Q) f t; 5 Î 2 Î - : « P S £ < 0 r—I ■s S S 1 « * îl' - & i l & s A 2 3 A■ 2 J u o " f t i a d in P iH 3 vi flo I £ i ^ s & f Ï woi f %*1 P .= :£ & in IS gj “1 ^ T -8 S8-88 8 8 3 !? % g n i i g ? 2 S * S 2 2 ^ ? 2 3 i 2 5 A S 5 ? (0 w Ü •H 4 J w U) *5 Ï •H i f ■P r 3 (d s s li CO I J #I 1 - 1 •fi* % r > 0 ) I? rH us I i s s s i s s s I 111 £ i à i i i à â £ Z T-3870 55 œ < c Csi CM 0 -» <_> 0 » C3 M 0 — J 0 » m z 0 0 (N II <=> <=> It OI * v -V CSI CU a. 00 CK M =0zxz 0 oT OQ 0 rx# CN in o>os in so Œ# cnÛU 0 0 <=* 0m z m 0 f) OJ CM 1 QJ § 8 ea Oi •H ca ro e 3 C H— 1 O i •M lit UJ Æ CO s ro a. 0 1 lit m •H t— lit OI e -*-• ou na LAICS',3 C0L0i^t\^0 SCi-iOOL c€ KIN32 GOLDDr?, COLORADO 80401 T-3870 56 Time of event start in relation to a diurnal wind shift. In order to study whether the diurnal wind shift was associated with either or both transported or locally gener ated poor VAQ events, the 5-minute traces of the wind direc tions from the Scott Carpenter Park and the Crossroads stations were examined and the times of the morning and afternoon wind shifts were determined when possible. For the study period December 1988-January 1989, 47 observations of the wind shifts were identified for this purpose. The frequency distribution of the times of the diurnal wind shifts is illustrated in Figure 12. It was hypothesized that the effects of the morning shift could be different from those of the afternoon shift. For this reason the poor VAQ events, both transport and local, were divided into two subgroups, A.M. events and P.M. events. Because the times of the peaks of the 51 identified poor VAQ events clustered into two distinct groups, as shown in Figure 13, whereas the times of the starts of the events did not, as shown in Figure 14; events were divided into the two subgroups depending on which cluster displayed their peaks. Events which peaked in the 0800 - 1400 hours were categorized as A.M. events in relation to the nearest morning diurnal wind shift. Events which peaked in the 1700 - 0400 hours were categorized as P.M. events in relation to T-3870 57 o en nj G 0 •H P <0 1 0) in 00 ja o Ul cn «T3 Q) CU P C (U ê p 0 cx> CO cn en 1 •H C Eh <0 m>- 0) M Q JG I P 00 eu 00 jr P m O ü C eu o Q c_ •H 3 p o G cn X P P •H P S P en â •H Q I P : O G o cr 0^ O) p (Z4 lf> m 2 & •H s;u3A3 jo jaquunN T-3870 59 un ru en P p (ü Pw o ru p c Q) > H P en O CO en en rH çn in CO g en •H C <0 c ro 0) 7 >. x: 3 ro p 00 CD 00 p en CO QJ o CO C en ü c QJ o Q u -H QJ p Q (_ 01 3 P O ■H X U S C p QJ (0 M > •H LU un O O ; C & > P c. § O o o & s CL (D M r H QJ Ph If) in 2 un , I & LO LD m ru •H Ph squ3A3 0 jaquunN T-3870 60 the nearest afternoon-evening diurnal wind shift. Statistical analyses were applied to the A.M. and the P.M. subgroups of the transport and local groups as well as to the larger groups with the subgroups combined. Data and test results for event timing in relation to the diurnal wind shift are summarized in Table 5. Conclusions must be tentative because of the limited data, subgroupings, and results. As stated above, since the classification methodology as applied to the January 1989 poor VAQ events was suspect, conclusions could be addressed to the December 1988 events only. The three transport events for which wind shift times could be identified start ed prior to the late afternoon diurnal wind shift, and therefore had to be considered as independent from the actual shift itself. Six of the seven local events for which wind shift times could be identified started after the late afternoon wind shift. This split of transport and local events on either side of the afternoon wind shift by more than 2.5 hours indicated that different meteorological variables were associated each type of event. These meteo rological factors were based in the dynamics of diurnal wind shift cycle: daytime upslope winds prior to the late after noon wind shift induced transport events; nighttime more variable, but likely weak or downslope, winds were T-3870 61 i o« a< -8 S cm X* i S ilsl I i TB - » 2 . I S * 3 ^ 1 - 5 - S I ! ~ | S : ^ S ^ T -s I 8-8 Î S 5 sis I s 73 0) g •H I I P A 3 G § II O O If) S 2 II Q> r H cn I ÇS I!M— »— « EH i n i l i z S zS z z T-3870 63 associated with locally generated events. Net Radiation. The 5-minute net radiation averages for the first 60 minutes of the identified start of a poor VAQ event were averaged from the Crossroads station data set. No net radiation data was available for any of the accept ably clear periods in December 1988, and so the Kruskal- Wallis H test could not be performed on this data group. Data and test results for net radiation are summarized in Table 6. No significant differences for the mean net radiations associated with transport or local poor VAQ events were found in the available data sets. Snow cover. Presence or absence of snow cover on the days of poor VAQ events was determined from National Weather Service records for Boulder. The sign test with binomial probability of 0.50 assigned to the presence of snow cover was performed on each of the three groups. Data and test results for net radiation are summarized in Table 7. On the basis of the December 1988 data set, the pres ence of snow cover might appear to have promoted both types of poor VAQ events. On the other hand, a low probability of snow cover on clear days could not be demonstrated. The high number of days with snow cover throughout the study T-3870 64 ad g 5 3sjO 5 CO i :S g % ST S s s Sf - 9 ii i I C o •H w (0 Ae £ •H §! 73 i » g & II I * if 4J sÇ •s % 9 ? > % ll «# w 1 ZI t: «3 k O (Q 4-» II rH s § 1 jg & «1 "2 •p Of tt 10 8 I S2 SÇ s i i i <0 10 Ü •H g (0 % •H 44 (d JJE 4J i ê CO VO l - i 0) r—I «Ô A i i i 111 I l i II S s J i â à J i T-3870 65 in so en i a* sO CM :g sO «o O' en un O» irt i :£ ug % S i co ill M O S % 44 *0» (0 S g s f g Ü I i 1 1 fli ■g (0 0) ü •H 44 g (0 %; •H 44 1 5 && s. (d «ô us period, 24 out of 31 recorded days in December 1988 and 17 out of 31 recorded days in January 1989, would have caused the appearance of a high probability of snow cover for all visibility conditions. Discussion For the month of December 1988, poor VAQ events caused by transported pollution were associated with wind speeds higher than those during locally generated poor VAQ events. This difference was on the average not more than 2 m/s. While winds from the northeast quadrant consistently caused the transported poor VAQ events, the winds that produced locally generated poor VAQ events were more variable in direction. The relative humidities during locally generated poor VAQ events were at least 10% higher than the relative humidities during periods of clear conditions. The relative humidities during transport poor VAQ events were not more than 5% higher than the relative humidities during periods of clear conditions. Local poor VAQ events were produced under conditions of lower inversions than transported events, at least 50 m lower. Transport events started in the early afternoon before the diurnal wind shift, while locally generated events, which peaked at night, started after the afternoon wind shift. The locally generated T-3870 67 events that peaked late morning or early afternoon started both before and after the morning wind shift. No signifi cant differences were found in the net radiation and pres ence or absence of snow cover for the two types of events. À summary of the conclusion of the statistical tests is presented in Table 8. From these statistical analyses, a generalized descrip tion of each type of poor VAQ event can be offered. Locally generated poor VAQ events outnumber transport poor VAQ events; they did by more than two to one for December 1988. The local events commence just about any time during the day but tend to peak either late morning or in the few hours around midnight. Winds are light, on the average of 1.5 m/s, and of variable direction. When locally generated poor VAQ events occur, there is likely to be a low inversion, as low as 30 m but on the average about 100 m. The relative humidity is at least 10% higher than during clear condi tions, which promotes the absorption of atmospheric water into fine particulate pollution and further degrades visi bility range. The poor VAQ events that are produced by transported pollution start in the early to mid-afternoon daylight hours when winds are out of the northeast quadrant up the Boulder Valley at speeds averaging 2.5 m/s. These transport events T-3870 68 Table 8. Summary of Statistical Test Results for Meterologic Variables MET FACTOR HYPOTHESES STAT METHOD RESULTS Dec 1988 Jan 1989 Dec'98&Jan Wind speed Ho:WSt=WSl=WSo Kruskal-Wallis Reject Ho Reject Ho Reject Ho Hl:WSt~=WSr=WSo H test Ho:WSt=WSl Mann-Whitney " Reject Ho Cannot Reject Ho Hl:WSt>WSl test reject Ho Ho:WSt-WSI=2m/s t-test, 1-tail, Cannot Cannot Cannot Hl:WSt-WSl>2m/s 2 sample reject Ho reject Ho reject Ho Wind direction No:WOt=WDl=WDo Kruskal-Wallis Reject Ho Reject Ho Reject Ho Hl:WDt~=WDr=WDo H test Ho:WDt’W01 Nann-Whitney Reject Ho Cannot Reject Ho H!:WDt~=WDl test reject Ho Presence of NE Ho:0 Ho:0 Relative Ho:RHt=RHl=RHo Kruskal-Wallis Reject Ho Cannot Reject Ho humidity Hl:RHt'=RHr=RHo H test reject Ho Ho:RHt=RHl Mann-Whitney Cannot Cannot Cannot Hl:RHt Ho:RHl-RHt=5% t-test, 1-tail, Cannot Cannot Cannot Hl:RHl-RHt>5% 2 sample reject Ho reject Ho reject Ho Ho:RHl-RHo=10% t-test, 1-tail, Reject Ho Cannot Reject Ho Hl:RHl-RHo>iOZ 2 sample reject Ho Ho:RHt-RHo=5% t-test, 1-tail, Cannot Cannot Cannot Hl:RHt-RHo>5% 2 sample reject Ho reject Ho reject Ho T-3870 69 Table 8. (continued) MET FACTOR HYPOTHESES STAT METHOD RESULTS Dec 1988 Jan 1989 Dec'88&Jan Inversion height Ho:Ht=Hl=Ho Kruskal-Wallis Reject Ho Cannot Reject Ho Hl:Ht~=Hr=Ho H test reject Ho Ho:Ht=Hl Mann-Whitney Reject Ho Cannot Reject Ho Hi:Ht>Hl test reject Ho Ho:Ht-Hl=50m t-test, 1-tail, Reject Ho Cannot Cannot Hl:Ht-Hl>SOm 2 sample reject Ho reject Ho Time of event Ho:Tl=Tt Nann-Whitney AM:insuff. AM; Cannot AM; Cannot start in rela Hi:Tl>Tt test data reject Ho reject Ho tion to a shift PM: PM;insuff. PM; Cannot Reject Ho data reject Ho AM&PM: AM&PM: AM&PM; not Cannot Cannot calculated reject Ho reject Ho Ho:Tl-Tt=150min t-test, 1-tail, AN;insuff. AM; Cannot AM; Cannot H1:T1-Tt>150min 2 sample data reject Ho reject Ho PH: PM;insuff. PM; not Reject Ho data calculated AM&PM; not AM&PM; not AM&PM: not calculated calculated calculated Ho:Tl-Tt=60min t-test, 1-tail, AM:insuff. AM; Cannot AM; not Hl:TI-Tt>SOmin 2 sample data reject Ho calculated PM; PM;insuff. PM; not Reject Ho data calculated AM&PM; not AM&PM; not AM&PM; not calculated calculated calculated T-3870 70 Table 8. (continued) MET FACTOR HYPOTHESES STAT METHOD RESULTS Dec 1988 Jan 1989 Dec'88&Jan Net Ho:NRt=NRl=NRo Kruskal-Wallis insuff. Cannot Cannot radiation HlîNRt'=NRP=NRo H test data reject Ho reject Ho Ho:NRt=NRl Mann-Whitney Cannot Cannot Cannot Hl:NRt~=NRl test reject Ho reject Ho reject Ho Ho:NRt-NRl=iOOW/m2 t-test, 1-tail, Cannot Cannot Cannot Hl:NRt-NRl>100W/m2 2 sample reject Ho reject Ho reject Ho Snov cover Ho:P(SCt)=O.S Sign test Reject Ho Cannot Reject Ho Hl;P(SCt)>0.5 rejectHo Ho;P(SCl)=0.5 Sign test Reject Ho Cannot Reject Ho H1;P(SC1)>0.5 rejectHo HoîP(SCo)=0.5 Sign test Cannot Cannot Cannot HlîP(SCoX0.5 reject Ho reject Ho reject Ho T-3870 71 peak in the few hours before midnight. It is feasible that a poor VAQ event could by initiated by a transport occur rence in mid-afternoon and then worsened by locally generat ed pollution sources as the evening progreses. This project was not able to make that distinction in the types of events. Transport events may or may not be accompanied by an identifiable inversion, but when there is one, it is above 100 m. The relative humidity at the time of a trans port poor VAQ event will be only slightly higher than the relative humidity recorded during clear conditions. Relationships Between Poor VAQ Events and High Carbon Monoxide Periods Since both fine particulates and carbon monoxide are pollutants generated predominantly by the same source cate gory, vehicles, any occasional difference in the levels of each pollutant may be due to météorologie factors. This project conducted a preliminary investigation into this question. Figure 15 shows a scatter diagram of the and carbon monoxide values for the month of December 1988. On most occasions when carbon monoxide levels were high, the fine particulate mass loading was correspondingly elevated. However, there were instances of exceptions on both sides, two of which are illustrated in the figures for specific T-3870 72 i_n OJ CO CO G\ o rH r\i u Q) a O) cn fD a c_ QJ g > ID ro o cu o D C in > O LO a CO E a. d Q) +4 TDQJ i X r H o (U c jC o E in S' z XX X X ^ X X oC d T3 c_ 0) m 44 o (Ü Q) ffi in 0) U & •H in o o o o O o o o 0 in o in o in o in in cn m nj CM 1 a5ejaA0 a^nuruj-og ‘T-UJ9-30T ‘Ajp-dsq T-3870 73 case study days. During the evening hours of December 3, 1988, both bgp.dry and carbon monoxide readings rose with the rush hour traffic (Figure 18). As the night proceeded, carbon monox ide levels dropped off while fine particulate levels re mained high. In this example, the meteorology was constant for both pollutants, so another cause must be sought for the different pollutant patterns. Possibilities include addi tional sources for fine particulate but not for carbon monoxide and different atmospheric chemistries between the two pollutants. On December 29, 1988 carbon monoxide levels started to rise gradually at least an hour before the abrupt jump in fine particulate levels at 1325 (Figures 20 a-b). The discussion below explains the arrival of an outside air mass at this time. With the incidence of this outside air mass, the carbon monoxide levels dropped, indicating that the new air mass was relatively carbon monoxide free. Because only a cursory inquiry into the relationships between micrometeorology and two different pollutants was possible in the project, any conclusions must be preliminary and tentative. These examples suggested that during air pollution events caused by immediately local sources, meteo rology played a minor role in the different temporal pat- T-3870 74 terns of fine particulate and carbon monoxide concentra tions. More likely causes were source or chemical differ ences. On the other hand, during transport events, meteo rology took a major role, but would not have if there had not been a characteristic of fine particulates which permit ted them to remain suspended and hence more subject to long distance transport. Phenomenological Descriptions of Selected Case Davs Acceptable Visibilitv Period; December 3. 1988. Saturday, December 3, 1988, occurred within a six-day period whose local weather pattern was consistently charac terized by the mountain-plains diurnal wind shift cycle rather than by the synoptic situation. NWS surface pressure maps for the morning showed fluctuating high pressure cen ters over southern Idaho, northern Utah, and western Colora do that merged with a high pressure center over the north ern Great Plains in the afternoon. A northeast-southwest oriented low pressure trough over Oklahoma, Kansas, and Missouri lay to southeast. The morning wind shift from downslope westerly drainage to upslope northeasterlies took place on December 3, 1988 between 0920 and 1005, depending on the particular location and elevation in the City of Boulder. During this same T-3870 75 period, a nighttime ground-based inversion that had ranged in height from 147 m at 0100 down to 107 m at 0400 dissi pated. The reversing late afternoon shift occurred between 1610 and 1645, and an evening inversion formed at 67 m, which elevated to 120 m by midnight. Figure 16 shows the wind directions along with speeds throughout the day. Wind speeds were usually 2-3 m/s except during early morning nighttime hours and again at the afternoon wind shift, when they dropped below 1 m/s. The surface pressure differential between the PROFS Boulder and Erie sites was examined to investigate the possible influence of the regional pressure pattern on local wind directions, speeds, and pollution transport. Ambient pressures at both stations were standardized to the Greeley station elevation. The pressure differential trace through December 3, 1988 is illustrated in Figure 17. Although values were adjusted to account for station offsets, a thorough quality assurance-quality control correction proce dure was well beyond the scope of this project. Consequent ly, the reference baseline must be considered relative, not absolute. A shift in the pressure gradient toward Boulder at about 0430 preceded a brief wind change a short time later. A midday change in the differential, Erie increasing relative to Boulder, concomitant with diurnal upslope flow. T-3870 76 o o on o 8 CO o nj OJ T3 0) 0) A CO "g A3 o o c OJ O •H 4J 0) ü -t-J 0) 3 M C •H E a I LD T3 c •H 00 CO 00 enCO a\ o m m o o u 0) ü Q g o- vo Q (U kl •Hg. Et, O o o o O ooo o o O in o ID o in o LD m m ru ru OTh s /lu ‘paads ' saauGap 'uoi oa ji o T-3870 77 o n U o 0) 13 iH 13 0 m 1 o (U o •H ai U M W cn i o ro o > ai c_ eu fd •H 4-> C OJ 0) u 0) (w o m •H in Q eu 0) U :3 ifl cn CD 0) 00 u CO o \ Pu ô “ o m o a CL CO O! 00 m ÜJ r—I m o o in ü d d g o LU o 0) 10 ai o OJ tû CD u 0 0 o 1 1 I g. •H quj 'ieiqu3J3:^ ^ çp ajnssajd T-3870 78 is evident. Yet, regional pollution transport from the lower Boulder Valley into the City of Boulder did not occur, probably because the atmospheric mixing during the daylight hours was excellent and also because the pressure differen tial resulted from the microscale heating differential between the two PROFS monitoring locations. Daytime temperatures reached 11-12® C. Nighttime tem peratures dropped to 1-3® C. Relative humidity was low, 16-21%, throughout the day; although it did rise to 32% at the afternoon wind shift and remained in the upper 30% level through the night of December 3-4, 1988. Pollution levels for the day are shown in Figure 18. Readings during the first hours of the early morning, when westerly winds were less than 2 m/s, reflected pollutant accumulations from the previous evening of December 2, 1988 that were backwashing down Boulder Canyon after being pushed there by that day's upslope flow. A brief rise of pollutant levels in the 0600 hour, most easily seen for carbon monox ide, also occurred during very low winds, less than 1 m/s, before the morning wind shift. This was a typical rise that would have been higher on a weekday when there is considerably more vehicular traffic (Gaynor et al., 1989). More significant was the fast rise in pollutant levels during the 1600 and 1700 hours, when wind speeds were slack, T-3870 79 o o m o Q g o O OJ U S CO o 0) o -p i OJ rH -U g O 00 0) I •H JOl o o o o o o o o o ooo en CD UD in en nj OTxLudd ‘apçxouQuu uoqjeo 'T-uJ9-30î 'A jp-dsq T-3870 80 due to local evening vehicular traffic sources. Carbon monoxide levels declined as night progressed but the parti cle scattering coefficient readings did not. The reason has not been determined. Heated nephelometer readings remained high, so the rise in relative humidity, still well below the critical 70% for atmospheric water association with fine particulates, at sunset is not accountable. Other sources such as wood burning stoves and fireplaces may have contributed after traffic decreased. During the 2300 hr the wind direction shifted to a slightly more westsouthwesterly direction and speeds fell, eventually to below 1 m/s during the early morning hours of December 4, 1988. A ground-based inversion at about 133m continued from the previous evening. These atmospheric changes caused the first poor VAQ event of the study period. However, since it happened during nighttime hours, its detection remained to instrumentation alone, not human observers. Transport Poor VAQ Event: December 29. 1988. December 29, 1988 provided an excellent example of the transport of an air mass laden with fine particulates from the lower Boulder Valley into the City of Boulder and an example of a localized air pollution event caused ultimately T-3870 81 by mesoscale and synoptic features. This Thursday was one day in an episode of poor visual air quality that followed a snowstorm on December 26, 1988, and the episode continued until January 1, 1989. That snow remained on the ground for the duration and beyond. A high pressure center over Nevada and Utah remained stationary. A north-south low pressure trough oscillated over eastern Colorado. Figures 19a-c, the surface pressure maps for 1100, 1400, and 1700 MST show the trough develop ment. The transport event was recorded at 1325 by the nephel- ometers situated at the Scott Carpenter Park site. Figure 20-a illustrates the abrupt rise in the b^p.dry readings accompanied by a dramatic wind shift from westerly to north easterly. The wind speeds associated with the air mass were certainly not high, 2.5-3 m/s, but they decreased to below 1.5 m/s in the two hours following. The immigrating air mass brought a 3® C fall in ambient temperature and an 18% rise in relative humidity. The vertical temperature structure for the 1300 hour is depicted by the sodar readings at the Crossroads site, reproduced in Figure 21, which shows the onset of a ground- based inversion with a depth of about 109 m. T-3870 82 Q) u :3 CA m 0) u cu 0) Ü (W u 0 W EH X o o r H r H CO 00 o> m CM Ü 0) o (d I G\ rH Q) Jh d O' •H T-3870 83 CA 0) U :3 CA CA 54 S o o rH 00 00 O' (N Ü g A I O' rH (U §> •H 84 T-3870 oc o ÜJ X o I- Q' U) CE z: Q) U 0 CO m m ü) o u 04 d) ü h- (d (w u O w E4 I O r- CO 00 I t- ü I ON I- 2 g. •H T-3870 85 o O o O o m m o o o Û in LD OJ ru Co •H ü •H ■O CL « O E c\j in I OJ -r-l in 8 CL o 00 CO (J\ a rH CL 4J 00 00 o ^ o a\ o en u CM OJ 00 ü OJ g a\ A CM 1 O a O LD CM 2 g. -H Em CM O O o o O O o O o O o o O O O O o LD o in O iD O o m o in m m CM (M m m •H - T-ujg-BOT "A jp-üsq ' Gap ‘qm Ol^ujdd ‘03 ' Gap ‘qm T-3870 86 o 4" 00 S i ^ a . 00\ M V. M m . : -=:n - •1.-r 1"4. % r::xrsJ \ "•• »•?)% * *:_ ♦ • 7-:L.* • • •• vr< » ^• • ••* • «#* •••«*• ^ •* ■i* M *. "1; »' L ." : r 4%^#^ :.% r » Ê Q _j 3 O m Figure 21. Dec 29, 1988, 1300hr. Vertical Temperature Profile, XRD. T-3870 87 The usual morning wind shift had not taken place on December 29, 1988. The warmer westerlies were strong enough to resist the movement of the colder polluted air mass up the valley until afternoon. The driving force behind the wind shift then was the movement of the regional pressure gradient commanded by the Front Range pressure trough. During the hours preceding the transport event, the trough was positioned between Boulder and Erie. With the sudden equalization of pressures between the two stations, approx imately 22 km apart, the trough has moved west of Boulder and allowed the polluted air mass to move into the City (Figure 22). Back-calculation of wind speeds and directions at monitoring stations down Boulder Creek indicated that a parcel of air in the transported air mass could have arrived from as far away as the intersection of St. Vrain Creek and the South Platte River, approximately 48 km from Scott Carpenter Park site. The ultimate source of the polluted air parcel could not be determined. The influx of fine particulate polluted air at 1325 was not accompanied by similar increases of carbon monoxide. In fact, carbon monoxide levels had begun to rise about two hours earlier and were actually dispersed by the arriving air mass. T-3870 88 o pn U 0) T3 r— I :3 o o QJ 2 QJ (W •H ID Q QJ QJ U a in in CD 0) CD k o \ P4 on CM d d \ 00 CM CO 00 ÜJ iH o- d LU CM OJ O Q> 00 UD CM O CM 00 0 0 1 1 & quu ‘iBiquaja:^ rp ajnssajd T-3870 89 The late afternoon wind shift was not definitive on this day. Pollutant levels, both fine particulate and carbon monoxide, rose as the evening progressed, due to vehicular sources, and then dissipated with changing winds during the 2200 hour. Surprisingly, as can be seen in Figure 20-a, this cleansing was initiated by winds with an easterly component, which arrived from an are^ where there had probably been no additional sources of pollution. Locallv Generated Poor VAO Event: December 30. 1988. The changing westerly winds around midnight December 29, 1988 purged the Boulder air space of both fine particu late and gaseous pollutants. However, this respite in the poor VAQ episode did not last long. Winds throughout the early morning hours of December 30, 1988, were light, 1.5 m/s and lower, and variable as seen in Figure 23. Pollutant concentrations are shown in Figure 24. Fine particulate levels again surged during the 0200 hour as the polluted air mass floundered back into the city. The stagnancy of the air mass is clearly indicated by the rise in carbon monoxide levels with the earliest beginnings of the morning traffic rush. The morning wind shift, documented at 0825 at the Crossroads station, with speeds approaching 2 m/s introduced a relatively cleaner air mass, although values stayed T-3870 90 o o m o u CO o 73 ID o UD' o Q OJ C o •H 4J Ü ? •H 3 o CO 00 o \ CO o o t-H fT) o CO Ü Q) O o m CM 0) o o •H& o o o oo oo o o o ID o D o D m m CM C\J OOT*s/u *paads putM 'saajpap "uoi^aajiQ puiM T-3870 91 o Q o m o g 0) 73 •H X o c o s o « (Q U CD o ro o > c_ g (D w J-J C QJ 0) 4-1 +J 3 C o E i rH ru- in I in Q) CD I % 73 TD 0) 00 +J 00 (0 o Q) o m m ai a. 00 o 00 TD CL a. o o ro (O- o CJ in ü g o o o o o o o o o o o in o ID o in o in n m nj ru 2 g. Otxujdd ‘03 • Î-UU9-30T 'Ajp-dsq •H h T-3870 92 above 66*10"* throughout the day. Another example of transport, this time relatively minor, may be illustrated in the brief rise and fall in the b^p.^^y readings around 1500. Although the extended flow of winds from the northeast would support the classification of this spurt in values as a transport event, the pressure differential between Erie and Boulder did not alter radically at this time. The apparent carbon monoxide spike around this time was due to equipment maintenance procedures. In the 1600 hour, as the wind shifted hesitantly to a dominant westerly component with very low speeds, particulate and gaseous pollutants again accumulated. Just as significant for the pollution buildup as the horizontal atmospheric variables of wind direction and speed, the vertical temperature profile aggravated the situation. Sodar records for the nighttime morning hours of December 30, 1988 indicated a ground-based inversion varying in height from 67 m to 187 m that did not break up until 1200. At 1815 another inversion was established at 53 m. Temperatures and relative humidities on December 30, 1988 showed a typical diurnal cycle. Ambient temperatures dropped to -9° C in the predawn hours, rose to only 3® C near noon, and fell consistently during the rest of the day to -7® C at midnight. The relatively humidity was high, 40% T-3870 93 during the day hours and 80-90% during the night hours. The high relative humidities at night aggravated the poor visual air quality as perceived by human observers. Synoptically the surface high pressure center on the windward side of the Rocky Mountains remained. The lee trough moved east and weakened. The pressure differential (Figure 25) showed that the 0200 hour movement of the air mass carrying the particulate loading was likely propelled back into the city by a gradient change similar to that initiating the transport event of December 29, 1988, 1325. Figure 25 does not provide evidence that the possible trans port occurrence at 1500 was caused by a regional pressure differential change. That the pressure differential between Erie and Boulder was neutral for the rest of the day ac counted for the increase and ebb of locally generated pollu tion in the evening and night hours. AKTHim LAKES LIBRARY COLORADO eCMOOL d UINEt GOLDEN. COLOR.qDO 8040) T-3870 94 ID C\J U Q) 'O rH D O CQ o C\J i <0 -H -P c c_ o 2 CM 0 0) D 1 (*4 •H Q 0) U D o to Œ1 01 CD Q) o \ O & m □ l CL CM a 00 CO UJ CO d D 0 n CL d Ü cc 1 ÜJ CO o o If) CONCLUSIONS This project presented evidence from two approaches, statistical and phenomenological, that during the month of December 1988 in the upper Boulder Valley, the météorologie conditions associated with periods of acceptable visibility, poor VAQ events caused by transported pollution, and poor VAQ events caused by immediate local sources were identifi- ably different. Seven météorologie variables were examined. Winds during the first 60 minutes of local poor VAQ events were variable in direction, but winds that initiated transport events were almost always from the northeast quadrant. Winds associated with selected periods of clear visibility often had a strong westerly component. The wind speeds of the transport events were somewhat higher than those of local events, but the difference was less than 2 m/s. Winds were higher during clear periods than during either type of poor VAQ event. The relative humidities were higher during poor VAQ events than during clear periods, at least 10% higher during local events but not more than 5% higher during transport events. A significant difference in the relative humidities between the two groups of poor VAQ events could not be detected. Inversion heights during T-3870 96 local events were likely to be at least 50 m lower than the inversion heights during transport events. The timing of the two types of poor VAQ events around the diurnal wind shifts was detectably different, but because of a limited data set, conclusions were tentative. Transport events occurred during early afternoon and were independent of a diurnal wind shift. Afternoon-evening local events often commenced after the afternoon-evening wind shift, and morn ing local events tended to begin before the morning wind shift. Low wind speeds were more critical for the buildup of local pollution than was timing with respect to a diurnal wind shift. No differences between the two types of poor VAQ events were found for net radiation or the presence of snow cover on the ground. This project suggested that meteorology was a major factor governing the levels of fine particulates, but not of carbon monoxide, in instances of pollution transport. In contrast, during local pollution events meteorology played a minor role in the different temporal patterns of the two pollutants. Phenomenological descriptions of three days selected as representatives of acceptable visibility, pollution trans port, and local pollution provided specific examples of the conclusions derived from the statistical analyses. Addi T-3870 97 tionally, the role of synoptic weather features in determin ing microscale pollution events was illustrated. In the cases discussed, the presence of a high pressure center on the windward side of the Rocky Mountains and a low pressure trough on the lee side corroborated earlier studies. The changes in pressure gradient on the immediate lee side of the Front Range mountains attributable to the east-west oscillation of this trough were seen as a driving force behind the transport of a polluted air mass in and out of the upper Boulder Valley. To track the movement of the trough more precisely than this study was able, would re quire a monitoring network more dense than the existing PROFS mesonet. A limiting factor in this research was that the catego rization scheme of poor VAQ events, based on a peak-to-mean ratio, was applicable to only one month of the two month data set. The reasons for this selective application remain unclear. Whether the scheme was inherently faulty, the data set collected during the BAQS was idiosyncratic, or the particular months of December and January are indeed charac terized by subtle meteorological differences that alter the nature of poor VAQ events within these months were possible reasons. Further improvement in this scheme is recommended, but additional data from other years would be needed. T-3870 98 REFERENCES CITED Appel, B.R., Y. Tokiwa, J. Hsu, E.L. Kothny, and E . Hahn. 1985. Visibility as related to atmospheric aerosol constituents. Atmospheric Environment. 19 (9): 1525-1534. Charison, R.J., N.C. Ahlquist, H. Selvidge, and P.B. McCready, Jr. 1969. The monitoring of atmospheric aerosol parameters with the integrating nephelometer. Journal of the Air Pollution Control Association. 9 (12): 937-942. Cassell, E.J. 1969. The health effects of air pollution and their implications for control. In Air pollution control, ed. C.C. Havinghurst, 1-20. Dobbs Ferry, New York: Oceana Publications, Inc. Colorado Air Quality Control Commission. 1989. Visibility Standard- adopted December 21, 1989. Covert, D.S., A.P. Waggoner, R.W. Weiss, N.C. Ahlquist, and R.J. Carlson. 1980. Atmospheric aerosols, humidity and visibility. In The character and origins of smoa aerosols. ed. G.M. Hidy, P.K. Mueller, D. Grosjean, B.R. Appel, and J.J. Wesolowski, 559-581. New York: John Wiley and Sons, Inc. Crow, L.W. 1976. Airflow study related to EPA field moni toring program Denver metropolitan area November, 1973. T-3870 99 In Denver Air Pollution Study - 1973 Proceedings of a Symposium. Vol. I, ed. Philip A. Russell, 3-29. Research Triangle Park, North Carolina: U.S. Environ mental Protection Agency. Ferman, M.A., G.T. Wolff, and N.A. Kelly. 1981. An assess ment of the gaseous pollutants and meteorological conditions associated with Denver's brown cloud. Journal of Environmental Science and Health, Part A . 3 (16): 315-339. Gaynor, J.E. A preliminary study of Boulder Valley's air quality meteorology. Preprint Volume, Sixth Joint Conference on Applications of Air Pollution Meteorology with the Air Pollution Control Association, January 3 0- February 3, 1989, Anaheim, California, American Meteo rological Society, Boston, Massachusetts, 253-255. Gaynor, J.E. and L.K. Wittenmeier. 1989. Short-term clima tology of Boulder Valley's high-pollution meteorology. Preprint Volume, Sixth Joint Conference on Applications of Air Pollution Meteorology with the Air Pollution Control Association, January 30-February 3, 1989, Anaheim, California, American Meteorological Society, Boston, Massachusetts, 250-252. Gaynor, J.E., D. Wolfe, R. Fritz, S. Summers, H. Sievering, G. Kenniston, and G. Mathews. 1989. Report to the City T-3870 100 of Boulder, Results of the Boulder Air Quality Study Winter of 1988-1989, submitted to the Boulder City Council, December 12, 1989. Groblicki, P.J., G.T. Wolff, and R.J. Countess. 1981. Visi bility-reducing species in the Denver "Brown Cloud" - I. Relationships between extinction and chemical compo sition. Atmospheric Environment, 15 (12): 2473-2484. Horvath, H. and K.E. Noll. 1969. The relationship between atmospheric light scattering coefficient and visibili ty. Atmospheric Environment, 3: 543-552. King, C.W. and C.A. Russell. 1989. Temperature structure effects on pollutant distribution in the Denver metro politan area. Preprint Volume, Sixth Joint Conference on Applications of Air Pollution Meteorology with the Air Pollution Control Association, January 30-February 3, 1989, Anaheim, California, American Meteorological Society, Boston, Massachesetts, 243-245. Lewis, C.W., R.E. Baumgardner, R.K. Stevens, and G.M. Russwurm. 1986. Receptor modeling study of Denver winter haze. Environmental Science and Technology, 20 (11): 1126-1136. Lundgren, D.A. and D.W. Cooper. 1969. Effect of humidity on light-scattering methods of measuring particle concen T-3870 101 tration. Journal of the Air Pollution Control Associa tion. 19 (4): 243-247. Malm, W.C. 1983. Introduction to visibility. Fort Collins, Colorado: National Park Service. Munn, R.E. 1970. Biometeoroloaical methods. New York: Aca demic Press, Inc. Pilinis, C., J.H. Seinfeld and D. Grosjean. 1989. Water content of atmospheric aerosols. Atmospheric Environ ment. 23 (7): 1601-1606. Pratt, J.F. and R.J. Clark. 1983. PROFS MESONET - Descrip tion and performance. Preprint Volume, Fifth Symposium on Meteorological Observations and Instrumentation, April 11-15, 1983, Toronto, Ontario, World Meteorologi cal Organization, American Meteorological Society, and Canadian Meteorological and Oceanographic Society. Ruby, M.G. 1985. Visibility measurement methods: I. Inte grating nephelometer. Journal of the Air Pollution Control Association. 35 (3): 244-248. Ruby, M.G. and A.P. Waggoner. 1981. Intercomparison of integrating nephelometer measurements. Environmental Science and Technology. 15 (1): 109-113. Summers, S., W.D. Neff, and C.W. King. 1989. Forecasting air pollution episodes over Denver. Preprint Volume, Sixth Joint Conference on Applications of Air Pollution T-3870 102 Meteorology with the Air Pollution Control Association, January 30-February 3, 1989, Anaheim, California, American Meteorological Society, Boston, Massachu setts, 235-239. Ruffieux, D. 1989. Simulation of solar radiation on downtown Denver. Preprint Volume, Sixth Joint Conference on Applications of Air Pollution Meteorology with the Air Pollution Control Association, January 30-February 3, 1989, Anaheim, California, American Meteorological Society, Boston, Massachusetts, 246-249. Waggoner, A.P., R.E. Weiss, N.C. Ahlquist, D.S. Covert, S. Will, and R.J. CharIson. 1981. Optical characteris tics of atmospheric aerosols. Atmospheric Environment. 15 (10/11): 1891-1909. Walpole, R.E. and R.H. Myers. 1989. Probability and statistics for engineers and scientists. 4th ed. New York: Macmillan Publishing Co. Watson, J.G. et al. 1988. The 1987-88 Metro Denver Brown Cloud Study, Volume III: Data Interpretation, prepared for 1987-88 Metro Denver Brown Cloud Study, Inc., Denver, Colorado. Wolfe, D.E. and J.E. Gaynor. 1989. A case study of a high- pol lution episode during the Denver Brown Cloud Study. Preprint Volume, Sixth Joint Conference on Applications T-3870 103 of Air Pollution Meteorology with the Air Pollution Control Association, January 30-February 3, 1989, Anaheim, California, American Meteorological Society, Boston, Massachesetts, 240-242. Wolff, G.T., R.J. Countess, P.J. Groblicki, M.A. Ferman, S.H. Cradle, and J.L. Muhlbaier. 1981. Visibility- reducing species in the Denver "Brown Cloud" - II. Sources and temporal patterns. Atmospheric Environ ment. 15 (12): 2485-2502. T-3870 104 APPENDIX A Visual Air Quality Events of December 1988. T-3870 105 e z e * a (S (VI m (VI CO c r . CO k S r - . (VI X o > u O M u O i r a (VI r ~ - C-. i r a CT» CO (VI r a m 1 o o ô o (VI i r a M CO o o c s i r a C -4 i r a (VI c n (VI (VI c r . CO ID i n o . i n JQ ■ o c (VI r » « c r * u a UO c r . (V I o - CO 1 ■$ O" I.S v o œ r ~ . r ~ - c r . s jO r ~ . c r . ( V | ID 0 - 1 m r - . r - . L S v S o COOQ (VI (VI C O o IDCO e 1 i r a i r â u O >j O c c o CO k S V < n i r a i r a CO (U -V <=> (VI «VICO(VI CO m a . 1/1 j a •X» m O M (VI«VI CO V k S CO (VI ID (VI cn r - ~ -rr* vjO CO CO (VI (VI (VI CO U 3 « = *9 > « ■ k. 1 o > i r a « = I D ID <=> o ID o o o ira O <=> o ID o o era a f IDID ID n* oo en CO cn cn cn ID a f cn so so > s OO CO n . -an o pa. (VI <=> "f- h_ Q| ira a f a«P -ca a r«a oo ID I — ID CO r o CO IDID ID CO a*- c cn ax 3 = X CM 0 4J OI lit c CO .a 0 0 >■ # Hi <=> tit 0 0 xa >■ S # m in 0 O i 0 II 0 0 ■o cn 3 C f in *-3 LU II 0 0 ■0 cn ac CD m LU z II æ X XX CO k. «0 3 C II f II rrw 1 r H Oi OJ -0 *» »— JO ja c e ca 3 C z 3 C s XX 4 C m m ro ro Q_ « H— a a OJ 0 — 0 Eh cn e c ■ lit 3 C ac X X XX