Managing Carbon Monoxide Pollution in Meteorological and Topographical Problem Areas (2003)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Contributions of Topography' Meteorology' and Human Activity to Carbon Monoxide Concentrations INTRODUCTION Topography, meteorology, and human activity contribute to high carbon monoxide (CO) concentrations in some areas that exceed the National Ambient Air Quality Standards (NAAQS). Despite the decline in national ambient CO concentrations, maintaining the 8-hour standard of 9 parts per million (ppm) has been a particular challenge for some locations (see Table 1-1~. Even when attainment of the standard has been achieved, there re- mains a vulnerability to future exceedances. Expressed mathematically, there is a nonzero probability of nonattainment in a future year. After a more detailed discussion of topography and meteorology, this chapter will discuss the seasonal, weekly, and diurnal patterns in CO con- centrations measured in some CO problem areas. These patterns help describe some of the physical and human factors contributing to the CO problem in these locations. The chapter discusses vulnerability to future exceedances, including a brief description of statistical approaches. The chapter concludes with illustrative examples of the factors contributing to the CO problems in CaTexico, California; Lynwood, California; Fairbanks, Alaska; Las Vegas, Nevada; and Denver, Colorado. 72

Contributions of Topography, Meteorology, and Human Activity 73 The committee identified four factors that contribute to the difficulties that the cities listed in Table 1-1 have had in meeting the NAAQS for CO: 1. Unfavorable topography. Low lying areas surrounded by higher elevations on three or more sides are vulnerable to CO buildup. 2. Unfavorable meteorology. Stagnant winter conditions character- ized by ground level temperature inversions (see definition below) and low windspeeds inhibit vertical mixing of CO. 3. Significant local CO emissions. 4. High concentrations of CO transportedirom nearby areas. Higher elevations with lower air densities tend to have higher CO con- centrations (in ppm) for a given emission flux.i Denver's average air den- sity is 85% of that at sea level. Lower oxygen density can increase CO emissions rates in older vehicles. Topography also can affect meteorologi- cal conditions in a variety of ways, as described below. Meteorology can influence pollutant concentrations through its effects on atmospheric mixing height, windspeeds and wind direction, and atmo- spheric water content (humidity). Humidity is a factor because dry climates and higher elevations tend to have lower total water columns overhead. Because water vapor is an important greenhouse gas (infrared radiation from the earth's surface is absorbedby wafer molecules end reradiated beck down, warming the surface), reduced water vapor allows infrared radiation to pass into space, producing ground level temperature inversions after sundowns and lower mixing heights. These lower mixing heights, com- bined with high evening traffic emissions, can lead to pollution buildup near the ground. It is noteworthy that all of the cities listed in Table 1-1, except Birmingham, are west of the Mississippi River, where the air tends to be drier. Two are in Alaska, where high latitudes and low winter temper- atures result in reduced solar heating at midday and atmospheric conditions are typically dry. ~ _ _ ~ Considering CO emissions as an area source, the emission flux is the mass of CO produced per square kilometer per hour. 2This radiation into space is the reason temperatures drop so rapidly at night in the desert and in the mountains.

74 Managing CO in Meteorological and Topographical Problem Areas METEOROLOGY AND TOPOGRAPHY Background The ease with which air can mix vertically to disperse pollutants de- pends critically on how the air temperature changes with altitude. Warmer air is less dense (more buoyant) and tends to rises and coot as the pressure decreases and volume expands. Tfthe vertical temperature profile decreases by 1°C/100 m (the adiabatic lapse rate) or more, the air mixes freely as warmer air from below moves upward. If the temperature decreases more slowly than 1°C/100 m, or increases with altitude (called an inversion),4 vertical mixing is inhibited. The faster the air temperature increases with altitude during an inversion, the more strongly mixing is resisted. Inversion Types There are several atmospheric processes that can form inversions, as illustrated in Figures 2-1 and 2-2. Cooling of the air near the ground as a result of infrared radiation into space after sunset can create a surface-based inversion, like that shown in Figure 2-la, and can produce a thermody- namically stable layer, which tends to trap pollution near the ground. Hori- zontal advection of warm air creates a high-altitude inversion and can simi- larly increase the stable temperature stratification aloft (Figure 2-l[b]~. Figure 2-~(c) shows the situation with both surface-based end high altitude inversions. In a subsidence inversion (Figure 2-2), a surface-based inver- sion can be strengthened by warm air that sinks and is warmed further as a result of compression. Each of the inversion types reduces the atmo- sphere's ability to mix through the inversion level, allowing pollution gen- erated below the inversion level to accumulate. 3This is the principle on which hot air balloons operate. Vehicle exhaust from a tailpipe is typically 50-100°C warmer than the surrounding air; however, it is rapidly diluted and cooled as it is mixed into the ambient air. 4An "inversion" in meteorology is defied as "a departure from the usual decrease or increase with altitude of the value of an atmospheric property" (Geer 1996~. The term is generally used to refer to a situation where temperature shows an increase with altitude rather than the usual decrease.

75 - . ·-o ~ Act \ \ \ ~ - ~ ·= ~ ~ + ~ -6 ~ \ cq.= g o (U1) aPrU,PIV ~ - o His i .= - ~ o Hi it Hi Hi En - o go l to (Ul) ~P~IlIV to Q Cal to to (Us) ~P~IlIV - - - - o en ~ ~ TV (V ° _ [,~A~ ~ , V-D o ~ tV ·~= Blot ~— AL O - J ID ~ ~ o ~~ · Cal . C) ~ A _—0 o _ ~ Ct V C~ ~ C'0 o .~o =4 S-d ·_1 Ct a~ ,~ _ UD Ct c~ a Ct -= _ - - _ ·~ ~d ao ·-' t; ~ ~ ~0 C) _ ao Ct - (~.) o o ._ a~ .t Ct C) ~Ci ·— Ct · _4 .= - g ~ U, ·~ 5 - , ~ O Cq V) g ._ ao O a~ U, · ~ Cd ,= ~ U, , ~ ~ a~ ~ — ~ ~ O c~ ~ au _~.= ~ ~ ~ C\: - 0 C~ ·~ ~ ~ Ct a., Cq ~ ·— O ~ _ · ~ Ct

76 Managing CO in Meteorological and Topographical Problem Areas 1 000— 500— · _ - surface \ ~ . \ Advectiarl of \ warm air aloft \~- ''.. Subside con ·. - )... +10 Temperature (Celsius) +20 FIGURE 2-2 Schematic of how an existing surface-based inversion (solid lined can be strengthened by subsidence (dashed lined or by advection of warm air aloft (dotted line). Recirculation Atmospheric flow eddies can recirculate the air one or more times. When pollution is emitted into these circulations, pollutant concentrations can increase over time. Figure 2-3 illustrates such a recirculation pattern in a trapping valley. Sea and land breezes represent an additional cause of atmospheric recirculation (Segal and Pielke 19811. As shown in the modeling study of Eastman et al. (1995), at least 70% of pollution recirculates with the sum- mer Lake Michigan sea breeze. These results mirror the observations of Lyons et al. (1995~. That study shows that Gaussian-type models fait to replicate the recirculation and the complex dispersion patterns that result when spatial variations in sensible heat fluxes exist at the surface (Pielke end Uliasz 1993~.

Contributions of Topography, Meteorology, and Human Activity 77 200 m - WARM AIR COLD AIR Resurface / / Ten~erature FIGURE 2-3 Schematic of a trapping valley. The temperature profile in the valley is shown on the right. Stagnation When air does not move significantly over tens of hours or more, the atmosphere is said to be stagnant. Stagnation can occur because of weak winds andlor the trapping of air (see Figure 2-3) When the atrno sphere is stagnant, emitted pollution can accumulate over time. Simple box models, such as those discussed in Pielke et al. (1991) and in Appendix C, can be used to estimate pollution buildup associated with stagnation. Figure 2-3 illustrates air stagnation in a trapping valley. Influence of Topography on Meteorological Conditions Pielke (2002) discusses the influence of terrain on atmospheric condi- tions under strong and weak large-scale winds. The following discussion illustrates that the direction of movement of air is actually quite compli- cated in complex terrain. In contrast to flat terrain, the wind flow in valleys can go in almost any direction, depending on the relative importance ofthe different forcing mechanisms. These mechanisms will affect the dispersion of CO. Under strong flow, for instance, large upward and downward mo- tions are produced, which can enhance pollution dispersion. Under weak

78 Managing CO in Meteorological and Topographical Problem Areas flow, mountain-valley winds that are a result of diurnal warming and cool- ing of the terrain surface can occur (Figure 2-44. These local winds trans- port and disperse pollution, although recirculation can also occur. Mountain-valley winds can occur on relatively small scales (in which case they are called upsIope and drainage flows) or on larger scales (where broad ascent and descent patterns occur). The resulting wind flows can be quite complex. Figure 2-5 illustrates the differences in the diurnal variation of the valley wind direction as a function of the wind direction above the valley (the geostrophic wind) for four distinct mechanisms that can control the wind direction. According to Pielke (2002), these four physical mecha- nisms operate as follows: (1) thermally driven winds are independent ofthe above-valley winds and are controlled by locally developed valley pressure gradients; (2) downward momentum transport ofthe above valley winds (as is associated with a deep convective boundary layer) produces similar wind directions at all levels; (3) forced channeling occurs when the valley flow alignment is dependent on whether the above-valley flow has a net flow down- or up-valley; and (4) pressure-driven channeling (which is out of phase with forced channeling) occurs when the winds in the valley respond only to large-scale horizontal pressure, not to the winds that occur above the valley. Without terrain, the airflow would be nearly parallel to the isobars. With other local flows involved (such as sea and land breezes), wind flow is more complex (Pielke 2002~. Large-Scale Meteorological and Climatological Events and Their Impact on Attainment Local air quality can be affected by large-scale meteorological and cTimatological events. CO exceedances may have patterns that are related to the occurrence of synoptic-scare meteorological events or cTimatological events, such as the El Nino Southern Oscillation (ENSO). Changes in the frequency of large-scale events could affect a location's ability to come into and maintain compliance with the NAAQS for CO. The committee ex- plored the potential effects of large-scare meteorological and climatological phenomena on local CO episodes in three cities: Lynwood, California; Fairbanks, Alaska; and Denver, Colorado. it should be noted, however, that the impact of climate and meteorological variability on air quality, including CO and related pollutants, is an area requiring more research.

Contributions of Topography, Meteorology, and Human Activity 79 A > > ~ > < < 4 ~ ~ ~ r ~ - ~ ~ ~ ~ ~ ~ ~ > ~ > o 1 1 5 10 15 20 25 Time - 4.0 hours 6 r, 81 ~ ~ 1 1 ~ ~ ~ ~ ~ ~ Y Y '~ ~-~' ~ ~ ~ ~ ~ '~ ~ ~ 7 7 'r- .', 8,` _ ~ ~ ~ _ ,,, =, _ 1 Oms-1 1 ~ _ ~ ~ ~ ~ .~ ,~ .~ .~ ~ _ ~ ~ ~ r B 6 ~ I l I I I I I I ', ~ I I I I l I I ~ -'r ~ ~ - 4 - y - ~ 3 .= a) I 2.1< < 1 ',0''l1t 30 ~ ~ , ~ l ~ ............ ~n _ _. ' 1Oms 1 ., ,l, . ~ * ~ ~ ~ ~ ~ '. ~ ~ 1 t r _ -, ,, >, ~, t t t ~; t ^~ ~ ~ ~ ~ ~ ~ . , ~ , '~ ~ ~ ~ ~ ~ ~ ~ ~ . . . , , · ~ · ~ ~ ~ ~ ~ ~ ~ ~ ~ _ ~ _ _~— ~ ~ · — 1 ~1 1 5 10 15 20 25 30 Time - 16.0 hours FIGURE 2-4 Two dimensional simulation of (a) nocturnal drainage flow and (b) upslope flow with no prevailing synoptic flow, with an input condition typi- cal of summer in midlatitudes. Source: Mahrer and Pielke 1977. Reprinted with permission; copyright 1977, American Meteorological Society.

80 Managing CO in Meteorological and Topographical Problem Areas NIGHT DAY it, TOTAL ' ~ , ~ ' . a, >. ~~— f ,,, . ~ . An. i ~ ~1 ,! i ~ ~ ~ of. ~—imp .. oil _ A a ~:04~. of t . ~ . i..~. _ ~ ~ I. ~ 90~ ~ ~~ 4, ^~_.~.~~ ~ i '. . :.' ~ .. t..~. t . ... .. i E S W N ~ £ ~ W ~ THERMALLY DRIVEN DOWNWARD MOMENTUM TRANSPORT FORCED CHANNELING PRESSURE DRIVEN CHANNELING FIGURE 2-5 Relationships between above-valley (geostrophic) and valley wind directions for four possible forcing mechanisms: thermal forcing, down- ward momentum transport, forced channeling, and pressure-driven channeling. The valley is assumed to run from northeast to southwest. Source: Whiteman and Doran 1993. Reprinted with permission; copyright 1993, American Meteorological Society. Lynwood, California Lynwood's local air quality may be influenced by ENSO. Historically, E] Nino recurs every 3-7 years when sea-surface temperatures (SSTs) in the equatorial Pacific Ocean offthe South American coast become warmer than normal. La Nina is essentially the opposite of El Nino, and exists when cooler-than-usual ocean temperatures occur near the equator between South America and the International Date Line. During an El Nino, the months of October through March tend to be wetter than usual in a swath extending from southern California eastward across Arizona, southern Nevada, Utah, and New Mexico, and into Texas. Almost all of the major flood episodes on main rivers in southern Califor- nia have occurred during El Nino winters. During La Nina years, dry con-

Contributions of Topography, Meteorology, and Human Activity 81 ditions are produced on the equator in the Pacific Ocean. La Nina gener- ally does not affect the United States as much as El Nino; however, strong La Ninas have been linked to dry seasons in southern California. In Lynwood, CO exceedances may tend to increase during strong La Nina years when dry and stable atmospheric conditions are produced: Con- versely, CO exceedances may tend to decrease during E] Nino years. Al- though some studies have explored the effects of E! Nino on ozone levels (e.g., Chandra et al. 1998), as of yet no studies have examined correlations between ENSO and CO exceedances. Further research may be needed in this area, including an assessment of how ENSO affects conditions that control concentrations of the pollutants associated with CO (i.e., air tonics and PM). Fairbanks, Alaska In Fairbanks, Alaska, all exceedances of the 8-hour CO standard from 1996 through 2001 occurred when a Tow-pressure system in or near the Gulf of Alaska produced southeasterly geostrophic winds. These winds, which travel over the Alaska Range, are associated with the counterclock- wise geostrophic flow around the low-pressure system. One hypothesis for the coincidence of CO exceedances with this synoptic-scare meteorological event is that the warm-air advection aloft reinforces the radiative ground- leve] inversion. The downward movement of air over Fairbanks also exerts a stabilizing influence on inversions. It is not known, however, what frac- tion of nonexceedance days has such meteorological conditions or how many of the exceedance days before 1996 had these conditions. Nonethe- less, the surface pressure gradient observed during all six exceedances from 1996 to 2001 must have some significance. However, further research over a longer period of time is needed to better understand the relationship. Denver, Colorado In the past, CO exceedances in Denver, Colorado, have coincided with the occurrence of lee troughs—lines of surface Tow pressure on the lee side (the side that is sheltered from the wind) of a mountain range. The air coming over the mountains sinks and warms and, at the same time, the lowering of pressure at the surface along the foothills draws colder air from

82 Managing CO in Meteorological and Topographical Problem Areas the plains and lowlands areas back towards the mountains. Thus, the air between the surface and 100-300 m becomes colder as the air above be- comes warmer, enhancing the inversion. Neff and King (1991) character- ized lee trough history for the 1980s and early 1990s. Lee troughs often occur several times each week during the winter months and are a key precursor to high CO levels in Denver. However, the effect of lee troughs on CO exceedances has not been studied since the mid-199Os because of the decline in exceedances. The decline in CO exceedances is mainly due to lower vehicle emissions, but Neff (2001) noted that there was also a decline in the occurrence of lee troughs during the late 1 980s, which per- haps reduced the frequency and the severity of conditions producing CO exceedances. Future studies also should explore the association of lee troughs, the Arctic Oscillation, and air quality. When cold-air arctic out- breaks occur, they usually provide a snow cover, which strengthens the ground-level inversion. Fewer arctic outbreaks over the Great Plains could help decrease the potential for pollution episodes in Denver. Denver has had a decade-Ion" period without the long-term snow cover and associated light winds that tend to promote atmospheric stagnation, which can lead to CO buildup. The lack of conditions conducive to high CO means less susceptibility to CO exceedances. TEMPORAL PATTERNS OF CO CONCENTRATIONS CO concentrations show seasonal, weekly, and diurnal patterns reflect- ing the temporal patterns in emissions and meteorology. Figures 1-1 and 1-2 show seasonal patterns in the numbers of days with exceedances ofthe 8-hour CO standard. Figures 2-6 and 2-7 also show patterns in the total numbers of exceedance days by month for Lynwood, California, and Fair- banks, Alaska, for periods of approximately 30 years. Lynwood exhibits a very symmetrical pattern, with the maximum number of exceedance days in December, when the winter solstice (shortest day, least solar radiation) occurs. Fairbanks exhibits the maximum number of exceedances in Janu- ary. The considerably greater numbers of exceedance days in Fairbanks in January compared with November, and in February compared with Octo- ber, are attributed to reduced cloud cover in the winter months compared with the autumn months.5 Clear skies in January and February contribute month. Wee Figure 2-3 in the interim report (NRC 2002) for Fairbanks cloud cover by

Contributions of Topography, Meteorology, and Human Activity 83 400 350 300 250 a O 200 D E 150 He 100 50 O- - July Aug Sept Oct Nov Dec Jan Feb l\~rch April N~y June Time (months) FIGURE 2-6 Number of days with exceedances of the 8-hour CO standard by month in Lynwood, California, between July 1973 and June 2000. (Data were missing for December 1997.) 300 - 250 - g!, 200- ci 0 1 s 0 - 7 100- . ~n _ O - ~ rat July Aug Sept ~.~. Nov Dec Jan Time (month) - Feb March April May June FIGURE 2-7 Number of days with exceedances of the 8-hour CO standard by month in Fairbanks, Alaska, between July 1972 and June 2001. Numbers in- clude days on which exceedances occured at any of the three monitoring sites.

84 Managing CO in Meteorological and Topographical Problem Areas Coo 160 to 140- 120- X 100- O 80- Q 60 ~ 40 At 20 _ . .~ Sun Man Tues Weds Day I: l = = Thug; Fri Sat FIGURE 2-8 Number of days with exceedances of the 8-hour CO standard by day of the week In Fairbanks, Alaska, over the period July 1972 Trough June 2001. The numbers include days on which exceedances occurred at any of the Tree monitoring sites. Source: Data provided by Paul Rossow, Fairbanks Norm Star Borough. to stronger ground-level inversions because of increased radiative heat Toss. Weekly patterns of CO concentrations and exceedance days occur because of patterns in vehicle use. Higher emissions typically are recorded on weekdays rather than weekends. Vehicle use is especially low on Sunday. This can be seen in Figure 2-8 (above) for Fairbanks. The diurnal patterns of CO concentrations in the cities with the most severe CO problems are discussed below. Comparison of Diurnal Variations in CO Concentrations in Ten Cities Insights into the fundamental physical and human processes involved in the buildup of CO in the atmosphere and how these processes vary among areas can be gained by comparing the diurnal variations of CO concentrations in the cities that have had the most difficulty meeting the 8- hour CO NAAQS (Table 1-1), excluding Birmingham, Alabama. The gen-

Contributions of Topography, Meteorology, and Human Activity 85 eral diurnal variation of CO is described in the most recent criteria docu- ment for CO (EPA 2000a). There are two maxima, a stronger one at 7:00- 9:00 a.m. and a weaker one at 6:00 p.m. (hour 18), and two minima, a deeper one at 5:00 a.m. and a shallower one at 2:00-3:00 p.m. This general pattern can be compared with the diurnal variations measured at the moni- tors listed in Table 1-1. Figure 2-9 shows the diurnal variations of mean and maximum 1 -hour average CO concentrations6 for nonholiday weekdays during the winter of 1999-2000 (the most recent winter for which hourly data were available for the 10 cities) shown in order of decreasing latitude (north to south). The mean concentrations (broad lines in Figure 2-9) show some inter- esting patterns, although these patterns may be due in part to local traffic flow. Although most of the monitors showed both morning and evening maxima, the one at the Post Office building in Fairbanks showed only one maximum in the evening a factor noted in the criteria document (EPA 2000a). A number of the more northern sites show indistinct afternoon minima. The very deep minima in Las Vegas and Calexico in the early afternoon suggest that solar heating at these southern latitudes is able to disperse the CO effectively at that time of day. (To verify that low after- noon CO concentrations are due to meteorology rather than to very Tow afternoon emissions would require an analysis of traffic flownear the mon- itors.) Fairbanks, Spokane, the Anchorage 1 site (AIRS ID 02-020-0037), Kalispell, and Denver each show a daily maximum at 6:00 p.m., attribut- able to local evening rush-hour traffic.7 The times of evening maxima in other areas varied widely; those in Lynwood, Calexico, and Phoenix were at 10:00-11 :00 n.m.8 Las Vegas showed an unusualIv broad evening maxi- ~, 6For each hour ofthe day (starting at midnight), the mean of 1 -hour average CO concentrations for the 85 nonholiday weekdays was determined, along with the standard deviation and maximum for that hour. Weekend days and major holidays (Thanksgiving, Christmas, New Year's Eve, and New Year's Day) were not in- cluded. 7Because the CO concentration may take time to build up, the maximum in concentration might be expected to lag the maximum in traffic flow somewhat. Modeling shows that the lag time depends on the windspeed but is typically an hour or less. Times of these maxima are much later than when people are likely to be out . . exerclsmg.

86 Managing CO in Meteorological and Topographical Problem Areas 9~ ~ a. ~ o 10: as,, ~ . ......... ~~ ~ ~~ hi. ~105 it, ~ :1 ~ -: C! ;: i4- 96f Is ~,4 12 t4 16 1~iB ~ z2 ~ ] `~9~i424 V ~ _ _ ~ ~'~ _ ~ :D ~ ~ At ~ ~ 9~.20 91~2 '1'4 128 398 =,= 2,ij;t - :' :3 t ~ ~ :L - Anchom~ 3,, 616 1~ ~ TIC _ -_~ . _ . :: ~,_~r9$cijif 9¢~, fff;_~_Yf~iSi~_~9iS9fffiffif9<:~& s ~ 9 o ~ 9 £ 0, ~ 4 ~ ^8 10- 12 14 1:6 18 ~ =. Or Alex . l t ~ _. :~- . , r 9 ~ ~ 9 , ; ~ - 9 ~ I] At (i ~ luff) '12> 1'4 14FiiP if'/i~ ;~ ~ ;~ - -Liz =1s5'~ ~ t l:' '1.. [Calispeli~ ~~ 1z ~ ~~ <~ ~ I_ , .. ,, ~ . She's' -, _ _ -, i ~ . · ~ ~ " 9 f 9 0 '2 Jl ~ 6' 1~0 offs i4' the 18; Afar 8,mk~' 4r I'm Am ~ ~~ fly 10~ Incur. 39~ 451 N - : 4 ' s ~ c~ 2' ~ ~ <5 10 12 '!4 fe 16 Flour FIGURE 2-9 Diurnal variations of mean (heavy line) and maximum (light line) 1-hour CO concentrations at monitoring sites listed in Table 1-1 for 85 nonholiday weekdays during the winter (November through February) of 1999- 2000. Sites are shown in order of decreasing latitude (north to south). The

Contributions of Topography, Meteorology, and Human Activity 87 it,= . - a:: an:: ~ Vag~ 36-- 1~ fJ ~ . . ~~ Hi. .~ _ ~ g ~ ' ~ . < A ' A ~ ~ ~ ~ ~ ~ ~ 1:2 1 4: 15 18 -I ~ 24 I'm rare ~:a, 15- . ~ 10- ~1~' 3~ N" ~< /\ ~ f I_ ~ ~ j · _ ~ ~ ~ ~ ~ 6: ~ To 12 1-4 18 ,8 .~ ~ 24 tow ]-l~g~ ~~ _ E15'- 6,'10 5- ~ 'ASH ~ ~ Ply st~ 4e ~ J~^ ~ T ~ . ~ ~ ~ ~ ~ g ~ ~ . O ~ ~ ~ ~ lD 12 14 16 18 .~ ~ 24 Itch o8> hi: Lo, at : ~< - ~ ~ - - ~ ~ - b' ~: g 5 5 5 ~ 5 5 5 g g 10 72 t~ 16 18 Alar eQ ~ AS a:: 5~ tO ~N O~xi=. 3~ 41'~ _ I 5 ' :', .' ~ 5 ~ . 5; ~ ~ ~ ~ ~ ~ to 12 14 lie 18 ~ 22 24 horizontal scale refers to the end of the 1-hour averaging period. There are breaks in the curves for Lynwood and Calexico because the period from 4:00 a.m. to 5:00 a.m. was used for instrument calibration.

88 Managing CO in Meteorological and Topographical Problem Areas mum that was essentially flat from 7:00 p.m to 12:00 p.m., possibly due to late-evening local traffic. The Anchorage ~ and Anchorage 2 sites, located downtown and in a residential neighborhood, respectively, differ in that the latter exhibits a morning maximum. This is likely due to cold-start emissions and periods of engine idle before motorists leave home for work. The average warm-up idle for morning commuters in Anchorage under these very cold conditions is approximately 10 min. some idling as long as 30 min (Morris and Taylor 1998). The highest I-hour CO concentrations observed in Lynwood, Phoenix, and CaTexico (the light lines in Figure 2-9) were unusual. They occurred at 10:00 p.m. or 3:00 a.m., and were consistent with CO transport into these areas from outside sources. Calexico is adjacent to Mexicali, a much larger Mexican city to the south, and Lynwood is surrounded by metropolitan Los Angeles. (See the illustrative examples of CaTexico and Lynwood later in this chapter.) The time of day when an area experiences its daily maximum CO con- centration depends on a variety of factors, including the time dependent diurnal emission rate in the area (including vehicle cold starts and traffics, meteorological variables (especially windspeed and inversion strength near the ground), and in some cases transport in from surrounding areas. Simula- tions of the six most recent exceedance days in Fairbanks (five of which occurred in February) showed that vehicles leaving downtown for home combined with strong evening temperature inversions at sundown yielded the highest CO concentrations at 5:00-6:00 p.m. (NRC 2002~. VULNERABILITY TO FUTURE EXCEEDANCES Once a nonattainment area is able to reduce its ambient CO concentra- tions and has eight clean quarters (no violations ofthe 8-hour CO standard over 2 years), it can apply for a change in attainment status. A state imple- mentation plan (SIP) is required and must show that maximum 8-hour average CO concentrations can be expected to remain below 9 ppm for the next 10 years. The confidence of the projection depends not only on ex- pected future emissions (for example, on the basis of expected vehicle- miles traveled (VMT) and fleet composition) but also on the variability of CO concentrations for a given annual emissions inventory. Some cities show more variability than others.

Contributions of Topography, Meteorology, and Human Activity 89 Effects of Meteorology and Emissions on Vulnerability to Future Exceedances This section examines the effects that meteorology and emissions have on the vulnerability of locations to future CO exceedances. Table 2-1 sum- marizes the characteristics of the mean diurnal patterns shown in Figure 2- 9, indicating the time of the daily maximum at each site for the average nonholiday weekday, the mean CO concentration at that time, and the stan- dard deviation. The next-to-last column provides the coefficient of varia- tion (COY) the ratio ofthe standard deviation to the mean for the daily maximum; the largest COVs (~0.7) are shown in italics. The last column gives the average CO concentrations for the 1999-2000 winter season, including weekends and holidays. The variability indicated in the COV column in Table 2-1 is the result of variability in both emissions rates and meteorological factors. Traffic volumes at 6:00 p.m. (hour 18) on significant roadways in Fairbanks, Alaska, during nonholiday winter weekdays show little variation, with a COV for traffic of only 0.08 ~ 0.01.9 Much greater variability is exhibited in CO concentrations than in traffic. Figure 2-10 compares the variability in daily average traffic on Cushman Avenue in Fairbanks with the variabil- ity in daily (24-hour) average CO concentrations measured at the Post Office monitor on the same road during the winter of 1999-2000. In each case, the data are sorted into two categories: (1) nonholiday weekdays, and (2) weekends and major holidays. Then the data are sorted by decreasing daily average value within each of those categories. The mean daily aver- age traffic for winter weekdays in Figure 2-10 was 616 vehicles per hour, with a standard deviation of 55 (COV = 9.0%), end the mean daily weekday average CO concentration was 2.2 ppm, with a standard deviation of 1.1 (COV = 50%~.10 These figures confirm that CO concentrations were con- siderably more variable than traffic flows. The highest and second-highest daily average CO values for that winter occurred on Tuesday, February 8, 2000, and on Monday, November 19, 1999. Both were exceedance days, 9This mean (0.08) and standard deviation (0.01) are based on an analysis of traffic on three roads in downtown Fairbanks during winters of 1995-1996 through 2000-2001 . For the locations of counters and monitors see Figure 2-4 in the interim report (NRC 2002~. Patois ratio of 0.50 differs from the 0.81 that appears in Table 2-1 because the former refers to a 24-hour mean and the latter to a 1-hour mean.

9o To to to .= 3 - .= z ~ .> ~ ~ ·Ct o ~ Ct Ct · I o Cal A au - o ~ ~ Oo ~ ~ o 0 Go ~ ~ Go _ ~ ~ ~ _ o Ct ~ . V) Ct A · ~ · - o do Do ~ ~ ~ ~ . . . . . _ At ~ _ _ _ To ~ ~ Oo o ,> . . . · . . · . o o o o ~ o o o r~ ~ ~ ~ cr . . . . . ~ - ~ - ~ - ~ oo . . . ~ - ~ ~ ~ ~ oo ~ ~ ~ ~ ~ ~ . . . . . . . . . . ~ oo ~ oo - ~ - ~ o o o o ~ o o - o ~ ~ o ol ol ol o o o ~ o o o ol ol - ~ ol ol o ~ oo ~ oo ~ - ~ - - t - oo ~ ~ ~ o o o o ol ol ol ol ol r~ O O ~ =\ o ~ ~ ~ ~ o o o ~ o 1 1 1 1 1 ~ ~ ~ oo o ~ o o ~ ~ oo - o o ol oo o - ~ a' .—~ ·— ~ cq - - o ~ ~ ~ ~ ~ ~ o ~ ~ o a~ - oo ~o au 'e .~ ~ u, ~ o Ct ~o .$ ~ ~ O.~-= g ~ ° g.° _ ,c~ ~ L~ == ~ °^ e C,: ~ V C,C, 0 s~ ~ ~ ~ ·~ O t's=m ~ _ ~ g O ^,= O V ~ Ct V ~ ~ {~0 0 .o -o ,~ ~ ~ {~e ~ Ce U, cq ~ =^ tC g .~ Ct . Ct o · C) o C) o v . ~ S~ g .~ Ct .~ a~ · u, u, - .O . 5- ct ~ - ~ ~ ~o .= uO . ~ au O ct ^ · - I. - c) · o ~ o ~ o o . - ct ~, · ct

Contributions of Topography, Meteorology, and Human Activity 91 Boo 700 400 300 200 100 (a) Weekdays Weekends and Holidays o 7.0 6.0 1.0 (b) E ~ 5.0 In ~ 4.0 o ~, 3.0 Weekdays Weekends and Holidays 0.0 FIGURE 2-10 Daily average (a) traffic on Cushman Avenue and (b) CO concentrations at the Post Office on Cushman in Fairbanks during the winter of 1999-2000. Values for weekdays are rank ordered, as are those for weekends and major holidays (Thanksgiving, Christmas, New Year's Eve, and New Year's Days). Days with data missing are indicated by missing bars. The daily average traffic was obtained by diving the total traffic count each day by 24 hours. with maximum 8-hour average CO values of 11.5 ppm and 11.2 ppm, respectively. Figure 2-11 shows a scatter plot of the data used for Figure 2-10. The two exceedance days (the top two points) had traffic flows (600-

92 Managing CO in Meteorological and Topographical Problem Areas 700 vehicles per hour) that produced many much Tower average CO con- centrations. An exceedance that occurred in Fairbanks on Saturday, January 11, ~ 997, highlights the importance of meteorological factors. The maximum S-hour average CO concentration that day was 13.3 ppm even though the daily mean traffic on Cushman Avenue was only 437 vehicles per hour. The daily average Tower inversion strength (measured between 3 and ~ 0 m above the ground)was 18.6°C/100 m, the average windspeed was 0.8 MPH, and the average temperature 5.5°F.~ Although traffic is not the only factor determining CO emissions rates (cold-start and idling emissions are also important in Fairbanks), such an exceedance indicates that meteorology may be able to produce exceedances despite emissions reductions. Two cities with the same average CO values in winter (e.g., Kalispell, Montana, and Denver, Colorado) can differ greatly in their vulnerability to future exceedances depending on their respective variability in CO concen- trations. Of the two Anchorage sites, site 2 (Turnagain) had a lower aver- age CO concentration in winter than did site ~ (~.2 and 2.0 ppm, respec- tively), but its much greater variability makes site 2 more vulnerable; in fact, the two most recent CO exceedances in Anchorage occurred at this residential site (see Table 1-1~. Characterizing the non-Gaussian distribu- tion of 8-hour average CO concentrations in a location might make it possi- ble to predict the probability of future exceedances based on projected emissions inventories. (Gaussian models are discussed in detail in Chapter 3 .) In addition, to adequately test the hypothesis that high variability in CO concentrations helps explain the difficulty that some areas have had in meeting the standard, the variability in CO concentrations in areas that met the standard relatively easily should also be examined. Assessment of Vulnerability The variability in CO concentrations leads to difficulties in predicting high CO episodes especially in geographical areas with unusually challeng- i~The ranges ofthe daily averages for these three meteorological variables for the November through February 1996-1997 winter season were as follows: lower inversion strength, -4.3 to 18.6°C/100 m; windspeed, 0.7-3.9 MPH; and tempera- ture, -43 to +33°F. Data were provided by Paul Rossow ofthe Fairbanks North Star Borough.

Contributions of Topography, Meteorology, and Human Activity 93 ~ ~ Weekends and Holidays ~ Weekdays ~ p7.0- 8 Q6.0_ ~ dV 5 0 ' ,04.0- ~3.0- ~ O 10 = ' ' ~ ·; · ·., ~ art; ~ Q go . . - aa i --at 0 100 200 _ 1 ' ' 300 400 500 Average Traffic on Cushman (Vehicles/Hr) 600 700 800 FIGURE 2-11 Scatter plot of Fairbanks traffic and CO during Me winter of 1999-2000. ing meteorological and topographical conditions. Variability in meteoro- logical conditions, such as strong temperature inversions or winds blowing from the direction of nearby communities with high levels of CO, contrib- utes to these difficulties. The status of problem areas could fluctuate be- tween attainment and nonattainment until further emissions reductions provide an adequate safety margin. Areas that have achieved attainment recently and do not yet have an adequate safety margin remain vulnerable to high CO episodes. Nonattainment might occur sporadically under unfa- vorable meteorological conditions, even when emissions rates remain at the levels projected in the SIP. Vulnerability can be expressed in terms of the probability of non- attainment in a future year or in terms of the reciprocal of this probability, which can be interpreted as the number of years that are likely to pass until the next nonattainment year takes place. The latter is analogous to a design condition in civil engineering, such as when a bridge is designed to with- stand a once-every-hundred-year's flood. It is also analogous to the con- cept in public health of the number needed to treat (NNT), defined as the reciprocal ofthe probability for a categorical change in outcome (e.g., from death to survival) for a randomly selected future patient. Given its stochastic nature, vulnerability is determined by both the central tendency (such as the median or mean) and the spread (such as the

94 Managing CO in Meteorological and Topographical Problem Areas interquartile range or standard deviation) of the air quality indicator (e.g., the annual second maximum nonoveriapping 8-hour average CO concentra- tion). An area with a large spread might have a substantial probability for nonattainment in future years even after achieving attainment for several years. To reduce vulnerability, emissions reductions must extend beyond the attainment threshold to provide an adequate safety margin. ILLUSTRATIVE EXAMPLES This last section provides five illustrative examples of locations that have had problems meeting the NAAQS for CO. The roles of topography and meteorology in concentrating CO and how those factors combine with patterns of emissions to produce episodes of high CO concentrations are briefly described. The committee realizes that the definition of a meteoro- logical and topographical problem area might be somewhat arbitrary, be- cause in all areas meteorology and topography are important factors in producing, concentrating, dispersing, or eliminating all air pollutants. In these examples, the committee primarily focuses on locations in the west-

Contributions of Topography, Meteorology, and Human Activity 95 em United States that experience winter temperature inversions and have topographical features contributing to the accumulation of CO. Calexico, California Calexico is located 125 miles (ml) east of San Diego in California's Imperial Valley, on the border with Mexico. This small city's population of 27,109 (in 2000) is similar to Fairbanks's (30,224) (U.S. Census Bureau 2000b).~2 However, it is across the border from Mexicali, a much larger Mexican city with a population of about 750,000. Motor vehicles on the MexicaTi side tend to be older and tend to have less sophisticated emissions control equipment that sometimes is not functioning properly or has been removed. In addition, Mexicali has no vehicle emissions inspection and maintenance program. The number of exceedance days recorded since 1995 at a monitoring site in Calexico (59) is surpassed only by the number recorded in Birmingham, Alabama. Although Calexico is a major border crossing point (an estimated 2,16S,000 vehicles crossed the border from Mexico in 1999 fCaTexico 19993), CO measurements in Mexicali indicate that the problem is not due to Tong lines of idling vehicles at the border. The committee initially thought that CO episodes in Calexico did not fit the profile of locations whose problems were created by meteorology and topography. However, monitoring of CO concentrations in Calexico indicated that the movement of a large, CO-rich air mass northward across the border is responsible for most of the CO pollution in CaTexico. This is especially true at night in winter when windspeeds are low and ground- leve] temperature inversions are strong. The situation is exacerbated by Calexico's topographical location in a valley down-slope from Mexicali. (The SaTton Sea, 30 mi to the north, is 235 feet below sea level.) Although Calexico does not have confining topography that traps or accumulates CO, its topography puts the city in the pathway of CO drifting across the border from MexicaTi, and its meteorology prevents CO from dispersing vertically. Further analysis is needed to assess the relative roles of cross-border trans- port of vehicles operating in MexicaTi compared with vehicles idling at the border in producing CO. To address the problem of cross-border air pollu- Tithe population ofthe Fairbanks North Star Borough, with a much larger area of about 7,000 square miles, was 82,840 in 2000 (U.S. Census Bureau 2002).

96 Managing CO in Meteorological and Topographical Problem Areas tion, the United States and Mexico have agreed to a new Border Air Quality Strategy, announced by EPA on November 26, 2002 (EPA 20026~. Lynwood, California Lynwood is a community of about 64,000 located approximately 12 mi east of the Pacific Ocean. It is south of downtown Los Angeles in the Los Angeles Basin, which has a total population of over 1 ~ million. Lynwood is a densely populated area with numerous freeways and highways, and many high-emitting vehicles. There were 58 exceedance days during the 7-year period from 1995 to 2001. Numerous studies of the CO problems in Lynwood, including those conducted by the California Air Resources Board (Nininger 1991; Bowen et al. 1996), have been undertaken to assess why CO concentrations are higher in Lynwood than other parts ofthe Los Angeles area. These studies are aimed to determine the relative contributions of local versus area sources of CO and the roles of meteorology and topography on CO concen- trations. Motor-vehicle emissions are clearly important. About half of Lynwood's CO emissions come from just 10°/O of the light-duty vehicle (LDV) fleet (Lawson et al. 1990~. Singer and Harley (1996, 2000) also noted CO emissions rates of vehicles registered in the Lynwood area were double those registered in higher income areas because of the prevalence of older vehicles. Nininger (1991) concluded that the entire Lynwood area is a CO hot spot, due not only to high vehicle emissions but also to Tower windspeeds and mixing volumes that occur in surrounding areas. Bowen et al. (1996) further qualified the role of meteorology and topography in contributing to the high concentrations of CO in Lynwood. A strong, surface-based inversion occurs in the Los Angeles area soon after sunset, and the strength of the inversion appears to be greater near Lynwood. In addition, the gradient of the terrain is smaller near Lynwood than at most other locations in the area, resulting in weaker nocturnal drainage winds. The study concluded that significant CO emissions originating in the Lynwood area are added to an urban air mass with high CO concentrations that is transported into the Lynwood area. These emissions sources com- bine with stable nighttime meteorological conditions to create high CO concentrations in Lynwood.

Contributions of Topography, Meteorology, and Human Activity 97 Fairbanks, Alaska Fairbanks, Alaska, is a small city in which topography, meteorology, and emissions conspire to produce air pollution in winter (NRC 2002~. The meteorological and topographical characteristics of Fairbanks's air pollu- tion problems are discussed in Bowling (1984, 1986~. The city has apopu- lation of about 30,000 and is located in central Alaska in the Fairbanks North Star Borough, a sparsely inhabited area of over 7,000 square mi with a total population less than 85,000. The city is a center for government, education, and distribution for the northern part of Alaska. Fairbanks is sheltered by hills to the west, north, and east and is situ- ated on Tow ground near the confluence of the Chena and Tanana Rivers. The terrain is open to the south, with the Alaska Range roughly 45 mi away. The meteorology is extreme continental arctic. Low winter tempera- tures are combined with unusually strong ground-based inversions and low windspeeds. The warmest point in a vertical temperature sounding is com- monly more than a kilometer above the surface, and near-surface inversion strengths often exceed 10°C/100 m. These factors greatly limit the amount of air available to dilute and disperse CO and other pollutants. Temperatures during the winter months are normally below 20°F bel- ow the limits of federal guidelines for cold-start emissions—so engine starting emissions can be substantially higher than normal. As tempera- tures drop below 0°F, automobiles become increasingly difficult to start. At temperatures below -20°F (not uncommon in Fairbanks during winter), most people use preheating "plug-ins," because it is nearly impossible to start a vehicle that has not been preheated. Engine preheating reduces cold- start CO emissions, so high ambient CO levels are rare at those low temper- atures. Encouraging the use of pretreating plug-ins at temperatures between 20 and 0°F, when unheated vehicles can be started but emit large amounts of CO, is the centerpiece of the borough's strategies to reduce CO emis- s~ons. Las Vegas, Nevada Las Vegas is a rapidly growing city in southern Nevada that had a population of nearly 41 S,000 in 1999, up from about 260,000 in 1990 (U.S. Census Bureau 2000c). Las Vegas is located in a valley surrounded by

98 Managing CO in Meteorological and Topographical Problem Areas mountains: the Spring Mountains to the west, the Pintwater, Desert, Sheep, and Las Vegas Mountains to the north; Frenchman Mountain to the east; and the McCullough and Big Spring Ranges to the south. Automobile and truck traffic go to and through the city 24 hours a day via three major high- ways. In late fall and throughout winter, coo] air drainage winds from the adjacent desert hills flow into the city and pool there, resulting in a local accumulation of CO. This pooling effect has resulted in a total of seven exceedance days recorded at two monitoring sites since 1995. Population growth is expected to continue, so future violations are a serious concern. Las Vegas undertook a significant CO saturation study to help assess the monitoring network and movement of CO (Ransel 20021. The study extensively augmented the 14 permanent monitoring sites with 63 tempo- rary fixed sites and mobile sampling at over 2,500 locations. The study concluded that the current monitoring network captures the peaks and extent of high CO. It also noted that a tongue of high CO appears to be caused by nocturnal drainage flow that follows the Las Vegas Wash. The study noted that, away from the urban core and effects from transport in the drainage flows, CO levels are relatively low (Ranse! 2002~. Denver, Colorado The city of Denver has a population of 501,700 (recorded in 2000) and is located in the South Platte River Valley, approximately 1 mi above sea level. To the west of Denver is the Front Range of the Central Rocky Mountains, with peaks above 14,000 feet. The Cheyenne Ridge (about 70 mi to the north) and the Palmer Divide (about 25 mi to the south) run east to west and are 1,000-2,000 feet above the plain; they combine with the Front Range to form a three-sided basin in which the city sits. Denver is a major national rail center, and two major interstate highways cross the city. This growing community had hundreds of CO exceedances in the 1970s and l980s. Since 1995, there have been only two. The decline is a result of local controls (including wood-burning bans), technological im- provements in motor vehicles and wood burning stoves, and favorable winter weather patterns that have permitted better ventilation of pollutants. However, the city remains vulnerable to future CO exceedances because of its steady population growth. The meteorological factors that contribute to elevated CO concentra- tions in Denver include: persistent light winds at the surface, a ground-

Contributions of Topography, Meteorology, and [Iuman Activity 99 level inversion, a lee trough along the foothills, snow cover, and warm air advection aloft (Neff and King 1991; King 1991; Neff 2001; Reddy 20011. Reddy (2001) associated elevated CO concentrations with winds at less than 1 MPH and an effective mixed layer less than 25 to 50 m that lasts for at least 3 hours. Neff (2001) also noted microclimatological factors in- volved in producing exceedances at the most problematic CO monitor. He noted that extensive shadowing by downtown buildings in the afternoon may exacerbate the trapping of pollutants in the area surrounding the moni- tor by prolonging the cooling ofthe surface during the winter (thus intensi- fying the ground-level inversion) and by blocking winds that could disperse pollutants.