Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Elevated ozone layers and vertical mixing in the lower fraser valley, British Columbia Lundgren, Jeffrey Ross Stanley 2000

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2000-0094.pdf [ 7.6MB ]
Metadata
JSON: 831-1.0089391.json
JSON-LD: 831-1.0089391-ld.json
RDF/XML (Pretty): 831-1.0089391-rdf.xml
RDF/JSON: 831-1.0089391-rdf.json
Turtle: 831-1.0089391-turtle.txt
N-Triples: 831-1.0089391-rdf-ntriples.txt
Original Record: 831-1.0089391-source.json
Full Text
831-1.0089391-fulltext.txt
Citation
831-1.0089391.ris

Full Text

Elevated Ozone Layers and Vertical Mixing in the Lower Fraser Valley, British Columbia By Jeffrey Ross Stanley Lundgren B.Sc. (Geophysics) 1992, Diploma of Meteorology 1994, University of British Columbia, Vancouver, B.C. A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF M A S T E R OF SCIENCE in THE F A C U L T Y OF G R A D U A T E STUDIES DEPARTMENT OF GEOGRAPHY ATMOSPHERIC SCIENCE P R O G R A M M E We accept this thesis as conforming to the required standard T H E U N I V E R S I T Y O F BRITISH C O L U M B I A © February 2000 Jeff Lundgren in presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, 1 agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of & gCyCQAPtiV The University of British Columbia Vancouver, Canada DE-6 (2/88) 11 Abstract The vertical distribution of tropospheric ozone in the Lower Fraser Valley (LFV) is examined. The criteria of what constitutes an elevated ozone layer is defined and a total of 105 profiles from aircraft, tethersonde and free ascent balloons obtained during the period of 1993-1996 were examined and a total of 110 elevated features identified according to these criteria. The results show a rich variety of structures present from the boundary layer up to the upper free troposphere. This richness is shown to be a result of a multitude of generation processes, ranging in scope from the local up to synoptic scales, that interact to produce elevated layers. The layers are typed and sorted according to a classification scheme based on the altitude of the layers with respect to significant atmospheric boundaries, such as the height of the mixed layer, and the specific processes of generation as inferred from other studies performed in similar locales, most significantly within the Los Angeles Basin. Layers are classified as Type I,n,ffl or IV ranging in height from within the mixed layer up to the upper troposphere, with Type I through IU being the result of pollution from within the L F V and Type IV layers being present as a result of long range transport from distant sources. Type I and II layers are shown to have the strongest potential for affecting ground level ozone and may be one of the causes of the strong temporal autocorrelation of diurnal near-surface concentrations, over a time scale of 2 to 5 days, during episode conditions. Slope and sea breeze flows are thought to be of great importance in the creation of Type II and HI elevated pollution layers from L F V sources to heights levels nearing 5000m a.g.l. and distances on the order of 100 km from Greater Vancouver. This suggests pollution exists on a vertical and ii iii horizontal scale that was not previously believed and is usually only associated with areas that are thought of as seriously polluted, such as the L .A . basin. Type IV layers are present at higher levels and appear to have their sources within the boundary layer, as opposed to being from stratospheric intrusions, possibly as a result of plumes from biomass burning. Trajectory analysis of upper level Type IV features indicates that long range transport of tropospheric ozone is occurring, perhaps on continental scales. This ability to be transported long distances is attributed to a much longer lifetime, up to 90 days versus 4-5 days, for ozone in the free troposphere as compared to ozone in the planetary boundary layer. Transilient turbulence theory is outlined as an appropriate means of mathematically describing mixed layer turbulence. A simplified transilient model, based on the available data, is presented to investigate the extent to which vertical downmixing of elevated ozone features may influence ground level ozone concentrations within the L F V . A case day, August 6 t h, 1993, is examined on which both a Type I and a Type II layer are present in the tethersonde profiles over Harris Rd. The results for both of these layers show that vertical downmixing of elevated ozone may account for as much as 50% of the mass increase in ground level concentration during the time of the model run. The influence of both or these layers is apparent in the ground level time series from Harris Rd. for this day. i n IV Table of Contents Abstract ii Table of Contents iv List of Figures vi List of Tables X Chapter 1 Introduction 1 Ozone Meteorology of the Lower Fraser Valley 16 Chapter 2 Data Sources 35 Transilient Turbulence 43 Chapter 3 Elevated Layers Within the Lower Fraser Valley 58 Layer Classification 59 Layer Definition 65 Lower Fraser Valley Layer Inventory 69 Case Studies of Elevated Layer Types in the Lower Fraser Valley 76 Summary of Lower Fraser Valley Layer Formation 106 Chapter 4 Transilient Turbulence Model 110 Chapter 5 Model Results: Vertical Downmixing of Elevated Ozone Layers 126 Chapter 6 Conclusions 146 Appendix A Derivation of the T K E Transilient Parametrization 153 Appendix B Listing of Turbulence Model FORTRAN Code Subroutines 156 IV Appendix C Sensitivity Analysis 162 References 168 V I List of Figures Figure 1.1.1 Diurnal Variation in the concentration of the components of a photochemical smog (From Bunce, 1991) 8 Figure 1.1.2 Contours of steady state ozone concentration vs. NOx and V O C (From Seinfeld, 1989) 9 Figure 1.1.3a and 1.1.3b N O x and V O C emissions in the Lower Fraser Valley (From Steyn, et. al, 1997) 11 Figure 1.1.4 Ozone isopleths for a severe episode in the L F V (From Joe et. al., 1995) 13 Figure 1.1.5 Hourly ozone concentration vs. cumulative frequency (From Steyn et.al.,\991) 15 Figure 1.2.1 Closed L F V circulation under condition of strong synoptic subsidence and a well developed sea breeze(From Oke, 1987) 17 Figure 1.2.2 Local pressure gradients and winds induced in sloping terrain (Whiteman, 1990) 18 Figure 1.2.3 Boundary layer evolution under anticyclonic conditions (Stall, 1988) 21 Figure 1.2.4 Development of modeled vs. observed near surface ozone (From Neu et. al, 1994) 23 Figure 1.2.5 Genesis of elevated layers in a coupled sea breeze/mountain wind domain (From Lu and Turco, 1994) 26 Figure 1.2.6 Vertical ozone sounding showing elevated layers over Los Angeles (From McElroy and Smith, 1993) 27 Figure 1.2.7 Tropospheric ozone from TOMS and SAGE Data 1979-1987 (Fishmanef. al, 1990) 31 Figure 1.2.8 Same as 1.12 for 1979-1989 (Fishman et. al, 1991) 33 Figure 2.1.1 Map of the L F V showing scope of Pacific 93 field program (Steyn, et. al, 1997) 37 Figure 2.1.2 Ascent and Descent AIR Ozonesonde Profiles Showing Distinct Hysterisis 39 vii Figure 2.1.3 Light Aircraft Measurement Flight Lines (McKendry et. al., 1998) 41 Figure 2.2.1 The Advective Nature of Turbulence as Shown by varying Grid Sizes (From Stull, 1993) 44 Figure 2.2.2 Examples of Atmospheric Eddy Structures (From Stull, 1993) 44 Figure 2.2.3 The Turbulent Energy Spectrum (From Stull, 1993) 45 Figure 2.2.4 Atmospheric Stability Determined by Non-Local Conditions (From Stull, 1993) 46 Figure 2.2.5 Example of Transilient Exchanges and the Corresponding Matrix for a N=5 Grid. (From Stull, 1988) 48 Figure 2.2.6 Examples of idealized transilient matrices showing non-local mixing possibilities. (Stull,1988) 51 Figure 2.2.7 Interpretation of Transilient Matrices (From Stull, 1988) 51 Figure 2.2.8 Destabilization and Mixing 'Half Steps' For Each Mixing Interval (Stull, 1988) 56 Figure 3.2.1 Sample Ozone Profile over Langley 13:00 PST Aug. 1, 1993 66 Figure 3.3.1 Elevated Layer height and Depth versus Above Background Ozone Concentration with Layer Types Identified a Priori 72 Figure 3.4.1.1 Ozone and Meteorology 05:58 PST July 21, 1994 at Harris Rd 78 Figure 3.4.1.2 Daytime Type I Layer over Pitt Lake, 13:52 PST Aug. 2, 1993 80 Figure 3.4.2.1 Aircraft Profiles of Type II Layers for Three Sites: (a) Boundary Bay (b) Pitt Meadows (c) Harrison Lake 83 Figure 3.4.2.2 Type U Layer over Langley 16:00 PST Aug. 4, 1993 84 Figure 3.4.2.3 Aircraft Measured Ozone from Haney to Stave Lake, 01:00 PST Aug. 24,1996 88 Figure 3.4.2.4 East-West Lidar Trace from South End of Pitt Lake, Aug. 5, 1993 (Hoffef. al, 1997) 88 Figure 3.4.2.5 a) Type U Layer over Langley 16:00 PST Aug. 2, 1993 viii b) Profile 04:00 PST Aug. 3 shows no trace of the Type H Layer 90 Figure 3.4.2.7 a) Type H Layer over Langley 13:00 PST Aug. 5 1993 b) Type H Layer still visible at 04:00 PST Aug. 6 91 Figure 3.4.2.6 Back Trajectories Showing Open Circulation of Aug. 2 16:00 PST vs. Stagnated Flow of Aug. 5 16:00 PST 92 Figure 3.4.3.1 Middle Tropospheric Profile over Langley, 16:00 PST Aug. 5, 1993 95 Figure 3.4.3.2 Ozone and Aerosol from NRCC Aircraft ascent over Hope 09:00 PST Aug. 5, 1993 96 Figure 3.4.3.3 Aircraft Ozone Profile over the North End of Pitt Lake 14:00 PST July 19, 1995 98 Figure 3.4.3.4 Aircraft Measured Ozone Transect over Pitt Lake, 14:30 PST Aug. 24, 1996 99 Figure 3.4.4.1 Upper Troposphere Ozone Profile over Langley 04:00 PST Aug. 6, 1993 101 Figure 3.4.4.2 Upper Troposphere Ozone Profile over Langley 04:00 PST Aug. 2, 1993 105 Figure 3.5.1 Type I Layer Generation for Nocturnal (Residual Layer) and Daytime Cases 107 Figure 3.5.2 Generation of Elevated Ozone Layers for an Idealized Coastal Region 108 Figure 4.1 Processes Affecting the Vertical Mixing Column 112 Figure 4.2 Sketch of Valley Geomoerty and its Effect on Advection Through Model Domain 113 Figure 4.3 Flowgram of the Transilient Model 124 Figure 5.1 Descent Ozone and Meteorology From Harris Rd. 08:00 PST, August 6 t h, 1993 128 Figure 5.2 Time-Height Section of Ozone From Harris Rd August 6 t h, 1993 129 Figure 5.3 Descent Ozone and Meteorology From Harris Rd. 12:37 PST, viii ix August 6 t h, 1993 130 Figure 5.4 Hourly Sequence of Modeled and Observed Potential Temperature 133 Figure 5.5 Hourly Sequence of Modeled and Observed Vertical Ozone for the Type I Elevated Layer '. 136 Figure 5.6 Hourly Sequence of Modeled Heat and Ozone Flux Profiles for the Type I Elevated Layer 137 Figure 5.7 Hourly Sequence of Modeled and Observed Ozone Profiles for the Type II Elevated Layer 139 Figure 5.8 Hourly Sequence of Modeled Heat and Ozone Flux Profiles for the Type II Elevated Layer 140 Figure 5.9 Ground Level Ozone vs. Time at Harris Rd. August 6 t h ,1993 141 Figure A.C.I Mixed Layer Model Development With Time Step 30s 162 Figure A.C.2 Mixed Layer Model Development With Time Step 180s 163 Figure A.C.3 Mixed Layer Model Development With Surface Heat Flux 0.25 Km/s 164 Figure A.C.4 Mixed Layer Model Development With Surface Heat Flux 0.10 Km/s 165 Figure A.C.5 Mixed Layer Model Development With Averaging Interval of 30 Minutes 166 Figure A.C.6 Mixed Layer Model Development With No Averaging Interval 167 ix X List of Tables Table 1.2.1 Classification of free tropospheric ozone features by air mass type (Browell, et. al. 1992) 30 Table 3.1.1 Summary of Classification of Elevated Ozone Layer Types 61 Table 3.3.1a Layer Occurrence by Data Source 70 Table 3.3.1b Layer Occurrences by Type 70 Table 3.3.2 Summary of Layer Type Features 75 Table 5.1 User Defined Model Parameter Values 132 x 1 Chapter 1 Introduction and Literature Review I Introduction Greater Vancouver is a metropolitan area of approximately 1.5 million people located at the western end of the Lower Fraser Valley, (hereafter LFV), on the southwest coast of British Columbia. A combination of climate, topography and pollutant emissions makes this region susceptible to episodes of photochemical smog (Steyn, elal., 1990). Smog is a common name given to an urban atmospheric contamination of emitted nitrous oxides and hydrocarbons and their secondary photochemical oxidants. The most important of these products is ozone. In many respects, ozone is perhaps the most significant urban pollutant of the 20th century. Smog episodes are common in many large cities that experience periods of warm stable weather. Such cities include Los Angeles, Athens and perhaps the most extreme example, Mexico City. The occurrence of elevated levels of near surface ozone is the result of a complex interplay between pollutant emissions, chemistry, topography and ambient meteorology. Bunce (1991) summarizes these processes into five 'conditions' that result in the accumulation of boundary layer ozone. 2 1) elevated atmospheric concentrations of oxides of nitrogen (NOx) 2) elevated atmospheric concentrations of VOC's (volatile organic compounds) 3) ambient temperatures above 18° C 4) strong ultraviolet insolation 5) a reduced mixed layer depth The first two of these conditions supply the reactants to produce ozone. The third and fourth are necessary to drive the photochemical reactions, and the last one allows the pollutants to accumulate. The reactants are a result of the pollutant emissions inventory for a given location while conditions 3 - 5 are a direct result of the ambient meteorology. In addition to these general conditions, the local synoptic conditions must be such that ozone will accumulated in the given locale. For example, high ozone concentrations are unlikely if the conditions 1-5 occur in the presence of 30 m/s winds. Until recently, understanding of the distributions of pollutants within the LFV has been based around a sparse surface monitoring network limited to the lowlands of the LFV basin. However, recent field studies have revealed the presence of distinct elevated pollutant layers over the LFV (Evans et. al, 1993), and demonstrated the potential for pollutants within elevated layers to be mixed to the ground (McKendry et. al, 1997). However, very little is known about the frequency of occurrence of these features, the specific mechanisms by which they are created and perhaps most significantly, the degree to which downmixing from these layers may influence near surface ozone concentrations. To this end, the archive of vertical profiles collected at several locations in the LFV over 3 the past six years by a variety of agencies affords an unprecedented opportunity to investigate the characteristics and occurrence of elevated pollutant layers and the impact of elevated ozone layers upon ground level air quality. Understanding of such elevated layer structures is vital as they: 1) Represent a potential sink for pollutants from the ABL, thereby effectively venting the Atmospheric Boundary Layer. 2) Have the potential to be mixed to ground and thereby contribute to ground level pollutant concentrations that influence human health. 3) Must be adequately resolved by air quality models if predictions are to be valid. 4) Influence tropospheric chemistry and therefore possibly the global climate system. Study Objectives After Steyn et. al. (1997), one of the objectives of Pacific 93 was to provide some insight into "an almost complete ignorance of the vertical structure of all pollutants and precursors" within the LFV. This study attempts to provide an inventory and classification of some of the more salient features present in the vertical distribution or layering of ozone pollution within the valley. Preliminary measurements involving lidar and balloon 4 sondes indicate that there is a complex structure to the vertical distribution of ozone above the LFV. It is not clear how these structures are created or what the fate is of the pollutants within them. Specifically, the lifetime of ozone within these layers is unknown and the potential for pollutants within these structures to be recirculated to ground level has not been determined. In addition, a model of boundary layer turbulent entrainment and mixing is presented to try to quantify the extent to which the pollutants within these elevated layers may be vertically down-mixed back to ground level. The remainder of Chapter 1 provides a review of the literature of ozone pollution meteorology, in particular of previous studies performed within the LFV and sites exhibiting a similar topographic setting. Most significant among these are studies identifying vertical features in ozone episodes within the Los Angeles Basin. Chapter 2 gives the methodological basis of the study. The wide variety of data sources exploited in the study are outlined, and an overview of the discrete form of Transilient turbulence theory as a method parametrizing the turbulent mixing of atmospheric tracers within the boundary layer is provided. Chapter 3 examines and attempts to classify the multitude of elevated ozone features present within the LFV from the boundary layer up to the upper free troposphere. The generation of these elevated layers is examined in terms of the wide spectrum of meteorological phenomena involved, ranging from local and mesoscale slope and sea breeze flows up to the synoptic scale circulation. In addition, the possible impacts of pollutants within these layers on surface level air quality is examined. 5 Chapter 4 presents a model of atmospheric mixing, based on Transilient Turbulence theory and adapted to suit available data, to calculate the vertical transport of ozone within the boundary layer. In Chapter 5, the transilient model is applied to two elevated ozone layers for a case day of August 6, 1993. Vertical fluxes of ozone implied by vertical mixing are calculated and the transilient matrices associated with this process are interpreted in terms of the type and scales of the mixing that is inferred. Lastly, Chapter 6 summarizes the study results and interprets their importance in terms of the ongoing air quality and atmospheric model studies being conducted within the valley. Photochemistry of Tropospheric Ozone Bunce (1991) describes the photochemistry of tropospheric ozone. The initiating species in ozone production is NO2. NO2 is important because it is the only gas present in the urban boundary layer that shows significant absorption of the solar radiation that penetrates into the troposphere. The reaction sequence is as follows: N02 *V-A<400nm >NO + 0 0 + 02+M >03 NO + 03 >N02 + 02 (1.1) 6 The net result of this sequence is N02+02<—>NO + 03 (1.2) which is a pseudo equilibrium that is driven to the right photochemically and to the left thermally. This results in a small steady state concentration, dependent on the ambient temperature and amount of sunlight, given by where k is the rate constant for reaction (1.1) and J is the photolysis rate of NO2. In reaction (1.1) one ozone molecule is required to regenerate a molecule of NO2. The rate of reaction for this sequence is not sufficient to produce on the level present in typical photochemical smog. A reaction sequence that converts NO to NO2 without consurning ozone will allow ozone to accumulate. The presence of hydrocarbons in the atmosphere provides such a path. The small concentration of ozone from equation (1.2) absorbs UV-B (~300nm) radiation by (1.4) This allows production of hydroxyl radicals from atmospheric water. 0(1D) + H20->20H (1.5) 7 In the presence of hydrocarbons, this concentration of OH can then catalyze the oxidation of a variety of reactions of volatile hydrocarbons in a sequence that produces more ozone. In particular, peroxy radicals (RQ2 where R is some alkyl group such as CH3 or C2H5) produced in the oxidation of hydrocarbon molecules react with NO to form NO2, shifting the photostationary state of equation (1.2) in favor of ozone production. The process is R02 + NO > N02 + RO N02+hv- >NO + 0 0 + 02+M >03+M Net. R02+02+hv >RO + 03 (1.6) The generation of the organic peroxy radical RO2 occurs as a result of attack from the OH produced by equation (1.5). Seinfeld (1989) outlines the reactions by which OH attack on a variety of hydrocarbons produces peroxy radicals. This sequence of reactions for a typical smoggy day is shown graphically in Figure (1.1.1). NO and hydrocarbons are emitted during the morning rush hour. As the sun rises, NO2 begins to form as NO is oxidized. From equation (1.1) the amount of ozone also starts to rise. The ozone absorbs the increasing solar radiation resulting in the production of the hydroxyl radical which in turn reacts with VOC's to produce ozone according to equation (1.2). Aldehydes rise later in the day as byproducts of the reactions between OH and VOC's. As the concentration of the OH radicals continues to rise, the NO2 is then 8 Figure 1.1.1 Diurnal Variation in the concentration of the components of a photochemical smog (From Bunce,1991) c o f——I 1 1 1 1 1 4 8 12 16 20 24 T i m e of day , hours consumed by N02 + OH+M >HN03 (1.7) There is no peak of NO x or hydrocarbons for the afternoon rush hour because the free radical chain reactions are fully underway and NO is quickly scavenged by the high ozone driving equation (1.1) to the left. As the sun begins to set, the "fuel" for the reactions is cut off and ozone begins to decrease as sinks become greater than sources. The dominant sink for boundary layer ozone is dry deposition; ozone is quickly destroyed in contact with the Earth's surface (Garland and Derwent, 1979). The primary photochemical sink for tropospheric ozone is the reaction of metastable atomic oxygen, 0(XD), one of the products of ozone photolysis, with water vapor, equation (1.5), to produce OH which in turn consumes N0 2 by reaction equation (1.7). The consumption of N0 2 shifts Jhe equUibrium of equation (1.3) and ozone concentrations ML (Seinfeld, 1989). After sunset, the dominant "dark" chemical sink for ozone is N02+03 >N03 + 02 (1.8) 9 Figure 1.1.2 Contours of steady state ozone concentration vs. NOx and VOC (From Seinfeld, 1989) I > » 1 I I I i i i | 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 VOC, ppm carbon N02 + 0 3 > N03 + 02 (1.8) Production of ozone is maximized when the ratio of [F0C]/[M2X] is near 5, (Bunce, 1991). Further it is claimed that as the level of this ratio in many chemical smogs is on the order of 10, reduction in V O C alone for a specific locale cannot significantly alter production of ozone. Similarly, if N0 X concentration is low, then ozone may actually be consumed by OH+03 >02+H02 H02 + 0 3 — - > OH+202 (1.9) R02+ H02 > ROOH + 02 and the steady state ozone is independent on N O X . The relation between [FcX?]/[M3x]and the steady state ozone is shown in Figure (1.1.2). From this graph it is apparent that depending on the concentration of each species, reducing one species 10 without the other may have no effect on, or even increase ozone. For example, at position 1, the reactions are NOx limited and lowering the VOC has no effect. Similarly at position 2, the atmosphere is relatively low in VOC and here decreasing NO x alone can actually increase ozone. This shows that a knowledge of roles that NO x and VOC play in the previously shown reactions for a particular location is crucial to the undertaking of any kind of emissions strategy aimed at reducing ozone levels. VOC and NOx Emissions Steyn et. al. (1997) describe the emissions of precursors to ozone for the LFV. The patterns of anthropogenic NO x and VOC emissions are given in Figures (1.1.3 a) and (1.1.3b), respectively. Data are in gm/day and are aggregated into 5 km grids for modeling purposes. These figures confirm what one might expect intuitively; that the patterns for both species are strongly correlated with population density. Both show maxima for the downtown core of Vancouver, with lesser maxima corresponding to secondary population centers, such as the municipalities of Burnaby and Richmond. The dominant source of NO x in the atmosphere is the combustion of fossil fuels. Pryor et. al. (1995) estimate that 75% of NO x emissions in the LFV are from mobile sources, primarily light duty automobiles. Most of the remainder comes from point sources such as space heaters and industrial stacks. They also claim that 50% of VOC emissions are from mobile sources. These are mainly from incomplete combustion of fuels that are then emitted as exhaust gases. Stationary sources of VOC's include chemical and petroleum industries and commercial food processing. 11 Figure 1.1.3a and 1.1.3b NO x and VOC emissions in the Lower Fraser Valley. The data are in g / day and are aggregated into 5 km grid squares. ( Steyn, et. al. 1997) 460 500 520 540 560 580 (a) UTM Coordinates (x103) 12 Of much concern in LFV ozone studies is the role of biogenic emissions of VOC's. According to Beisenthal et. al. (1997), in a case study of an episode in August 1993, about 13% of the ozone produced at a site near Pitt Meadows was from the reaction of hydroxyl radicals with biogenic VOC's, most likely isoprene. It is well known that some species, such as certain types of pine trees, can produce significant amounts of pinene and isoprene. However, the concentration and speciation of biogenic VOC's during ozone episodes is not well understood. Compounding the problem of estimating biogenic VOC emissions is the fact that they depend not only on how much a certain species emits but also on the populations of the various species in both the rural and urban environments, and the dependence of plant metabolism on quantities such as temperature and sunlight. Drewitt et. al. (1998) examines the rates of VOC emissions from several common tree species. They found emissions of isoprene and monoterpenes from native species such as Western Red Cedar, Douglas Fir and Coastal Hemlock were quite low. Cottonwoods also showed low emissions of monoterpenes but very high emissions of isoprene. Spatial and Temporal Distribution of Ground Level Ozone in the Lower Fraser Valley Ground level ozone exhibits a very distinct seasonal cycle. The highest values occur in summer and the lowest in winter. The analysis of Pryor et. al. (1995) showed that for the period of 1984-1992, all exceedances of the federal standard for ground level ozone occurred between April and October with the most in July and August. The statistics of ozone are dominated by the vast majority of low pollution values while the 13 Figure 1.1.4 Ozone isopleths for a severe episode in the LFV (From Joe et al, 1995) extreme values represent relatively infrequently occurring cases of severe pollution. Also, Robeson (1987) showed that ozone episodes are strongly intermittent; that is, neither periodic nor chaotic but with a distinct scale of variance. Furthermore, the time scale of their duration is about the same as that for traveling weather patterns, i e . 2-5 days. The spatial pattern of an extreme episode from August of 1988, as measured by the Greater Vancouver Regional District (GVRD) air quality monitoring network, is shown in Figure (1.1.4). This pattern is typical of ozone concentrations in the L F V . The highest concentrations are seen to the east of Vancouver, usually in the regions of Port Moody and Port Coquitlam, and large values often persist well east into the municipalities of Langley and Abbotsford. This contrasts sharply with the pattern of emissions shown in Figure (1.1.3) which shows highest values where ozone values are among the lowest. 14 These figures demonstrate a common feature of ozone pollution: The highest concentrations occur 10's of kilometers downwind of the emissions sources. Such a pattern is also observed for ozone in the Los Angeles basin, and this pattern is a result of the reaction sequence summarized by Figure (1.1.1). The time lag between the appearance of NO x and VOC and the rise in ozone along with onshore winds result in the spatial displacement in precursor emissions and ozone concentrations that is evident between Figures (1.1.3) and (1.1.4). The chemical sequence also actively keeps ozone low near emissions sources. When NO is emitted into an atmosphere in which ozone is present, the ozone is consumed by driving equation (1.1) to the right. McKendry (1993) states that the ozone stays low near emissions because the destruction of ozone by this reaction is faster than the production through the sequence summarized by equation (1.2). Figure (1.1.5) shows the cumulative frequency of hourly ozone concentrations for three stations in the LFV. This diagram shows that the National Ambient Air Quality Objective (NAAQO) is exceeded 0.2% of the time. An episode like that in Figure (1.1.4) occurs about once every 5 years. This shows that ozone contamination in the LFV is not yet on the scale of more polluted centers such as, for example, Los Angeles where in 1985 alone 100 ppb was exceeded on 107 days (Elsom, 1989). Figure (1.1.5) also demonstrates another feature of ozone pollution specifically and air pollution episodes in general. The ozone concentration varies over orders of magnitude while the emissions over the same period vary, on the long term with population and over the shorter term seasonally and hebdomadally, on the order of 10-25%. This shows that while emissions and topography may predispose a particular site to pollution episodes, the dominant factor in producing a given ozone episode is the ambient meteorology. Within 15 Figure 1.1.5 Hourly ozone concentration vs. cumulative frequency (Steyn et. al. 1997) 100 90 s o z o o HI z 8 o 10 ~\—i—i—r -i 1 1—i—i—i—i—r ~i 1 1—r Cumulative frequency of hourly ozone data for Rocky Point Park, Port Moody; Kitsilano; Chllllwack Works Yard for 1984 • 1992. Port Moody / / / I I I I I I I I - I I I I I I I I I I L_ 0.06 0.1 02 0.5 I D 2.0 5.0 10 20 30 40 50 60 70 80 90 95 98 99 CUMULATIVE FREQUENCY, (%) 99.8 99.9 99.99 the LFV, during weather conditions conducive to the arcumulation of atmospheric pollutants, there is a broad spectrum of meteorological phenomena present. Though the large scale synoptic controls of an ozone episode are fairly well understood, (McKendry, 1994), there is much interest in the effects that the more localized thermally forced flows such as the sea/land breeze and upslope and valley winds within the North shore mountains may have on ozone distribution in the LFV. 16 n Air Pollution Meteorology of the Lower Fraser Valley The synoptic conditions that result in ozone episodes in the LFV are described by McKendry (1994). In general, they consist of an upper level anticyclone and a surface thermal trough. Subsidence in the anticyclonic flow leads to clear skies and an elevated inversion. The thermal trough is associated with high surface temperatures from strong insolation. This enhances the ozone reactions while the inversion limits the depth of the mixed layer and weak synoptic winds are less able to transport pollutants out of the region. In addition, the 850 mb temperature has been shown to be highly correlated with the production of ozone (Hanna, 1991). These conditions are most likely in the summer months when the Alaska Low forms over the North Pacific producing an anticyclonic flow in the coastal region. (Pottier et. al., 1997). Resulting ozone episodes are characterized by a strong 500 mb ridge over B.C., a surface thermal trough from Oregon to B.C. with temperatures of 30-35° C, and the 850 mb 20° isotherm extending over southern B.C. Thermally Forced Mesoscale Circulation Although synoptic meteorology determines the occurrence of ozone episodes, the spatial and temporal variability within an episode is greatly affected by topographic and thermally forced mesoscale flows. These phenomena are best developed under the conditions that also favor ozone production. Whiteman (1990) summarizes how the thermal gradients caused by complex topography result in various local wind systems. Most important in the coastal environment is the land/sea breeze. Oke (1987) shows how 17 advection or marine air over Vancouver leads to strong internal boundary layer (IBL) between air of marine origin and are that is adjusted to the land surface. This boundary effectively restricts dispersion of pollutants and further restricts mixed layer depth to on the order of one half what it would be over homogeneous terrain. During stable conditions, the sea breeze cycle reduces mixed layer depth in the LFV to about 600-800m (Steyn, 1980) and this suppression persists for many kilometers inland. When combined with the steep topography to the north and south (the Coast and Cascade mountains) the shallow mixed layer effectively closes the atmospheric circulation within the valley and allows high concentrations of pollutants to collect. This closed circulation is shown in Figure (1.2.1). In addition to the sea breeze, the steep topography of the LFV drives local wind systems. Whiteman (1990) provides a review of the literature of slope and valley winds and explains the diurnal cycle. In a manner roughly analogous to sea/land breezes, valley winds are driven by thermal differences between valley and plain. The process is depicted Figure 1.2.1 Closed LFV circulation under condition of strong synoptic subsidence and a well developed sea breeze. (Oke, 1987) 18 Equal Pressure 850 mb Q24 ,P res sure G r a d i e n t s - ' - - - ' • lOOOjrib _ 1003jTib P la i n Va l l e y Nighttime Figure (1.2.2). A column of air in the valley experiences a larger diurnal range of temperature than does a similar column over a fiat surface. During the day, air in the valley is warmer, and therefore less dense, causing a local thermal trough as compared to the plain. This then causes a pressure gradient toward the valley and drives winds up slope. At night the situation reverses; air in the valley is colder, and thus more dense, and winds flow down slope. There are two proposed features that produce the valley-plain thermal differences. The first is the slope of the valley floor. As heating (or cooling) is strongest at the surface, Le. the potential temperature is constant along the surface, the slope of the valley floor itself will cause the horizontal temperature, and therefore pressure, gradients to force the 19 winds. Sutton (1953) gives a solution for the up slope component of this situation and derives the Prandtl profile of wind speed that shows the familiar low level jet. The second is a result of cross-valley geometry and is known as the topographical amplification factor (TAF), McKee and O'Neal (1989) explain this process and solve for TAF adjusted pressure gradient. A unit length of a valley cross-section contains a smaller volume than does a cross-section of similar height over the plain. As the cross sections are of equal width at the top, they will each gain or lose energy through the same size area. Because the valley contains less air it will heat up faster during the day, causing upvalley flow, and will be less able to retain heat at night, resulting in down valley flow. It is important to note that this mechanism will produce wind even if there is no change in elevation along the valley floor. For example, if the valley floor is covered by a lake. It is believed that these flows have an important role determining the final destination of pollutants within the LFV. In the Los Angeles basin, upslope flows have been cited as a possible means by which elevated layers of pollution above th&mixed layer may be created. (Ulrickson and Mass, 1990; McElroy and Smith, 1993; Lu and Turco, 1994). Modeling studies have suggested the presence of similar phenomena within the LFV. Miao (1993) employed the RAMS model to study the possible trajectories of emitted pollutants and found that parcels released in Vancouver and advected inland by the sea breeze were then deflected up deep valleys to the North of the LFV, most notably Pitt Lake. Depending on the height, location and time of release, some parcels traveled up valley and escaped the LFV while others were recirculated back to lower levels with the nighttime outflow. Lidar observations along the Pitt River (Olivier et. al., 1994) confirmed the existence of a distinct daytime up-valley flow, suggestive of some topographic 20 forcing, and also showed significant backscatter from aerosols within this flow, indicative of an advecting urban plume. In addition aircraft based measurements have shown the transport of ozone up Pitt Valley and up Harrison Lake further to the east (McKendry et. al, 1998), and Banta et. al, (1997) have documented nocturnal flows out of Pitt Valley. There is evidence that concentrations within these valleys during episodes are among the highest in the region (McKendry, 1997)_0'Kane (1997) has shown that the maximum isopleth in Figure (1.1.4) likely extends up over Pitt Lake. Boundary Layer Structure Within the LFV The depth, structure and stability of the boundary layer is of central importance in air quality studies as these characteristics define the vertical extent to which pollutants emitted into the lower troposphere may be dispersed. Figure (1.2.3) shows the diurnal evolution of the planetary boundary layer during anticyclonic conditions over homogeneous terrain. During the day, surface heating drives convective turbulence keeping the layer well mixed. Vertical gradients of atmospheric quantities in this layer will tend to zero. At night, the surface cools radiatively and a stable layer with suppressed turbulence is formed. Above, with less coolings turbulence decays more slowly. This residual layer tends to stay fairly well mixed throughout the night. The difference in state causes a MecGurjhng' of these two layers, with energy and mass exchanges between them greatly reduced. As a result of this decoupling, and as most sources and sinks of pollution are near the ground, pollutants, including ozone, within the residual layer will tend to persist overnight. 21 Figure 1.2.3 Boundary layer evolution under anticyclonic conditions (From Stull, 1988) 2000 1000 -J.—H - ^.irs^ y^A\ En t rapment Zone ' a * C l o u d TTaye?r^N—j" "*3 zmz r r — — i' ^ •}-iuii^fV-1<'t-*'-'":^ .Convect ive , « Mlxod Layer .Surface Lnycr Free Atmosphere Capping Inversion : Entralnment Zone Residual Layer Stable (Nocturna l ) Boundary L a y e r . S4 SS S6 Within the LFV, boundary layer variations are similar to those described in Figure (1.2.3) except that sea breeze advection will tend to reduce the height of the capping inversion (Steyn, 1980). In addition, the stable stratification of the nocturnal inversion is intensified by katabatic downvalley drainage flows. This intensification results in an even stronger decoupling of the stable drainage flow from the residual layer above with mass and energy within the residual layer thus being able to travel long distances with very little turbulent losses (Blumen,1990). However, even under strongly stable conditions, sporadic bursts of turbulence do occur within the nocturnal layer, allowing some iriixing with the residual layer, and therefore allowing some depletion of residual layer stored pollution. Using a regression analysis under similar conditions within a valley in the Swiss Alps, Neu (1995) estimates that sporadic nocturnal turbulence bringing ozone to ground and dark 22 chemical losses result in residual layer ozone being depleted by about 20-50%. Or in other words, about 50-80% of residual layer ozone is conserved overnight. Over mountain ridges, however, boundary layer depth is increased compared to the plain (Stull, 1988). Differential heating over slopes compared to plains creates amplified convective turbulence and drives a deeper mixed layer. In the Los Angeles basin, mixed layer depths at ridge top may approach 2-3 km (Ulrickson and Mass, 1990; Wakimoto and McElroy, 1986). This is important in that pollutants trapped in upslope flows and dispersed into the deepened mixed layer may reach much greater heights above ground than they would in simpler terrain. Down-mixing of Residual Layer Pollutants. During morning, the rising entrainment zone, (see Figure (1.2 3)), erodes the nocturnal inversion and the overnight residual layer. Any pollutants within the residual layer will then be re-mixed throughout the growing mixed layer. This of course allows for the possibility of day to day retention of pollutants within the boundary layer and is perhaps a factor in the strong daily temporal autocorrelation seen within ozone episodes. Neu et. al, (1994) used a transilient turbulence model (Stull, 1988,1993) to investigate the down-mixing of residual ozone within a valley in the Swiss Middlelands. A transilient model is deemed more appropriate than a local closure scheme because studies of morning breakup in complex terrain have shown that a wide spectrum of various sized eddies interact to erode the nocturnal inversion (Sakiyama, 1990; Whiteman, 1990). Also, a transilient scheme is better able to handle transfers of mass and energy within a well 23 Figure 1.2.4 Development of modeled vs. observed near surface ozone for 5 and 6 July at Ibach Switzerland. (From Neu et. al, 1994) 110 100 7 . 5 8.0 8.5 9.0 9.5 10.0 10.5 1 i .0 11.5 12.0 12.5 13.0 13.5 14.0 14.5 15.0 15.5 16.0 16.5 17.0 17.5 Time (CEST) 5/7/91 measured - - ~ - 6/7/91 measured 5/7/91 calculated 6/7/91 calculated mixed layer where all vertical gradients are tending toward zero (Stull, 1993). Figure (1.2.4) shows the ground level ozone concentration as predicted by the transilient model vs. the actual measured concentration for two case days. The authors claim the figure shows that on high ozone days, nearly all of the rise in surface ozone during the morning hours comes from down-mixing of residual ozone (Le. before local chemical production becomes significant) and similarly 50-60% of the subsequent daytime ozone maximum is accounted for from re-Dirxing of 'stored' ozone. Kleinman et. al. (1994) found similar results using a chemical model for a rural site in the Southeastern U.S. Elevated Ozone Layers 24 In addition to the residual layer, atmospheric processes within the complex terrain of the LFV (i.e. slope flows) allow for the possible generation of elevated polluted layers above the mixed layer shown in Figure (1.2.3). Elevated layers are a well documented feature of Los Angeles smog episodes (Lu and Turco, 1994; McElroy and Smith, 1993; Ulrickson and Mass, 1990), have recently been identified over the southern coast of the Iberian peninsula (Millan et. al., 1997) and have also been observed in the LFV (Olivier et al, 1994; McKendry et. al., 1997). These layers are seen from just above the mixed layer up to around 3000m a.g.l. As the oxidants within these layers are separated from surface based oxides of nitrogen sources, which tend to scavenge ozone, the ozone concentrations in these layers are often higher than those observed in the mixed layer (McElroy and Smith, 1993). Consequently, they form a 'reservoir' for ozone and have, similar to residual layer contaminants, the potential to greatly affect surface concentrations. Ulrickson and Mass (1990) demonstrate how elevated layers that are intercepted by the mixed layer may fumigate their pollutants to lower levels. In addition, depending on local wind kinematics, transport of pollutants within these layers may impact air quality significant at distances away from the source regions (Millan et. al., 1997; Moore et. al, 1991; Blumenthal et. al., 1978). Elevated layers over Los Angeles demonstrate very distinct vertical structuring. This structuring has been linked to the varying processes that may be responsible for the genesis of these layers (McElroy and Smith, 1993), and these processes are themselves a result of the strongly coupled dynamics of the local thermally forced wind systems that 25 occur within complex terrain during ambient conditions that are conducive to producing ozone episodes. McKendry et. al., (1997)^  summarizes three mechanisms for the generation of elevated layers that may important within the LFV. The first is pollutants being imbedded into the marine inversion layer by strongly buoyant convective thermals (Edinger, 1963). Under conditions of strong surface heating, occasional thermals will experience enough buoyancy to 'overshoot' the capping inversion and escape the mixed layer. This mechanism is amplified in complex terrain by intensified differential heating over the valley slopes and by the TAF, as outlined previously with respect to slope and valley winds. Second is the injection of pollutants into the free atmosphere by upslope flows. (Ulrickson and Mass, 1990; Blumenthal et. al, 1978). This so called "chimney effect" allows pollutants to travel upslope and be dispersed into the deeper ridge top boundary layer, where they may then be advected back over the basin into the free atmosphere. The third is undercutting of polluted air by the passage of the sea breeze front. The sea breeze is analogous to a cold front and as it advances the warmer, land adjusted, more polluted air is forced up and over the cooler, cleaner marine flow (Blumenthal, 1978; McElroy and Smith, 1993). It is important to note that these processes do not occur singly; they are closely coupled, as are the local winds that result in their genesis. Lu and Turco (1994), in a schematic representation of an idealized coastline, show these processes as developing in three stages as represented in Figure (1.2.5). In the first frame, the atmosphere is undisturbed and shows three layers: A marine boundary layer, an inversion layer on top and the free atmosphere at higher levels. The 26 Figure 1.2.5 Genesis of elevated layers in a coupled sea breeze/mountain wind domain. Elevated layers are generated in three stages. (From Lu and Turco, 1994) Early Morning Free atmosphere Inversion layer Mar ine layer Noon, Mixed Layer Growth Free atmosphere Inversion layer Mar ine layer =i> Afternoon, Sea-breeze Inland Intrusion Free atmosphere Inversion layer Mar ine layer Evening. Mixed Layer Siabilizition Free atmosphere inversion layer first stage is mixed layer growth. As the sun rises, convective thermals are produced by surface heating. The mixed layer grows by entraining warm air from the inversion layer. The sea breeze and mountain winds begin to form but are still quite weak. The next stage is sea breeze intrusion. As thermal differences between the sea and land increase, the sea breeze builds and pushes inland. Strong heating continues to drive turbulence in the mixed layer with some thermals penetrating through the entrainment zone into the inversion layer. Upslope "chimney effect" winds are well developed, venting pollutants into the free atmosphere. Some polluted surface air is lofted and advected back 27 into the stable layer by the sea breeze circulation. Elevated layers of pollution are formed in the inversion layer and in the free atmosphere at ridge top. Last is the mixed layer stabilization. As the sun sets, turbulence in the mixed layer is suppressed and thermal gradients are weakened. Warm air in the mixed layer moves up the mountain slopes and is trapped in the inversion layer. When the downslope winds form pollutants within the inversion are 'decoupled' from the surface. Elevated layers form over the entire plain. Figure (1.2.6) shows a vertical ozone sounding from McElroy and Smith (1993). Three distinct elevated features are present. The authors claim that each peak in ozone can be related to one of the processes above. Beside is a schematic of the coupled sea breeze mountain slope winds. The circulation in the schematic corresponds roughly to the layers visible in the sounding. The base of the elevated ozone layer at 500m is a result of marine air penetration. Below is relatively clean marine air. The middle peak is deemed to be the result of convective elements penetrating above the top of the mixed layer, which for this locale was about 750m, into the inversion layer above. The upper layer, which shows an ozone maximum at about 1250m, is associated with pollutants being vented by upslope flows and then advected over the basin. This judgment is made based on the winds at this height being in contrast with the onshore winds seen in the layers below. Based on this analysis, it would appear that layers caused by marine penetration or convective debris are more likely to exist at levels that might be intersected by a deepening boundary layer and thus be of more concern to surface air quality while slope flows might indeed allow ozone to escape the boundary layer and reach a height where they will not be fumigated to lower levels. 28 Figure 1.2.6 Vertical ozone sounding showing elevated layers over Los Angeles July 13, 1987 along with an idealized view of the coupled sea breeze and mountain circulation. (From McElroy and Smith, 1993) 2000.0" 1500.0 • to 2 5 1 0 0 0 . 0 - ^ 500.0 —J 0 . 0 . Deg. C -10 PPB Ozone Temperature 10 100 I" 20 200 30 300 40 400 50 500 Lastly, subsidence may bring vented layers back down to lower levels where they may again become significant in terms of air quality, Millan et al (1997). This process is important in producing the 'stacked layer system' with the oldest layers closest to the surface that is observed in the vertical ozone distribution over the Spanish Mediterranean coast. Subsidence may occur as a result of synoptic anticyclonic circulation and it may be amplified by the sea breeze mesoscale circulation and by near shore subsidence acting as a compensatory response to a local surface thermal trough. Free Tropospheric Ozone Ozone pollution is generally considered to be a regional scale phenomena, (unlike, for example, acid rain which can occur thousands of km's from emission sources). This is because most of the processes involved have relatively short time scales. Even a strong 29 anticyclonic ridge will seldom persist any longer than a week at most and the time scale for the chemical destruction of boundary layer ozone is no more than 2-5 days (Seinfeld, 1989). When combined with weak synoptic winds, there is little chance for lower level ozone episodes to persist for any more than distances on the order of 100 km downwind. However, ozone within the free troposphere can persist for much longer. The primary photochemical sink for boundary layer ozone involves the reaction of 0(!Z))with water vapor, equation (1.5). Thus the lifetime of tropospheric ozone is most dependent on the amount of water vapor and ultraviolet radiation. As water vapor is mOst abundant in the planetary boundary layer, the rate of destruction for ozone that reaches higher levels will be greatly increased. Fishman et. al. (1991) claim that the lifetime for ozone in the free troposphere is as high as 90 days. When combined with higher upper level winds, the potential exists for ozone to persist for distances on the order of 1000's of km. Ozone is present in the free troposphere at background levels of 10-50 ppbv and increases with height at a rate of about of 5-7 ppb/km up to the lower stratosphere where concentrations quickly rise above lOOppbv (Browell et. al., 1994). Instances of enhanced ozone within the free troposphere have two sources: stratospheric intrusions and pollutants escaping the boundary layer. Shapiro (1980) summarizes the various mechanisms by which air and its chemical constituents may be transported across the tropopause. Most important in the consideration of the downward entrainment of stratospheric air to lower heights is tropospheric 'folding'. This process occurs in association with extratropical cyclones as a result of the discrepancy in tropopause height on opposite sides of the jet stream. 30 Compared with the troposphere, stratospheric air is very dry and relatively devoid of atmospheric aerosols. Ozone rich air of recent stratospheric origin exhibits frost points < -60°C and H 2 0 mixing ratios an order of magnitude lower than the free troposphere. Water and aerosol content are thus very useful as tracers of origin (Broweli et. al., 1992). In addition, positive tropospheric ozone perturbations from stratospheric sources are both strongly correlated with higher potential temperature and potential vorticity. Ozone and its precursors that originate at or near the surface may escape into the free troposphere by the mechanisms outlined earlier. Millan et. al., (1997) postulate that elevated ozone layers generated over the Iberian coast may be transported toward the Intertropical Convergence Zone over Northern Africa in summer and be pumped directly into the upper troposphere. Boundary layer air that is high in ozone also carries other Table 1.2.1 Classification of free tropospheric ozone features by air mass type (Broweli, et. al., 1992) Percentage of Troposphere Containing Air Mass Types Altitude Range Tropospheric Coverage, Air Mass Type 0-2 km . 2-4 km 4-6 km 6-8 km 8-10 km 10-12 km 0-12 km Background (free troposphere) 17 40 Boundary layer 36 8 Plumes 26 24 Stratospheric intrusion 8 24 Low ozone air 4 1 Convective outflow 9 3 Low tropopause 0 0 Number of Flight Legs 33 30 pollutants that are only produced near the surface that may be used as tracer of air mass origin Most significant are aerosols and CO. Each of these has its only sources within the boundary layer and thus when matched with high ozone features gives conclusive evidence as to origin (Fishman et. al., 1991). This pollution may or may not be anthropogenic. Biomass burning (forest fires) releases significant amounts of precursors to ozone. In 43 40 34 23 33 0 0 0 0 7 2 2 2 I 10 45 41 31 51 33 9 13 26 19 12 1 4 5 3 4 0 0 2 3 I 15 27 , 25 15 31 addition, measuring the non-methane hydrocarbons to determine how many reactive species remain in the air mass can give a measure as to the relative age of the air mass. Figure 1.2.7 Tropospheric ozone from TOMS and SAGE Data 1979-1987. Tropospheric residuals >40 DU have been shaded. (Fishman et. al., 1990) Tropospheric Residual ( W E S T ) L O N G I T U D E ( E a S T ) ( W E S T ) L O N G I T U D E i*-^) 32 Browell et. al, (1994) used airborne lidar to examine tropospheric ozone over Eastern Canada. They found evidence of both stratospheric intrusions and boundary layer sources. They attempted to classify the observed air masses as to origin. (See Table 1.2.1) As one might expect, the source region is highly correlated with air mass height. Higher levels (6-12km) are more likely to be affected by stratospheric ozone while at the lower levels (0-4km) boundary layer sources are dominant, most notably plumes from biomass burning. Using a back trajectory analysis, the authors were able to identify elevated layers of ozone up to 20 ppb above background that they associated with forest fires in Alaska. Studies in other areas, (i.e. the Amazon basin, Fishman, et. al, 1990; Browell et. al, 1988) have found a similar degree of tropospheric ozone enhancement from biomass burning. The detection of an Alaskan forest fire plume in Eastern Canada demonstrates the scale on which ozone pollution may affect the free troposphere. Data from the SAGE (Stratospheric Aerosol and Gas Experiment) and TOMS (Total Ozone Mapping System) satellites may be used to estimate the extent of this enhancement. TOMS measures the total column ozone, while SAGE provides profiles of stratospheric gases. Subtracting the integrated SAGE profile from the TOMS total column for instances where the data are contemporaneous gives a measure of the tropospheric component. Fishman et. al, (1990) used more than 22,000 SAGE profiles and 9 years of TOMS data to examine the global climatology of tropospheric ozone. The seasonally averaged results are presented in Figure (1.2.7). They show maxima downwind of the eastern U.S., Europe and Eastern Asia in the Northern hemisphere summer of about 40 DU. This corresponds roughly to 33 about 60 ppb and appears to correlate well with areas of increased emissions of precursors from industrialization. The figure also shows a maximum off the coast of West Africa in Sep-Nov. In a follow up study the authors take a closer look at this area using a slightly longer data set Figure 1.2.8 Same as 1.12 for 1979-1989 (Fishman et al 1991) January-Feo-uary Ju ly -Augus t May-June 180 -12C -60 0 60 120 I V M Lcctuoe Ess. 18C I8C 18C -120 -60 0 6C Long i tude September -October -180 -120 -60 0 60 V9S Long i tude November -December 0 60 Long i tude 180 12C 18C 80 1 5 20 25 30 3? 40 45 50 Dobson un i ts 34 (Fishman et. al., 1991). The results are shown in Figure (1.2.8). Again the polluted areas of the Northern hemisphere are present. The spatial continuity of the ozone maxima visible in each of these, which is not resolvable over long distances using single points and wind speeds from balloon-borne sondes, clearly shows that the pollution clouds, particularly in the southern hemisphere, must originate within the troposphere and are not a result of stratospheric intrusions. In the case of the West African peak, balloon measurements of aerosols and CO in the area confirm a biomass burning source on the African continent. 35 Chapter 2 Methodology I Data Sources Representative sampling has long been problematic to studies of the atmosphere. Atmospheric quantities vary in three spatial dimensions and with time. When the area of interest is the size of the LFV and exhibits the complexity of topography and local wind system as does the LFV, representative sampling becomes nearly impossible or at least very expensive. As a result there has been a paucity of data pertaining to the three-dimensional ozone distribution within the region. Recently, a multiple-agency agency field program, Pacific '93, was undertaken in July/August to improve understanding of ozone episode meteorology (Steyn et. al, 1997). Most of the data used here were obtained as part of this program. In addition, data from ongoing complementary programs run by the GVRD (Greater Vancouver Regional District), Atmospheric Environment Service (AES) and U.B.C. Department of Geography are used. Because of the difficulty of representative atmospheric sampling and the complexity of elucidating the three dimensional processes that may be affecting the vertical distribution and flux of ozone, this study draws from a wide variety of data sources. These sources include: 1) The GVRD near-surface air quality monitoring network 36 2) AIR tethered balloon ozone and meteorology sondes 3) AES free ascent balloon ozone and meteorology sondes 4) Airborne and ground based LIDAR measurements 5) Aircraft borne ozone measurement In addition, AES archival synoptic charts and Vancouver airport climatic records are utilized. GVRD Surface Ozone Measurements The GVRD measures ground level ozone as part of its ongoing air quality monitoring program. Ozone concentrations are recorded as 5 minute averages and later reduced to one hourly averages for presentation here. The measuring sites are labeled T l through T23 in Figure (2.1.1). In addition to ozone, the GVRD also monitors carbon monoxide (CO), nitrogen dioxide (N02), nitric oxide (NO), sulphur dioxide (S02) and total suspended particles. Munn (1993) provides details of the operations and instrumentation of the network. AIR Tethersondes Profiles of ozone and meteorology obtained by tethered balloons were recorded during Pacific 93 and in ongoing field programs by the U B C . Department of Geography. Flights were made from Harris Rd. and Little Goose Island in the area of Pitt Lake in the 37 Figure 2 . 1 . 1 Map of the L F V showing scope of Pacific 93 field program (Steyn, et. al, 1997). Harris Rd., Little Goose Island and Langley measurement sites are labeled. 38 summer of 1993 and 1994 and from a site in Greater Vancouver during 1997. Soundings of ozone, temperature, pressure, wind vectors and relative humidity and flight time were obtained using an Atmospheric Instrumentation Research Inc. (AIR) tethersonde (TS-3A-SPH) with ozonesonde (OZ-3AT) carried by a 5m3 helium filled balloon. Ascent and descent were controlled by an electric winch with a 1000m tether line. Surface to surface time for a typical flight is about 45 minutes with a sampling interval of around 10 seconds. The data collected are then reduced on spreadsheet to give profiles of ozone concentration, potential temperature, wind speed and direction and water vapor mixing ratio as a function of height a.g.l. along with the decimal time of each measurement. During approximately half of the Harris Rd. flights during Pacific '93 a custom built, miniaturized Luminol chernUuminescence (Scintrex/Unisearch Inc.) NO2 sensor was flown in parallel with the ozone and meteorology sondes. One feature present in the AIR profiles is a distinct hysterisis between ascent and descent ozone values, with descents typically on the order of 10 ppb greater at ground level. An example is shown in Figure (2.1.2). The AIR ozonesonde uses a gas bubbler system within a potassium iodide solution to measure a voltage proportional to the concentration of ozone in the bubbled air. The manual for the ozonesonde (AIR, 1992) recommends that it be operated for at least 20 minutes on the ground to allow the solution time to equilibrate with the ambient air before a flight is made. It is possible that in the field this requirement was not always met and that the sonde was still adjusting when the balloon left the ground. As the time taken during the ascent should guarantee that the sonde has become properly adjusted, for purposes of modeling, descents are used whenever possible. 39 Figure 2.1.2 Ascent and descent ADR. ozonesonde profiles showing distinct hysterisis. Descent is 'skewed' downward and to higher ozone values than the ascent. August 6,1993 Harris Rd. 09:51 PST ascent and 10:06 PST descent ozone Ozone Concentration (ppb) Free Ascent Balloon Sondes During Pacific 93 and in the summer of 1992, the Atmospheric Environment Service (AES) released free ascent balloons at two sites within the LFV. Profiles were obtained from Clearbrook in 1992 and from Langley in 1993. The measuring system was a PC based VIZW-9000 Meteorological Processing System with a VIZ model 1543-523 Mark II meteorological sonde equipped with an ECC-5 A ozonesonde and interface card for ozone measurements. Meteorology and ozone were sampled at 1 second intervals by the sonde rising at 150m-200m per minute. Balloons were released on high ozone days at 40 5 a.m., 11 a.m., 2 p.m., 5 p.m., and 8 p.m. Pacific daylight time and tracked by aLORAN receiver (Evans et. al, 1993). The profiles give ozone concentration, temperature, relative humidity and wind speed and direction from the surface up to 12,000m a.g.l. Lidar Measurements Both surface and aircraft based lidar measurements were made during Pacific 93. A ground based scanning Doppler lidar instrument developed by the U.S. National Oceanic and Atmospheric Administration Wave Propagation Lab (NOAAAVPL) was situated at the Pitt Meadows airport. The NOAAAVPL lidar emits infrared light from a CO2 laser at 10.59um. A small amount of the emitted light is detected by the lidar as it is backscatterred by aerosols. The instrument records both the backscattered intensity and the Doppler shift of the return signal allowing a measurement of both the pollution turbidity and the along beam aerosol velocity (Olivier et. al, 1994). The lidar scans in both a Range Height Indicator (RHI) and Plan Position Indicator (PPI) modes and has a maximum operating range of 30 km. The narrow beam width of 90 ur with no ground clutter or side lobes allows excellent scanning into complex terrain. In addition, a 1.064 urn downward pointing Nd:YAG lidar was operated from aboard the National Research Council of Canada (NRCC) Convair 580 aircraft. (Hoff et. al, 1997). The instrument was flown at 4.2-4.8 km height along the lines labeled L1-L14 in Figure (2.1.1). The system scans downward at 8.2° of nadir and has a good sensitivity for particles of size on the order of 1p.m. Again, measurements of the backscatterred 41 intensity and Doppler shift allow determination of the distribution and movements of lower tropospheric aerosols. Aircraft Based Measurements Airborne measurements of ozone concentration along various flight lines within the LFV were made during the summer months of 1994-95 by the U.B.C. Department of Geography using a fast response chemiluminescent ozone sonde aboard a lightweight aircraft. The system employs a GFAS (Gesellschaft Fur Anglewandte Systemtechnik) OS-42 B-2 ozonesonde, originally intended for use with stratospheric balloons, mounted on a Cessna 172 (McKendry et. al., 1998). Outside air is tunneled into the sensor located inside the cockpit via Teflon tubing extended through the open cockpit window. The inlet of the tube is deployed into the propwash on the opposite side of the plane from the engine exhaust. This configuration allows for measurements of both horizontal transects and, by flying an ascending or descending spiral, vertical profiles of ozone concentration. The time constant for the OS-B-2 is approximately one second. With a horizontal airspeed of ~50 m/s the system can resolve ozone fluctuations on wavelengths of the order of 100m. For vertical profiling, a vertical velocity of ~3 m/s allows resolution to about 6m. Figure (2.1.3) shows the flight lines followed by the aircraft on a typical flight over the Fraser Valley. In addition, the Convair carried a TECO 49 UV Photometric Absorption ozonesonde and a Particle Measuring Systems (Boulder, CO) Active Scattering Aerosol Sizing Probe. These instruments provide profiles of aerosol and ozone concentration from the surface up to 5000m a.s.l. n Transilient Turbulence 43 Transilient turbulence theory is a first order closure scheme that is based on a quasi-advective, non-local approach to atmospheric turbulence as opposed to a quasi-diflusive local closure scheme such as K- theory. The advective approach means that mixing can be modeled between an array of points and on a variety of scales and not just through transfer between adjacent points (Stull, 1984, 1988, 1993; Stull and Driedonks, 1987). Under this framework, unknowns at one point in space are parametrized using known quantities from many other points in the vertical dimension, hence the term non-local. This allows for the inclusion of a wide spectrum of eddies and can thus model mixing in problems where a diffusive approach breaks down, such as in zero-gradient or counter gradient situations as might be found in a convective mixed layer. As a result the method is extremely useful for modeling pollutant entrainment and/or dispersion. Advective vs. Diffusive Nature of Turbulence Turbulence and molecular diffusion both cause dispersion of atmospheric quantities. However, they are different physical processes. Figure (2.2.1) shows how the advective nature of turbulence becomes evident when viewed on an appropriate scale. The motions on the left appear to be inter-grid diffusion. However, when the grid size is reduced, the transfers between the grids, i.e. advection, become apparent. Advection is thus effected by turbulent eddies as well as by the mean flow. Figure (2.2.2) shows several atmospheric examples of large eddies, or 'coherent structures', that result in advective 44 Figure 2.2.1 The Advective Nature of Turbulence as Shown by varying Grid Sizes.(From Stull, 1993) The same turbulent motions that might appear diffusive-like for large grid boxes (a) show their advective nature when the grid resolution is made finer (b). (a) (b) Figure 2.2.2 Examples of Atmospheric Eddy Structures. (From Stull, 1993) (a) looping smoke plumes; (b) swirls of leaves or snowflakes; (c) aerosol-laden thermals; and (d) cellular cross sections. zl (a) (O turbulent transport. An appropriate parametrization for turbulent transfer must be consistent with this advective nature. 45 Turbulent Energy Spectrum Figure (2.2.3) shows the turbulent energy spectrum for a typical boundary layer. This shows the wide range of eddies of varying sizes that are present in a turbulent atmosphere. Most of the spectral energy is in larger scale eddies and they make the greatest contributions to the vertical turbulent transport and flux. Their contributions to Figure 2.2.3 The Turbulent Kinetic Energy Srjectrum. (From Stull, 1993) In (wavenumber) the total flux are also more precisely characterized as advective (Stull, 1993). The largest eddies are usually anisotropic and have a size comparable to the boundary layer scale. For example, convective thermals have most of their energy in the vertical, because of buoyancy, and are of about the same scale as the depth of the mixed layer. Any method that involves the description of turbulence in terms of mixing along some characteristic length or depth, even if the characteristic length is allowed to change with height, will misrepresent this spectral nature. Local vs. Non-Local Layer Stability 46 Atmospheric stability is commonly defined in terms of the local lapse rate. The atmosphere is classified as stable, neutral or unstable depending on whether the vertical gradient of potential temperature, -ff , is greater than, equaL or less than zero, respectively. However, if by 'unstable' one means 'turbulent', then the local lapse rate is an inaccurate measure of the state of the atmosphere. For example, in a convective mixed layer, -ff goes to zero, implying a neutral atmosphere, but a weU-defined heat flux profile Figure 2.2.4 Atmospheric Stability Determined by Non-Local Conditions. Static stability is based on non-local parcel movement, not on local lapse rates. (From Stull, 1993) t . parcel movement subadiabatic 1 turbulent adiabatic statically unstable 44 4 4 I f superadiabatic • subadiabatic is present and the layer is turbulent. The first order closure for turbulent heat flux may be expressed as \M = K(dO/dz) (2.2.1) 47 If the potential temperature gradient vanishes but heat fluxes are still present, K becomes undefined. This is because the turbulent state of a point in the mixed layer is determined by the layer profile as a whole. That is, the local stability at one point is dependent on the properties of the atmosphere at other, non-local points within the layer. For this reason, one should not use a stability quantifier such as 'stable' or 'unstable' for the local lapse rate. The terms 'super-adiabatic', 'adiabatic' or 'sub-adiabatic' are more appropriate for lapse rate that are negative, zero, or positive, respectively. Figure (2.2.4) demonstrates the nature of non-local layer stability. The static stability of the adiabatic region is influenced by the superadiabatic layer below it. A parcel experiencing positive buoyancy in the super-adaiabtic layer will continue to rise through the adiabatic region even though the local lapse rate becomes neutral. This is because the stability of this parcel at some point in the mixed layer is determined by the non-local conditions in the super-adiabatic region where the parcel, or thermal, was generated. Again, an accurate parametrization for mixed layer turbulence must include non-local effects. Stull (1984) provides a brief history of some of the approaches to describing atmospheric turbulence (i.e. K-theory, similarity theory, higher order closure) and outlines how each fails to deal with either the advective or spectral nature of turbulence. Transilience theory is developed as a. more appropriate alternative. Two forms of transilient turbulence theory have evolved: one in discrete form for numerical work and another in analytical integral form for theoretical work. For purposes of brevity, only the discrete case is presented here. For the details of the continuous form the reader is referred to Stull, (1984, 1993). 48 Transilient Framework The basic tenet behind a non-local turbulence scheme is that any point in the vertical is influenced not only by transfers between adjacent points, but by transfers with all other points in the vertical domain. The word transilient is latin for "to jump over". If one assumes a vertical grid of N boxes, see Figure (2.2.5), then for a given grid box transfers will exist between that box and all others in the column, i.e. the quantities within one box may "jump" to the others by means of turbulent advection. The arrows in Figure (2.2.5a) show all the possible exchanges between boxes for a grid of size N=5. The dark arrow highlights the movement of air from grid 5 to grid 2. Figure 2.2.5 Example of Transilient Exchanges and the Corresponding Matrix for a N=5 G r i d A turbulent field consisting of a superposition of mixing between all possible pairs of of N grid cells (a) is described by an N x N transilient matrix (b). Mixing between source cell 5 and destination cell 2 is highlighted in both (a) and (b). (From Stull, 1988) (b) 5 4 c 51 c 52 C 53 C 54 s55 C 41 C 42 C 43 jg44 C 45 C 31 C 32 s33 C 34 c 35 C 21 J*22 °23 C 24 C 25 M11 c 12 C 13 C 14 C 15 49 If we consider that each transfer relates a percentage of air traveling from the source (i) box to the destination (j) box, then we can define an NxN matrix of 'mixing coefficients', dj, that describes this mixing. This is the transilient matrix. Figure (2.2.5b) shows the 5x5 matrix of all transfers for the N=5 grid. The matrix element corresponding to the bold arrow is highlighted. The diagonal elements, i.e. c„ = cy, represent the amount that stays in a grid box and may be thought of as the intergrid mixing or diffusivity. If we let ^ be the amount of some atmospheric quantity, (tracer concentration, heat momentum etc.), in a certain box (i), the change in that box in a certain time, At, is then the sum of all the contributions from the other grid boxes (j) as represented by ?(' + AO = Zc ,(f,A/K(0 (2-2-2) y=i where the transilient matrix is determined by the state of the atmosphere at time t, and the interval, At over which the mixing is occurring. A wide variety of physical processes may be modeled with the use of transilient matrices. Figure (2.2.6) gives idealized examples of several atmospheric situations and a corresponding example form of the transilient matrix form. Conversely, the coefficients of the matrix may be interpreted as to the types of mixing they imply. Figure (2.2.7) shows how the locations of the coefficients within the matrix are related to the type of mixing present. For example, rapid downward mixing is indicated by large mixing coefficients in the bottom right hand corner of the transilient matrix. Note that this diagram has been 50 inverted so that the indices along the vertical axis are analogous to the height above ground. The mixing implied by the transilient matrix can not violate any expression of the law of conservation of mass. The first is the "conservation of air mass" (Stull, 1988). Since the total amount of air in the reference box does not change with time, as much air must leave the box during At as enters. If C y is the fraction of air entering box (i) from box (j), then this requires that the sum over j of all mixing fractions must be unity. 1 = SS (2.2.3a) .7=1 Also the 'conservation of tracer amount' requires that 1 = I>, (2.2.3b) i=l This ensures that the amount of tracer originally in box (j) is conserved as mixes out of (j) and into the other boxes (i). In addition, the elements of Cy must be non-negative for randomness and entropy to be increasing. Negative elements will still give a solution to equation (2.2.2), but it would be non-physical. That is, it would imply un-mixing. Each element of the matrix must be less than or equal to 1.0. Again, as cy represents the fraction of air arriving at destination from some other source, it is impossible to have more than 100% of the 51 Figure 2.2.6 Examples of idealized transilient matrices showing non-local mixing possibilities. The arrows are not physical eddies, but they represent the net mixing effect of many real eddies acting in 3-D space. From Stull (1988). No Mixing OridBoxn TrafWJMflt Matrix = 1 2 1 4 4 1*1 % 0 0 0 a 2 0 \ 0 0 2 3 0 \ 0 1=1 4 0 0 Small Eddy Mixing (K-theory) Orid Bora* TraraMwit IMrtx HH o o H^Ji^H o o o H% Complete) Mixing Orld Box** TmMNhHit Matrix XYA^YA YA YA H% YA YA YA YAYAYA YA YA YAYA Patchy Turbulence H. * o o' yi%o 0 0 0 1 0 0 0 0 1:: Detraining Updraft Core c C! 0 1 0 0 H 0 % 0 H o H H H 0 0 % Eddies Triggered by One Layer YA YAYAYA YA YA. 0 0 YA 0 H 0 YA 0 0 % Figure 2.2.7 Interpretation of Transilient Matrices. Mixing processes can be interpreted from the relative location of an element within the matrix. (From Stull, 1988) This figure has been inverted compared to those in Fig. 2.2.6 so that the bottom of the matrix corresponds the bottom of the transilient model mixing column. Bottom Source Top c o ca c w a Bottom Top Rapid Upward Mixing Rapid Downward Mixing destination air arriving from anyplace. flux Determination Under the transilient frame, vertical fluxes are defined at and occur across the boundaries between grid boxes, rather than in the middle of the boxes. Atmospheric 52 quantities pass through these boundaries as they are transferred from source to destination during the transihent mixing. The calculation of the flux of £ through some level k during a time step At is then a matter of adding up all the contributions to the flux from all the eddies that cross this level. From Stull and Dreidonks (1987) this can be written as The flux across any level k should depend only on the eddies that cross that level. This implies that the source and destination grid boxes of these eddies are below and above k, respectively, so equation (2.2.4a) can be rewritten as Calculation of Transilient Matrix Coefficients. Any approach to turbulence closure consists of two parts: the conceptual framework and the mathematical parametrization. As outlined, the conceptual framework of transilient theory is the quasi-advective process. We must now define a method for mathematically determining the coefficients of the transilient matrix. This may be achieved either by a 'a priori' method that utilizes assumptions or knowledge about the turbulent spectrum or the frequency distribution of turbulence, or by a 'responsive' approach that f ( * ) = W ) 4 l & , ( ^ § ) 1=1 7=1 (2.2.4a) 0 1 i=i /=*+il E. - c i . (2.2.4b) 53 allows the matrix to change in response to changes in the mean flow. Two responsive parametrizations are given here. For an example of an a priori method see Stull (1993). Let us define a matrix Ay that is the potential for mixing from (j) to (i). This potential is then dependent on the amount of instability in the flow, which is assumed to govern the amount of turbulent mixing that results. The coefficients of the transilient matrix are obtained from the mixing potentials by c ' = i4T f o r i i j ( 2 ' 2 5 ) where WAW^ is the matrix norm of Ay defined by ML = max, A , ] (2.2.6) That is, \\A\\X is the maximum sum over any row of Ay The diagonal elements are found using the requirement for conservation of mass c,=l-2>, (2.2.7) The mixing potentials are parametrized as a function of the dynamic instabihty of the mean flow. Stull (1993) gives two responsive parametrizations: one based on the turbulent energy equation and the other based on a non-local approximation to the Richardson number. Stull and Dreidonks (1987) show how the turbulent kinetic energy equation may be used to find the mixing potentials from 54 <Atf)} + (AF)}-V L g DAt fori^j (2.2.8) where U, V are the Cartesian velocity components, 6X is the virtual potential temperature and Az is the grid spacing. This equation is derived in Appendix A. The symbol Ay represents a non-local difference; For example, Aa0 — Oj - 0j. D, T„ and Rc are scaling parameters relating to the dissipation, time scale of turbulence and the critical Richardson number respectively. These Y y value are the first guesses to the elements of the Ay matrix. Stull (1988) recommends that the mixing potentials for eddies of a certain size be no less than for eddies of larger sizes. That is, the mixing between, say, 100m and 200m above ground be at least as strong as the mixing between 100m and 500m. Thus the elements of Y must increase monotonically from the upper right and lower left corners toward the main diagonal. The diagonal elements Y„ = Yjj are set equal the largest immediate element on the same row plus a Y ^ value that represents the potential for subgrid mixing within one box. YM=um(yt„,Yt„) + Y« (2.2.9) Then the mixing potential matrix Ay is A i j = A j i =Y i j (2.2.10) And the transilient matrix Cy is obtained from equations (2.2.5) to (2.2.7). The recommended values of the parameters in equation (2.2.8) are T 0 = 1000s, Y r ef = 1000, D = 1.0, Re= 0.21. 55 Alternatively, Ay may be defined in terms of the Richardson number of the mean flow (Stull and Zhang, 1991). A non-local analogy to the Richardson number is g(A,ftO(V) 'r» (A, (7) 2 +(A,n 2 (2.2.11) with ry = rji. If ry > R« then the Richardson number of the flow is non-critical. No turbulence is initiated and Ay = 0. Otherwise, turbulence will cause mixing until the flow reaches a termination value for the Richardson number, RT. Then the potentials of matrix A are i - ' r ' for i * j (2.2.12) wy is a distance weighting function from wtj = U0At/[Az]f - j]] for i * j (2.2.13) Both wy and Ay are bounded between zero and 1.0. The coefficients of the transilient matrix are obtained from for i^j (2.2.14) Nd is the number of grid points within the subdomain which is turbulent. Again the diagonal elements are found from equation (2.2.7). Recommended values of the three parameters, derived in Appendix A, are: critical onset Richardson number R« = 1.5, 56 termination Richardson number Rr = 2.0 and velocity length scale U 0 = 0.5 m/s (Stull, 1993). Numerical Implementation Stull (1993) quotes Le Chatelier's principle as 'If some stress is brought to bear upon a system in equilibrium, a change occurs, such that the equilibrium is displaced in a direction which tends to undo the effect of the stress". This principle applies to turbulence within the boundary layer. Dynamic forcings such as surface heating or shear stress cause instability, thereby perturbing the atmospheric equnibrium, and then turbulent transport erupts to try to restore the equiHbrium state. In the context of our discrete transihent framework, this process may be thought of as occurring in two 'half steps' within each interval At (see Figure (2.2.8)). In the first part of the step, the dynamic forcings and boundary conditions are applied to destabilize the mean flow. These forcings may be the surface fluxes of heat and water vapor or advection and fluxes into/out of the grid at higher levels. In the second part of the step the flow is restabilized by calculating the transilient matrix and rjerforming the mixing as per equation Figure 2.2.8 Destabilization and Mixing 'Half Steps' For Each Mixing Interval (Stull, 1988) Part 2: Turbulent stabilization. Part 1: Destabilization of the flow. 57 (2.2.2). By stepping through time in this manner, the evolution of turbulence and the resulting transport of atmospheric tracers in the boundary layer may be modeled. If turbulence is stationary, the matrix need not be recalculated. Otherwise, cy must be recalculated at every step in response to changes in the mean flow. The transilient mixing described by equation (2.2.2) is absolutely stable to any step size At. Because of the conservation of mass constraints placed on the mixing coefficients by equations (2.2.3a) and (2.2.3b) each row and column of the transilient matrix must sum to one. After Stull (1988) this means that the largest eigenvalue modulus of cy is no greater than one and thus ensures the stability of the calculation. The choice of grid size, Az, and step size, At, will depend on the specifics of the particular dataset and the situation being modeled. As a guideline, Stull (1993) recommends that the ratio of grid size to step size, Az/At, be greater than 0.1 m/s. 58 Chapter 3 Elevated Ozone Layers Within the Lower Fraser Valley: A Classification In this chapter the vertical distribution of tropospheric ozone over the LFV from the boundary layer to the tropopause is examined. Until fairly recently there has been very little knowledge of the distribution of ozone above the ground. However, Pacific '93 and other recent field programs have provided a significant amount of ozone data in various forms. The various methods of data collection of these programs were outlined in Chapter 2. This study draws upon vertical profiles from both tethered and free ascent balloons, ground and aircraft based lidar measurements and aircraft derived ozone transects and profiles in order to classify the elevated ozone structures that are observed over the LFV. It should be noted that because most of these observation programs were conducted during summertime anticyclonic conditions, the results give a biased view of lower atmospheric ozone occurrence. This study, therefore, is not aimed at trying to quantify the general level of pollution within the LFV. i.e. the ratio of polluted days to clean air days. Rather, the objective is to examine the vertical structuring of tropospheric ozone that occurs on days when high pollution levels are already present. 59 This chapter attempts to classify the types of layers present and identify their distinguishing characteristics. These include such properties as the height at which the layers are found and the suspected mechanisms by which they are created. Also of central importance is the scale of pollution within the layers. That is, how much do ozone concentrations within the layer exceed 'background' levels and how deep or thick are they. In addition, the horizontal extent of the pollution layers and their persistence is examined. Lastly, case studies of each type are presented and the implications for storage or transport of pollution within these layers and their potential for impacting upon ground level ozone concentrations is examined. I Layer Classification In this section a classification of elevated layers is proposed on the basis of: 1) Altitude of the layers with respect to significant atmospheric boundaries. (For example, the height of the mixed layer). 2) The processes of generation as inferred from other studies performed in similar locales, most significantly within the Los Angeles Basin. First, the height of the mixed layer capping inversion should provide a means by which to divide layer structures, as this usually constitutes a clear dynamic boundary between that region of the atmosphere that is dominated by surface or boundary layer 60 processes and the free atmosphere above. This then provides a basis for defining layers as occurring either below (boundary layer) or above (free atmosphere) the mixed layer capping inversion. Boundary layer features are well described, e.g. the Residual Layer of Figure (1.2.3), and several studies have documented the generation of pollutant layers above the inversion top in the L.A. basin (Ulrickson and Mass, 1990; Lu and Turco, 1994). Secondly, within pollutants that escape the boundary layer there will likely be a differentiation on the basis of the specifics of layer formation and the height at which the layers exist. For example, McElroy and Smith (1993) show a distinction between layers that occur just above the mixed layer, usually generated by convective debris from the polluted mixed layer, and those that occur at slightly higher levels, usually as a result of pollutants venting up mountain slopes. Finally, studies of free tropospheric ozone show that ozone pollution is occurring on continental and even global scales, (Browell et. al., 1992; Fishman et. al., 1990, 1991) and that consequently ozone appearing at higher elevations over the LFV may be present as a result of transportation from distant sources. Thus, a final distinction may be made between free tropospheric ozone pollution being advected over the LFV from distant sources and pollution that has its source within the LFV. With these distinctions in mind, in order to provide a framework through which to discuss the distribution of elevated polluted layers within the LFV, The following classification scheme of layer 'Types', as summarized by Table (3.1.1), is proposed. 61 H << -a a 3 73 a 73 CT •o V "2 M S f i 2 B" i s* o o -O co fT P 1^ De O n 73 CO o sTro Haz c 73 CT 3 p O ( A O CA f P 73 73 «3 he he CT 2. CO o " o " «< CT o r -•-i 09 O eg] «< -t CT <g" CO S" 00 » . OQ H CD C0 3 D a o 3 CT do r 1 g o ? £? » — O CT * ° 8 13 CT r> | 1 S'-5 «. cr — 2. OQ" c r o >-l> o 3 P CO CD > •a 73 3 X o o o I •*>. o o o .3 o" cT co ET 3 o> CT ? a. 3- P CT ST s. 2, |" o cr «i c « » ET o 5. 3 Is 1-1 3 tt 3 3- > o cr cr ° X a of* g o • a R t o o o °- 2. CT 2 o e •-»» 3 r o o » o" 3 O 3 O • a o o P P CO OQ cr ciq" 3 o P 73 cr 73 < 5' CT CW cr c Cu & ^ cr 3 2 n < o cr i f g. e d = =- S CT co 73 cr CT >-f , o ' 3- . 0 0 P 3.73 <5 cn 3" a i § 8 CO o r/3 -*> »< <D 3 3," o CT -O « Cf. o cT s c t» 3 — P < g cr c „ 73 co J2.T3 « a. 72 3 c r ™ o 5^ i s- 2-OQ < 2 ? S I S 52. re 73 O O- < S » 3 .5" 00 COO* p a. cr 3 TO 3 o. cr « 3 5. e. 3 73 c o 3 "2 3 ° c 3 ^ Ci m l-H CO 2 » S cr « 2 •-• Cu 0Q 0Q O 3 OQ ?d 1 2 § & Cu " B'S, 3- S C 73 CTT T3 3 O CT co a> S. 2 3 3 CT •• P O o 3 CT a. TJ CT O 73 CO CT co § . § CT CT 3 co 2 CT "•8 o 3 P cr o. c: ou •S f 3 5 CT •3>'c? S-3 . 3 73 2 O C T ° 0 O g * 1 -80" CT o a a 7r • § o O 3 P S5 00 — 2 CT «< CT o. g o c g 2 ° ; 3 _ 5- 3. CO *^ o1 3. 5' 0Q Co if-CT 3 5: H "° o co o O o « 3 O CT o O O c . "1 CT 3 O CT c < =i CT CT = 3 2-CT VC 3 cT 3 < CT § s-1 CT CO 2 73 ^ 3 O C T p c= g 5 S. >-> =.• 3 CT o —. 73 < O a £ o 3 2 3 r> 3 CS o. ?r CT CT 2. 3 0 P 1 c? r o co > 3-0Q CT 5" o CT Cu sr c;- o. cr CT , . 73 % O 73 3 O •ft *T> 3 Ju • CT 3 3 OQ g. S1" Sf CT O 3. CT — -&. w : P a CT Cu cr 8 o. o 3 3- p co 3 CT O . CT t-i 3" B> •5' 5-— " o 3. 3> x § Cu CT _ X oa O <Z O CT 2 -i Cu - ^ 2* o O 62 Type I: Elevated Layers within the Boundary Layer Type I layers are designated as those features which occur above the ground but below the height of the daytime mixed layer. In the mid-latitude coastal environment of B.C. this is usually in the range of from 200-600m a.g.l. They form in situations where atmospheric turbulence near the ground becomes suppressed, resulting in a reduction in atmospheric transfers or mixing between the air near the ground and that at higher levels. Because most sinks of ozone are at ground level, the most important being surface dry deposition, this "decoupling" of the elevated layer from the surface air means that there will be less destruction of ozone in the upper air mass compared to the air below resulting in higher ozone concentrations in the elevated layer. The suppression of near ground turbulence can occur as either an increase in shear stability caused by a low level jet or from a decrease in buoyant instability from a surface inversion. They may occur nocturnally or during the day. The most significant nighttime feature is the Residual Layer, as described by Stull (1988), that forms as a result of surface based radiative cooling. In complex terrain, drainage jets may contribute to the formation and intensification of the decoupling of the Residual Layer (Blumen, 1989). They may also form during the day overtop of upslope venting jets. Banta (1994) has observed what appeared to be an elevated polluted layer advecting mountainward over Pitt Lake in the LFV. As these features are separated from ground based sinks, they have a large propensity for storage and transport of boundary layer ozone and because they are located below the height of the daytime mixed layer they have a strong potential to affect ground level ozone concentrations, (Neu et. al., 1994). 63 Type II Inversion Top or Fumigating Layers Type II layers are found within or immediately above the mixed layer capping inversion. These are similar to the fumigating or inversion layers observed in several studies over the Los Angeles Basin, (Ulrickson and Mass, 1990; McElroy and Smith, 1993), and over the Iberian peninsula (Millan et. a l . , 1997). They usually appear in the mid to late afternoon and are likely a result of an interaction of pollution debris being convectively injected into and above the inversion, upslope venting of pollution at ridge top and undercutting of polluted air by the onshore marine sea breeze flow as described by Lu and Turco (1994). These layers may be found as low as 300m a.g.l in areas close to the ocean or as high as 2000m a.g.l. over mountainous terrain. As with Type I layers, because Type II layers are also separated from ground based sinks they often exhibit ozone concentrations that exceed those within the mixed layer below. Under stable synoptic conditions, complex coastal topography and a strong capping subsidence can cause the local circulation to become closed for extended periods, allowing layers of this Type to persist for longer than just the present day. Also, as they exist at a height where they may be entrained into the top of the mixed layer, they have the potential to be fumigated to lower levels and may affect ground level ozone concentrations. (Ulrickson and Mass, 1990) Type DI 'Deep Haze' Layers Above Type II layers, during extended episodes, may develop Type HI or what have been termed 'Deep Haze' layers. These layers are a result of mountainous 64 topography and of elevated subsidence inversions that may result in atmospheric circulations in areas of complex terrain to become relatively closed up to heights of 4000-5000m a.s.l. Millan et. al, (1997) describes creation of middle tropospheric layers at these altitudes over the Southern coast of Spain. They are most likely produced by upslope venting of ozone rich air within coastal mountain ranges and its subsequent advection back over the valley plain by the upper level return flow (Lu and Turco, 1994). Pollutants vented into this region of the atmosphere have a larger volume into which to be mixed than they would within the surface mixed layer, thus ozone concentrations are not as high as at lower levels, but they persist over depths of 2000-3000m (Evans et. al, 1993). They exist on the same time scale as the ozone episode itself, as they are dependent on the same synoptic scale subsidence that triggers the episode in the first place. These layers are effectively separated from the ground and are very unlikely to affect ground level concentrations. However the depth and breadth of ozone contamination that they demonstrate suggests pollution on a scale that is present only in what are considered to be highly polluted areas, such as the Los Angeles basin. Type IV Free Tropospheric Layers These are layers of enhanced ozone concentrations that occur in the upper troposphere between approximately 5000m a.s.l. and the tropopause, above the region of the atmosphere that is dominated by local mesoscale dynamics. Ozone at these levels of the atmosphere has a hfetime approaching 90 days (Fishman et. al, 1991), and these layers are likely present over a given locale as a result of synoptic scale advection from 65 distant sources, either boundary layer emissions most probably from biomass burning, or stratospheric intrusions. These layers are very unlikely to influence surface concentrations, but their appearance on continental scales over large portions of the middle to upper troposphere provide further evidence of the global scale ozone pollution clouds that have been observed by several studies within the free troposphere (Fishman et. al., 1990,1991; Browell, et. al., 1992) and also are evidence of the potential for global impact of pollution that is created by local scale anthropogenic sources (Millan et. al., 1997). II Layer Definition The definition of exactly what constitutes an ozone layer is arbitrary at best. Qualitatively, one may identify features of enhanced ozone in a vertical profile, but to systematically define these features in a manner that will allow quantitative analysis is a difficult task. First, most of the available data consists of one-dimensional vertical profiles, (that is, a series of measurements at only one point at any level). However, to define a feature as a layer requires some measure of horizontal extent, (i.e. at least two points at any one level), and strictly speaking therefore requires at least two simultaneous profiles. This means that any feature observed in a profile could very well be a 'blob' or 'puff instead of an ozone layer. Fortunately, lidar measurements have demonstrated the stratified nature of elevated pollutants in the LFV (Hoff, et. al., 1997). Consequently, the assumption made in this study is that enhanced ozone features in vertical profiles are indeed layers. In addition is the question of whether apparent ozone layers that are visible due to the surrounding air being depleted in ozone as opposed to the layer itself having 66 Figure 3.2.1 Sample Ozone Profile over Langley 13:00 PST Aug. 1, 1993 Specific Humidity (g/kg) 0 5 10 15 Ozone Concentration (ppb) and Potential Temperature (degrees Celsius) enhanced ozone should be included. Lastly, any arbitrary definition, no matter how soundly based in the physical processes at hand, will undoubtedly both exclude some features that one might mtuitively label as a layer and include some that otherwise would be discarded as uninteresting. The ozone profile shown in Figure (3.2.1) demonstrates some of the difBculties in defining what constitutes an elevated layer of enhanced ozone concentrations. The graph shows ozone concentration vs. height as.l over Langley at 13:00 PST on Aug. 1, 1993. At first glance, two large enhanced ozone features are present: one between 1000m and 2500m with ozone of about 50 ppb, (B), and one between 4000m and 5500m with ozone 67 of about 60 ppb, (C). However, upon closer inspection the profile becomes somewhat more complicated. Throughout the profile, fluctuations in the ozone signal are present. These are the result of both instrument error and of natural variations about the local mean ozone, and are generally on the order of about 5ppb and seem to increase with height above ground. The point at which these fluctuations cease being noise and become actual ozone signal is not immediately clear. For example, in the feature between 1000m and 2500m, (labeled B), there are slightly larger positive fluctuations at both the top and bottom of the layer. It is difficult to determine whether these 'subpeaks' are distinct features of their own accord or if they are merely local maxima within the larger layer. Similarly, the layer between 4000m and 5500m, (C), seems to also have two 'subpeaks', with the higher one having ozone about 10 ppb greater. Again, it is difficult to determine if this is one or two layers. Another problem is illustrated by what appears to be a thin layer of enhanced ozone at about 7200m, (D), with concentrations of about 45 ppb. Given that background ozone concentrations increase with height throughout the troposphere it is difficult to discern if this is indeed an enhanced ozone layer or if it is a region of roughly normal ozone with regions of depleted ozone above and below. Finally, there is the question of what role ambient meteorological factors have in defining ozone layers. At 600m, (A), there is a small increase in ozone that might easily be overlooked. However, this height closely corresponds to an inversion that is likely the top of the mixed layer and therefore this increase in ozone, though small, might indeed be a real feature, possibly related to convective lofting forcing pollutants up through the capping inversion. 68 Despite these difficulties, some sort of a layer definition is required to provide a framework by which to examine the occurrence of these features within the LFV. As a starting point, one may look to the criteria employed by other studies to identify layer of enhanced ozone. Browell et. al., (1992), in studying the occurrence of middle tropospheric ozone over central Canada, used a difference of 20% in ozone levels compared to the surrounding air as means of identifying ozone layers. Next, the level of enhancement that is used to define a layer must be greater than the ozone sonde measurement error to ensure that any exceedance is real and not a result of instrument noise. Also, the definition must ensure that ozone maxima are in fact due to enhancement within the layer and are not just apparent maxima resulting from ozone depletion above or below, i.e. from ozone scavenging in an advecting NO plume. Lastly the definition must be such that it includes only elevated layers and not surface based features of ozone pollution. Within these guidelines, and for the purposes of this study, a layer is defined as: ua region of finite vertical extent occurring above the ground in which ozone concentrations exceed 20 ppb above background and differ from those above and below by at least 20% " The figures of 20ppb and 20% are the most arbitrary points of the definition and were arrived at in a somewhat interative manner to try and include as many interesting features as possible in the analysis. However, they conform to the criteria used in the Browell et. al., (1992) study and should be sufficient to satisfy the conditions outlined above. 69 The definition as it stands still requires several clarifications. First is the question of what is "background" ozone. In this framework background is arbitrarily considered to be 10 ppb up to 1000m a.s.l., increasing above this level at 7ppb/km, according to figure cited by Browell, et. a i , (1992) for lower Canadian latitudes during the summer months and corresponding to the background trend visible in the upper troposhere of Figure (3,4.4.1) over Langley. This definition applies only to layers occurring "above ground", (i.e. a well mixed polluted boundary layer extending from the surface upward is not included), and uses the phrase "differs" rather than "exceeds" in relation to the surrounding air. This is to accommodate situations in which one layer of enhanced ozone occurs directly on top of another. Between two such layers there may be an increase in ozone that clearly defines the region as a separate layer without ambient ozone levels first dropping to background levels. Also, there is the question of "double peaks". That is, whether two adjacent ozone maxima are individual layers or just "sub-peaks" within the same layer. As the creation of these layers in closely linked to the local meteorology, adjacent peaks are considered separate layers only if there is a clear dynamic atmospheric boundary between them, such as an elevated inversion or a distinct wind shift. HI LFV Layer Inventory Applying the definition described above, the available archive of LFV profile data was examined to obtain an inventory of elevated ozone features that occur within the 70 region. The data sources that comprise the archive include, as outlined in chapter 2, AES free ascent sondes from 1993, AIR tethersondes from 1993 and 1994 and light aircraft profiles from 1995 and 1996. Again, the profile data are biased toward high ozone days Table 3.3.1a Layer Occurrence by Data Source Data Source Year Number of Profiles Number with Layers Multiple Layers A E S 1993 22 21 18 Tethersonde 1993 38 18 2 1994 13 6 1 Aircraft 1995 21 7 1 1996 11 6 0 Total 105 58 22 Table 3.3.1b Layer Occurences by Type Layer Type Number of Occurences Number with Clear Inversions Type I 21 17 Type l l 34 32 Typem 7 2 TypeTV 23 13 Unclassified 15 9 Total 110 73 and reflects the occurrence of elevated features on these days rather than ozone pollution in general. 71 Various profiles were examined and features that fit the layer definition were identified. The following data were recorded. - height, z, of the centre of the layer - depth, Az of the layer from top edge to bottom edge - maximum above background ozone concentration, A O 3 , within the layer In addition, any significant meteorology associated with each layer, (i.e., temperature inversions etc.), was noted. Once the individual layers were identified they were then classified as Type I, n , HI, IV or unclassified as per the classification scheme outlined previously in section (3.1). The results of the inventory are given in Table (3.3.1a). In all, 105 profiles from various sources were examined and a total of 110 elevated features identified. Of the AES data, 96% of the 22 profiles showed elevated layers with 81% having multiple layers. The tethersonde data showed elevated layers in 48% of 51 profiles with 3 occurrences of multiple layers while the aircraft sampled elevated layers in 41% of 32 profiles with 1 occurrence of multiple layers. Of the 110 layers the breakdown according to Type was as summarized in Table (3.3. lb). Type I, II and IV layers appear to be very common features of ozone episodes, with 21, 34 and 23 instances respectively while Type i n layers are more rare with only 7 examples identified. (It is important for the purposes of comparison to note that because the AES flights sampled more of the atmosphere they are more likely to encounter layer at higher levels and multiple occurrences of elevated layers than either 72 Figure 3.3.1 Elevated Layer height and Depth versus Above Background Ozone Concentration with Layer Types Identified a Priori. 10000 8000 to £ 03 X 6000 4000 2000 10 LtJ Ltl LtJ P Lp [] -I] T LtJ I; + > voe'" - i — • "Typell" I X l "Typelll" 1—*—I "TypelV" l • I The plots shows the height of the ozone maximum versus tie peak above background ozone value, with the depth of the layer shown at each point Each point is also displayed as per its classification Type. • 4- I i - i - I ± m t | j i 20 30 40 50 60 70 Ozone Concentration Above Background (ppb) 80 90 100 73-the tethersonde or aircraft profiles.) Most of the layers typed as 'unclassified' are probably most closely linked with Type TV layers . That is to say that they are likely due to advection from distant sources but their classification is less clear because they exist at levels that are more often identified with Type I and II layers. This inventory is displayed graphically in Figure (3.3.1). Layers are plotted with the maximum AO3 shown at the middle of the layer depth, although in reality layers are most likely asymetric about the A0 3 maximum. This is done to simplify cases where the exact height of the AO3 maximum is not clearly defined , i.e. where AO3 is constant over some altitude range or where one elevated feature shows two separate peaks at different heights. The plot shows that elevated ozone layers occur at virtually all levels of the atmosphere from the surface up to the tropopause, with the more polluted layers are located closer to the surface. Intuitively this makes sense as most of the ozone layers visible over the LFV valley have surface based emissions as their source, and the higher they reach in the atmosphere, the more chance they have of being dispersed. In addition, the graph shows that layer depths tend to increase with height above ground. Again, this is what one might expect as the "characteristic" size of the layers is going to increase as they move further away from any limiting physical boundaries such as the ground surface or the mixed layer inversion. The hierarchy of layer types is also evident in the figure. Type I and n layers are seen clustered at lower levels (between 500m and 1500m a. si), associated with tighter stratification and higher pollution values. Type II layers displaying lower ozone values but existing over a larger range of altitudes are visible in the lower middle range of the graph. 74 Finally, in the top half of the figure are the Type TV layers that are virtually ubiquitous throughout the AES profiles. The most prevalent meteorological condition associated with the occurrence of elevated layers appears to be the presence of an inversion in the potential temperature profile. Of the 110 features identified, 75 are accompanied by a distinct inversion. However the specifics relationship between inversion and layer occurrence is markedly different for each layer type. Inversions are most common with Type I and II layers, with 17 of 21 and 32 of 34 of the layers showing inversions, respectively (Table 3.2.1b). Both of these features are strongly dependent on the mixed layer capping inversion that is a result of the anticylonic subsidence that triggers the episode in the first place, either as an upper (Type I) or lower (Type IT) layer boundary, and a surface based inversion is an important aspect of Type I layer generation. The correlation with an elevated inversion is less clear for layers at higher elevations with Type HI and Type IV showing only 2 of 7 and 13 of 23 with distinct inversions. There are not enough instances of Type n layers with distinct inversions to speculate as to a spatial relation, but the profiles seem to show that Type IV layers most often display inversions at the bottom of the layer. Though these inversion are still most likely caused by subsidence, in terms of layer origins they probably indicate a change in air mass origin, as opposed to a cap that prevents dispersion. The strongest inversions are those encountered with Type I and II layers nearer to the ocean, (i.e. at Boundary Bay) and are related to the TTBL between the adverted marine air and the land-adjusted air below. Inversion strength seems to decrease both with distance inland and with height above ground with the weakest inversions being found 75 Table 3.3.2 Summary Table of Layer Type Features Type Number Observed z o z A z o(Az) ° 3 OO, Inversion Occurence Type I 21 490 135 176 55 45 15 17 inversions in 21 layers, usually at the top and bottom of the layer TypeH 34 830 440 173 100 59 19 32 inversions in 34 layers, usually bottom of layer, occaisionaiiy on top also Type XSi 7 3440 1060 860 401 24 3.6 2 of 7 layeis with inversions but no conlusive pattern TypelV 23 6140 1760 630 330 30 53 13 inversions in 23 layers, usually at the bottom of layer z, oz the mean and standard deviation layer height, in meters. (Az), o~(Az) the mean and standard deviation layer depth, in meters 03 , o03 the mean and standard deviation above background ozone concentration, in ppb with Type I and II layer over Pitt and Harrison Lakes and with upper tropospheric Type IV layers. Table 3.3.2 summarizes the average characteristics of the various layer types. The values in the table conform to the classification scheme of Table 3.3.1. The Type I layers encountered in the study have an average depth of 176m with the average ozone maximum of 45 ppb above background at 490m. Type II layers have a roughly similar depth of 173m but have an average ozone maximum of 59 ppb at 830m. As expected, these layers straddle the typical LFV mixed layer height of 600-800m and show the highest pollution levels, with the Type n layers having higher ozone concentrations as result of being separated from mixed layer sinks of ozone. Type I layers are often 'sandwiched' between a surface based inversion and the mixed layer capping inversion, while the mixed layer inversion usually forms the bottom boundary of Type II layers. Type HI have the greatest vertical extent, averaging about 860m in depth, with an 76 average ozone maximum of 24 ppb at around 3400m. This appears to be the upper range of the region of the atmosphere affected by LFV pollution. There is no distinct relation between elevated inversions and Type HI layer boundaries, (only 2 of 7 layers are associated with inversions). The lack of an inversion with a Type i n layer may be due to the different processes by which they are created, (i.e. mountain top venting as opposed to convective debris or nocturnal radiative cooling), and is probably the reason that the pollutants of Type III layers are so much more vertically dispersed. That is, the pollution is spread over a greater depth because there is no specific boundary to reduce dispersion. Type TV layers average around 630m in depth with an average ozone maximum of 30 ppb above background at 6140m. These layers are often accompanied by inversions, but the relation between elevated inversion occurrence and layer origin is different from that of Type I and II layers. Though these inversions will restrict transfer of atmospheric tracers across their depths, at the altitudes associated with Type IV layers they are most significant as a marker of a boundary between air masses of differing origins, (e.g. between a clean air mass below and distant source advecting ozone plume overhead.) IV Case Studies of Elevated Layer Types in the LFV IV. 1 Type I Layers The most significant Type I layer, in terms of influencing near-surface ozone concentrations within the LFV is the ozone bearing residual turbulent layer that forms in the evening above the nocturnal inversion. This is the phenomenon described by Figure 77 (1.2.3) in chapter one and examined by Neu et. al., (1994, 1995) . Within the L F V and specifically within the North Shore tributary valleys, most notably Pitt Lake, residual layer decoupling may be amplified by nocturnal drainage winds. Structures of this type are very effective at preserving and transporting boundary layer pollutants (Blumen,1990). Figure (3.4.1.1) shows an example of a Type I residual layer observed over Harris Rd. on the morning of July 21, 1994 The profile is a tethersonde descent obtained just after sunrise, but the nighttime flow regime and vertical structure are still evident. Potential temperature increases with height throughout the lower atmosphere, but there is a fairly clear discontinuity in the lapse rate at approximately 200m a.g.l. indicating the boundary between the nocturnal surface inversion and the residual turbulent layer above. Ozone concentrations are about 35 ppb at the surface and increase slightly with height through the inversion with one sharp jump to near 45 ppb at around 80m ag.l. Above the nocturnal inversion, ozone concentration shows a rapid increase to around 65ppb as the residual turbulent layer is encountered. This rise marks the zone where transfers between the air masses above and below are reduced allowing ozone in the layer above to persist through the night. Ozone remains approximately constant through the residual layer up to a height of 500m and then drops slightly at 600m. This height corresponds to an inversion in the potential temperature profile that is probably the previous day's mixed layer cap. Ozone in this layer must have been 'stored' from the previous day because the profile is from too early in the morning for it to have been produced on the present day. Mixed layer ozone concentrations for the previous day at Harris Rd. exceeded 100 ppb, and the value of 65 ppb throughout the residual layer fits well with the claim of Neu et. al., (1995) that 50-80% of mixed layer ozone survives the night in the residual layer. This 78 Figure 3.4.1.1 Ozone and Meteorology 05:58 PST July 21 , 1994 at Harris Rd. Wind Direction (azimuth) 180 270 to CD -*-» CD E, ri) cd •4-" JZ O) 'CD X 1000 800 600 400 200 ~l 1 wind_speed" "direction" 360 2 3 Wind Speed (m/s) 2 CD CD E. ri) cd D) X 1000 800 h 600 400 200 0 Specific Humidity (g/kg) 8 9 "potential_temperature" "specif ic_humidity" ozone 11 30 40 50 60 70 80 Ozone Concentration (ppb) and Potential Temperature (degrees Celsius + 30) 79 ozone within the residual layer represents a large reservoir of pollution which can potentially impact surface level ozone concentrations when the residual layer is eroded and entrained by the next day's deepening mixed layer. Within the wind speed profile, there is some evidence of a drainage jet with maximum wind speeds of about 3m/s at 200m a.g.l. that correspond to the height of the nocturnal inversion, indicating that there is likely some negative shear induced intensification of the decoupling of the elevated layer occurring. Of perhaps more importance in terms of air quality, however, is the wind direction throughout the profile. Wind direction is from about 180° at the surface , shifting to 360° at 100m and remaining fairly constant through the residual layer until shifting again to 90° at around 600m. The 360° wind direction is significant because the roughly northerly direction implies ozone transport within the residual layer that is directed back downvalley, (back toward Greater Vancouver). This shows that although the Pitt Lake valley has been identified as a net sink for boundary layer ozone, (O'Kane, 1997) and although Banta et. al, (1997) have shown that the nocturnal down valley flow is relatively clean up to about 300m, advection and transport within the nocturnal residual layer may nonetheless provide a path for recirculation of pollutants back to ground level within the LFV. Evidence of residual layers is virtually ubiquitous throughout the LFV ozone profile data. They can be seen in tethersonde profiles from several days at both Harris Rd. and Goose Island arid in the free ascent profiles over Langley. The preponderance of these features and their potential for storage and transport of ozone means that knowledge of the processes of their creation and of the turbulent muring within them will be crucial to air quality studies in the region 80 Figure 3.4.1.2 Daytime Type I Layer over Pitt Lake, 13:52 PST Aug. 2, 1993. Specific Humidity (g/kg) 2 3 4 5 6 7 8 9 10 11 ) 45 50 55 60 65 70 Ozone Concentration (ppb) and Potential Temperature (degrees Celsius + 20) Wind Direction (azimuth) 0 90 180 270 360 1 0 0 0 I — | ) I • I I — I L J — - — I 1 2 3 4 5 6 7 Wind Speed (m/s) 81 A less common but interesting feature within the data is what appears to be a daytime version of the elevated Type I layer that is associated with the afternoon upvalley low level jet. Figure (3.4.1.2) shows a tethersonde profile obtained over Pitt Lake on the afternoon of Aug 2, 1993. The wind profile shows a distinct up-valley low level jet with maximum wind speed of 5.5 m/s at 150m a.g.l. and what appears to be the valley return flow between 600m and 800m. Potential temperature shows a weak inversion from the surface up to the height of the jet max. This is probably from the advection of the air mass across the relatively cooler surface of Pitt Lake. Above this surface inversion the increase in potential at around 200-250m a.g.l. corresponds to the top of the inversion and the 'downside' of the jet temperature is virtually constant up to the capping inversion at 600m. This mixed layer height is also marked by a sharp drop in the specific humidity. Ozone is approximately constant at 50 ppb near the surface; at the height of the jet maximum and the top of the inversion layer ozone rises quickly to more than 60 ppb and remains high until the capping inversion is reached whereupon ozone drops to about 40 ppb in the return flow aloft. The region of the increase in ozone concentration between 200-250 a.g.l. corresponds to the top of the inversion and the 'downside' of the jet maximum, (i.e where dU/dz is large negative). As with the residual layer, this rise in ozone occurs in a region where turbulence is suppressed by both the negative wind shear and the weak lake surface inversion. Again, the zone of reduced turbulence will cause the air above to become kinetically decoupled from the air near the ground, separating the elevated layer from surface based ozone sinks and allowing the ozone concentration in the decoupled layer to remain higher. 82 This layer is likely related to an elevated plume observed advecting up valley by Banta et. al. (1994) and may be the flow that gives the trajectories modeled by Miao (1993). In addition, this phenomenon may be one of the pathways by which ozone is transported into the North Shore mountains and vented to higher levels and may be one of the factors resulting in Pitt Lake exhibiting among the highest boundary layer ozone concentrations within the LFV. (McKendry et. al., 1998) Interestingly, the Aug. 2, 1993 layer appears to be a special case. This elevated structure is not apparent on other days during Pacific 93 or during the more typical ozone episode of July 1994. The meteorological forcings were not as strong during Pacific 93 as they are during most ozone episodes (Pottier et. al, 1997). During a stronger episode, higher ozone levels throughout the boundary layer might blur the distinction between the upper and lower air masses, or stronger surface heating might cause buoyancy to dominate the negative shear region above the jet maximum, thus preventing the decoupling of the elevated air mass. IV.2 Type II Layers Type II layers are defined as zones of enhanced ozone that occur within and immediately above the mixed layer capping inversion. These features are found at heights ranging from around 300m as.l. near water to 2000m over mountain ridges. Figure (3.4.2.1) shows three aircraft derived profiles of ozone and potential temperature for three sites in the LFV: Boundary Bay airport, Pitt Meadows airport and the North end of Harrison Lake. At Boundary Bay, close to the ocean, mixed layer depth is about 300m; at 83 Figure 3.4.2 1 Aircraft Profiles of Type II Layers for Three Sites: (a) Boundary Bay (b) Pitt Meadows (c) Harrison Lake (a) Boundary Bay, 16:25 PST 26 Aug. 26, 1996 (b) Pitt Meadows, 15:45 PST 26 Aug. 26, 1996 to » so 7 ° *° so ioo 10 ,10 30 SO 90 70 »0 » '00 Ozone (ppb) Specific Humldity#5 (g/kg) Potential Temperature (C) -Ozone • - • • • SpecificHumidlly PotentialTomperalure | (c) Harrison Lake, 14:15 PST 19 July, 1995 3000 T 84 gure 3.4.2.2 Type II Layer over Langley 16:00 PST Aug. 4, 1993 35 40 45 50 55 60 Ozone Concentration (ppb) and Potential Temperature (degrees Celsius + 15) 85 Pitt Meadows the depth is 600m, typical for most of the LFV, and the Harrison Lake profile shows a mixed layer of just over 2000m, likely as a result of stronger differential heating over the valley slopes. In each case, the base of the inversion corresponds to a virtual step change increase in ozone concentrations, and in each case the increase is on the order of 40-50 ppb in the inversion layer compared to the values in the mixed layer below. In the case of the Pitt Meadows profile, mixed layer ozone is only on the order of 40 ppb, much too low to be considered an 'episode', but the ozone concentration at 600m is almost lOOppb, or significantly greater than the NAAQO standard of 82 ppb. In fact, Type II layers commonly exhibit the highest concentrations of any of the elevated features present within the LFV. Type II layers are visible in the AES free ascent balloon profiles over Langley on almost every day of the Pacific '93 field program. Figure 3.4.2.2 shows the ozone and meteorology profiles for the 16:00 PST flight on Aug. 4, 1993. Again, a Type II layer is visible at the top of the mixed layer at about 600m. In this profile a second elevated inversion is visible at about 800m, seeming to trap the pollutants from above and below. The height of this ozone layer also appears to be correlated with a shift in the wind direction from a westerly flow at about 270° to a more northerly flow off of the North Shore mountains at about 360°. This layer appears on virtually every afternoon during the period of observation of the AES profiles, usually in the 16:00 PST flight, and the layer is usually accompanied by the northerly shift in wind directions. A particularly interesting feature of these layers in the horizontal scale of pollution with the LFV that they imply. From Boundary Bay to the north end of Harrison Lake is a distance on the order of 100 km. That urban ozone pollution can affect areas on this scale 86 is not unprecedented. Fishman (1987) showed that an ozone episode in Atlanta resulted in atmospheric contamination over similar scales and was clearly detectable in the TOMS total column ozone signal. Millan et. al, (1997) documented elevated layers existing over southern Spain on horizontal scales approaching 300 km. Previously, with ozone monitoring conducted primarily by the GVRD near-surface sampling network, ozone pollution was beleived to be confined to the boundary layer within the LFV plain. However, the appearance of ozone over a hundred km downwind of the emissions sources of Greater Vancouver implies pollution on a much greater scale than was previously believed to exist in the LFV. The manner in which these layers are generated is not fully understood. The fact that they occur at the top of the mixed layer, regardless of what that depth actually is, implies that they are created at least in part by convective injection of pollutants into and above the inversion. Ozone or its precursors will be carried aloft by thermal plumes; some of these plumes will have sufficient kinetic energy to overshoot the inversion and be trapped above the mixed layer. Separated from ground based sinks, this ozone will be preserved or further increased by continuing precursor reactions. As a result, these layers become large photochemical reactors and ozone concentrations within them will often greatly exceed those in the mixed layer below. When mixed layer turbulence begins to wane after noon, the height of the entrainment zone will fall and the Type II layer will be left above. However this mechanism alone cannot account for the occurrence of Type II layers. Firstly, these layers often form when mixed layer pollution is not particularly severe. For example, in the Pitt Meadows profile of Figure (3.4.2.1), it seems unlikely that 87 convective thermals in a mixed layer where ozone is scarcely 40 ppb could generate concentrations of 100 ppb immediately overhead. Second, the Boundary Bay profile is at a site that is, under sea breeze conditions, upwind of most emissions sources. This means that the pollutants visible above the mixed layer at this locale must have been lofted somewhere further inland and then somehow advected back oceanward. The lower height of this layer as compared to the Type JJ layers located fiirther inland implies that significant subsidence has occurred, either from synoptic or local scale circulations. Thirdly, the existence of the wind shift, though subtle, implies that some form of differential advection is occurring within the Type II layers compared to the mixed layer below. Further, the northerly direction of the winds in the layers over Langley tends to support the possibility that venting over the North Shore mountains may be one of the sources of this pollution There is considerable evidence of pollutant venting out of the North Shore mountains. Figure (3.4.2.3) shows an aircraft transect across the ridges between Haney and Stave Lake. The aircraft height is approximately constant at 1500m a.s.l. as the aircraft crosses a series of three ridge tops. The ozone concentration along this flight line shows a peak in accordance with the passage of each ridge, implying that pollutants are being lofted to higher levels in the deeper mixed layer over the mountain slopes compared to the valley plain. Figure (3.4.2.4) is an east-west lidar section from the Pitt Valley just north of Harris Rd. Most of the aerosol in the valley is confined below 800m but on the eastern side of the valley an upslope plume can be seen exiting the valley flow and escaping to higher levels. In addition, Hoff et al (1997) detected aerosol plumes above the 88 Figure 3.4.2.4 East-West Lidar Trace from South End of Pitt Lake, along flight line T l , Aug. 5, 1993. Aerosol laden air is seen escaping the Pitt Valley to the right of the figure.(Hoff et. al. 1997) tofl X 122.7 122.6 Longitude (W) 122.5 89 mixed layer in the nighttime flow out of the valley. The differential advection in the Type II layers over Langley is consistent with a change from onshore sea breeze flow at lower levels to a valley return flow aloft above the mixed layer. This is suggestive of both the cooler and denser marine air of the sea breeze front pushing under the polluted layer aloft and with advection of pollutants lofted by slope flows. In addition, the lower boundary of the Type TJ layer is typically associated with a sharp drop in specific humidity, further supporting the idea that the elevated layer air mass is of a differing origin from air in the mixed layer. It is more likely that Type II layers are created by an interaction of several processes. Pollutants are convectively injected into the inversion, both over the valley plain and, to higher levels, over the valley slopes. Upslope plumes, like the one visible in Figure (3.4.2 4) vent pollutants out of the LFV mixed layer. Ozone vented over the mountains is then advected back over the valley plain appearing in the profiles over Pitt Meadows and Langley. This flow is probably the northerly wind over Langley in the layers of Figure (3.4.2.2) and a valley return flow is also present above the Type I layer over Pitt Lake in Figure (3.4.1.2). Once over the LFV plain, these layers are then undercut by the onshore sea breeze flow. Marine undercutting is probably more important in the case of Type II layers that have persisted overnight or are located closer to the ocean, because the onset of any marine influenced air, at around noon, would precede the appearance of Type II layer, mid to late afternoon, over the middle of the valley. It is further possible that ozone advected back over the LFV by the valley return flow may in turn be caught in the upper level return flow of the of the sea breeze. This would account for the occurrence of the layer over Boundary Bay. Unfortunately, there are no wind profile data available over 90 Figure 3.4.2.5 a) Type II Layer over Langley 16:00 PST Aug. 2, 1993 b) Profile 04:00 PST Aug. 3 shows no trace of the Type II Layer (a) 16:00 PST August 2,1993 Specific Humidity (g/kg) 5 6 7 8 9 10 20 30 40 50 Ozone Concentration (ppb) and Potential Temperature (degrees Celsius) (b) 04:00 PST August 3, 1993 Specific Humidity (g/kg) 0 I . L I . I : I I 0 10 20 30 40 50 Ozone Concentration (ppb) and Potential Temperature (degrees Celsius) 91 Figure 3.4.2.6 a) Type II Layer over Langley 13:00 PST Aug. 5 1993 b) Type II Layer still visible at 04:00 PST Aug. 6 (a) 13:00 PST August 5, 1993 Specific Humidity (g/kg) 10 20 30 40 50 60 70 Ozone Concentration (ppb) and Potential Temperature (degrees Celsius) (b) 04:00 PST August 6, 1993 Specific Humidity (g/kg) Ozone Concentration (ppb) and Potential Temperature (degrees Celsius) 92 Figure 3.4.2.7 Back Trajectories Showing Open Circulation of Aug. 2 16:00 PST vs. Stagnated Flow of Aug. 5 16:00 PST. Each mark shows the air mass position at 12 hour intervals for the previous three days for each of the levels indicated as calculated by A E S . On the 2 n d , al levels displayed show a source region som where over the middle of the Pacific Ocean. However, on the 5 t h, the same analysis shows source regions within southwestern B . C . or northwestern Wash, for all levels up to 600 mb. this site to support this hypothesis. Lastly, subsidence is also likely important in bringing layers created by mountain venting closer to the ground. Once aloft, Type II layers will either be transported out of the area by synoptic winds or will be entrained back into the mixed layer below. As they are separated from most sinks, these layers are more likely to be dispersed by meteorological processes before they are consumed chemically. Under episode conditions, when the LFV circulation becomes closed, Type II layers may persist longer than the day on which they were 16:00 PST Aug. 2 16:00 PST Aug. 5 -B-750mb -e-700mb -£r650mb ^~600mb -&750mb -e-700mb -£r650mb -X-.660mb-93 created. Early in the Pacific 93 episode week, synoptic scale winds were well developed and there is no trace of the layers on the mornings after they were produced. Figure (3.4.2.5) shows the profiles for Aug 2 16:00 PST and the following day at 04:00 PST. On the 2nd there is a distinct Type II layer present, but there is no trace of this feature at 0400 on the 3rd. As the synoptic gradients weaken and the LFV becomes closed, (that is, synoptic winds are not suuficient to transport polluants out of the area and mesoscale flows serve mostly to recirculate pollutants within region without allowing them to escape), later in the week, (see Figure (3.4.2.7)), there is evidence that these layers persist for longer. Figure (3.4.2.6) shows the Aug 5 13:00 PST and Aug 6 04:00 PST profiles. There is a Type II layer visible on the afternoon of the 5* that is still present on the morning of the 6th. These layers then may preserve ozone at least overnight and, under stronger anticyclonic conditions than those present during Pacific 93, probably longer. Because these layers occur at inversion top and exist within a subsiding atmosphere, it is possible that layers that persist overnight into the next day may be entrained into the top of the mixed layer and fumigated to lower levels. Considering the horizontal scale on which these layers are present and the levels of contamination within them, combined with the possibility that they may be intersected by the mixed layer, Type II layers present a tremendous store of pollutants that may affect ground level air quality within the LFV. IV.3 Type m Layers 94 When the L F V circulation becomes closed under strong synoptic subsidence, atmospheric pollutants may accumulate over several days up to depths of a few thousand meters a.s.l. Enhanced ozone features at these heights are deemed Type JB. layers or, as they have been referred to in other studies, 'Deep Haze' layers. These layers occur in the lower free troposphere, above the mixed layer and Type I and H layers, but are still within that region of the atmosphere that is affected by local pollution sources of the L F V . Figure (3.4.3.1) shows the middle tropospheric profile over Langley on Aug. 5, 1993 at 16:00 PST. At the bottom of the figure a polluted mixed layer and a Type II layer above are clearly visible. Between 2000m and 4500m there is a deep layer of moderately high ozone, with values between 40-50ppb. The upper boundary of the layer appears to be capped by an elevated inversion in the potential temperature profile and there is another inversion associated with the local ozone maximum at 2500m. This layer is also observed in an aircraft profile over Hope earlier on the same day. Figure (3.4.3.2) shows the ozone and aerosol concentration profiles on Aug 5 at just after 09:00 PST. The profile is an ascent from the boundary layer up to 5000m measured by the NRCC Convair 580. Again ozone levels are moderately high to almost 5000m a.s.l. Aerosol concentrations are also high to a very large depth, exceeding 500 cm" 3 at about 500m and remaining over 100 cm"3 to 4500m Normal values for a clean atmosphere are on the order of 25 cm"3. High aerosol concentrations tend to be strongly correlated with tropospheric ozone pollution. The presence of enhanced aerosol concentrations 95 Figure 3.4.3.1 Middle Tropospheric Profile over Langley, 16:00 PST Aug. 5, 1993. 0 8000 2 3 "55 E <A ro —• sz D) '© X 7000 6000 5000 4000 3000 2000 1000 h 0 Specific Humidity (g/kg) 5 10 i r "ozone" "potentiaLtemperature" "specific_humidity" 15 20 30 40 50 60 70 Ozone Concentration (ppb) and Potential Temperature (degrees Celsius + 15) Wind Direction (azimuth) 180 270 Wind Speed (kts) 20 360 30 96 Figure 3.4.3.2 Ozone and Aerosol from NRCC Aircraft ascent over Hope 09:00 PST Aug. 5,1993. a I • • i i i • • i • i i i i i i i i i r • • i • t i i i • i I • , i I a ' — • — • f . I 3fl 35 M 35 O 43 56 55 I M 1SI 3 M 25a 799 35* 4 M « 5 0 » • 55a 4*0 Ozone (ppb) Total Aerosol (cm-3) throughout the layer in question is strong evidence that the ozone at these heights is from boundary layer sources. The three-day back trajectories for the air mass over the LFV on Aug 5 at 16:00 PST, (Figure (3.4.2.7)) reflect stagnation up to 650 mb (about 4000m a.s.1.) over the previous two days. This means that the aerosol contamination could not have been transported from a distant source, for example, from a forest fire plume. This figure and the aerosol concentrations of Figure (3.4.3.2) confirm that the middle tropospheric ozone 97 pollution over Hope, and similarly over Langley, must be pollution from local sources within the LFV. One curious feature of this layer, to date unexplained, is the apparent opposite trends of the ozone and aerosol concentrations with height. As ozone and aerosol tend to be correlated, both species should show similar trends within their respective profiles. However, in Figure (3.4.3.2) the aerosol concentration is highest nearer the ground and decreases with height, while ozone does the opposite, showing the highest values nearer to the top of the profile. The source of Type HI layers is almost certainly upslope venting of pollutants over the North Shore mountain slopes. In essence this is the only feasible explanation. The layers are too high to be the result of convective debris; even the most buoyant plumes are unlikely to reach 4000m in a subsiding atmosphere. Similarly, sea breeze undercutting will affect only the lowest 1000m of the layer and Hope is too far inland to be dominated by sea breeze mechanics. Aerosols may be lifted to these levels in convective cloud structures, but the water vapor would destroy the ozone. Upslope venting was previously discussed in relation to Type II layers. The question is whether or not this process will allow pollutants to reach the heights apparent in Figures (3.4.3.1) and (3.4.3.2). Figure (3.4.3.3) is another aircraft profile from the north end of Pitt Lake. The figure shows an elevated layer that would probably best be described as a Type II feature. What is significant with respect to the creation of Type HI layers is that the figure shows ozone levels of around 80 ppb reaching heights in excess of2300m. Figure (3.4.3.4) is an aircraft transect flying north over Pitt Lake at about 500m. The figure shows a series of peaks in ozone concentration toward the north end of the lake, 98 that again appear to show venting of ozone plumes in excess of 100 ppb. The back trajectories of Figure (3.4.2.7) show a gentle northerly flow at 650 mb over this proposed source toward the LFV plain for the previous two days, suggesting a means of advecting these plumes back over the measuring sites at Hope and Langley. This gentle northerly flow at this height is likely a common feature of the LFV stagnant air mass as it would be a natural result of the 500mb ridge over the west coast of B.C. that is characteristic of ozone episodes. As these layers are trapped by an upper level subsidence inversion, and persist over Figure 3.4.3.3 Aircraft Ozone Profile over the North End of Pitt Lake 14:00 PST July 19, 1995. 2400 | 2300 + <n 2200 + £ 2100 | 2000 + 1900 + 1800 •! 1 1 — 1 ! 1 1 1 : 0 10 20 30 40 50 60 70 80 90 OZONE (PPB) 99 Figure 3.4.3.4 Aircraft Measured Ozone Transect over Pitt Lake, 14:30 PST Aug. 24,1996 120 the LFV due to weak anticyclonic winds, the lifetime of these layers will probably be of the same span as the traveling anticyclone, that is from 2 to 5 days. There is evidence of the layer in Figure (3.4.3.1) over Langley on both the afternoon of Aug 4. and the morning of the 6th, suggesting that the layer was present for about 2 days before the episode passed and the pollutants within the layer were carried away by the renewed synoptic winds. Again, the fact that the episode captuerd in Pacific 93 was not particularly strong implies that under stronger anticyclonic conditions, Type III layers would persist even longer and perhaps show higher levels of pollution. The height at which these layer exist means that they have very little chance of ever being recirculated back to the surface, thus they are not of specific or direct interest to ground level air quality. However, as with Type II layers, the spatial scales on which 100 they appear to occur is striking. As with the Type II layers over Harrison Lake, Hope is over 100km downwind from the emission sources of Greater Vancouver, and the depth of the atmosphere that is affected, even at this large distance, is at least two to three kilometers. Again, this layer implies pollution within the LFV occurs to a degree that was not previously anticipated and is on a scale that is usually only observed in what have been considered to be highly polluted areas. IV.4 Type IV Layers The upper tropospheric profiles over Langley show several features of enhanced ozone concentration, from heights of 4000m a.s.l. upward, above the level of the atmosphere that is directly influenced by pollution from sources witJiin the LFV. These 'Type I V upper free tropospheric ozone layers are neither caused by LFV emissions nor are they of any influence on surface air quality in the region. However they are interesting and important in terms of tropospheric pollution in general and, more specifically, they are representative of the extent to which ozone pollution is present on continental and even global scales in the upper free troposphere. Figure (3.4.4.1) shows the upper level ozone profile over Langley for Aug. 6 at 04:00 PST. The figure shows two distinct elevated ozone features in the middle to upper free troposphere. The lower of the two is suspected to have originated from the boundary layer while the higher layer is most likely from a stratospheric intrusion of ozone. Between 3500m there is a layer of enhanced ozone with a peak value of 65 ppb at 3800m. There are distinct inversions at the top and bottom edges that appear to constrain the layer 101 Figure 3.4.4.1 Upper Troposphere Ozone Profile over Langley 04:00 PST Aug. 6, 1993. Dew Point Temperature (degrees Celsius) -60 -50 -40 -30 -20 -10 0 10 I 40 50 60 70 80 90 Ozone Concentration (ppb) and Potential Temperature (degrees Celsius+20) 102 vertically, and a third inversion at the peak ozone concentration. This layer is first visible in the 10:00 PST profile on the 5th and can be traced through the 13:00 and 16:00 PST profiles as it subsides and the inversions strengthen. This layer occurs directly above the Type m layer that is present on the 5th and 6th and at first glance appears to be the upper extension of the layer below. However, there are several clues to the fact that the two features are part of different air masses. First, specific humidity drops sharply from 6 g/kg to under lg/kg between 3300 and 3700m, indicating a change from the moister LFV influenced air mass below to a drier air mass advecting overtop. Also, the lower boundary is associated with a subtle veering of the wind from 270° to 315°, and the back trajectories of Figure (3.4.3.3) for 16:00 PST Aug 5 show an 'opening' of the circulation over the LFV between the stagnated flow at 650mb and the 600mb level, roughly corresponding to the height of the layer. As shown previously, the 650mb flow at this time is mostly confined to the LFV, but the 650mb level shows advection from over north-central British Columbia over the previous three days. With these facts in mind, the lower boundary of the elevated ozone feature at 3500m almost certainly marks the boundary between the region of the atmosphere directly affected by LFV sources and the free troposphere above. The first clue Of the origin of this layer is the height at which it occurs. Table (1.2.1) in Chapter 1, from Broweli et. al. (1992) shows most of the ozone layers encountered in that study at heights below 5000m were believed to be of boundary layer origin. More quantitatively, the best tracers for ozone of boundary layer origin are the presence of either aerosols or carbon monoxide, both of which are produced only at the earth's surface. Unfortunately, the NRCC aircraft profiles topped out at around 4500-103 5000m, or just below the height of the Type IV region. However, some evidence of the origin of this air mass may be found at the top of aerosol profile in Figure (3.4.3.2). Just below the top of the aircraft ascent, there is a spike in the aerosol concentration. The height of this spike is at roughly the same level at the bottom of the first appearance of this layer in the 10:00 PST profile over Langley. If this spike in the aerosol concentration is real, as opposed to some 'glitch' or noise from the end of the measurement, then it may be from a jump in aerosol concentrations as the Type IV layer is entered. If real, this aerosol is strong evidence that the pollution within the layer originated at the earth's surface. RecaUing the back trajectories of Figure (3.4.3.3), this aerosol laden ozone plume may be from biomass burning, (i.e. a forest fire), somewhere over the northern interior of B.C. Above 4500m, the ozone concentration increases almost linearly at about the typical 6-7 ppb/km for the unpolluted troposphere at middle latitudes (Browell et. al., 1992). Between 6500m and 8500m there is a broad zone of enhanced concentrations. The highest value observed is 84 ppb at 7200m, or about 35 ppb above background levels at this height. Again there are weak inversions marking the upper and lower boundaries of the layer. This feature first appears in the 04:00 PST profile on the 5th at a height of about 9500m as what appears to be the bottom edge of the stratosphere and subsides throughout the subsequent profiles. As no aerosol or carbon monoxide measurements are available, the origin of this layer is not clearly evident, but it is most likely some sort of stratospheric intrusion. The first clue to the layer's origin is the height at which it appears. The first appearance is at over 9500m, which is close to the tropopause in an anticyclone, and the layer remains above 6500m. All the layers of enhanced ozone encountered by Browell et. al. (1992) 104 above 6km were found to be of stratospheric origin. Stratospheric air is very dry compared to air in the troposphere, exhibiting dew points as low as -60°C. Dew point at around 7500m dip to -50°C and specific humidity is less than 0.3 g/kg. Winds in this layer are steady for the previous day at 25 kts. from 315°. This suggests a source region somewhere over the Gulf of Alaska, which makes a boundary layer origin unlikely. Winds of 25 kts. for 24 hours give a horizontal layer extent of a least 600 nautical miles. A feature of this size is probably created by forcings on roughly the same scale, that is, by some synoptic scale phenomenon. Most likely this layer is a stratospheric intrusion originating over the Pacific Ocean from some sort of folding episode in the air mass history. Two distinct layers are also visible over Langley in the 04:00 PST profile on Aug. 2. (see Figure (3.4.4.2)) Between 3500m and 5000m ozone concentrations are over 60 ppb with a maximum of 73 ppb at 4500m. The occurrence of this layer appears very similar to the middle tropospheric Type TV layer of Aug. 6. Again, there is a strong inversion and drop in water content at the lower boundary of the layer. There is also a strengthening and backing of the wind at this level. As with the Type TV layer of the 6th, this indicates the presence of a distinct air mass boundary between the ozone rich layer and the relatively unpolluted air below and, again, the height at which the enhance ozone layer occurs supports a boundary layer source region. However, in this instance, the three-day back trajectories for Aug. 2 (Figure (3.4.2.6)) show a path almost entirely over the North Pacific, except for a brief crossing of the northern tip of Vancouver Island within the previous day. It is very unlikely any forest fire plume encountered at this crossing could reach this height in less than a day while adverting over water. As a result, the 105 Figure 3.4.4.2 Upper Troposphere Ozone Profile over Langley 04:00 PST Aug. 2, 1993. Dew Point Temperature (degrees Celsius) -80 -70 -60 -50 -40 -30 -20 -10 0 20 30 40 50 60 70 80 Ozone Concentration (ppb) and Potential Temperature (degrees Celsius) 106 source of this layer is not immediately evident. Recalling that free tropospheric ozone may persist for up to 90 days, it is difficult to speculate as to a specific source. Figure (1.2.8) from Fishman et. a l . (1991) shows large clouds of ozone in the free troposphere over the North Pacific during the North Hemisphere summer. These clouds were attributed to the eastward transport of pollution from intense industrialization or biomass burning over Eastern Europe and Siberia. The back trajectories of Figure (3.4.2.6) place the source of this Type TV layer firmly in the middle of the Northern Pacific ozone cloud and, therefore, the appearance of the layer over Langley may be related to the long range free tropospheric transport of this pollution. There is also a Type IV layer of enhanced ozone centered at 9500m at which height ozone reaches 68 ppb. This layer is almost certainly of stratospheric origin. The layer occurs very close to the tropopause and the ozone rich air is very dry. Dew points in the layer are less than -70°C and specific humidity is approximately 0.02 g/kg, representing an order of magnitude drop from the more typical free tropospheric value of around 0.25 g/kg below 8000m. This layer is a much more obvious instance of stratospheric ozone than the upper tropospheric Type IV layer of the 5th and 6th, but the exact mechanism of this intrusion is unclear. Again, it is most likely the result of an episode of folding somewhere over the Pacific in the air mass history. V Summary of LFV Layer Formation This inventory demonstrates the wide variety of elevated structures of enhanced ozone of varying spatial and temporal scales that occur in the atmosphere over the LFV. 107 Figure 3.5.1 Type I Layer Generation for Nocturnal (Residual Layer) and Daytime Cases. Daytime Type I Layers - "Decoupled" Polluted layer Upslope Low Level Jet Mixed Layer!nversion Top ~600magl Decoupled" Polluted Layer dU/dz < 0 Region of Suppressed Turbulence Nocturnal Type I Layers (Residual Layer) Residual I-ayet Mixed Layer Inversion Top —600m agl Decoupled" Polluted Layer (Residual Layer): Drainage jet and Nocturnal Inversion Region of SuppressedTurbulence / ///////////////// This variety is a reflection of the broad spectrum of atmospheric phenomena that coexist in the region during episode conditions, from microscale radiative transfer, through mesoscale slope and sea breeze flow up to the synoptic scale circulation that interact to generate the vertical structures. Results show a continuum of features, from the boundary layer upward, in which 108 Figure 3.5.2 Generation of Elevated Ozone Layers for an Idealized Coastal Region. 109 the temporal and spatial scales of both the polluted layers and the scales of their mechanisms of generation increase with height above ground. For example, the creation of near ground layers is governed by local effects such as slope flows and radiative cooling, while layers in the middle to upper troposphere are a result of synoptic scale circulations. Figures (3.5.1) and (3.5.2) graphically summarizes these processes for an idealized coastal region. Figure (3.5.1) shows the generation of Type I layers for the nocmrnal (residual layer) and daytime cases. Although the figure shows the up valley winds being forced by the sloping land surfaces, in reality, over the flat surfaces of Pitt and Harrison Lakes, the topographical amphfication factor (TAF) between the valley and plain is likely at least equally important. Figure (3.5.2) summarizes the generation of layer above the valley mixed layer; that is, Type JJ JI and IV layers. The dorninant forces of Type III and Type IV are fairly obvious. With Type JJ layers, the specific topographic locale likely dictates which process is most dominant. For example, near the ocean, marine undercutting surely plays a larger role, while upslope venting becomes more important with proximity to the mountains. In terms of air quality, Type I and Type JJ layers are the most significant. Each of these features has the possibility of being down-mixed to ground level if the dynamics of the mixed layer result in the entrainment zone intersecting the polluted layer overhead. Though Type HI and Type IV layers do not appear to directly impact surface air quality within the LFV, they are important in fully understanding both pollution in the LFV and in the free troposphere on global scales. 110 Chapter 4 Transilient Turbulence Model Overview In this chapter, a model of turbulent mixing within the growing daytime boundary layer is developed to investigate vertical ozone fluxes and the resulting influence of the elevated ozone features examined previously in Chapter 3 on ground level ozone concentrations. Specifically, the effects of residual ozone entrained by the developing mixed layer and layers just above inversion top fumigating into the mixed layer in the early afternoon are examined. The model is based on the approach used by Neu et. al., (1994). Turbulence is parameterized using the method of transilent matrices shown in Chapter 2. This method was chosen because alternative mixed layer models based on closing the turbulence equations through first or higher order closure approximations either dorit represent heat fluxes within the mixed layer properly or they dont accurately predict the entrainment through the inversion zone. Secondly, a transilient model is easily driven by the data obtained from balloon profiles, and does not require the solution of a system of differential equations. Consequently, the numerical stability of the calculation is assured. The transilient model constructed here is a 1-D vertical ozone mixing model. It should be noted that although transilient theory can be used to fully predict transfer of Ill heat, momentum and water vapor in three dimensions within a developing boundary layer, data limitations dictated a more simple approach in this case. However, although a fully three dimensional analysis was beyond the scope of this study, processes occurring in three dimensions that will affect the 1-D vertical mixing grid must be considered. The most notable of these are advection in and out of the model domain and the surface fluxes of heat, water vapor and momentum. Unfortunately, the data used in this study were collected without the construction of such a model in mind. As a result adequate estimates of the affect of advection on the meteorological profiles or of the surface fluxes into and out of the grid are not available. There is also not a large enough vertical extent to most of the data to calculate subsidence. For these reasons the model uses measured wind, temperature and humidity at each time step to drive turbulence and adjust the developing mixed layer while the ozone profile is allowed to evolve undisturbed throughout the model run. That is, measured values are used to force the distributions of heat, momentum and water vapor to approximate reality and this information is used to examine the vertical transport of ozone, or any other passive tracer, though out the boundary layer. This of course means that nothing is known of the horizontal advection of ozone in or out of the column or of any chemical transformations or surface dry deposition that may be occurring. Therefore in constructing the model, the following assumptions are made. 1) In the early morning the horizontal gradient of ozone within the boundary layer is smaller than it is in the afternoon because the present day's production has not yet become significant (i.e. examining Figure (1.1.1) in Chapter 1, the diurnal cycles of ozone and its 112 Figure 4.1 Processes Affecting the Vertical Mixing Column Advection (heatozone, water vapor) Surface buoyancy flux 1 Subsidence N i Solar Heating Surface momentum stress precursors from Bunce (1991), in situ ozone production does not appear to be important until the late morning). According to Hastie (1994) advection and production can be related by the idea that "one hours' advection may be thought of as the previous hours' production". This leads to the conclusion that while production in the morning is low, advection should be also 2) In the early morning, the winds in the valley are relatively stagnant as they are shifting between the nocturnal downvalley and offshore regime and the daytime upvalley and onshore flow (i.e. Sakiyama, 1991) 3) The profile sites are far enough removed from NO x sources that morning scavenging as low level NO x reacts with the down-mixing ozone, (i.e equation (1.1.1) in chapter 1), will be minimized 113 Figure 4.2 Sketch of Valley Geometry and its Effect on Advection Through Model Domain. Valley Wall A T Dominant wind direction is up/down valley 1 2 3 4 5 6 7 8 9 Valley Wall The model vertical column is grid 5. If up/down valley winds are dominant, only transfers through 2 and 8 will be important 4) It is assumed that, as the sites are roughly in the middle of either the Pitt River or Fraser River valleys, up or down valley winds will be dominant thus reducing the extent to which advection on the cross axis is a concern. This situation is depicted in Figure 4.2 As seen from above, the model column would be grid 5. If up/down valley winds dommate, only transfers through grids 2 and 8 will be important. 5) On days when ozone levels are significant, the dry deposition of ozone at the surface will be much smaller than fluxes into the surface layer, at the times when the model is valid. Parametrization of the Transilient Matrix Mixing coefficients for the transilient matrix are obtained using the TKE parametrization for mixing potentials from equation (2.2.8) as described in chapter 2. 114 AtTa T\{A,u)\(^v)2-g{Mj^) ( v) l (2.2.8) (evRc) T» The coeffeicients of the transilient matrix are then obtained from equation (2.2.5) (2.2.5) with Ajj-Aji-Yij. Equation (2.2.8) requires a profile of the virtual potential temperature, © v This is found from the potential temperature ®, and the water vapor content, q, from where q is in units of g/kg. As with the parametrization based on equation (2.2.11) for the bulk Richardson number, the differences in equation (2.2.8) are easily obtainable from the vertical profile of ozone and from meteorology, unlike a turbulence spectral parametrization which would require fast response wind speed measurements However, the TKE approach is deemed more appropriate because it has within it a dependence on wind shear, from the mean wind terms, AU $ and AVp, that is not present in the equation for the bulk Richardson number. This is particularly important during the first part of the morning when the layer is still stably sfratified and turbulence is driven mostly by shear. Indeed in the early morning, equation (2.2.11) would imply no mixing at all, because the transilent matrix would go to zero for a stable layer. The TKE parametrization, while it may suggest very little turbulent energy within the layer, can never go to zero because of the empirical constants in the last term of the equatioa Thus, even in a stably stratified atmosphere, equation (2.2.8) will allow for the occurrence of some turbulence, which is of course a more realistic result as even a strong stably stratified nocturnal layer will exhibit occasional bursts of turbulence. 0V = 0 ( 1 + (0.61- q-0.001) (4.1) 115 The coefficients obtained from equation (2.2.8) are adjusted using the rules for diagonalization, symmetry and normalization of the transilent matrix as discussed in chapter 2. Discretization of the Data Profiles The use of transilient matrices as parametrized by equation (2.2.8) requires profiles of wind speed, potential temperature and water vapor content that are discretely spaced in height and time. Data profiles are of course collected as the balloon rises, and as the rate of ascent, or descent varies, (with atmospheric stability, horizontal wind speed etc.), the values obtained are not evenly spaced in height and are collected over about an hour. Profiles must therefore be binned and averaged over a vertical grid and this grid must be interpolated over time at each height to give values for the time required for each model step. Vertical Averaging Creating a vertical grid is a relatively simple matter of binning the raw data into a uniformly spaced vertical grid and then taking the arithmetic average of the values in each grid box. This is done for wind speed, potential temperature, water vapor content and ozone and for decimal time of the samples. Grid size, or averaging interval, is chosen to be of the same scale as the surface layer, so that destabilization of the bottom grid box by surface fluxes is of a similar scale 116 to reality. It must also be small enough to give a good resolution to features in the layer, but be large enough for each grid to contain enough points to make the averaging meaningful. Computing time is also a concern in determining grid size because, as the model is basically a series of matrix calculations, the number of computer operations varies as N squared, (i.e. halving the grid size will roughly quadruple run time). The depth of the surface layer in the study area is on the order of 10m. The grid size was chosen as 20m to ensure enough data in the individual grids. For this size, with the AIR data the speed of ascent and sampling rate give about 3-4 points per grid box. However, occasionally the balloon rise was so fast that no data points fell in a certain grid. In this case, values for the grid are determined by linearly interpolating between the grids immediately above and below. With the choice of 20m, very rarely were there two successive grids that required interpolating. The AES data are sampled much more quickly, but for consistency, 20m is used for these profiles as well. The height of each box is taken as the grid midpoint, (e.g. if the spacing is 20m then the lowest grid is assigned a height of 10m). AIR profiles are averaged up to about 800m (N=40) while the AES are averaged to about 2000m (N=100) Temporal Interpolation Equation (2.2.8) also requires vertical profiles of wind speed, potential temperature and water content that are instantaneous. That is, all values in the model column are for the same time. This is of course not the case for a balloon rising through the atmosphere. Therefore, during the model run, at each time step, an instantaneous 1 1 7 vertical data column for the required model time is obtained by linear interpolating for each profile parameter, at each height, between the values of the two profiles immediately before and after the required model time. Each value at each height is obtained from where: x is the desired model parameter tmod is the model ran time Xj,Xj+i are the values of the desired parameter from the profiles immediately before and after the present model run time respectively tj,tj+i the before and after profile times respectively For instances where the model time is either before the first profile time or after the last, a value is obtained by linearly extrapolating from the closest two profiles using or *(,-) = *j +J-^4(t^ ~ tj) (4.4) depending on whether the model time is before the first or after the last profile respectively. 118 Boundary Conditions The boundary conditions required for the model are the fluxes in or out of uppermost and lowermost grids. For the lower boundary these are: Again, as the data used was not collected specifically for this model, measurements of these quantities were not available to go along with the profiles used. They must then be estimated, or assumptions must be made that will allow their exclusion from the model. The most important of these fluxes is the sensible heat flux, Qh. This is the dominant driving force of the turbulence within the model and, except in the early morning, is the most important factor in the destabilization of the bottom grid layer. With no time series of Qj, measurements from the model site available, this quantity can be fairly accurately estimated using a sine wave and a representative value for the maximum value, i.e. Qh - the surface sensible heat flux Q c - the surface latent heat flux x* - the surface shear stress G*(/»*) = 6*-«-an (4.5) sunset where: t is time of the local sunrise t sunset is time of the local sunset 119 Qhmax is a characteristic value of the maximum heat flux over the study site This is applied to the bottom grid as A * ( 0 = & ( O ( | 9 ( 4 6 > Qh(t) is the sensible heat flux at time t expressed in units of °Km/s. Az and At are the grid size and time step, respectively and A0i is the change in the potential temperature of the bottom grid box during the time step At. An estimate of Q h may be obtained by taking one for a similar site from other heat flux surveys done in the Fraser Valley. Values for tams* and tanm* are obtained from solar tables for the study day. The estimated value of Q j , is adjusted up or down to improve the modeled boundary layer evolution as compared to reality. This is done to account for factors that are not handled by the sine wave estimate, (e.g. mountain shadows, passing clouds, advection etc). An estimate of the momentum flux into the ground is obtained by using equation (2.2.4b) for the flux from the transilient matrix and the measured interpolated wind profile to calculate the flux between the first and second grids (i.e. the momentum flux through the plane' at 20m for the previous time step). This value is then used as an estimate of the surface momentum flux in the next time step. This is applied to the bottom grid value of the wind speed in a manner analogous to equation (4.6) An estimate for the surface latent heat flux can be obtained in the same manner as for the momentum flux. However, model runs using this method were found to produce too much water vapor in the layer. For this reason, evapotranspiration is ignored as a 120 factor in the destabilization of the bottom grid during each time step. Changes in the water vapor content within the layer are still accounted for in the real measurements of humidity that are part of the input profiles. The water vapor content is only important in the calculation of the virtual potential temperature profile that is necessary for evaluation of equation (2.2.8). Neu et. al., (1994) claim that ignoring the flux of water vapor for a given step causes at most a 3% difference in the mixing potentials. Adjusting the Model Potential Temperature Profile As stated previously, measured profiles are used through the model run to adjust the results for factors not included in the model construction. This is done by periodically taking an arithmetic average between the modeled potential temperature profile and the measured interpolated profile for the same time. It is important to again mention that the modeled ozone profile is not adjusted during the run, only the meteorological profiles that control the mixing of ozone witfiin the layer. The underlying conceptual model of the transilent theory is that of atmospheric thermals being created at the surface and then rising through and mixing with the rest of the layer. The smoothing interval is then chosen to be long enough to allow several of these to be created and mixed out before the profile is readjusted. This allows the model atmosphere and the output profiles produced, (i.e. heat and ozone fluxes), enough time to settle and smooth themselves out. The time scale, U, for a change to propagate through the layer can be estimated using 121 h (4.7) where h is the height of the layer and w» is the characteristic vertical velocity scale within the layer given by Typical values for the Lower Fraser Valley are h = 800 m, w* = 1.75m/s (with 0 — 300° K, y^ff = .20 mK7s (Steyn, 1980). The time scale is then 450 seconds or about 6 minutes. The averaging interval was then chosen as between 15-30 minutes to allow for several 'thermals' to be generated and rnixed out. The value of the time step is of central importance in determining mixing potentials because it is included explicitly in equation (4.6) for the adjustments of the surface fluxes to the bottom grid, and is therefore implicitly included in equation (2.2.8) for and subsequently in the coefficients on the transilient matrix cy. If the time step is too small, the destabilization will be too small and turbulence will be suppressed; too large and the mixing potentials will be overestimated. Stull (1988) recommends that the ratio of grid height to time step, h/t, be greater than 0.1 m/s. With a grid height, h, of 20m the value of must be less than 200s. In most cases, a time step of 60 seconds is used. This is small enough to satisfy the above condition and large enough to minimize the model run time. (4.8) Time Step 122 Model Outputs The model generates vertical profiles of ozone and potential temperature. It can also produce the interpolated profiles of water vapor, wind speed and measured potential temperature and ozone at any time. The coefficients of the transilient matrix for a given time step may also be output. Profiles of vertical flux can also be calculated using the transilient matrix and equation (2.2.4b) for the vertical flux of a given tracer at grid level (k). ^ ( * ) = £ Z (2-2-4°) where Q is the profile quantity being mixed. Using this equation, vertical profiles of fluxes of ozone, temperature and water vapor are calculated. The value of the ground level ozone implied by the model may be estimated by using the ozone concentration in the bottom grid. This is really an estimate of the average ozone between the ground and a height of 20 m and is perhaps more accurately called the surface layer ozone. However, for purposes of comparison with ground level ozone measurements from stations within the GVRD network, this value is deemed the modeled ground level ozone. Model Algorithm These elements are incorporated into the final model code according the algorithm summarized by the flowgram of Figure (4.3). The model is coded in FORTRAN77. The 123 subroutines that do the linear interpolation between profiles, calculate the transilient matrix and the vertical fluxes and perform the transilient mixing are supplied in Appendix B. Firstly, the averaged meteorological data profiles and model run parameters are read from file. The run parameters include the grid height and time step size; the value of Qh and time of sunset and sunrise required for the heat flux paramaterization; and the model start and end times and averaging interval. At this point the times for which model outputs are desired are also specified. Next, the model ozone and potential temperature profiles are initialized. For the starting ozone profile the descent profile closest to the model start time is used. The potential temperature is initialized using the interpolated field for the model start time. The model now enters the loop over time through which the boundary layer simulation is advanced. The model runs from the starting time, tstart, to the end time, tend, advancing by interval of tstep. At each step, an interpolated field of instantaneous data profiles is calculated from the input averaged profiles. This is done by the routine ntrp.f and the function serintf listed in Appendix B. Next, the boundary conditions are applied to destablize the model potential temperature profile and initiate turbulence. This is the first 'half of the step as described by "Le Chatelier's Principle" in Chapter 2. The virtual potential temperature profile is then found using equation (4.1) with the modeled potential temperature and the measured interpolated water content. From this profile and the interpolated wind speeds the transilient matrix for the current time step is determined as per equation (2.2 8). This is done in the routine trans./. 125 This matrix is then used to calculate the turbulent fluxes of heat and ozone for the the current step from equation (2.2.4b). This done by the subroutine_^ acc./ Now, the 'second half of the step, the turbulent mixing, or restablization, is performed by applying the current Transilent matrix to the modeled ozone and potential temperature profiles using equation (2.2.2). This completes the time step and gives the new model ozone and temperature profiles. The nibring is performed by the subroutine mix.f. At the specified intervals, the modeled and measured ozone and potential temperature profiles, the model flux profiles and the current Transilient matrix are saved to a file, and the averaging, or 'nudging', of the measured and modeled potential temperature profiles is performed. Lastly, the condition of the time loop is evaluated. If the current time is before the model end time, the model is advanced by the interval timestep and the next turbulent step is simulated. Otherwise, the loop is exited, the output files are closed and the model run ends 126 Chapter 5 Model Results: Vertical Down-Mixing of Elevated Ozone Layers In this chapter the transilient model of Chapter 4 is applied to two distinct elevated ozone layers observed over Harris Rd. on August 6, 1993. Firstly, the model is run starting hi the early morning on a Type I layer that is a remnant of the residual layer from the previous night and, secondly, on a Type JJ layer that was advected over me Harris Rd. site in the early afternoon. The results of both model runs show that significant amounts of ozone from each of these layers are errtrained into the mixed layer below and down-mixed to the surface and that the influence of both of these cases of downward transport is clearly reflected in the ground level ozone concentration time series for this day. Meteorological Context August 6 was the final day of the episode week of Pacific'93. As mentioned in Chapter 3, the synoptic forcings present during Pacific '93 were not as well developed, in the sense of producing stagnant anticyclonic atmospheric conditions conducive to the accumulation of boundary layer ozone, as generally occur during more typical episodes (Pottier et al, 1997). The 500 mb ridge was located too far off the west coast of B C. to establish a typically stagnant air mass (Steyn et. al, 1997). During the week, conditions were most conducive to ozone production on the 5th. Subsidence under the strengthening 127 ridge brought clear skies and warm temperatures. Highest concentrations for Pacific '93 were observed on this day with a reading of 83 ppb recorded in the central LFV and concentrations of 111 ppb observed over Pitt Lake (McKendry et. al, 1997). On the morning of the 6th, the east side of the ridge collapsed, to be replaced by a light and variable cyclonic circulation (Pottier et. al, 1997). This was accompanied by a low level surge of marine stratus from a northward propagating coastatty trapped disturbance. Such disturbances are quite common along the west coast during the summer months and are associated with the transition from warm and sunny anticyclonic conditions conducive to the accumulation of low level ozone to cooler cloudy conditions with a more southerly now. Despite the collapsing of the ridge, wind in the area remained light, around 2-3 m/s, from the south to southeast over most of the lower atmosphere. This permitted short trajectories that could keep pollutants in the LFV for the period of a day. Interestingly, even though the synoptic forcings had weakened, August 6 still exhibited fairly high ozone concentrations, with surface readings at Harris Rd. exceeding SO ppb in the early afternoon and clear examples of Type I and Type II layers manifested in the atmosphere above. (McKendry et. at.y 1997) Vertical Ozone Distribution at Harris Rd. on August 6,1993 Data used for the model study are the AIR tethersonde profiles from Harris Rd. for August 6, 1993. Due to problems with hysteresis discussed in Chapter 2, descending profiles only are used. The first flight for the case day was from 07:20 PST in 128 Figure 5.1 Descent Ozone and Meteorology From Harris Rd. 08:00 PST, August 6 t h, 1993. Wind Direction (azimuth) 0 1 2 3 4 5 6 7 Wind Speed (m/s) 8 . 5 1000 800 600 400 200 0 9 T" Specific Humidity (g/kg) 9.5 10 "ozone" "potentiaLtemperature" "specific_humidity" J_ 10.5 35 40 45 50 55 Ozone Concentration (ppb) and Potential Temperature (degrees Celsius+30) 129 Figure 5.2 Time-Height Section of Ozone From Harris Rd August 6*. 1993. Contours are ozone concentration in ppb. Concentrations higher than 80 ppb are shaded to highlight the appearance of the early afternoon Type II layer. 7 9 11 13 15 17 19 T I M E the morning and flights continued approximately hourly throughout the day. The last balloon flight on this day was at 18:15 PST in the afternoon. The data reveal the presence of two distinct elevated layers of differing origin appearing at different times that each have the potential to increase on surface ozone concentrations. Figure (5.1) shows the tethersonde descent ozone and meteorology for 8:00am PST on August 6th. Though this profile is approximately 2 hours after sunrise and mixed layer development is well under way, a residual Type I layer is still present Ozone concentrations are below 40 ppb from the surface up to about 250m and rise to over 50 ppb at 300m. This height corresponds to a weak inversion in the potential tennperature profile at 300m that likely marks the height of the developing mixed layer. Above this at 500m is a second stronger inversion that is 130 f i g u r e 5.3 Descent Ozone and Meteorology From Harris Rd. 1237 PST, August 6* 1993. 8.5 1000 800 600 400 200 Specific Humidity (g/kg) 9.5 10 10.5 1— T ozone 30tential_temperature" "specific_humidity" T i i 11 11.5 50 60 70 80 90 Ozone Concentration (ppb) and Potential Temperature (degrees Celsius+50) 1000 800 f-600 400 h 200 h 90 Wind Direction (azimuth) 180 T 270 "wind_speed" "direction" 360 Wind Speed (m/s) 131 the top of the residual layer and the previous day's mixed layer. In the developing mixed layer, winds are southerly at 4m/s and are associated with the low level stratus surge, indicating that marine undercutting and overnight radiative cooling are the dominant processes in creating this layer. Within the remnant of the residual layer winds are much lighter at under 2m/s. Again this is significant because low winds at this level mean elevated pollutants may remain within the LFV despite the fact that synoptic conditions are no longer conducive to in situ ozone production. In the early afternoon another elevated layer appeared over Harris Rd. Figure (5.2) shows the ozone time-height section for August 6 at this site. An apparent Type 13 layer is clearly visible developing around noon. Figure (5.3) shows the descent ozone and meteorology profile at 12:37 PST. The mixed layer is now well developed to about 600m with ozone approximately 50 ppb from the surface up to about 500m. Across the entrainment zone ozone jumps sharply to over 80 ppb. As with earlier in the day, winds were fairly strong from the south within the mixed layer, but weakened above the inversion arid shifted to a more easterly direction. Again, mixed layer conditions on the 6th were not conducive to producing ozone concentrations as high as 80 ppb and winds at this level indicate that this was likely an aged air mass that had advected back over Pitt Meadows from a source somewhere in the eastern LFV. Model Parameters As detailed in Chapter 4, user defined variables for each run were the maximum heat flux, time of sunrise and sunset, the model run start and end times, the time step and 132 Table 5.1 User Defined Model Parameter Values maximum heat flux: 0.15 Km/s sunrise: 5.867 decimal hours sunset: 20.75 decimal hours timestep: 60 seconds averaging interval: 15 minutes the profile averaging interval. For each of the Type I and Type H runs values for the user defined variables were as Usted in Table (5.1) above. Sunrise and sunset in Vancouver for August 6, 1993 were 5:52 PST and 20:45 PST (source: the Vancouver Sun). Heat flux was chosen as a characteristic estimate of the kinematic sensible flux over a grassy site for a fairly cool and cloudy day in August at Vancouver's latitude and was adjusted in an iterative manner to better approximate, in a qualitative sense, the development of the model mixed layer as compared to the real observation mixed layer. The times and flux values are for a flat plane at Vancouver airport and do not account for the presence of mountain shadows, topographic heating effects or passing clouds, but without direct heat flux measurements at the site these values are the best estimates possible. For the Type I model run, the model ozone profile was initialized with the 8:00 PST measured descent profile and potential temperature with the measured interpolated profile at this time. This run was started at 8:00 PST and carried through to 13:00 PST. The Type U model run used the 12:00 PST measured interpolated profiles to inWalize both the ozone and potential temperature profiles. This run represented the time period 133 Figure 5.4 Hourly Sequence of Modeled and Observed Potential Temperature Modeled Observed 800 700 600 500 -400 -300 -200 -100 -288 09:00 PST entrainment zone 289 290 291 292 293 Potential Temperature (degrees Kelvin) 294 £2 (D a> E, ri> to 3= _g> a X 800 1 1 - 1 \^ A' I s 700 1 / — 600 I f -500 - -400 - J "» entrainment -300 1 s j zone 200 — c \ v. I / I 100 0 — \ \ K. i I I 1 289 290 291 292 293 294 Potential Temperature (degrees Kelvin) 295 800 700 600 500 400 300 200 f -100 11:00 PST aitrainment zone I 290 291 292 293 294 295 Potential Temperature (degrees Kelvin) 296 <D E. CO a> X 800 I 1 1 I s^r ~-700 — 600 — j***"" \ entrainment 500 - s*Z>^' l zone _ r 400 • 300 / / 200 v. i t . — 100 • i — 0 i V. i i i 290 291 292 293 294 295 Potential Temperature (degrees Kelvin) 296 12:00 PST 13:00 PST 800 700 •S, 400 100 I 1 .1 - - • - • • .r'' / > entrainment J zone — • -> -\ . ' < . ^ " * ^ > -~~ *v • ~" s \ — t v. \ V , I l l 291 292 293 294 295 296 Potential Temperature (degrees Kelvin) o E £ O) CD X 297 292 293 294 295 296 Potential Temperature (degrees Kelvin) 297 134 from 12:00 PST to 15:00 PST. Transilient parameters of the TKE parametrization of equation (2 2.8), as derived in Appendix A. 1, were chosen as T D = 1000s, Y^f = 1000, D = 1.0 and Rc= 0.21 as suggested by Stull (1993) and outlined in chapter 2. For a discussion Of the senshivity of transilient nrixing on the values of these parameters the reader is referred to Stull and Dreidonks (1984) Mixed Layer Development Modeled vs. observed mixed layer development for August 6 is given in the hourly sequence of profiles displayed in Figure (5.4). Once several 'thermals' have been generated and mixed out, (e.g. starting with the 09:00 PST profile), the model produces an idealized mixed layer with % tending to zero, though sHghuy superadiabatic. A distinctly superadiabatic surface layer is associated with the bottom grid box. In the early hours, the model slightly overestimates potential temperature below the inversion and underestimates it above. The two profiles achieve a better convergence in the early afternoon. However, for purposes of estimating the downward fumigation of elevated ozone layers, the most important feature is the height of the entrainment zone. For this variable the model shows strong agreement between observed and calculated values. Type I Model Results The hourly sequence of modeled vs. observed vertical ozone profiles for the Type I model run is shown in Figure (5.5). Figure (5.5f) shows the starting ozone profile from 8:00 PST and the 12:00 PST modeled and observed profiles. At ground level, this model 135 suggests that the change in concentration due to vertical down-mixing accounts for approximately one half of the total increase in surface ozone. The remainder is most likely a result of the advection of ozone into the area as chemical production on this day was minimal. It should also be noted that the morning down-mixing of ozone may in fact be significantly greater than that suggested by Figure (5.5) as this simulation starts well after the days' mixed layer development has already begun. Several studies, most notably Neu et al (1994), indicate that down-mixing is strongest in the hours immediately after sunrise. The model sequence shows that in the upper portion of the mixed layer between 300m and 600m, vertical mrxing causes a reduction in ozone concentration as elevated pollutants are transported from the polluted air mass above down into the cleaner air below. The effect at each model step is to essentially redistribute or vertically 'average out' the total ozone below the height of the mixed layer. The process of entraining and redistributing ozone from above the mixed layer is supported by the hourly sequence of modeled vertical heat and ozone fluxes as shown in Figure (5.6). The heat flux profile shows the surface value as determined from the sine wave parametrization, reducing with height and showing a negative crossover below the top of the mixed layer associated with the entrainment of the inversion above. Modeled ozone flux is negative throughout the mixed layer indicating downward ozone transport at all levels. For each hour it shows a peak magnitude that corresponds to the negative heat flux maximum at the top of the mixed layer which then decreases linearly to zero at the ground. This confirms again that the dominant process described by the model is that of entraining ozone laden air from the polluted layer above and redistributing the ozone under me mixed layer height. As the mixed layer height stabilizes around noon, 136 Hourly Sequence of Modeled and Observed Vertical Ozone for the Type I Elevated Layer Modeled Observed 08:00 PST 09:00 PST 40 45 50 Ozone Concentration (ppb) CD m £. ri> cd £ cn CD I 30 40 50 Ozone Concentration (ppb) 10:00 PST 11:00 PST 50 60 Ozone Concentration (ppb) 12:00 PST 60 70 80 Ozone Concentration (ppb) CD s O) cd £ D) CD I 2 CD E c i cd £ g> "CD I 40 50 60 70 Ozone Concentration (ppb) 800 700 600 500 400 300 200 100 0 0800 PST Observed and 12:00 PST Modeled and Observed 30 T T •08:00PST_Observed" - — — "12:00PST_Modeied" - | "12:00PST_Observed" • 50 60 70 Ozone Concentration (ppb) 80 90 137 Figure 5.6 Hourly Sequence of Modeled Vertical Heat and Ozone Fluxes for the Type I Elevated Layer Heat Flux Ozone Flux 52 co CD E_ D) CO JZ D> '<B X -0.002 800 09:00 PST Vertical Ozone Flux (ppbm/s) -0.001 0 T 0 0.02 0.04 0.06 0.08 Vertical Kinematic Heat Flux (Km/s) 0.001 to © E. di to £ 2> 'CD X 800 700 600 500 400 300 200 -0.4 10:00 PST Vertical Ozone Flux (ppbm/s) -0.3 -0.2 -0.1 0 -0.05 0 0.05 0.1 Vertical Kinematic Heat Flux (Km/s) 0.1 1 I I I l | • -v. i 0.15 800 700 _ 600 <D 2 500 D) td JZ O ) 'co X 400 300 200 [-100 h 11:00 PST Vertical Ozone Flux (ppbm/s) -o.5 •' ; 0 T -0.1 0 0.1 Vertical Kinematic Heat Flux (Km/s) 0.2 -0.03 800 12:00 PST Vertical Ozone Flux (ppbm/s) -0.02 -0.01 0 T -0.05 0 0.05 0.1 Vertical Kinematic Heat Flux (Km/s) 0.01 0.15 138 the magnitude of the downward ozone flux decreases as less and less of the overhead layer is entrained. Type II Model Results The hourly sequence of modeled and observed ozone profiles for the Type n layer starting with the observed interpolated profile for 12:00 PST is shown in Figure (5.7). Again, approximately half of the increase in surface ozone may be attributed to downmixing of the elevated layer. Also the Type II flux profiles, (Figure (5.8))^  confirm that entrainment of polluted air and of warmer inversion top air occurs across the top of the mixed layer ami that the entrained air is then redistributed evenly below the mixed layer height. Further, measured profiles between 12:00 and 14:00 show very little change in ozone concentration at heights from 550 to 700m. This indicates advection at this height was not significant over this time and offers solid evidence that advection was not the dominant cause of increased concentration at ground level over the same time. As with the Type I model run, the magnitude of the downward ozone flux decreases toward the end of the model run as most of the ozone advected into the region of the upper mixed layer has already been redistributed by turbulent mixing and, in this case^  as the destabilizing surface heat flux begins to decrease in the afternoon Despite neglect of chemical or advective processes within the model, results for August 6th suggest two separate time frames, (early morning and early afternoon), during which down-mixing of ozone from a Type I and a Type II layer respectively, significantly influences ground level concentrations. These two separate events are reflected in the 139 Figure 5.7 Hourly Sequence of Modeled and Observed Vertical Ozone for the Type II Elevated Layer Modeled Observed • 12:00 PST 13:00 PST i2 2 a> E. d> cd JZ g> "<D X 800 700 600 500 400 300 200 100 0 1 7 \ - ( x \ > ) J i ^ \ \ -s . -\ " * • 1 / / / . i —m 1 \ < •s. • / • 1 1 1 50 60 70 80 Ozone Concentration (ppb) 90 50 60 70 80 Ozone Concentration (ppb) 90 140 Figure 5.8 Hourly Sequence of Modeled Vertical Heat and Ozone Fluxes for the Type II Elevated Layer. Heat Flux Ozone Flux 800 - 700 600 a> a> 500 E ti> 400 CO 300 ai CD X 200 100 0 13:00 PST Vertical Ozone Flux (ppbm/s) -0.4 -0.3 -0.2 -0.1 0 1 — T -0.05 0 0.05 0.1 Vertical Kinematic Heat Flux (Km/s) 0.1 0.15 14:00 PST Vertical Ozone Flux (ppbm/s) -0.4 -0.3 -0.2 -0.1 0 ~]—r 0 0.05 0.1 Vertical Kinematic Heat Flux (Km/s) 0.1 0.15 141 Figure 5.9 Ground Level Ozone vs. Time at Harris Rd. August 6 t h, 1993. 60 I r 1 1——1 i 1——" " r I • l i t i j i i L — _ i _ — i 1 0 4 8 12 16 20 24 TIME (PST) surface concentrations observed at Harris Rd. on August 6* as shown in Figure (5.9). The ground level trace shows three distinct phases events as the concentration of ozone rises: first, a sharp increase of 9.5 ppb/hr for 2 hours in the early morning, a period of gentler increase of around 3ppb/hr for 5 hours and a second period of sharper increases of about 5 ppb/hr for 3 hours in the early afternoon. Model results suggest that the times of the two periods of sharper increases correspond to the two phases of down-mixing outlined above. That is, ozone concentrations rise quickest in the early morning as the nocturnal inversion is eroded and the growing mixed layer begins to entrain residual layer pollution. Concentrations then increase more slowly with reduced in situ chemical production under the cooler, cloudier atmospheric conditions of the 6th . Then in the late morning as the 142 mixed layer depth stabilizes and concentrations again rise as the Type H layer is adverted overhead and then fumigated throughout the layer. Sensitivity Analysis The only independent user defined variables in the transilient turbulence model are the maximum heat flux, the size of the time step and the averaging intervals. All other parameters are either based on recommended values, such as the empirical transilient constants To, Y«f, D and Rc, or are constrained by the assumptions made in the model construction, (e.g. sunrise, sunset and grid size). Even the time step is recommended, as in Chapter 2, to be chosen so that the ratio of the grid size to time step, Az/At, is greater than 0.1 m/s. In all cases attempted in this study the transilient model overestimates the surface layer temperature. However, as mentioned previously^  for purposes of modeling the downward entrainment of elevated polluted layers, the most important model feature is the time evolution of the height of the mixed layer. Appendix C shows several sequences of modeled vs. observed mixed layer for several values of the heat flux, time step and averaging interval. Of the three independent variables , the parameter with the smallest effect on the model results appears to be the time step. As long as At is chosen to satisfy the above condition, there is very little difference between model runs, though a longer time step does provide for a slightly deeper model mixed layer. The longer step size probably allows 143 for a greater destabilization in the first 'half 'of each step by giving a larger flux of sensible heat into the bottom grid. Results are not as dependent on the heat flux as one might expect. With the periodic vertical averaging of the observed and modeled profiles in place, a larger flux does not dominate the mixed layer development because the averaging will 'hold back' any exaggerated heating of the model layer and once the 'idealized' model profile has formed about an hour after the model starts, the smaller heat flux is still sufficient to keep the layer well mixed. The lower value of 0.10 Km/s actually appears to give better agreement between observed and modeled results than the value of 0.15 Km/s used in the study, but this is again because with the averaging in place the only factor causing the modeled and observed potential temperature profiles to differ is the sensible heat flux into the bottom grid. If this is reduced then the two profiles will of course converge. Indeed, the parameter with the most influence on the model results appears to be the averaging interval. With the interval increased or removed altogether, even the chosen value of 0.15 Km/s for the surface heat flux results in a model mixed layer that is too deep or too warm. The fact that the process of adjusting the model mixed layer by using the observed mixed layer data is so important to the model results shows that the 1-D grid model domain is influenced more strongly by advective processes than was assumed in Chapter 4. It is also possible that the assumptions did not hold at the times when the model was actually run. For example, the model was designed to be an approximation of down-mixing at or near sunrise, when neither the up-valley nor down-valley flow regimes are well developed. In reality, the earliest model output is for at least two hours after sunrise when this is likely not the case and it is certainly not true of the Type H run which 144 starts at 12:00 PST. As a consequence the model results are perhaps best interpreted not as the actual amount of downward transport that is occurring but rather as the potential degree to which each pollutants from each layer may be down-mixed to the surface. Summary of Results Results for the morning down-mixing of a Type I residual layer are similar to those of by Neu et. al. (1994). That study found 50-60% of the diurnal rise in surface ozone could be attributed to down-mixing of residual layer pollutants. However, comparison between these two studies is complicated by the fact that the Type I simulation began well after sunrise, while the model of Neu et. al. (1994) was initiated at dawn. Neu et al. (1994) showed the strongest down-mixing immediately after dawn and comparison of the model surface concentration with the measured ground level trace showed that virtually all the observed rise in surface concentrations was from down-mixing. Though the Type I results also show that around half the rise in surface ozone can be accounted for by down-mixing, the agreement in modeled and observed surface values as they rise in the early morning is not as strong for the Type I simulation as it is with the Swiss study. This may be because the Type I simulation omits the period of strongest down-mixing in the early morning, or there may be other processes (i.e. advection, in-situ production) that were present during the Type I down-mixing that did not become important in the Swiss study until after mixed layer growth was complete. The Type II simulation illustrates a process that Neu et. al. (1994) did not model: Specifically, the down-rnixing or fumigation of an elevated layer located just above the 145 mixed layer inversiori. (i.e. a Type JJ layer). This process has been described for elevated layers over the L.A. basin (Ulrickson and Mass, 1990). Results here show that fumigation of a Type JJ layer may significantly affect ground level concentrations, even after the mixed layer has reached its maximum depth in the early afternoon. Chapter 3 shows that Type II layers are ubiquitous in the LFV under episode conditions and ozone concentrations wrthin them are often very much higher than those within the mixed layer below. These layers represent a very large reservoir of pollutants. Given the ubiquity of Type II layers^  down-mixing Of this reservoir of pollutants is likely a very important process in LFV ozone episodes. 146 Chapter 6 Conclusions Vertical Ozone Distribution in the LFV In this study 105 profiles from aircraft, tethersonde and free ascent balloons obtained during the period of 1993-1996 were examined for the presence of elevated ozone layers. Results show a rich variety of structures present from the boundary layer up to the upper free troposphere. Applying the criteria for defining elevated ozone features developed in Chapter 2 a total of 110 elevated features were identified within the profiles. These layers were typed and sorted according to the classification scheme presented in Table (3.1.1). This scheme attempts to categorize layers according to the altitude of the layers with respect to significant atmospheric boundaries, such as the height of the mixed layer, and the specific processes of generation as inferred from other studies performed in similar locales. Layers were identified as Type I, II, m or IV with Type I layers located within the mixed layer ranging up to Type IV layers in the upper troposphere. The richness of structure as demonstrated by the presence of these varying layer types is 147 shown to be a result of a multitude of interacting generation processes, ranging in scope from the local up to synoptic scales. Specifically, Type I layers are formed predominately by local scale radiative cooling that produces a surface based inversion that decouples this surface from the Type I layer (or residual mixed layer) overhead. Slope and sea breeze flows appear to be of great importance in the creation of Type U and UJ elevated pollution layers, while trajectory analysis of upper level Type IV features indicates these layers are present over the L F V as a result of long range transport of free tropospheric ozone, perhaps on continental scales. In all cases, occurrence of these layers appears to be strongly correlated with the occurrence of atmospheric inversions. These inversions either inhibit the dispersion of pollutant or they mark the boundary between air masses of different sources. In terms of L F V air quality, Type I and II layers are shown to be the most significant. These layers exhibit the highest concentrations of ozone, particularly the Type II, and both exist at heights, either just above or below the top of the mixed layer, respectively, where the pollutants within them have the potential to be entrained into the mixed layer and possibly fumigated to ground level. Type IU layers exist too high above the mixed layer to pose a significant threat to surface air quality. They provide clear evidence of the scale on which ozone pollution from emissions predominantly within Greater Vancouver occurs within the L F V . Trajectory analysis of Chapter 3 shows that even under relatively weak atmospheric conditions, circulation within the L F V may become closed up to 5000m a.g.l. and the AES and Convair profiles show Type HI layers at heights over 4000m. Both Type JJ and Type HI layers are observed at locales, such as 148 Harrison Lake and Hope respectively, that are on the order of 100 km downwind from Greater Vancouver. This suggests pollution on horizontal and vertical scales that is usually associated with what may be thought of as heavily polluted areas such as Atlanta (Fishman et al, 1987) or Los Angeles (McElroy and Smith, 1993). Vertical Down-mixing of LFV Ozone Layers Of the various layer Types identified and examined in Chapter 3, the two with the most significant implications for ground level air quality are Type I and H Both of these occur at levels in the atmosphere where the may be intercepted by the top of the mixed layer and the pollutants within them entrained and fumigated to levels below. Chapter 2 examined the nature of atmospheric turbulence and demonstrated that Transilient Turbulence Theory (Stull, 1988,1993) is an appropriate mathematical model by which to examine this potential transport. In Chapter 4 a simplified transilient model, modified to be of use with available data, was constructed for this purpose. A case day of August 6 t h 1993, was chosen to test the model. On this day, both a Type I and Type II layer were visible in the AIR tethersonde profiles over Harris Rd. Furthermore, ground level concentrations exceed 50 ppb on a day not particularly conducive to high ozone concentrations. The transilient model showed that down-mixing of elevated pollutants at the lower levels of the atmosphere does in fact significantly influence ground level ozone. For both the Type I and Type II layers, the model shows as much as 50% of the increase in surface level ozone over the duration of the model run may be attributed to down-mixing of the specific elevated layer. Further, for the Type I 149 layer this is quite possibly an underestimate of the influence of the elevated layer as the model starts about two hours after sunrise at a time when the case day's mixed layer is about 300m deep and large amounts of ozone have already been mixed to ground. In general, the overall effect of the model at each stage of its development is to take the amount of ozone from the starting profile that is below the current mixed layer height and redistribute or average it across the entire depth of the mixed layer. As a result of this averaging, the most important feature of the model in determining the amount of down-mixing that has occurred at any one time is the height of the mixed layer. Consequently, an estimate for the potential for down-mixing may be obtained from the ozone profile showing the residual or Type II layer and a simpler model for the mixed layer height such as that outlined by Steyn (1980). As the model is constructed, there are very few'independently chosen model parameters. The sensitivity analysis of Appendix C shows that the model is relatively stable to changes in the magnitude of the surface sensible heat flux and size of the time step and is most sensitive to how well the model accommodates actual observations of the mixed layer. That is, the interval at which the modeled and observed potential temperature are averaged. This strong dependence of the measured potential temperature field on the model development demonstrates that the simplified one-dimensional model grid as outlined in Chapter 4 is actually much more strongly influenced by up and down valley advective processes than was assumed during model construction or that the assumptions did not hold during the times when the model was actually run. 150 Implications/Further Research In general, uncertainties associated with generation and origin of the various layer types increases with their respective height above ground. Type I layer generation is the best elucidated. With 21 occurrences within the data set, they are less frequent than either type TJ or TV layers, but they exist at lower levels and are usually observed within the tethersonde profiles. Most examples therefore include both ozone and full meteorology. However, the Type I data set is sparse with respect to measuring these layers when they are most significant. Specifically, at night when the residual layer is best developed and in the hours during and immediately after sunrise when the nocturnal inversion is being eroded. In addition, the daytime case of the Type I layers is something of a rarity, with only a couple of examples observed over Pitt Lake. Whether these layers are more common than this suggests and whether or not they are also present over Harrison Lake is not clear. For modeling purposes, a series of profiles from before sunrise through to the afternoon along with specific measurements of the various surface fluxes would allow for a better transilient model of the early morning residual layer down-mixing. This would help to confirm the assertion that as the Type I model of Chapter 5 did not begin until well after sunrise, Type I layer down-mixing in the early morning may be under-estimated by the model results. Type U layers are the most prevalent with 34 examples present in the data. However, many of Type U layer examples are from light aircraft flights and these profiles show ozone, water vapor and potential temperature but unfortunately do not provide any 151 information about wind speed or direction. This information would greatly help elucidate some of the questions about Type II layer origins. For example, whether the Type II layers that occur over Pitt Meadows and Harrison lake display the same subtle northerly wind shift as do those visible in the AES profiles and whether winds over Boundary Bay confirm that Type II layers there have been transported back ocean ward from somewhere further inland. At higher levels, the data set of Type HI layers is very sparse, with only 7 instances recorded and all of these occurring during the latter half of the episode week of Pacific '93. It is unknown if these are ubiquitous features of L F V episodes or i f they are a phenomenon that is a special occurrence associated with the atypical conditions of Pacific '93. If these features are commonplace, they likely exist in a more distinct form with even higher pollution levels during more stagnant synoptic conditions. In addition, further measurements of other atmospheric pollutants such as NO, , GO or other carbon compounds would help to determine the age and pinpoint the origin of these layers. Upper level Type IV features are present throughout the AES profiles and their occurrence appears to be unrelated to lower level air quality. The source of the layers that appear over the L F V is not clear. Back trajectory analysis of one example in Chapter 3 shows only a brief crossing of Vancouver Island in the past three days, and it is not likely that emissions from this region could have reached the upper troposphere in so short a time. These layer may be related to the continental scale ozone clouds that are present over portions of the Northern Hemisphere during summer (Fishman et al, 1991) .Again, measurements of oxides of nitrogen and carbon would help determine the photochemical age of these air masses and their boundary layer sources. 152 Model results show elevated layers are indeed an influence on ground level concentration and an accurate fully three dimensional mesoscale prognostic model of the L F V airshed will have to include the effects of these features. Results also show that though transilient theory is an appropriate means by which to model atmospheric turbulence and gives an estimate of the amount of downward transport, the one-dimensional model is probably only accurate in the early morning hours. At other times advective processes are too important and the assumptions made in the one-dimensional model construction are too extreme. For more representative results, the full three dimensional transilient scheme including advective processes will be required and should be nested within a regional mesoscale model such as R A M S . 153 Appendix A Derivation of the T K E parametrization for the transilient mixing potential (From Stull and Driedonks, 1987) Consider the equation for the turbulent kinetic energy, E, in simplified form, neglecting the transport terms. — = —({/' w ) — -{v'w')— + ^ -{W0- ) - s ( A . l ) a di di 0, 1 Where: U, V are the Cartesian velocity components («V) , (v ' iv ' ) are the kinematic momentum fluxes Q ~ —(w'f9;) is the kinematic buoyancy flux 0V 8 is the rate of dissipation and 0V is the virtual potential temperature Terms I and II represent the mechanical production of turbulent kinetic energy and III is the buoyant production term. Any Static or dynamic instability between any two grid points i and j wil l generate turbulent on that same scale, I j - i I Az , in order to remove the instability. A fraction of this energy is dissipated through e. Now let Eij be the turbulent kinetic energy on the scale |j - i | Az , and (u'w')^ be the turbulent flux on the same scale. One can then write a non-local analogy to (A. l ) A ^ . 7 _ 77T7-.V-AL/ - T T - A V , g A,r v /,J Az " Az 0 154 A t denotes a difference over time and Ajj a difference over space in the vertical levels i and j . Integrating over time and normalizing by •£/,- gives M i {u'w')9 f AU Az (v'w1).. (AV Az + g W ) „ £ V • 9„ Eu A,t (A.3) Now define three scaling parameters: a time scale of turbulence, T 0 ; a dimensionless parameter, Re, relating buoyancy to shear; and a dimensionless dissipation, A as follows: Eu Az r„ (A0V K \ Az E. T •J o (A.4a) (A.4b) (A.4c) (A.4d) This closes the equations at first order. The right hand side of A.3 are the forcings that generate the turbulent energy between grids, Ey. Conceptually, the right hand side of A.3 is then the 'mixing potential' energy between the grids. If we denote Yy as the mixing potential between any two grid boxes i and j , with i * j , and substitute A.4 into A.3 we have. 155 AU Az + Z AV 0„ / ? , V A z J , T„ IJ V c \A.t (A.5) Which simplifies to y = T0At " ~ (Az) | (AU)lMAV)l-[-^-](A0v)ij Z)Af For i ^ j (A.6) 156 Appendix B Fortran Subroutines Nrtp.f subroutine ntrp(d,In,ti) *** given the profiles^  finds interpolated field at given time Pajameter(NgridMAX=200,NprofM AX=40,NparMAX= 15) Real d(NgridMAX;NprofMAX,Npar^ Real ser(NprofMAX,2) Integer NgridMAX,NparMAX,NprofMAX Integer Ngrid,Nprof,Npar Gommon /one/ Ngrid ,Nprof,Npar doi-l,Ngrid In(i,l)=ti do k=2,Npar do j=l,Nprof serG,l)=d(ij,l) seri j^dO.j .k) end do In(i,k)=serint(ser,ti,Nprof) end do end do Return End •• . '.' Serintf • **********#****************************** Function serint(s,t,Nser) Parameter(NgridMAX=200,NprofMAX=40,NparMAX=l 5,NserMAX=40) Integer NgridMAX,NparMAX,NserMAX,NprofMAX Integer Ngrid,Npar,Ngrad,Nprof,Nser . real s(NserMAX,2) real t,serint integer j Common/one/Ngrid,Nprof,Npar Common/two/Ngrad 157 *** find interpolated o3 gradient If (t-s(l.l)) 310,320,330 310 serint=s( 1,2)- (s(2,2)-s( 1,2))/(s(2,1 )-s( 1,1 ))*(s( 1,1 )-t) go to 370 320serint=s(l,2) go to 370 330 If (s(Nser.l)-t) 340,350,360 340 serint=s(Nser,2)+(s(Nser,2)-s(Nser-1,2))/ c (s(Nser, 1 )-s(Nser-1,1 ))*(t-s(Nser, 1)) go to 370 350 serint=s(Nser,2) go to 370 360doj = l,(Nser-l) If (t.eq. s(j,l))then serint=s(j,2) Else If ((t .gt. s(j,l)) -and. (t It. s(j+l,l))) then serint=s(j ,2)+(s(j+1,2)-s(j ,2))/ c (sO+l,l)-sG,l))*(t-s(j,l)) End If end do 370 Return End Trans.f Subroutine trans(forc,pt,c) *** given the met forcings, computes the transilient matrix Parameter(NgridMAX=200,NprofMAX=40,NparMAX= 15) Realforc(NgridMAX,NparMAX),c(NgridMAX,NgridMAX) Real y(NgridMAX,NgridMAX),pt(NgridMAX),vpt(NgridMAX) Integer NgridMAX,NparMAX,NprofMAX Integer Ngrid,Nprof,Npar Real To,Dy,Rc,g Real dz,du,dtemp,dt,dh,Ymax,yrow,sum,yref Common /one/ Ngrid,Nprof,Npar Common/four/dt,dh,Nd 158 *** set empirical parameters **** To= 1000.0 Dy=1.0 Rc= 0.21 g = 9.8 yref= 1000.0 *** convert to virtual potential temp. ******* do i=l,Ngrid vpt(i)=pt(i)*(1.0+(0.61*forc(i,4)*001)) end do do 110i=l,Ngrid do 105j=l,Ngrid If (i .eq. j) go to 105 dz=Real(i-j)*dh du=forc(i,2)-forc(j,2) dtemp=vpt(i)-vpt(j) *** tke parametrization ****** y(i,j)=(dt*To/dz**2)*(du**2-(g*dtemp*dz/(forc(i,3)*Rc))) c -Dy*dt/To if(y(ij).lt.0.0)y(ij)=0.0 105 continue 110 continue *** make y(ij) increase toward the diagonal ******* do 120i=l,Ngrid *** left side of diagonal ******* do 115j=l,i If ((j .ne. Ngrid) .and. (j ne. i))'then if (y(ij) gt. y(i,j+D) y(i,j+i)=y(i,j) endif 115 continue *** right side of diagonal ******* do 117j=Ngrid,i,-l If ((j .ne. 1) .and. (j .ne. i)) then lf(y(i.j)gt.y(ij-D) y(ij-D=y(i,j) endif 117 continue 120 continue *** make symmetric ******* do 140i=l,Ngrid do 135 j=l, Ngrid If (y(ij)-yG,0) 125,135,130 125 y(ij)=y0.i) 130 y(j,i)=y(i,j) 135 continue 140 continue *** set diagonal ******* do 150 i=l,Ngrid If (i eq. 1) then y(l,l)=y(l,2) go to 150 Else If (i .eq. Ngrid) then y(Ngrid,Ngrid)=y(Ngrid,Ngrid-l) go to 150 Endif y(i,i)=y(i,i-l) If (y(i,i+l) .gt. y(i,i)) y(i,i) = y(i,i+l) y(i,i)=y(i,i)+yref 150 continue *** find the normalization factor ******* Ymax=0.0 do 160 i=l, Ngrid yrow=0.0 do 155j=l,Ngrid yrow=yrow+abs(y(i,j)) 155 continue If (yrow .gt. ymax) ymax = yrow 160 continue *** off diagonal elements ******* do 170i=l,Ngrid do 165 j=l,Ngrid If (i .ne. j) c(i,j)=y(i,j)/ymax 165 continue 170 continue *** diagonal elements ******* do 180 i=l,Ngrid sum=0.0 do 175 j=l, Ngrid If (i .ne. j) sum=sum + c(i,j) 175 continue c(i,i)=1.0-sum 180 continue Return End 160 Flux.f subroutine flux(p,c,f) *** given a profile p and the transilient matrix c, finds the flux *** using the recursive method of Stull, 1988 Parameter (NgridMax=200) Real p(NgridMax),c(NgridMax)NgridMax),f(NgridMax) Real dz,dt,ksum,jsum Integer Ngridmax Integer Ngrid,Nprof,Npar Common /one/ Ngrid,Nprof,Npar Common/four/dt,dz,Nd do k=l,Ngrid ksum=0.0 do i=l,k jsum=0.0 do j=k+l,Ngrid jsum=jsum+(c(j,i)*p(i)-c(i,j)*p(j)) end do ksum=ksum+jsum end do f(k)=(dz/dt)*ksum end do Return End Mix.f Subroutine mix(x.c) * applies a transilient matrix to the profile x ******* Parameter(NgridMAX-200) Real x(NgridMAX),temp(NgridMAX),c(NgridMAX,NgridMAX) Integer Ngrid Integer NgridMAX Common /one/ Ngrid,Nprof,Npar do i=l,Ngrid temp(i)=x(i) end do do i=l,Ngrid sum=0.0 do j=l, Ngrid 161 sum=sum+(c(i,j)*temp(j)) end do x(i)=sum end do Return End ************************************************ 162 Appendix C Sensitivity Analysis Mixed Layer Model Sequences Figure A.C.I Mixed Layer Model Development With Time Step 30s. Modeled Observed • 800 700 600 500 400 300 -200 -100 -0 289 290 291 292 293 294 Potential Temperature (degrees Kelvin) 2 © E. (0 '(D X 295 290 291 292 293 294 295 Potential Temperature (degrees Kelvin) 296 11:00 PST I •••• -'i™ 1 1 ^ -700 / / ~ x * 600 J S 500 -400 - / s s -300 i \ s 200 . . 1 I — 100 - • / / 1 -0 / h 1 1 290 291 292 293 294 295 Potential Temperature (degrees Kelvin) 296 co £ co •E ri> «' O) co X 800 700 600 500 400 300 200 h 100 J -JL X X 291 292 293 294 295 296 Potential Temperature (degrees Kelvin) 297 163 Figure A C . 2 Mixed Layer Model Development With Time Step 180s. Modeled Observed 09:00 PST 800 289 290 291 292 293 294 Potential Temperature (degrees Kelvin) 295 CO d) «' g> 'co X 290 291 292 293 294 295 Potential Temperature (degrees Kelvin) 296 164 Figure A.C 3 Mixed Layer Model Development With Surface Heat Flux 0.25 Km/s. Modeled Observed 09:00 PST 289 290 291 292 293 294 295 Potential Temperature (degrees Kelvin) 10:00 PST 291 292 293 294 295 Potential Temperature (degrees Kelvin) 165 Figure A.C .4 Mixed Layer Model Development With Surface Heat Flux 0.10 Km/s. Modeled Observed 09:00 PST 10:00 PST 289 290 291 292 293 294 295 290 291 292 293 294 295 296 Potential Temperature (degrees Kelvin) Potential Temperature (degrees Kelvin) 11:00 PST 166 Figure A C 5 Mixed Layer Model Development With Averaging Interval of 30 Minutes. Modeled Observed 09:00 PST 10:00 PST Potential Temperature (degrees Kelvin) Potential Temperature (degrees Kelvin) 167 Figure A.C.6 Mixed Layer Model Development With No Averaging. Modeled — — Observed 09:00 PST 10:00 PST 800 S2 co a> E , ui cd D) "<D X 289 290 291 292 293 .294 Potential Temperature (degrees Kelvin) 295 52 © 15 E . D) cd -•—• _c at '<D X 800 290 291 292 293 294 295 Potential Temperature (degrees Kelvin) 296 11:00 PST 12:00 PST 800 700 600 I 500 d> 400 300 200 100 1 1 — s. / • 1 S y • •. * / y • • • — s - s • y V — / / \ •X N y / / \ / 1 / k , V, v , 52 <5 E, CO cd £ *<D I 800 700 600 500 400 300 200 100 / / r — y rf- *"* "~ ~ ' \ 1 s i \ — • -• * _ / / / V / _ / s. N > 290 291 292 293 294 295 Potential Temperature (degrees Kelvin) 296 291 292 293 294 295. 296 Potential Temperature (degrees Kelvin) 297 168 References A I R . Inc. 1990: A D A S Models ATR-3A, AIR-3B and AIR-3C: Operation and Technical Reference Manual, Boulder,CO. A.I.R. Inc., 1990: Tethersonde Ozone Sounding System O Z - 3A -T: Operation Manual, Boulder, CO. Banta, R .M. , Harwood, M . , Sheppard, A . , Froude, F., Martin, J.B., and Strapp, W., 1997: Nocturnal cleansing flows in a tributary valley, Atmospheric Environment, 31, 2147-2163. Biesenthal, T., Wu, Q., Shepson, P.B., Wiebe, H.A., Anlauf, K .G . , and Brickell, P., 1997: Production and distribution of organic nitrates, and their relationship to carbonyl compounds in the urban environment, Atmospheric Environment, 31, 2049-2058. Bunco, N . , 1991: "Environmental Chemistry", Wuerz Publishing Ltd., Winnipeg. Blumen, W., 1990: Mountain Winds, Chapter 1 in Atmospheric Processes over Complex Terrain, W. Blumen, ed. Meteorological Monograph 23 (45), 1-5. 169 Blumenthal, D.L., White, W.H. , and Smith, T.B., 1978: Anatomy of a Los Angeles smog episode: pollutant transport in the daytime Seabreeze regime, Atmospheric Environment, 12, 893-907. Browell, E.V., Gregory, G.L., Harriss, R.C., and Kirchoff, V.W.J.H., 1988: Tropospheric ozone and aerosol distributions across the Amazon Basin, J. Geophys. Res., 93, 1431-1451. Browell, E.V., Fenn, M . A . , Butler, C.F., Grant, W.B., Harriss, R.C., Shipman, M.C . , 1992: Ozone and aerosol distribution in the summertime atmosphere over Canada, J. Geophys. Res., 99, (DI), 1739-1755. Browell, E.V., Danielsen, E.F., Ismail, S. Gregory, G.L., and Beck, S.M., 1987: Tropopause fold structure determined from airborne Lidar and in situ measurements, J. Geophys. Res., 92, (D4), 2112-2120. Drewitt, G.B., Curran, K. , Steyn, D.G. , Gillespie, T.J., Niki , H. , Measurement of biogenic hydrocarbon emissions from vegetation in the lower Fraser Valley British Columbia, Atmos. Envir., 32 (20), 3457-3466. Elsom, D., 1989: "Atmospheric Pollution", Basil Blackwell, Oxford. 170 Evans, C.J., Martin, B. , Froude, F.A., and Thomson, R.B., 1993: Ozone and meteorological profiles in the Lower Fraser Valley of British Columbia, Report PAES-93-7, Atmospheric Issues and Services Branch, 12 pp Fishman, J., Vukovich, F .M. , Cahoon, D.R., and Shipman, M.C. , 1987: The characterization of an air pollution episode using satellite total ozone measurements, J. dim, Appl. MeteoroL, 26, 1638-1654. Fishman, J., Watson, C.E., Larsen, J.C., and Logan, J.A., 1990: Distribution of tropospheric ozone determined from satellite data, J. Geophys. Res., 95 (D4), 3599-3617. Fishman, J., Fakhruzzaman, K., Cros, B. , and Nganga, D., 1991: Identification of widespread pollution in the southern hemisphere deduced from satellite analyses, Science, 252, 1693-1696. Garland and Derwent, R.G., 1979: Destruction at the ground and the diurnal cycle of ozone and other gases, Quart. I. Met. Soc, 105, 169-183. Hanna, S.R., 1991: Characteristics of ozone episodes during S C C C A M P 1985, J. Appl. MeteoroL, 30, 534-550. 171 Hoff, R .M. , Harwood, M . , Sheppard, A. , Froude, F., Martin, J.B., and Strapp,W., 1997: Use of airborne lidar to determine aerosol movement in the Lower Fraser Valley, B C , Atmospheric Environment, 31, 2123-2134. Kleinman, L. , Yin-Yan, L. , Springston, S.R., Nunnermacker, L. , Xianliang, Z., Brown, R., Hallock, K., Klotz, P., Leahy, D., Lee, J.H., and Newman, L. , (1994): Ozone formation at a rural site, J. Geophys. Res., 99 (D2), 3469-3482. Lu, R., and Turco. R.P., 1994: Air pollutant transport in a coastal environment, Part I: Two-dimensional simulations os sea-breeze and mountain effects, J. Atmos. Sci., 51, 2285-2308. McElroy, J.L., and Smith, T.B., (1993): Creation and fate of ozone layers aloft in Southern California, Atmospheric Environment, 27A, 1917-1929. McKee, B . M . , and O'Neal, R.D., 1989: The role of valley geometry and energy budget in the formation of nocturnal valley winds, J. Appl. Meteorol., 28 (6), 445-456. McKendry, I.G., 1993: Ground level ozone in Montreal, Canada, Atmospheric Environment, 27B, 93-103. McKendry, I.G. 1994: Synoptic circulation and summertime ground level ozone concentration at Vancouver, B.C., Canada. J. App. Met. 33 (5), 627-641. 172 McKendry, I.G., Steyn, D.G., Lundgren, J., Hoff, R . M . , Strapp, W., Anlauf, K , Froude, F., Martin, J.B., Banta, R . M . , and Olivier, L.D. , 1997: Elevated ozone layers and vertical downmixing over the Lower Fraser Valley, B.C. , Atmospheric Environment, 31, 2135-2146. McKendry, I.G., Steyn, D.G., O'Kane, S., Zawar-Reza, P. and Heuff, D., 1998: Lower tropospheric ozone measurements by light aircraft equipped with chemiluminescent sonde, Journal of Atmospheric and Oceanic Technology, 15, 136-143. Miao, Y. , 1993: Mesometeorological Modeling and Trajectory Studies During an Ai r Pollution Episode in the Lower Fraser Valley, B.C. , Canada, The University of British Columbia, Vancouver, Canada, M.Sc. Thesis. Millan, M . M . , Salvador, R., Mantilla, E. and Kallos, G., 1997. Photo-oxidant dynamics in the western Mediterranean in Summer: Results from European Research projects. J. Geophys. Res. 102 (D7): 8811-8823. Moore, G.E., Douglas, S.G., Kessler, R. C , and Killus, J.P., 1991: Identification and tracking of pollutant/air masses in the South-Central Coast Air Basin, J. Appl. Met., 30, 715-732. Munn, R.E., 1993: GVRD Air quality monitoring network review. Published by Greater Vancouver Regional District, Burnaby, B.C. , July 14, 1993. 31 p. 173 Neu, U . , Kunzle, T. and Wanner, H. , 1994: On the relation between ozone storage in the residual layer and daily variation in near-surface ozone concentration - a case study, Boundary Layer MeteoroL, 69, 221-247. Neu, U . , 1995: A parametrization of the nocturnal ozone reduction in the residual layer by vertical downward mixing during summer smog situations using sodar data, Boundary Layer Meteorol.,73,189-193. O'Kane, S., 1997: An Investigation of Ozone Distribution Downwind of Greater Vancouver, British Columbia Using a Novel Aircraft Measurement System, The University of British Columbia, Vancouver, Canada, M.Sc. Thesis. Oke, T.R., 1987: Boundary Layer Climates, 2 n d ad., London and New York, Methuen, 435pp. Olivier, L.D. , Banta, R . M . , and Hardestry, M.R. , 1994: Wind flow in the Fraser Valley as measured by a pulsed C02 Doppler lidar, Eighth Joint Conference on the Applications of Air Pollution meteorology with AWMA, 23-28 Jan. 1994, Nashville, Tennessee. Pottier, J.L., Pryor, S.C., Banta, R . M . , Synoptic variability related to boundary layer and surface features observed during Pacific '93. Atmospheric Environment, 31, 2163-2173. 174 Pryor, S.C., and D.G. Steyn, 1995: Hebdomadal and diurnal cycles in ozone time series from the Lower Fraser Valley, B.C. Atmospheric Environment, Vol . 29, No. 9, 1007-1019. Sakiyama, S.K., 1990: Drainage flow characteristics and inversion breakup in two Alberta mountain valleys, J. App. Meteorol. 29, (10), 1015-1030. Seinfeld, J.H., 1989: Urban air pollution: state of the science, Science, 243, No. 10, 745-752. Shapiro, M . N . , 1980: Turbulent mixing within tropopause folds as a mechanism for the exchange of chemical constituents between the stratosphere and troposphere, J. Atmos. Sci., 27, 994-1004. Steyn, D.G. , Roberge, A .C . and Jackson, C , Anatomy of an Extended Air Pollution Episode in British Columbia's Lower Fraser Valley. Repared for B C Ministry of Environment, 49 pp. Steyn, D.G. , Bottenheim, J.W. and Thompson, R.B., 1997: Overview of tropospheric ozone in the Lower Fraser Valley, and the Pacific 93 field study. Atmospheric Environment, 31, 2025-2035. 175 Steyn, D.G.,1980: Turbulence, Diffusion and the Daytime Mixed Layer Depth over a Coastal City, The University of British Columbia, Vancouver, Canada, Ph.D. Thesis. Stull, R.B., 1984: Transilient turbulence theory. Part I: The concept of eddy-mixing across finite distances, 7. Atmos. Sci.,41, 3351-3367. Stull, R.B., and Driedonks, A . G . M . , 1987: Applications of the transilient turbulence parametrization to boundary-layer simulations, Boundary Layer MeteoroL, 40, 209-239. Stull, R.B., 1988: An Introduction to Boundary Layer Meteorology, Dordrecht, Kluwer A C A D E M I C Publishers, 666pp. Stull, R.B., 1993: Review of non-local mixing in turbulent atmospheres: transilient turbulence theory. Boundary Layer MeteoroL 62, 21-96. Sutton, O.G., 1953: Micrometeorology, McGraw-Hill,271 pp. Ulrickson, B .L . and Mass, C.F., 1990b: Numerical investigations of mesoscale circulation over the Los Angeles basin. Part U: Synoptic influences and pollutant transport, Monthly Weather Re vie w, 118, 2162-2184. Wakimoto, R . M . , and McElroy, J.L., 1986: Lidar observations of elevated pollution layers over Los Angeles, J. Clim. Appl. Met., 25, 1583-1598. 176 Wanner H. , Kunzle, T., Neu U. , Uhy B., Baumbach, G. and Steisslinger, B. , 1993: On the dynamics of photochemical smog over the Swiss Middlleland - results of the first P O L L U M E T field experiment, Meteorol. andAtmos. Phys. 15, 117-138. Whiteman, C D . . , 1990: Observations of thermally developed wind systems in mountainous terrain, Chapter 2 in Atmospheric Processes over Complex Terrain, W. Blumen, ed. Meteorologic Monograph, 23 (45), 5-42. } 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0089391/manifest

Comment

Related Items