UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The morphology of, and impact of wildfire smoke on the development of, the convective boundary layer… Ferrara, Madison R. 2018

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

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata


24-ubc_2018_november_ferrara_madison.pdf [ 17.71MB ]
JSON: 24-1.0372879.json
JSON-LD: 24-1.0372879-ld.json
RDF/XML (Pretty): 24-1.0372879-rdf.xml
RDF/JSON: 24-1.0372879-rdf.json
Turtle: 24-1.0372879-turtle.txt
N-Triples: 24-1.0372879-rdf-ntriples.txt
Original Record: 24-1.0372879-source.json
Full Text

Full Text

The morphology of, and impact ofwildfire smoke on the development of,the convective boundary layer overSouthwestern British ColumbiabyMadison R. FerraraB.Sc., Indiana University, 2016A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF SCIENCEinThe Faculty of Graduate Studies and Postdoctoral Studies(Geography)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)October 2018© Madison R. Ferrara 2018The following individuals certify that they have read, and recommend tothe Faculty of Graduate and Postdoctoral Studies for acceptance, a the-sis/dissertation entitled:The morphology of, and impact of wildfire smoke on the development of,the convective boundary layer over Southwestern British Columbiasubmitted by: Madison R. Ferrara in partial fulfillment of the requirementsfor the degree of Master of Sciencein GeographyExamining Committee:Ian McKendry, GeographySupervisorRoland Stull, Earth, Ocean, and Atmospheric SciencesSupervisory Committee MemberBrett Eaton, GeographyAdditional ExamineriiAbstractThe Weather Research and Forecasting (WRF) model was used to under-stand spatial and temporal variations in convective boundary layer (CBL)height over Southwestern British Columbia. The model was evaluated withseveral vertical profiles collected by Windsond weather balloons and wasfound to be in good agreement with observed data. A comparison betweenterrain height and CBL height showed that the CBL is more terrain followingin the morning and less terrain following in the afternoon. This behaviourwas further quantified by calculating r and T values for each hour using meanCBL height data for each month. The least terrain following behaviour oc-curred at 1400 PST for each month. Mean CBL depth (above ground level)was contoured over the study region and showed that the CBL tended togrow deepest in the eastern half of the Fraser Valley, however, several moun-tain peaks had a CBL depth similar to those observed over the valley. MeanCBL height (mean sea level) showed that, for all mountain peaks, the CBLwas higher than all locations over the Fraser Valley. The large spatial vari-ations in CBL height were shown to be able to result in the formation ofelevated layers over the valley via advective venting. A thick layer of wild-fire smoke that was present in the lower atmosphere during summer 2017was documented to understand the impact it had on CBL development.The plume had an aerosol optical depth (AOD) of about 4 during the mostiiiAbstractintense period and PM2.5 concentrations exceeded 50 µg/m3. Windsondprofiles collected before and during this event showed a more stable atmo-sphere existing on the smoke day. The potential temperature gradient washigher near the surface and lower aloft when smoke was present in the at-mosphere compared to the clear day. It has been speculated that the impacton atmospheric stability due to wildfire smoke can act as a feedback loopby suppressing CBL growth, allowing pollutants to accumulate in the loweratmosphere, thus further degrading air quality.ivLay summaryPollutants emitted near-ground level are primarily limited to the convec-tive boundary layer (CBL) over flat terrain. Various circulations inducedby complex terrain, however, can provide mechanisms that transport air,and pollutants, from within the CBL to the free troposphere. This altersthe distribution of pollutants in the atmosphere by allowing air to exist atlevels unattainable by convection alone. The morphology of the CBL overa landscape affects the locations that these ‘venting’ mechanisms can exist.This study focuses mainly on understanding the morphology of the CBL overSouthwestern British Columbia to better understand the distribution of pol-lutants in the atmosphere. Additionally, this study focuses on how wildfiresmoke may alter CBL development. As the CBL is largely controlled bysurface sensible heat fluxes and wildfire smoke reduces surface sensible heatfluxes, the atmosphere tends to be more stable when smoke is present.vPrefaceThe research topic and study design was developed by Madison R. Ferraraand Ian McKendry. Data from Windsond balloon launches was collectedby Madison R. Ferrara, Dylan Weyell, and Carrington Pomeroy. Qualityassurance and quality control for LiDAR data was perfomed by Paul Cottle.Chapter 3 and Chapter 4 form the basis of a paper that will be submit-ted for publication, in collaboration with Ian McKendry and Roland Stull.All other data analysis and processessing was informed by Ian McKendryand implemented by Madison R. Ferrara. The manuscript was prepared byMadison R. Ferrara with editing by Ian McKendry, Roland Stull, and BrettEaton.viTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiLay summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiiList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ixAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . x1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Motivation for study . . . . . . . . . . . . . . . . . . . . . . . 11.2 Project overview . . . . . . . . . . . . . . . . . . . . . . . . . 32 Background and methods . . . . . . . . . . . . . . . . . . . . . 62.1 Characteristics of CBL . . . . . . . . . . . . . . . . . . . . . 62.2 Description of venting mechanisms . . . . . . . . . . . . . . . 82.3 Topography and CBL height . . . . . . . . . . . . . . . . . . 102.4 Description of study area . . . . . . . . . . . . . . . . . . . . 122.4.1 Geography of Southwestern British Columbia . . . . 12viiTable of Contents2.4.2 Climate of Southwestern British Columbia . . . . . . 142.4.3 Mesoscale circulations within Southwestern British Columbia152.5 Datasets and data analysis methods . . . . . . . . . . . . . . 182.5.1 Datasets . . . . . . . . . . . . . . . . . . . . . . . . . 182.5.2 Data analysis methods . . . . . . . . . . . . . . . . . 263 Spatial and temporal variations in convective boundary layerheight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 273.2 Model evaluation . . . . . . . . . . . . . . . . . . . . . . . . . 293.3 Spatial variations in CBL height over Southwestern BritishColumbia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353.3.1 Terrain following behaviour of the CBL . . . . . . . . 363.3.2 CBL morphology . . . . . . . . . . . . . . . . . . . . 433.3.3 Evidence of advection venting . . . . . . . . . . . . . 483.4 Temporal variation in CBL height over Grouse Mountain . . 553.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 624 Case Study: Wildfire smoke impacts on Convective Bound-ary Layer Development . . . . . . . . . . . . . . . . . . . . . . 664.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 664.2 Observations of the smoke layer . . . . . . . . . . . . . . . . 684.3 Impacts on stability . . . . . . . . . . . . . . . . . . . . . . . 734.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 795 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 825.1 Project outcome . . . . . . . . . . . . . . . . . . . . . . . . . 82viiiTable of Contents5.2 Suggestions for future work . . . . . . . . . . . . . . . . . . . 85Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87ixList of Tables2.1 1981 - 2010 Canadian Climate Normals for Vancouver Inter-national Airport (Environment Canada) . . . . . . . . . . . . 172.2 1981 - 2010 Canadian Climate Normals for Chilliwack, BritishColumbia (Environment Canada) . . . . . . . . . . . . . . . . 173.1 Dates and times of data used in correlations. . . . . . . . . . 313.2 Correlation coefficients for model and observed values. . . . . 353.3 r and T values at 1400 PST for each month. . . . . . . . . . 403.4 Mean CBL depth (above ground level) for each month for theentire model domain. . . . . . . . . . . . . . . . . . . . . . . . 443.5 Maximum CBL depth AGL for each month. . . . . . . . . . . 60xList of Figures2.1 Illustration of venting mechanisms that can be induced byorography. Dashed line (h) represents the CBL top (adaptedfrom Kossman et. al., 1999). . . . . . . . . . . . . . . . . . . 92.2 Illustration of venting mechanisms and the structure of thelower atmosphere adapted from De Wekker (2002). Dashedline h is CBL top and solid line ha is the aerosol layer top. . . 122.3 Digital elevation map of Southwestern British Columbia. Shadedvalues are terrain height in meters. WGS-84 pseudo-mercatorprojection. Scale: 1/50,000. (Source: http://maps.canada.ca/czs/index-en.html). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132.4 DEM of the WRF model domain. . . . . . . . . . . . . . . . . 232.5 Vertical resolution of the WRF model. . . . . . . . . . . . . . 242.6 Model peak (red) and topographic peak (yellow) overlaid onmodel terrain. . . . . . . . . . . . . . . . . . . . . . . . . . . . 253.1 Scatter plots of predicted values determined by the WRFmodel compared to observed values measured by the Wind-sond’s for potential temperature, water vapour mixing ratio,wind speed, and wind direction. The dotted grey line repre-sents a perfect fit and the solid red line is the linear regression. 32xiList of Figures3.2 Windsond (black) and model (grey) profiles for July 25, 2017. 343.3 Modeled mean CBL height MSL vs. terrain height at 0700PST. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383.4 Modeled mean CBL height MSL vs. terrain height at 1500PST. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393.5 Histogram of terrain height from WRF output (only cells overland). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413.6 r and T values from mean CBL height vs. terrain height. . . 423.7 Monthly mean CBL depth (above ground level) at 1400 PSTmodel output segregated by month. Shaded values are inmeters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463.8 Monthly mean CBL height (mean sea-level) at 1400 PSTmodel output segregated by month. Shaded values are inmeters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473.9 Synoptic maps for August 04, 2017 at 1100 PST from NCEP/NCARReanalysis 1. 500 mb geopotential height (top) and mean sea-level pressure (bottom). . . . . . . . . . . . . . . . . . . . . . 503.10 Station plots from (a) Vancouver International Airport (b)Mahon Park, North Vancouver (Environment Canada). (c)is WRF model 10 m winds at 1400 PST. . . . . . . . . . . . . 513.11 Vertical cross-sections for August 04, 2017 at 1400 PST fromWRF output. (a) location of transect (b) cross-section ofwater vapour mixing ratio (c) cross-section of vertical velocity.Grouse Mountain is peak at 49.39°. . . . . . . . . . . . . . . . 533.12 Time series of PBL depth (above ground level) over GrouseMountain for all days by month. Red line is average PBLdepth. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57xiiList of Figures3.13 Maximum daily PBL depth (above ground level) for each dayin the study period. . . . . . . . . . . . . . . . . . . . . . . . 583.14 Boxplot of daily maximum PBL depth (above ground level)by month. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 593.15 Average evolution of PBL depth AGL (top) and average PBLgrowth rate (bottom). . . . . . . . . . . . . . . . . . . . . . . 614.1 MODIS imagery of wildfire smoke residing over SouthwesternBritish Columbia. . . . . . . . . . . . . . . . . . . . . . . . . . 694.2 LiDAR imagery of the smoke layer from LiDAR stationed onGrouse Mountain. . . . . . . . . . . . . . . . . . . . . . . . . 704.3 Particulate matter 2.5 concentrations (top) and irradiancevalues (bottom) measured at Vancouver International Airport(YVR) for the duration of the smoke event. . . . . . . . . . . 724.4 Vertical profiles collected by a Windsond weather balloon forJuly 31, 2017 and August 09, 2017. . . . . . . . . . . . . . . . 754.5 Vertical profile output from the WRF model for July 31, 2017and August 09, 2017. . . . . . . . . . . . . . . . . . . . . . . . 764.6 Potential temperature gradients of the atmosphere on July31, 2017 (clear day) and August 09, 2017 (smoke day) forWindsond data (left) and WRF output (right). . . . . . . . . 78xiiiAcknowledgementsI would like to thank my research supervisor, Dr. Ian McKendry, for hisexpertise and unwavering support throughout my degree program. I wouldalso like to thank the other members of my committee, Dr. Roland Stulland Dr. Paul Cottle, for providing their assistance with the execution ofthis project. My sincerest gratitude is also extended to Dr. Brett Eaton forhis help along the way. This project would not have been possible if not forthe willingness of the management of Grouse Mountain to allow our researchequipment to be stationed here. I would like to thank Erik Bowkett and theentire staff of Grouse Mountain for being so accommodating.xivChapter 1Introduction1.1 Motivation for studyAir pollution meteorology emerged as a field of study when it was recog-nized that changes in synoptic patterns can have a drastic effect on localair quality. The connection between large-scale weather phenomena andpollutant concentrations experienced near-ground level lies in the way thesepatterns impact the stability of the lower atmosphere. As stability is alsolargely controlled by surface forcings, interactions between earth’s surfaceand the planetary boundary layer (PBL) are at the forefront of air pollutionmeteorology research.The PBL is the layer of the atmosphere directly adjacent to earth’s sur-face therefore the behaviour of this layer has important implications forhuman and environmental health. The PBL is dominated by turbulent con-vection during daytime hours and is called the convective boundary layer(CBL) when this is the case. Many studies have worked to characterize thestructure and morphology of the CBL in different parts of the world howevera significant portion of this research has focused on regions with relativelyflat and homogeneous terrain. As populations increase near coastal and oro-graphic areas it is important to gain a better understanding of the connectionbetween geography and CBL dynamics.11.1. Motivation for studyThe unique geography of Southwestern British Columbia makes for aninteresting area to study boundary layer meteorology. This region is char-acterized by complex terrain such as mountains, valleys, and coastlines. Alarge majority of the studies in this region have focused on the behaviour ofthe CBL over the lower Fraser Valley (LFV) however a broader view of thespatial variations in the CBL over the area as a whole would help explain thedistribution of pollutants observed over the valley and the daily variationsin local air quality.Air quality in the LFV affects the over 2 million residents that residein this region. The City of Vancouver is the largest city in the LFV witha reported population of 631,486 according to 2016 census data (Statis-tics Canada, 2016 Census of Population). As with any large city, pollutionis emitted from the various forms of transportation (trucks, buses, ships,trains, planes, etc.) and industrious operations (construction, power gener-ation, combustion processes, etc.). Air pollution can be defined as “a stateof the atmosphere in which substances that harm human or plant health orcorrode materials exist in concentrations above their normal or backgroundlevels” (Steyn and McKendry, 2002). Therefore, along with anthropogenicemissions, air quality is heavily impacted by natural sources such as biolog-ical decay, volcanoes, and wildfires.In 2017, British Columbia experienced a particularly active wildfire sea-son that led to several air quality advisory warnings for the City of Vancou-ver. Along with the direct effects of pollution from wildfire smoke, the largeamount of particulate matter emitted from these events can alter the stabil-ity of the CBL thus affecting the way pollutants disperse in the atmosphere.Due to an increased threat of wildfires resulting from climate change (IPCC,2014) and a rapidly growing population in the LFV, an ever increasing effort21.2. Project overviewis being put towards understanding the impact of wildfire smoke on CBLdynamics and the resulting effects on public and environmental health.1.2 Project overviewThis thesis presents a meteorological study of the general behaviour of thelower atmosphere over Southwestern British Columbia using model and ob-servational data. The overarching theme of this research is to characterizethe behaviour of the CBL in relation to the geography of the region. Anemphasis will also be put towards understanding the impact of a dense layerof wildfire smoke in the lower atmosphere on CBL development.Spatial and temporal variations in CBL depth (above ground level) andCBL height (mean sea level) over Southwestern British Columbia will beexamined using the Weather Research and Forecasting (WRF) numericalmodel. Numerical models are advantageous in that atmospheric propertiescan be investigated over a larger spatial and temporal domain than whatcan be done by observational measurements alone. However, models lackthe full complexity observed in nature and therefore provide a simplifiedview of atmospheric dynamics. Nonetheless, the use of numerical modelsis prevalent in air pollution meteorology research because they can provideinsight to the behaviour of the lower atmosphere. With the assistance ofthe WRF model, the following two research questions about CBL behaviourwill be addressed in this thesis:1) What is the relationship between terrain height and CBL height overSouthwestern British Columbia?2) What is the general morphology of the CBL over Southwestern BritishColumbia?31.2. Project overviewThis thesis will also present evidence for wildfire smoke suppressingboundary layer development using observational data collected by Wind-sond weather balloons. This analysis has the advantage of using a datasetthat contains realistic values for points in space. However, often it is hardto conclude causal relationships between variables using observational mea-surements alone. This section will be largely speculative in nature and willrelate the findings here to the existing literature on the topic. The thirdresearch question of the thesis will then be discussed:3) How does a thick layer of wildfire smoke affect the development of theCBL?This thesis is organized into five chapters. Chapter 2 will provide a moredetailed description of the characteristics of the CBL over complex terrainand the mesoscale circulations that can alter CBL behaviour. A completedescription of the WRF model used in this study will be discussed herealong with a summary of the geography and climate of the study area. Thischapter will conclude by outlining the data analysis methods and datasetsused in the following chapters.Spatial and temporal variations in convective boundary layer heightover Southwestern British Columbia will be investigated using the WRFmesoscale model in Chapter 3. The performance of the WRF model will beevaluated with observed data and then used to examine the terrain followingbehaviour of the CBL. A picture of the general morphology of the CBL overthe study area will be presented here. The remainder of the chapter willexamine the temporal evolution of the boundary layer by focusing on theatmosphere over Grouse Mountain to provide insight to the development ofa CBL over a mountain range.Chapter 4 will examine the effect of wildfire smoke on CBL development41.2. Project overviewusing observed vertical profile data from Windsond weather balloons. Pro-files will be compared between a clear day and a day when a significantamount of wildfire smoke was present in the atmosphere. Potential temper-ature gradients of the atmosphere on these days will be compared to analyzethe stability of the atmosphere for each case. The results from this chapterwill be related to existing literature on the topic to conclude the chapter.The concluding chapter of this thesis will be presented in Chapter 5. Thischapter will summarize the results from the previous chapters and relate thefindings to each other. Suggestions for future work will be presented at theend of this chapter.5Chapter 2Background and methods2.1 Characteristics of CBLA qualitative definition of the planetary boundary layer (PBL) is “the layerof the troposphere that directly interacts with earth’s surface and respondsto surface forcings within a timescale of about an hour or less” (Stull, 1988).This results in a rapidly evolving PBL that exhibits different characteristicsthroughout the day in reaction to variations in surface properties. Thestability of the lower atmosphere has large control on wind patterns withinthe PBL and is often determined by analyzing potential temperature profiles.Several terms exist to describe the different states of the PBL based on thesedifferences in thermodynamic structure.Between sunset and sunrise the lower atmosphere cools more rapidlythan air at upper-levels. This can result in the formation of a potential tem-perature inversion beginning at the surface which is indicative of a stableatmosphere. During this time, the PBL can be referred to as the noctur-nal boundary layer (NBL) and very little vertical mixing occurs when theseconditions are present (Nieuwstadt, 1984). Many of the poor air qualityepisodes experienced around the world are a result of stable conditions pre-venting mixing processes thus allowing pollutants to accumulate in the loweratmosphere (Schwartz, 1993).62.1. Characteristics of CBLDuring daytime convective conditions, the PBL is dominated by tur-bulent convection resulting in a well-mixed boundary layer. When theseconditions are present, the PBL can be referred to as the convective bound-ary layer (CBL). To better understand the behaviour of the CBL it can befurther divided into the following three components: surface layer, mixedlayer, and entrainment zone. Shortly after sunrise the input of sensible heatresults in a superadiatic surface layer causing instability in the lower atmo-sphere. Thermal eddies convecting throughout the CBL result in a mixedlayer defined by an adiabatic lapse rate. The entrainment zone is the layer ofthe CBL that is the intersection between the stable free troposphere and themixed layer of the CBL and is marked by a potential temperature inversion.This layer is largely responsible for the growth of the CBL by entrainingless turbulent air from above (Stull, 1988). Several definitions exist for de-termining the depth of the CBL, many of which are based on the lapserate of the atmosphere. Depending on the method chosen CBL depth canvary; however, most definitions result in the top of the CBL being placedsomewhere within the entrainment zone.Most pollutants released near the surface during the day are emitted di-rectly into the CBL. Turbulent mixing results in relatively constant concen-trations of pollutants within the CBL and a sharp decrease in concentrationthroughout the entrainment zone before reaching the low background con-centrations experienced in the free troposphere. Many studies have shownthat CBL depth based on the potential temperature profile of the atmo-sphere often agree with the height pollutants released at the surface canrise over flat and homogeneous terrain (Hennemuth and Lammert, 2006;Lammert & Bo¨senberg, 2006). If this is the case, observations of pollutantconcentrations using LiDAR or other instruments can be used as a proxy for72.2. Description of venting mechanismsstability and can therefore be used to determine CBL depth in the absencethermodynamic observations (Seibert et. al., 2000).Orographic or complex landscapes can alter this behaviour, however.Thermally induced flows induced by variations in surface properties, suchas orography or coastlines, can induce mechanisms to vent air out of theCBL and into the free troposphere (Kossman et. al., 1999). The presence ofthese mechanisms allow air, and thus pollutants, to reach levels unattainableby convection alone. This alters the distribution of pollutants in the loweratmosphere and can act as a sink for polluted boundary layer air by ventingpollutants to levels where it does not directly affect human and environmen-tal health. The following section will describe these venting mechanisms inmore detail and emphasize the importance of these processes on the observeddistribution of pollutants in Southwestern British Columbia.2.2 Description of venting mechanismsThe importance of mesoscale circulations on air pollution dynamics was firststressed in studies conducted over the Los Angeles (LA) basin. Lu and Turco(1994) found that sea-breeze and mountain-valley circulations in the LAbasin are able to penetrate the potential temperature inversion that marksthe CBL top. These circulations provide a mechanism for boundary layer airto reach levels otherwise unattainable than if subject to convection alone.Mesoscale circulations that allow exchange between the boundary layer andthe free troposphere are termed ‘venting’ mechanisms and the existence ofthese mechanisms in a region can significantly affect near-ground level con-centrations. Studies such the one conducted by Lu and Turco have clearlyshown that topography must be considered when addressing air pollution82.2. Description of venting mechanismsentrainmentmountain ventingadvectiveventingmountain/cloud ventingℎFigure 2.1: Illustration of venting mechanisms that can be induced by orog-raphy. Dashed line (h) represents the CBL top (adapted from Kossman et.al., 1999).dynamics over complex terrain.A complete description of the most prevalent venting mechanisms thatcan be induced by orography was provided by Kossman et. al. (1999). Thethree mechanisms described by this study were mountain venting, cloudventing, and advective venting. Mountain venting occurs when differentialheating of elevated surfaces result in upslope or ‘anabatic’ winds convergingat a mountain peak or ridge. This adds additional momentum for air topenetrate the CBL top leading to an exchange between the boundary layerand the free troposphere. Depending on the saturation of the atmosphere,this process may be accompanied by cloud formation. If this is the case,the process is termed cloud venting. Advective venting is the result of spa-tial variations in CBL height and a wind vector that is not parallel to theboundary layer top. Exchange between the CBL and the free tropospherecan occur by basic flow or secondary circulations under these conditions.The mechanisms described by Kossman et. al. (1999) are illustrated inFigure Topography and CBL heightVenting mechanisms alter the structure of the atmosphere by introducingboundary layer air into the free troposphere. This can result in elevated pol-lution layers existing at levels significantly above the boundary layer thusaltering the distribution of pollutants in the atmosphere. Mountain andcloud venting processes can alter the morphology of the CBL which canthen affect the locations where advective venting can occur. Hence, vari-ations in CBL height over regions with complex terrain must be analyzedto fully understand the connection between topography and the observeddistribution of pollutants in the atmosphere. The next section will brieflyreview the literature of the effects of topography on CBL height.2.3 Topography and CBL heightSeveral studies have investigated the degree to which the CBL follows ter-rain, however, different conclusions were reached based on whether datawas collected with thermodynamic observations or LiDAR data. Nyeki(2000) analyzed LiDAR data over the European Alps from the EuropeanUnion’s“Scientific Training and Access to Aircraft for Atmospheric ResearchThroughout Europe” (STAAARTE) field project that used a nadir-pointingaerosol LiDAR attached to an aircraft that flew several transects over theJungfraujock High-Alpine Research Station in Switzerland. In this analysis,CBL height was assumed to be equal to the top of the aerosol layer andit was concluded that CBL height exhibits no significant terrain followingbehaviour on a scale of 20 - 25 km. The CBL did tend to be slightly higherover areas with a higher mean elevation indicating some terrain followingbehaviour but only on a scale of roughly 50 km. This finding contrasted withfindings from previous studies based on radiosonde data. Holzworth (1964)102.3. Topography and CBL heightanalyzed archived radiosonde data throughout the contiguous United Statesand showed that mean CBL height is generally higher over mountainousterrain and experiences greater variation throughout the year than the CBLheight over flat terrain. Also using radiosonde data, Lenschow et. al. (1979)reported that, over hilly terrain, CBL height closely follows topography inthe morning and exhibits some terrain following behaviour during the day.Similarly, Banta (1984) observed that CBL height over mountainous terrainis very inhomogeneous during the morning, but becomes more homogeneousthroughout the day. The discrepancies between the findings was found tobe due to the existence of mesoscale circulations venting particles out ofthe CBL and into the free troposphere (De Wekker, 2002). This creates anaerosol layer that is not equal to CBL height. When this is the case, LiDARmeasurements do not accurately represent CBL height.De Wekker (2002) provided a model for understanding the discrepan-cies observed between CBL depth over complex landscapes with LiDARand radiosonde measurements. This work showed that a CBL dominatedby turbulent convection forms over a landscape and closely follows terrain.However, venting mechanisms can allow pollutants to penetrate the CBL topand reside at levels above the CBL. This results in an aerosol layer that doesnot follow terrain as closely as the CBL. This is illustrated in Figure 2.2.This thesis will focus on the CBL which will help understand observationsof the aerosol layer.112.4. Description of study areaentrainmentmountain ventingadvectiveventingmountain/cloud ventingℎ"ℎFigure 2.2: Illustration of venting mechanisms and the structure of the loweratmosphere adapted from De Wekker (2002). Dashed line h is CBL top andsolid line ha is the aerosol layer top.2.4 Description of study area2.4.1 Geography of Southwestern British ColumbiaSouthwestern British Columbia is characterized by complex terrain such ascoastlines, valleys, mountains, and varying land-uses. The Fraser Valley iscomprised mainly of flat agricultural land and is bordered by steep valleywalls on both the northern and southern edges. The City of Vancouver is thelargest city in the region, however, Richmond, Surrey, Abbottsford, Burnaby,and Chilliwack all have relatively large populations and urban landscapes.The North Shore Mountains, including Grouse Mountain, Cypress Moun-tain, and Mount Seymour, form the northern boundary to the Fraser Valleyand have peak elevations between 900 - 1449 m which is reached within 10km of the valley floor. The western edge of the valley is bordered by the122.4. Description of study areaStrait of Georgia, a large body of water separating Vancouver’s Lower Main-land and Vancouver Island. The presence of the coast has a large impact onSouthwestern British Columbia’s climate by moderating temperatures andbeing the source of the majority of precipitation that falls in this region(Mass, 2008). Additionally, the mountains have a large effect on precipita-tion via orographic uplift resulting in significantly higher precipitation overthe valley slopes than in the middle of the LFV (Oke and Hay, 1994). Adigital-elevation map of the region is shown in Figure 2.3.Figure 2.3: Digital elevation map of Southwestern British Columbia. Shadedvalues are terrain height in meters. WGS-84 pseudo-mercator projection.Scale: 1/50,000. (Source: http://maps.canada.ca/czs/index-en.html).132.4. Description of study area2.4.2 Climate of Southwestern British ColumbiaSouthwestern British Columbia is a mid-latitude location comprising partof the Pacific Coast. The climate of the region is primarily controlled byseasonal shifts in the position and strength of the jet stream. In late falland winter, a strong westerly jet stream results in the frequent passage offrontal systems over the region which often bring rain to lower elevationsand snow to the mountains (Oke and Hay, 1994). The Pacific Ocean hasa moderating effect on near-surface temperature throughout the year mak-ing snow in the LFV a rare phenomena as temperatures are often abovefreezing. However, snow can occur when cold continental air gets pushedover the Rocky Mountains and undercuts a marine air-mass (Mass, 2008).The majority of precipitation falls during winter months, in part definingSouthwestern British Columbia as a Mediterranean climate.Summers are usually characterized by fair weather anti-cyclonic condi-tions due to the northward movement of the jet stream. An upper-level ridgeover Southwestern British Columbia is common during summer months re-sulting in stable conditions near the surface. Climate normals recorded atthe Vancouver International Airport (YVR) between 1981 - 2010 show anaverage temperature of 18°C for July and August and 14.9°C for Septem-ber. The station at YVR is about 2 km from the coast and a few me-ters above sea-level, therefore, inland locations further away from the coastsuch as Chilliwack experience a slightly higher average temperatures duringsummers and high elevation areas experience slightly lower average tem-peratures. Summers are typically dry in Southwestern British Columbiacompared to winters with only about 10% of annual precipitation occurringbetween June - August (Environment Canada). The climate normals for142.4. Description of study areaYVR and Chilliwack can be seen in Table 2.1 and Table Mesoscale circulations within Southwestern BritishColumbiaMesoscale circulations are most prominent during anti-cyclonic conditionswhen local forcings dominate forcings from synoptic patterns (Kossman et.al., 1999). These circulations are thermal in nature and are commonly theresult of pressure gradients caused by differential heating of a landscape.Land/sea breeze circulations are common in the LFV due to the proximityof the coast and play a significant role in the distribution of pollutants whenpresent (McKendry and Lundgren, 2000). Bodies of water have large ther-mal inertia compared to land resulting in land heating much more rapidlyduring the day. When synoptic flow is weak, a horizontal pressure gradi-ent can form perpendicular to the coast during the day bringing marine airinto the valley. Sea-breezes in this region have been observed extending toAbbottsford about 60% of the time the circulation is present and can evenreach Chilliwack. Sea-breezes in this region are generally around 3 m/s (Okeand Hay, 1994). At night, the direction of the pressure gradient switches dueto the land cooling much more rapidly than water resulting in near-surfacewind flowing from the valley towards the coast. Land-breezes in this regionare generally weaker than the sea-breeze, often around 2 m/s (Oke and Hay,1994).Similar mechanisms are responsible for the formation of mountain-valleybreezes. Elevated or sloped terrain heats near-surface air more rapidly thanair directly adjacent to the slope during the day. Again, this results in a hor-izontal pressure gradient pointing towards the terrain and wind moving up152.4. Description of study areathe mountain slope. When these conditions are present, the flow is referredto as upslope or ‘anabatic’ wind. As discussed in the previous section, ups-lope winds are the dominant driver of mountain venting by allowing air toreach levels otherwise unattainable than if this circulation was not present.To compensate for the upslope flow, subsidence is required over the valleycore which brings air from higher elevations towards the surface (Rampanelliand Zardi, 2004). During the night, the direction of the pressure gradientswitches resulting in ‘katabatic’ or down-slope flow, reversing the circulationMountain-valley breezes generally occur on a scale of individual slopes,however, larger circulations can exist penetrating further into the LFV. ‘Up-valley’ circulations can form in the LFV due to the subsidence in the valleycore resulting from mountain-valley breezes. Subsiding air warms adiabati-cally as it is brought towards the surface, creating a pressure gradient up theLFV and individual tributaries. Drainage occurs at night when the directionof the pressure gradient and flow reverses. This circulation in conjunctionwith land/sea breezes and mountain-valley breezes all effect the way pollu-tants disperse in Southwestern British Columbia.162.4.DescriptionofstudyareaTable 2.1: 1981 - 2010 Canadian Climate Normals for Vancouver International Airport (Environment Canada)Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov DecAverage Temp. (°C) 4.1 4.9 6.9 9.4 12.8 15.7 18 18 14.9 10.3 6.3 3.6Precipitation (mm) 168.4 104.6 113.9 88.5 65.0 53.8 35.6 36.7 50.9 120.8 188.9 161.9Table 2.2: 1981 - 2010 Canadian Climate Normals for Chilliwack, British Columbia (Environment Canada)Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov DecAverage Temp. (°C) 3.3 4.9 7.3 10.5 13.7 16.4 18.8 18.7 15.7 10.8 6.2 3.3Precipitation (mm) 233.5 125.8 154.7 116.3 93.1 91.7 48.1 56.7 75.2 178.5 283.8 210.1172.5. Datasets and data analysis methods2.5 Datasets and data analysis methods2.5.1 DatasetsThis study uses both model and observational data to examine CBL be-haviour over Southwestern British Columbia. The WRF numerical modelwas used to evaluate spatial and temporal trends in CBL behaviour anda mini-micropulse LiDAR (mMPL) and Windsond weather balloons wereused to investigate the effect of wildfire smoke on CBL development. Thissection will provide a more detailed description of these datasets along withthe data analysis methods used in subsequent chapters.WindsondWindsond’s are a uniquely small weather balloon system that can be usedto observe local conditions at different altitudes. The sonde instrumenta-tion is contained in a small 12 g Styrofoam cup which is attached to anapproximately 30-litre helium balloon. The Windsond measures several at-mospheric variables, such as temperature, relative humidity, pressure, windspeed, wind direction, as well as altitude and GPS location. The Windsondcollects data at 3-second intervals and generally has an ascension rate be-tween 1 – 2 m/s. This results in an average vertical resolution of about 3 –4 m. The maximum altitude the sonde can reach is dependent on the initialvolume of the balloon as well as air pressure at upper-levels. All balloonsreached at least 5 km above ground level (AGL) for this study with severalgoing above 7 km AGL. The temperature sensor has an accuracy of 0.3°Cand resolution of 0.01°C. Humidity is recorded with an accuracy of 2.0 %RHand resolution of 0.05 %RH. Air pressure can be measured between 300 –1100 hPa with an accuracy of 1.0 hPa (at 700 – 1100 hPa) and resolution of182.5. Datasets and data analysis methods0.02 hPa. The resolution for wind direction measurements is 0.1°.Several calculations have to be made to convert temperature to potentialtemperature and relative humidity to water vapour mixing ratio. Potentialtemperature was calculated using Equation 2.1.θ = T(pop)κ(2.1)where θ is potential temperature [K ], T is temperature [K ], po is standardpressure [hPa], p is pressure [hPa], and κ is the Poisson constant [unitless].Relative humidity was converted to water vapour mixing ratio by firstcalculating the saturated vapour pressure using the Clausius-Clapeyron equa-tion (Equation 2.2) and then deriving values for vapour pressure (Equation2.3).e∗ = eo · exp[LvRv·(1To− 1T)](2.2)where e∗ is saturated vapour pressure [kPa], eo is the reference vapourpressure [kPa], Lv is the latent heat of vaporization [J kg−1], Rv is the gasconstant for moist air [J K−1 kg−1], and To is the reference temperature[K ].RH =ee∗· 100 (2.3)192.5. Datasets and data analysis methodswhere RH is relative humidity [% ] and e is vapour pressure [kPa].Water vapour mixing ratio can then be calculated using Equatio ??:q =0.622eP − 0.378e (2.4)where q is water vapour mixing ratio [kg kg−1].LiDARFor this study, a Sigma mini-Micropulse LiDAR (mMPL) was used to exam-ine the vertical structure of pollutants above Grouse Mountain. The mMPLis a cost-effective and portable system that is housed in a steel enclosurethat makes it suitable to be deployed in adverse environments. The mMPLcontinuously emits thousands of low-energy pulses at 532 nm and integratesthem into a single profile. Each pulse has an energy of 4 µJ which is severalorders of magnitudes less than larger, non-portable systems. To make upfor the weaker signal, the mMPL has a pulse frequency of 4 kHz which isseveral orders of magnitude greater than larger systems and an expandedbeam diameter of 7.62 cm. All of this combined makes this system ratedeye-safe at 3.5 m. The mMPL can profile from the surface to 15 km high;however, the first 150 m above the LiDAR cannot be profiled due to incom-plete overlap between the emitted beam and the receivers field of view. Thetime of integration can be set anywhere between the range 1 s - 60 min andthe vertical resolution can be set to 5,15,30, or 75 m. For this study, anaveraging time of 5 min and vertical resolution of 30 m was chosen.The main output of the mMPL is a value termed Normalized Relative202.5. Datasets and data analysis methodsBackscatter (NRB). NRB is a range corrected product that is derived fromthe standard LiDAR equation (Equation 2.5).S(z, λ) = Eo(λ)Az2γO(z)βpi(z, λ) exp−2 z∫0α(z′, λ)dz′+ SB (2.5)where S(z, λ) is the signal measured at detector [counts/µs], E0 is theenergy of laser pulse [J ], A is usable area of receiver entrance pupil [km2],z is range [km], γ is the overall system efficiency constant [counts/µs*µJ ],O(z) is the incomplete overlap correction function [unitless], βpi is volumebackscatter coefficient [km2 sr−1], α is extinction coefficient [unitless], andSB is the background signal [counts/µs].The first term in this equation is called attenuated backscatter becauseit is a function of backscatter as well as the attenuation of the emitted beam(described by the extinction coefficient) while the second term describes thesignal received by the LiDAR due to background radiation. In order tocalculate NRB a few corrections must be made to the raw signal received bythe LiDAR. After accounting for the deadtime resulting from the nature ofthe photon-counting module (d(s)) and removing the effects of the afterpulse(a(z)) and background radiation (SB), the signal is normalized relative tothe beam energy at each height to get NRB (Equation 2.6).NRB(z) =S(z) ∗ d(S)− a(z)− SB(z)E0 ∗O(z) ∗ z2 (2.6)212.5. Datasets and data analysis methodswhere NRB is normalized relative backscatter [counts*km2/µs*µJ ], d(S)is the dead time correction factor [unitless], a(z) is the afterpulse correctionfactor [counts/µs], and O(z) is the overlap correction factor [unitless].Weather Research and Forecasting modelA nested WRF model was used to analyze local conditions and boundarylayer heights for this study. The WRF model operates on a 279 x 279 curvi-linear grid with a 1333.3330078125 m horizontal resolution. The nested gridis oriented -33.01874° from a pure north-south alignment to better matchthe orientation of the North Shore Mountains. The top-left corner of thegrid is located at 48.91369°N and -127.321°W and the bottom-right corneris positioned at 48.07262°N and -120.7709°W (Figure 2.4). There are 40vertical levels in the model starting at the surface and extending to the topof the model atmosphere at 10 hPa. Vertical grid cells are linearly relatedto pressure causing them to be closer together nearer the surface and spreadout as they get closer to the top of the atmosphere (Figure 2.5).The model sees a“smoothed”version of terrain due to the horizontal gridresolution. This results in a difference between the location and height of“Grouse Mountain” as seen by the model (49.40683°N, -123.0771°W) and theactual topographic peak for Grouse Mountain (49°22’43”N, -123°04’57”W).The model peak is shifted approximately 3.162 km southwest of the to-pographic peak (Figure 2.6). The actual elevation of Grouse Mountain is1089.355 m whereas for the model it is 1115.915 m. For all analysis of GrouseMountain with the WRF model, the location of the model peak was used.Planetary boundary layer height is a direct output from the WRF modeland is determined using the Yonsei University (YSU) scheme (Hong et. al.,222.5. Datasets and data analysis methodsFigure 2.4: DEM of the WRF model domain.232.5. Datasets and data analysis methodsPressure [mb] Elevation [m]200 400 600 800 0 10000 20000 Vertical grid number010203040010203040Figure 2.5: Vertical resolution of the WRF model.242.5. Datasets and data analysis methodsFigure 2.6: Model peak (red) and topographic peak (yellow) overlaid onmodel terrain.252.5. Datasets and data analysis methods2006). This method takes into account vertical sub-grid-scale fluxes due toeddy transports in the whole atmospheric column and includes an explicittreatment of the entrainment zone at the top of the PBL. Entrainment ismade proportional to the surface buoyancy flux and PBL height is definedas the elevation where a critical bulk Richardson number of zero is firstreached, starting from the surface. The YSU scheme generally gives betterestimates of PBL height than other schemes that can be implemented by theWRF model, such as Mellor–Yamada–Janjic (MYJ) scheme or asymmetricconvective model, version 2 (ACM2) (Hu, 2010).2.5.2 Data analysis methodsThis thesis is an observational study using data from the WRF model aswell as in-situ measurements. Spatial and temporal variations in CBL heightover Southwestern British Columbia will be analyzed to better understandthe behaviour of the CBL in this region. The focus of this study is tounderstand the general morphology of the CBL over Southwestern BritishColumbia, therefore, a large section of this thesis will compare mean CBLheight values between each month.26Chapter 3Spatial and temporalvariations in convectiveboundary layer height3.1 IntroductionIn this chapter, the spatial and temporal variations in CBL height overSouthwestern British Columbia will be investigated using the WRF mesoscalemodel. The WRF model will be used to examine atmospheric propertiesover a large domain that would not be possible by observational measure-ments alone. With the aid of the model, locations where mountain ventingand advection venting mechanisms may be most prevalent will be identifiedby assessing the terrain following behaviour of the CBL over the inhomo-geneous landscape of the study area. Additionally, the temporal variationof the boundary layer will be examined over a fixed location (the peak ofGrouse Mountain) to better understand the development of a CBL over amountain range.Mountain circulations can vent polluted air out of the top of the bound-ary layer and deposit that vented air as polluted layers aloft. These circu-lations are especially prevalent in coastal cities with complex terrain where273.1. Introductionlocal scale circulations dominate the observed distribution of pollutants inthe lower atmosphere. Wakimoto and McElroy (1986) have shown that com-binations of sea breeze, slope flows, and convective activity can explain theobserved layers of pollutants in the Los Angeles Basin. In fact, it has beenshown that pollutant concentration in elevated layers can be greater thanboundary layer concentrations (McElroy and Smith, 1993). This has im-portant implications on ground-level concentrations when considering theeffects of vertical downmixing due to boundary layer growth in morninghours that can mix elevated pollutants to the surface.Elevated layers are also common phenomena in the LFV in SouthwesternBritish Columbia (McKendry et. al., 1997; McKendry et. al., 2011). Thischapter primarily focuses on examining CBL height over the study area toidentify locations where mountain venting and advection venting may beoccurring, however, several other mechanisms exist in this area that canalso give rise to elevated layers. McKendry and Lundgren (2000) have out-lined eleven processes in total specific to this region that can result in theformation of elevated layers. Most of the studies regarding elevated layersin the LFV have relied on observational data fixed in either space (groundbased LiDAR/tethersondes) or time (downward pointing LiDAR) or both(radiosonde). Additionally, these studies focused on the structure of thelower atmosphere over the valley rather than a mountain peak. The spatialvariation in CBL height examined in this chapter will provide a much neededpicture of the general morphology of the CBL in this region which will helpto understand the observed distribution of pollutants in the atmosphere.It has been speculated that mountain venting is significant along theNorth Shore Mountains that comprise the northern edge of the LFV (McK-endry, 1997), although little data has been collected to understand the evo-283.2. Model evaluationlution of the boundary layer over the mountains themselves. The interactionbetween these mountains and the boundary layer air in the LFV have impor-tant consequences for air quality in the City of Vancouver. These circulationscould act as a potential sink to the polluted valley air by venting pollutantsout of the boundary layer and into the free troposphere. Grouse Mountain(49°22’46”N, 123°4’54”W) is a popular tourist destination for outdoor activ-ities and welcomes more than 1 million visitors annually. The existence ofventing mechanisms induced by this mountain would have important healthimplications for its visitors as well as for air quality in the LFV. For this,Grouse Mountain was chosen as the study site to investigate the temporalevolution of the CBL over a mountain peak. This chapter will attempt to ad-dress the existence of venting mechanisms in Southwestern British Columbiaas well as compare the relative magnitude of these mechanisms in relationto specific mountain peaks.First, the performance of the WRF model will be evaluated by comparingvertical profiles collected by radiosonde instruments to model output to showthat the model can be reasonably used to examine CBL characteristics forthe study area. The model will then be used to examine the spatial variationin daily maximum CBL height over the Greater Vancouver area and themonthly variation of this data. Lastly, the model will be used to examinethe temporal variation of the atmosphere directly over Grouse Mountain andprovide insight to the development of a CBL over a mountain range.3.2 Model evaluationTo evaluate the performance of the WRF model the model atmosphere wascompared to vertical profiles obtained by Windsond weather balloons to as-293.2. Model evaluationsess the agreement of these data. Differences in the nature of these datasets,however, slightly perturb the representativeness of these comparisons. AllWindsond’s were launched from the topographic peak of Grouse Mountain(49°22’46.40”N, -123°4’54.49”W); therefore, only data from the atmosphereabove the model peak (49°24’24.59”N, -123°4’37.56”W) were used from theWRF output to better simulate the effect of the mountain on the verticalprofile. The rise speed of a Windsond is typically around 2-3 m/s and cantake upwards of 1.5 hours after launch before the instrument reaches itsmaximum altitude. The WRF model, on the other hand, outputs valuesfor the beginning of each hour in the day. Therefore, model output differsslightly in time to the observed values by the Windsond. In addition todifferences in time, model output differs from the observed values in space.Depending on wind speed and wind direction during the Windsond flight,the instrument can travel upwards of several kilometres horizontally whilecollecting data. This results in a skewed profile over the mountain obtainedby the Windsond opposed to the model profile which outputs a volume av-eraged value at 1333. m x 1333. m grid cells directly above the modelpeak. Despite these differences, relative similarity between the atmospheresrecorded by these datasets allow general comparisons to be made.Only data from clear-sky conditions with no significant amount of wild-fire smoke or other aerosols present in the atmosphere greater than normalbackground conditions were used in this analysis. This allows a more realis-tic comparison of data as the WRF model does not account for the radiativeeffects of aerosols if they are present in the atmosphere. The variables usedto evaluate the performance of the WRF model are potential temperature,water vapour mixing ratio, wind speed, and wind direction, which are thevariables most commonly used for boundary layer detection (Stull, 1988).303.2. Model evaluationTable 3.1: Dates and times of data used in correlations.Date Windsond (PST) Model (PST)July 25 0908, 1254, 1655, 2018 0900, 1300, 1700, 2000July 31 0908, 1256, 1659 0900, 1300, 1700August 220709, 0855, 1139,1412, 1713, 19590700, 0900, 1200,1400, 1700, 2000The dates and times of the Windsond data and WRF model output used inthese comparisons are listed in Table 3.1.Thirteen vertical profiles collected on three different days were comparedto evaluate WRF output. Windsond profiles were linearly interpolated tomodel levels for a total of 540 data points used in the comparisons. Scatterplots of predicted by the model values compared to observed by the Wind-sond values are shown in Figure 3.1. The r-squared correlation coefficient(r2) for each variable can be found in Table 3.2.Potential temperature has minimal spread and is best predicted at lowervalues for this data, representing the atmosphere nearer the surface. Athigher potential temperatures the model tends to underpredict the observedvalues. This difference may be due to the difference in location and timebetween the model and the Windsond once the instrument reaches upper-levels. The r2 value for potential temperature is 0.9736.Water vapour mixing ratio and wind speed exhibit more spread thanpotential temperature values; however, the regression line is close to a per-fect 1-1 fit for both. The r2 value for water vapour mixing ratio is 0.8491.For wind speed, r2 = 0.9073. Lower wind speeds are better predicted bythe model as exhibited by minimal spread observed for these values com-pared to model output for higher wind speeds which tend to be slightly313.2. Model evaluationlllllllllllllllllllllllll lll lllllllllllll llllllllllllllllllllllllllllllllllllllll lllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll lll llllll290 310 330 350290310330350potential temperaturepredictedobservedRegression1−1 fitlllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll0.000 0.004 0.0080.0000.0040.008water vapour mixing ratiopredictedobservedRegression1−1 fitllllllll l llll ll lllllllllll llllllllllll lllllllllll ll ll llllllllllllllll lll llllllllllll llllllllllllllllllllllllllllllllll lllllllllllllllllllll lllllllllllllllllllllll ll llllllllllllll lll ll llllll ll llllllllll llllll llllllllll l llllllllllllll l ll lll l l lllllll0 5 10 15 20 250510152025wind speedpredictedobservedRegression1−1 fitllllllllllllllllllll lllllllllllllllllllllllllllllllll llllllllllllllllll llllllllllllllllllll lllllllllllll lllllllllllllllllllllllllllllllllllllllllllllllll lllllllll ll0 100 200 300 4000100200300400wind directionpredictedobservedRegression1−1 fitFigure 3.1: Scatter plots of predicted values determined by the WRF modelcompared to observed values measured by the Windsond’s for potential tem-perature, water vapour mixing ratio, wind speed, and wind direction. Thedotted grey line represents a perfect fit and the solid red line is the linearregression.323.2. Model evaluationunderpredicted.Wind direction is not very well predicted by the model (r2 = 0.2834).Most of the data points collected for this variable are between 200° - 300°.This is a result of the predominant westerly flow at upper-levels of the atmo-sphere during the days compared. Data collected at other wind directionscan be predicted to be upwards of 115° off the observed value. This is mostlikely due to the erratic nature of wind on short time scales, especially withinthe boundary layer, in conjunction with the sampling time of the instrument.The Windsond measures instantaneous values of wind direction making ithighly influenced by individual wind gusts whereas the model outputs a 1-hour volume averaged value of wind direction, representing larger-scale bulkflow.An overlay of one of the model profiles on a Windsond profile can be seenin Figure 3.2. This Windsond was launched from the top of Grouse Mountainon July 25, 2017 at 1755 PST and the model output is from 1800 PST. Themodel does not predict the small-scale fluctuations in atmospheric propertiesthat are observed by the Windsond resulting in discrepancies between theprofiles. Potential temperature profiles agree in gradient although the modeloutput is approximately 3 - 4°C cooler than the Windsond profile. Theobserved water vapour mixing ratio profile has greater variance than modeloutput and has a maximum discrepancy of 0.000968 kg/kg. Wind speed iswithin 1 m/s for most levels and has a maximum discrepancy of 2 m/s at3600 m. Wind direction is highly variable in the bottom 2 km in this profileand model output can differ up to 138° off-direction in this layer. However,the general shape of the profiles are well predicted by the model, which isthe most important feature for boundary layer detection (Seibert. et. al.,2000).333.2.Modelevaluation290 310 3300100020003000400050006000Potential temperature [K]windsondemodel0.000 0.0060100020003000400050006000Mixing ratio [kg/kg]0 5 10 15 200100020003000400050006000Wind speed [m/s]0 100 2500100020003000400050006000Wind direction [degrees]Figure 3.2: Windsond (black) and model (grey) profiles for July 25, 2017.343.3. Spatial variations in CBL height over Southwestern British ColumbiaTable 3.2: Correlation coefficients for model and observed values.Variable r2Potential temperature 0.9736Water vapour mixing ratio 0.8491Wind speed 0.9073Wind direction 0.2834This section has shown that the WRF model is able to predict the gen-eral shape of the vertical profile over Grouse Mountain quite well, with theexception of wind direction. High correlation coefficients exist between thevertical profiles obtained by WRF model output and observed values by theWindsond weather balloons for potential temperature, water vapour mixingratio, and wind speed (Table 3.2). These results provide confidence thatthe WRF model can be reasonably used to examine the variation of the at-mosphere over a larger spatial and temporal domain. The following sectionwill use WRF model output to examine the spatial variations in CBL heightover Southwestern British Columbia to better understand the nature of theboundary layer in this region.3.3 Spatial variations in CBL height overSouthwestern British ColumbiaThermally-driven flows induced by mountainous or complex terrain can leadto large variations in CBL height over inhomogeneous landscapes (Kossmanet al., 1999). The degree of spatial variation in CBL height over a landscapeis determined by both convective instability in the lower atmosphere andmesoscale circulations that allow air to reach levels otherwise unattainable.353.3. Spatial variations in CBL height over Southwestern British ColumbiaThis section will focus on CBL height using output from the WRF model tobetter understand the terrain following behaviour of this layer over South-western British Columbia. These results will help understand some of theLiDAR findings from other studies in this region.3.3.1 Terrain following behaviour of the CBLVariations in CBL height over a landscape is related to the degree CBLheight follows terrain. It has been found that advection, entrainment, andfriction contribute to making the CBL follow terrain while gravity works tomake the CBL more level (Stull, 1998). This results in a tendency for theCBL to follow the terrain in the morning and become less terrain followingthroughout the day. Variations in CBL height over a landscape is the mostimportant factor when considering the effects of advection venting (Kossmanet. al., 1999). It is therefore important to understand the influence of terrainon CBL height in a region to understand how pollutants disperse in theatmosphere. This section will quantify the terrain following behaviour ofthe CBL throughout the day and relate it to the resulting morphology ofthe CBL over Southwestern British Columbia.The degree to which the CBL follows terrain can be assessed by plottingCBL height (mean sea level) as a function of elevation. To examine thegeneral behaviour of the CBL in Southwestern British Columbia mean CBLheight was calculated for each month for every grid cell in the model domain.Grid cells over water were removed from the dataset leaving 43,405 pointsfor comparison. The relationship between CBL height and elevation over themodel domain at 0700 PST is shown in Figure 3.3. CBL height is directlyproportional to elevation and minimal spread in these graphs indicate the363.3. Spatial variations in CBL height over Southwestern British ColumbiaCBL closely follows terrain at this time. The same analysis done for 1500PST (Figure 3.4) exhibits much greater scatter indicating the CBL becomesless terrain during the day. The larger variation in values nearer the surfaceis probably a result of having the most data points at this elevation. Morethan half the grid cells have a terrain height less than 500 m resulting in thehigh density of points at lower elevations (Figure 3.5). The larger scatterin CBL height in relation to elevation shown in this analysis makes it clearthat CBL height is more terrain following during the morning and is lessterrain following in the afternoon.373.3.SpatialvariationsinCBLheightoverSouthwesternBritishColumbiaMay June JulyAugust SeptemberMay June JulyAugust Septembere lyAug t Se berElevation [m] Elevation [m] Elevation [m]0 1000 2000 3000 0 1000 2000 3000 0 1000 2000 3000Elevation [m]0 1000 2000 3000Elevation [m]0 1000 2000 3000PBL height MSL [m]0100020003000PBL height MSL [m]0100020003000Figure 3.3: Modeled mean CBL height MSL vs. terrain height at 0700 PST.383.3.SpatialvariationsinCBLheightoverSouthwesternBritishColumbiaMay June JulyAugust SeptemberElevation [m]May June JulyAugust SeptemberElevation [m] Elevation [m]0 1000 2000 3000 0 1000 2000 3000 0 1000 2000 3000Elevation [m]0 1000 2000 3000Elevation [m]0 1000 2000 3000PBL height MSL [m]0100020003000PBL height MSL [m]0100020003000ay June JulyAugust temberFigure 3.4: Modeled mean CBL height MSL vs. terrain height at 1500 PST.393.3. Spatial variations in CBL height over Southwestern British ColumbiaTable 3.3: r and T values at 1400 PST for each month.May June July August Septemberr 0.91 0.93 0.93 0.92 0.93T 0.38 0.37 0.45 0.48 0.32The terrain following behaviour of the CBL can be further quantifiedby calculating the parameters r and T = (σtopo − σh)/σtopo, where r is thelinear correlation coefficient and σtopo and σh are the standard deviations ofterrain height and CBL height, respectively. A CBL that exhibits completeterrain following behaviour will have a T = 0 and r = 1. A complete non-terrain following CBL will have a T = 1 and r = 0. These parameters werecalculated using the mean CBL height data used in the previous analysis foreach hour in the day to better understand the evolution of the boundary layer(Figure 3.6). For all months, CBL height closely follows terrain between 0000- 0900 PST. The CBL then begins to exhibit less terrain following behaviourbetween 0900 - 1900 PST indicated by an increase of T and decrease of r.The least terrain following behaviour of the CBL is seen at 1400 PST for allmonths. T values at this time range between 0.38 and 0.48. The r and Tvalues for each month at 1400 PST are listed in Table 3.3.The analysis above quantified the terrain following behaviour of the CBLby using mean CBL height values predicted by the WRF model in compar-ison to terrain height. The parameters T (= (σtopo − σh)/σtopo) and r werecalculated to provide quantifiable evidence of this finding and to track thediurnal cycle of boundary layer behaviour. The following section will nowprovide a picture of how this behaviour affects the overall morphology ofthe CBL over Southwestern British Columbia for each month of the studyperiod.403.3. Spatial variations in CBL height over Southwestern British ColumbiaTerrain height [m]Frequency0 500 1500 2500050001000015000Figure 3.5: Histogram of terrain height from WRF output (only cells overland).413.3.SpatialvariationsinCBLheightoverSouthwesternBritishColumbiaMayTime [PST]2 6 10 14 18● ● ● ● ●● ● ●●●●●●● ● ●●●●● ● ● ● ●●rTJuneTime [PST]2 6 10 14 18 22● ● ● ● ● ● ● ● ●●●●●●● ●●●●● ● ● ● ●●rTJulyTime [PST]2 6 10 14 18 22● ● ● ● ● ● ● ●●●●●●●● ●●●●●● ● ● ●●rTAugustTime [PST]2 6 10 14 18● ● ● ● ● ● ● ●●●●●●●● ●●●●●● ● ● ●●rTSeptemberTime [PST]2 6 10 14 18 22● ● ● ● ● ● ● ● ●●●●●● ● ●●●● ● ● ● ● ●●rTAll monthsTime [PST]2 6 10 14 18 22● ● ● ● ● ● ● ●●●●●●●● ●●●●●● ● ● ●●rTTime [PST] Time [PST] Time [PST]02 06 10 18 2214MayTime [PST]2 6 10 14 18● ● ● ● ●● ● ●●●●●●● ● ●●●●● ● ● ● ●●rTJuneTime [PST]2 6 10 14 18 22● ● ● ● ● ● ● ● ●●●●●●● ●●●●● ● ● ● ●●rTJulyTime [PST]2 6 10 14 18 22● ● ● ● ● ● ● ●●●●●●●● ●●●●●● ● ● ●●rTAugustTime [PST]2 6 10 14 18● ● ● ● ● ● ● ●●●●●●●● ●●●●●● ● ● ●●rTSeptemberTime [PST]2 6 10 14 18 22● ● ● ● ● ● ● ● ●●●●●● ● ●●●● ● ● ● ● ●●rTAll monthsTime [PST]2 6 10 14 18 22● ● ● ● ● ● ● ●●●●●●●● ●●●●●● ● ● ●●rTTime [PST] Time [PST] Time [PST]02 06 10 18 2214 02 06 10 18 221402 06 10 18 2214 02 06 10 18 2214 02 06 10 18 22140. y J e JulAu st Sep ber All m thsFigure 3.6: r and T values from mean CBL height vs. terrain height.423.3. Spatial variations in CBL height over Southwestern British Columbia3.3.2 CBL morphologyThe degree to which the CBL follows terrain affects the overall morphologyof the CBL over a mountainous landscape. Inhomogeneities in CBL heightcan add a mechanism to vent boundary layer air into the free troposphere viaadvection. The introduction of boundary layer air to the free tropospherecan then alter the stability of the atmosphere at those levels and also have alarge impact on the distribution of pollutants in the atmosphere (Kossmanet. al., 1999). A better understanding of where advection venting may beoccurring in Southwestern British Columbia will help understand the ob-served distribution of pollutants in this region and also provide insight tothe formation of elevated layers that have previously been observed over theLFV (McKendry, 1997). This section will use the WRF model to examinethe morphology of the CBL over Southwestern British Columbia by firstanalyzing mean CBL depth (above ground level) over the model domainto identify locations that tend to develop a deeper CBL. Then, mean CBLheight (mean sea level) will be analyzed to provide a picture of what theCBL looks like for an average month. This analysis will provide a muchneeded picture of the variations in CBL height experienced over Southwest-ern British Columbia.Mean CBL depth for each month at each grid cell was calculated andthen contoured over the model domain (Figure 3.7). This analysis was donefor the time 1400 PST to understand CBL morphology when it exhibits theleast terrain following behaviour during the day. For all months, mean CBLdepth increases moving eastward into the LFV. This is expected as CBLgrowth is dependent on surface sensible heat flux which increases movingaway from the coast and the atmosphere becomes less stable. Mean CBL433.3. Spatial variations in CBL height over Southwestern British ColumbiaTable 3.4: Mean CBL depth (above ground level) for each month for theentire model domain.May June July August Septembermean CBL depth [m] 462 510 427 367 366depth for locations near the coast, such as the City of Vancouver, rangebetween 200 - 600 m.There are several locations over the mountainous terrain in the northhalf of the domain that exhibit a similar mean CBL depth as those seenin the Fraser Valley. Appian Mountain in Squamish, BC (49°31’29.48” N,123°06’26.03” W) sticks out as a “hotspot” for having a deeper CBL thanits immediate surroundings. This may be the result of greater convectiveactivity being present at this location or a result of thermally-induced flowsbringing air to levels higher than it would reach from convection alone. MeanCBL depth AGL over this mountain varied between 700 - 1000 m duringthe study period with the maximum mean CBL depth being achieved inJune. June has the highest mean CBL depth of all the months analyzed andseveral high-elevation areas achieve a mean CBL depth of over 700 m.The mean CBL depth for each month averaged over the entire modeldomain for this time (1400 PST) are listed in Table 3.4. June has the highestmean CBL height for the entire domain at 510 m. Each subsequent monthdecreases in value with September having the lowest mean CBL depth AGLat 366 m. The greatest change in mean CBL depth occurs between July andAugust where there is a 60 m difference between these months.A better visualization of the morphology of the CBL is seen by contour-ing mean CBL height (mean sea level) over the model domain (Figure 3.8).Variation in CBL height is more important when considering the effects of443.3. Spatial variations in CBL height over Southwestern British Columbiaadvection venting because it is a result of the overall inhomogeneity in theCBL rather than in relation to terrain height. Although the CBL tends togrow deeper in the Fraser Valley, when considering terrain height at eachpoint mean CBL height MSL is several hundred meters lower than those ob-served over the mountains. In June, the month with the highest mean CBLheight over the entire domain, mean CBL height in the Fraser Valley reachesa maximum of approximately 1300 m. The highest mean CBL height in thedomain is seen over Mount Garibaldi (49°51’02”N, 123°00’17”W). Terrainheight at this location is 2,678 m and mean CBL height reaches a maximumof over 3000 m MSL in August and a minimum of of 2800 m MSL in May.The CBL does not tend to grow deeper over this mountain; however, thehigh-elevation here leads to the highest mean CBL height in the entire studydomain.There are large spatial variations in CBL height even at the time whenCBL experiences the least terrain following behaviour during the day. TheCBL tends to grow deeper (above ground level) in the eastern half of theFraser Valley and can be several hundred meters lower at areas near thecoast. CBL height (mean sea level), however, is consistently lower in thevalley than over any of the mountain peaks. All of the North Shore Moun-tains exhibit a CBL that reaches at least 1500 m MSL for this time. Thisresult shows that the North Shore Mountains (including Grouse Mountain)are capable of venting boundary layer air into the free troposphere if thereis advection towards locations with a lower CBL height.453.3.SpatialvariationsinCBLheightoverSouthwesternBritishColumbia0 150 300450 600 750900105012001350 0 150 300450 600 750900105012001350 0 150 300450 600 7509001050120013500 150 300450 600 750900105012001350 0 150 300450 600 7509001050120013500 150 300450 600 750900105012001350 0 150 300450 600 750900105012001350 0 150 300450 600 7509001050120013500 150 300450 600 750900105012001350 0 150 300450 600 75090010501200135049N49N 49N49N 49N123W 123W 123W123W 123WMay June JulyAugust SeptemberFigure 3.7: Monthly mean CBL depth (above ground level) at 1400 PST model output segregated by month.Shaded values are in meters.463.3.SpatialvariationsinCBLheightoverSouthwesternBritishColumbia0 300 600 9001200150018002100240027003000 0 300 600 900120015001800210024002700300049N49N 49N49N 49N123W 123W 123W123W 123WMay June July0 300 600 9001200150018002100240027003000 0 300 600 900120015001800210024002700300049N49N 49N49N 49N123W 123W 123W123W 123WAugust SeptemberFigure 3.8: Monthly mean CBL height (mean sea-level) at 1400 PST model output segregated by month. Shadedvalues are in meters.473.3. Spatial variations in CBL height over Southwestern British Columbia3.3.3 Evidence of advection ventingThe previous sections have shown that the terrain following behaviour ofthe CBL leads to large variations in CBL height over Southwestern BritishColumbia. The inhomogeneity in CBL height due to this behaviour pro-vides evidence that high-elevation terrain and the resulting CBL height canbe responsible for some of the elevated layers observed in this region viaadvective venting. This section will provide a more in-depth look at howelevated layers in the LFV can be sourced from the mountains just northof the City of Vancouver by analyzing wind patterns in conjunction withknowledge of CBL behaviour. The goal of this section is to provide a moremechanistic view of the formation of elevated layers by examining modeloutput from a single day in the study period.August 04, 2017 has been chosen as the case study day to more closelyexamine the venting mechanisms previously described. The synoptic condi-tions for the day from the NCEP/NCAR Reanalysis 1 dataset are shown inFigure 3.9. A 500 mb ridge aligned at 137°W and a surface low residing overthe majority of Washington brought northwesterly flow at upper-levels andto the majority of Vancouver’s mainland at the surface. Observed hourlywind data from a station located at Vancouver International Airport (YVR)(49°11’41”N, 123°11’02”W) shows this northwesterly flow existing through-out the day until 1900 PST when wind shifts to a northeasterly flow (Figure3.10a). Evidence of a mountain-valley circulation existing on this day isprovided by observed wind data from a station located at Mahon Park inNorth Vancouver (49°19’27”N, 123°04’57”W) (Figure 3.10b). This station isapproximately 1.5 km north of Vancouver Harbour and 4 km south of thebase of Grouse Mountain. Wind direction from this station shows a northerly483.3. Spatial variations in CBL height over Southwestern British Columbiaflow existing in pre-dawn hours and a southerly flow existing between 0600- 1800 PST. Wind shifts back to a northerly flow after these hours. Sunriseon this day was at 0536 PST which corresponds with the time the southerlyflow was first observed. Wind flow switches back to northerly a few hoursbefore the sunset at 21:02 PST. Surface wind patterns on this day outputby the WRF model are shown in Figure 3.10c. Model data agrees well withthe observed values at the two stations referenced above. The shift in winddirection during daytime hours at locations near the base of the North ShoreMountains provides evidence that a mountain-valley circulation was presenton this day.Vertical cross-sections from the model were obtained to provide evidenceof advective venting giving rise to an elevated layer. The WRF model usedin this analysis does not have an option to add aerosols as a passive tracer;however, water vapour mixing ratio is a conserved variable that can beused to track boundary layer air (Stull, 1988). A north-south transect fixedat 123.0771°W from 49°N to 49.8°N is shown in Figure 3.11. This cross-section is from August 04, 2017 at 1600 PST and passes directly over themodel peak of Grouse Mountain, which is illustrated by the left-most peakseen at 49.39°N. The solid red line is modeled CBL height MSL and thecontours represent potential temperature. This image is a view of CBLheight at its maximum value for the day. CBL height is around 200 mMSL for most of the valley and approximately 1900 m MSL over GrouseMountain. The highest CBL height in this figure is over the peak at 49.60°Nand reaches approximately 2900 m MSL. CBL height over these mountainsis approximately 2 km higher than that over the valley. Shaded values inFigure 3.11c is vertical velocity in the atmosphere. Because of the modeloutput being a volume averaged value for a 1333. m x 1333. m grid cell,493.3. Spatial variations in CBL height over Southwestern British Columbia500mb GEOPT HEIGHTMEAN SEA-LEVEL PRESSUREFigure 3.9: Synoptic maps for August 04, 2017 at 1100 PST fromNCEP/NCAR Reanalysis 1. 500 mb geopotential height (top) and meansea-level pressure (bottom).the values shown here do not represent individual thermals, rather it is thegeneral flow of the atmosphere that is represented. All locations within theboundary layer over the mountains have a positive vertical velocity. CBL503.3. Spatial variations in CBL height over Southwestern British Columbia5 10 15 200200Vancouver International AirportTime [PST]Wind direction [degrees]5 10 15 200200North Vancouver − Mahon ParkTime [PST]Wind direction [degrees]a)b)c)Figure 3.10: Station plots from (a) Vancouver International Airport (b)Mahon Park, North Vancouver (Environment Canada). (c) is WRF model10 m winds at 1400 PST.513.3. Spatial variations in CBL height over Southwestern British Columbiaheight is seen to be related to vertical velocity where larger vertical velocityvalues give rise to a higher CBL height. Over the tallest peak at 49.60°Nvertical velocity exceeds 0.6 m/s. The high CBL height at this locationand over the mountains in general can be attributed to the strong verticalvelocities seen over these locations.523.3.SpatialvariationsinCBLheightoverSouthwesternBritishColumbia b) c)a)Height [km] 49.39 49.6649.13 49.39 49.66Water vapour mixing ratio [kg/kg] z-wind component [m/s]-0.6 -0.4 -0.2 0 0.2 0.4 0.6.001 .0028 .0046 .0064 .008249N 123W200 1000 1800 2600Figure 3.11: Vertical cross-sections for August 04, 2017 at 1400 PST from WRF output. (a) location of transect(b) cross-section of water vapour mixing ratio (c) cross-section of vertical velocity. Grouse Mountain is peak at49.39°.533.3. Spatial variations in CBL height over Southwestern British ColumbiaWind data, indicated by the black wind vectors, clearly show air fromwithin the boundary layer over the mountains being advected into the freetroposphere over the valley due to the inhomogeneity in CBL height. Theshaded values in Figure 3.11b are water vapour mixing ratios. Air withinthe boundary layer over the mountains have the highest water content withmixing ratio values reaching 0.0075 kg/kg. Over the valley, similar valuesto this are observed near the surface, however, a dry layer exists between800 - 1800 m MSL. Another moist layer is seen between 1800 - 2600 m MSLbefore transitioning to very low mixing ratio values in the upper atmosphere.The location of the elevated moisture layer over the LFV agrees with thedirection of the wind vectors within the CBL over the mountains indicatingthat advective venting due to the difference in CBL height MSL betweenthe mountains and the valley is the most probable source of this air.In this section, it has been shown that large variations in CBL heightexist in Southwestern British Columbia in part due to the terrain followingbehaviour of the CBL. Even when the CBL exhibits the least terrain fol-lowing behaviour in the afternoon, CBL height MSL varies several hundredmeters between the mountainous areas and the low elevation terrain found inthe Fraser Valley. Analysis of the vertical wind profiles using model outputin relation to CBL height have shown that elevated layers observed over theLFV can be sourced from the mountains on the northern edge of the valley.Furthermore, evidence has been provided that Grouse Mountain is capableof venting boundary layer into the free troposphere, potentially acting as asink for pollutants emitted near the surface. The next section will focus onthe evolution of the CBL over Grouse Mountain and provide deeper insightto the development of a CBL over a mountain peak.543.4. Temporal variation in CBL height over Grouse Mountain3.4 Temporal variation in CBL height overGrouse MountainGrouse Mountain is a popular tourist destination and venting mechanismsinduced by this mountain have important health implications for the Cityof Vancouver. The temporal evolution of the CBL over the peak of GrouseMountain will be analyzed in this section to provide insight into the generaltrend of CBL behaviour at this location. First, the daily variance in CBLdepth (above ground level) will be analyzed. Monthly mean CBL depth willthen be calculated to identify monthly trends within the study period. Oncemean CBL depth for this location has been calculated, the general behaviourof the CBL will be discussed.CBL depth AGL over Grouse Mountain for each day in the study periodsegregated by month can be found in Figure 3.12. The black lines in thisfigure are CBL depth for each day in the month and the red line is mean CBLdepth for the month. Maximum CBL depth varies upwards of over 1 kmwithin a single month. The largest difference in maximum CBL depth withina month occurred in June where the largest CBL depth was 1758 m AGL seenon June 23, 2017. The lowest daily maximum CBL depth in this month was357 m which occurred the previous day (June 22, 2017). This result showsthat maximum CBL depth over Grouse Mountain can range upwards of 1400m within a single month. The least amount of variation in daily maximumCBL depth occurred in July where the lowest daily maximum CBL depthwas 410 m AGL on July 20, 2017 and largest CBL depth seen was 1110 mAGL on July 17, 2017. A time series illustrating the trend in daily maximumCBL depth AGL is shown in Figure 3.13. Additionally, the maximum CBLdepth AGL experienced for each month in the study period is listed in Table553.4. Temporal variation in CBL height over Grouse Mountain3.5. To better visualize the temporal variation of CBL depth a boxplot ofdaily maximum CBL depth by month is shown in Figure 3.14. It should benoted that the YSU PBL scheme implemented by the WRF model alwaysoutputs a value greater than zero for CBL depth. This may not be physicallyrealistic, as some mountain peaks may, at times, be above the influence ofboundary layer air (Ghallagher et. al., 2011). Specifically, the PBL depthsduring night outputted by the WRF model may be higher than the truevalues.563.4. Temporal variation in CBL height over Grouse MountainMayAltitude [AGL]01:00 05:00 9:00 13:00 17:00 21:0001000JuneAltitude [AGL]01:00 05:00 9:00 13:00 17:00 21:0001000JulyAltitude [AGL]01:00 05:00 9:00 13:00 17:00 21:0001000AugustAltitude [AGL]01:00 05:00 9:00 13:00 17:00 21:0001000SeptemberTime [PST]Altitude [AGL]01:00 05:00 9:00 13:00 17:00 21:0001000Figure 3.12: Time series of PBL depth (above ground level) over GrouseMountain for all days by month. Red line is average PBL depth.573.4.TemporalvariationinCBLheightoverGrouseMountainMay Jun Jul Aug Sep Oct05001500Daily maximum PBL depthAltitude AGL [m]Figure 3.13: Maximum daily PBL depth (above ground level) for each day in the study period.583.4.TemporalvariationinCBLheightoverGrouseMountain0500100015002000MayAltitude AGL [m]ll0500100015002000Junell0500100015002000July0500100015002000Augustllll0500100015002000SeptemberFigure 3.14: Boxplot of daily maximum PBL depth (above ground level) by month.593.4. Temporal variation in CBL height over Grouse MountainTable 3.5: Maximum CBL depth AGL for each month.May June July August Septembermaximum CBLdepth [m]1390.48 1757.77 1122.96 1213.2 1660.69The general evolution of the CBL over Grouse Mountain can be furtherassessed by analyzing mean CBL depth AGL throughout the day for eachmonth. Figure 3.15a shows mean CBL depth for each month overlaid forbetter comparison. The highest mean CBL depth was experienced in Junewith a value of 750 m. All of the other months in the study period havea mean CBL depth within 150 m AGL of this value except for Septemberwhich experienced a maximum mean CBL depth of 510 m AGL. For allmonths, CBL depth is near the surface up until sunrise. The CBL thenbegins to grow in depth starting at 0600 PST. A maximum CBL depthwas reached between 1400 - 1600 PST for all months and then begins tosubside after this time. The derivative of these lines were calculated tofind mean CBL growth rate [m/hr] for each month (Figure 3.15b). Thefastest boundary layer growth is seen in the morning hours between 0700 -0900 PST for all months. At maximum growth, the boundary layer can beseen to reach a mean growth rate of just over 100 m/hr. The growth rategradually declines until approximately 1500 PST where the boundary layerthen begins to decrease in depth due to subsidence. The quickest decrease inboundary layer depth is seen between 1700 - 2000 PST where the maximumdecrease can be upwards of -150 m/hr.603.4. Temporal variation in CBL height over Grouse MountainAverage PBLHTime [PST]Height AGL [m]00:00 03:00 06:00 09:00 12:00 15:00 18:00 21:0002004006008001000MayJuneJulyAugustSeptemberPBL growth rateTime [PST]PBL growth rate [m/hr]00:00 03:00 06:00 09:00 12:00 15:00 18:00 21:00−150−50050100MayJuneJulyAugustSeptembera)b)Figure 3.15: Average evolution of PBL depth AGL (top) and average PBLgrowth rate (bottom).613.5. Discussion3.5 DiscussionThis chapter has worked to characterize the spatial and temporal variationsof the CBL over Southwestern British Columbia using WRF model output.The model was evaluated with Windsond profiles and was found to be ingood agreement with observed values. To understand the degree to whichthe CBL follows terrain CBL height was plotted in relation to elevation andthe parameters r and T were calculated. The CBL exhibited a diurnal cyclein which the CBL was more terrain following in the morning and becomesless terrain following during the day. This finding agrees with the large-eddysimulations performed by Lenschow et al. (1979) and Banta (1984) thatreported similar behaviour over hilly terrain. A similar analysis on CBLbehaviour over the Swiss Alps (De Wekker, 2002) and over SouthwesternGermany (Kalthoff et al., 1988) showed the same general behaviour as thisanalysis, although r and T values differed. De Wekker (2002) modeled CBLheight over Jungfraujoch, Switzerland and found the least terrain followingbehaviour at 1500 LST with a T value of 0.35 and correlation-coefficientof 0.78. In addition to climatic differences between these regions, many ofthe elevations analyzed are much higher than what was used in the analysisover Southwestern British Columbia. The larger variance in elevation maybe responsible for the lower r values found here. Furthermore, De Wekker(2002) reported values from a single day whereas this study focused on meanCBL height for each month. For any single day the CBL over SouthwesternBritish Columbia may be more or less terrain following than the valuesreported here; however, this analysis has shown the general trend of theCBL is to become less terrain following in the afternoon.Mean CBL height MSL values provided by the WRF model show the623.5. DiscussionCBL is significantly higher over the mountains than over the LFV. Thelargest variation in CBL height through the LFV appears to be due to theproximity to the coastline where locations further up the LFV experiencehigher mean CBL heights (MSL) and depths (AGL) than areas nearer thecoast. The values output my the WRF model agree well with values previ-ously observed in this region. Van der Kamp and McKendry (2010) analyzedthe boundary layer in Metro Vancouver using sodar data and found a meanCBL depth of 500 m AGL on days with convective clouds present duringsummer months (June, July, August). On days with no convective clouds ob-served, a mean CBL depth of 450 m AGL was found. The mean CBL depthsanalyzed in this section were calculated using all days for each month and istherefore comprised of days both with convective clouds and no convectiveclouds observed. CBL depth varies from day-to-day depending on local andsynoptic conditions; however, the values reported here represent the generaltrend in CBL height over this region.Mesoscale circulations have been previously observed in the LFV (Okeand Hay, 1994, McKendry and Lundgren, 2000), however, this chapter usedWRF model output to provide a more mechanistic view of venting mecha-nisms. The results from this chapter have shown that the North Shore Moun-tains are capable of venting boundary layer air into the free troposphere viathe formation of mountain-valley and other mesoscale circulations. Addi-tionally, the large variation in CBL height MSL due to the Pacific Coastmountains allows for the formation of elevated layers via advective vent-ing. The highest mean CBL height MSL over the study region was overMount Garibaldi, the location with the highest elevation in the domain.This mountain may be the source of the high-altitude elevated layers pre-viously observed in the LFV (McKendry et. al., 1997; McKendry et. al.,633.5. Discussion2011). The venting mechanisms discussed here are the same mechanismsresponsible for the formation of elevated layers in the similar regions aroundthe world, such as the LA basin (Wakimoto and McElroy, 1986) and Beijing,China (Chen et. al., 2009).CBL depth AGL over Grouse Mountain was analyzed to examine thebehaviour of the CBL over a mountain peak. On average, May tended tohave the highest maximum CBL depth out of all months (1760 m). Averagemaximum CBL depth decreased with each proceeding month with Septem-ber having the lowest at 678.26 m AGL. Interestingly, May and Septemberexperienced the most variation in daily CBL depth. As CBL depth is de-pendent on atmospheric stability, a more in depth analysis of the synopticconditions that occurred for each month of the study period would help elu-cidate these findings. Mean CBL depth AGL exhibits a similar diurnal cycle,although different values, for each month with the maximum depth beingachieved at 1400 PST. The time with the fastest CBL growth varied butusually occurred somewhere between 0700 - 0900 PST. The fastest growthoccurred in August at 0800 PST at over 100 m/hr. A similar study usedmodel output over southern India reported a CBL growth rate of 200 m/hr,however, this was during the Spring, rather than Summer. The results fromthis chapter agree with the previous findings and has provided a baseline forfurther investigatations of CBL behaviour in the future.This chapter has worked to show how the degree the CBL follows ter-rain affects the overall morphology of the CBL over Southwestern BritishColumbia. The various mesoscale circulations that can exist in this regionprovide a mechanism for boundary layer air to penetrate into the free tropo-sphere, thus altering the distribution of pollutants in the atmosphere. Thefollowing section will now focus on how a thick layer of wildfire smoke in the643.5. Discussionlower atmosphere may alter this behaviour.65Chapter 4Case Study: Wildfire smokeimpacts on ConvectiveBoundary LayerDevelopment4.1 IntroductionWildfires emit large amounts of gases and particulate matter into the atmo-sphere. The direct effect of these particles existing in high concentrations asa result of a wildfire have an obvious impact on public and environmentalhealth. However, there is an insufficient amount of research on how smokeplumes can indirectly affect air quality by altering the stability of the loweratmosphere and CBL development.Wildfire smoke is primarily composed of carbon dioxide, water vapour,carbon monoxide, and particulate matter. Particulate matter is mainly re-sponsible for the reduced visibility through a smoke plume by scatteringradiation resulting in more diffuse radiation and less direct radiation at thesurface. Black carbon, or soot, makes up a significant portion of the partic-ulate matter in a smoke plume and is mainly responsible for the absorbing664.1. Introductionproperties of smoke. Atmospheric stability can be altered by a thick layerof smoke due to these scattering and absorbing properties. Less radiationreaching the surface results in a cooler surface layer and radiation beingabsorbed at higher levels heats the atmosphere. This works to stabilize theatmosphere and can suppress CBL growth. In theory, these effects wouldreduce the depth of the CBL or hinder the development of a CBL altogether.British Columbia experienced a particularly active wildfire season in2017. Approximately 1.3% of BC’s total area was burned marking thelargest total area burnt in recorded history (Environment Canada). Themajority of these wildfires originated from BC’s interior in part due to drierthan normal patterns for this region. On several days throughout Summer2017, a thick plume of wildfire smoke traversed over the LFV. The longestof these events was a smoke plume that was advected to the region on July31, 2017 and remained in place until August 09, 2017. This unique eventprovided an excellent opportunity to investigate the radiative effects of asmoke plume and how this may alter the thermodynamic structure of theCBL.This chapter will use observations of a thick layer of wildfire smoke thatsettled over Southwestern British Columbia in 2017 to examine how thisaffected CBL development in this region. First, the synoptic overview willbe described as well as observations of the smoke plume with LiDAR and airquality measurements. Vertical profile data collected by Windsond weatherballoons will then be compared between a clear day and a smoke day toaddress the radiative effects of a smoke plume on atmospheric stability.674.2. Observations of the smoke layer4.2 Observations of the smoke layerA thick smoke plume settled over Southwestern British Columbia late atnight on July 31, 2017. Satellite imagery shows smoke residing throughoutthe LFV and extending into Puget Sound and the southern part of Vancou-ver Island (Figure 4.1). The plume remained in place until August 11, 2017due to a persistent upper level trough over the eastern Pacific. LiDAR im-agery from the mMPL stationed on Grouse Mountain shows the smoke firstarriving at around 2100 PST. The first signals of smoke were recorded atapproximately 1 km above ground level (Figure 4.2). This is most likely dueto the smoke “wall” coming in at an angle rather than being mixed down-wards or subsiding. High backscatter values are seen throughout the bottomkilometre of the atmosphere for the entire period the smoke is present. TheLiDAR on Grouse Mountain was at approximately 1110 m MSL and thesmoke extended to the valley floor. The smoke layer can be approximatedto be roughly 2 km above ground level over the LFV.684.2. Observations of the smoke layerFigure 4.1: MODIS imagery of wildfire smoke residing over SouthwesternBritish Columbia.694.2.Observationsofthesmokelayer012345Range (km)Time [PST]00:50:0008/01/201721:40:0008/02/201718:40:0008/04/201715:30:0008/06/201712:20:0008/08/201709:10:0008/10/201706:15:0008/12/201703:05:0008/14/2017Counts km #/ (µs µJ) 4.2: LiDAR imagery of the smoke layer from LiDAR stationed on Grouse Mountain.704.2. Observations of the smoke layerA meteorology station at Vancouver International Airport (YVR) showsPM2.5 increasing an order of magnitude coinciding with the arrival of thesmoke layer (Figure 4.3). PM2.5 was measured at about 1 µg/m3 beforethe arrival of the smoke and a maximum of 52.9 µg/m3 was measured mid-afternoon on August 09, 2017. At this time, aerosol optical depth (AOD) at500 nm measured at Saturna Island (48.775°N, 123.128°W) was recorded at1.018. However, the maximum AOD at 500 nm recorded during this eventwas approximately 4 measured on August 04, 2017.Irradiance measured at YVR significantly decreased during this event.The day before the smoke arrived the maximum irradiance was measured atover 800 W/m2. Values dropped down to approximately 600 W/m2 at thesame time the next day. No clouds were present on this day indicating thatthe drop of irradiance is directly attributable to the presence of the smokeplume. The least amount of irradiance reaching the surface coincides withthe highest AOD on August 04, 2017. On this day the maximum irradiancemeasured was 488 W/m2. This represents a nearly 40% drop in irradiancefrom what was measured before the arrival of the smoke plume.714.2.ObservationsofthesmokelayerAug 02 Aug 07 Aug 1201020304050PM2.5 [µgm3]Aug 02 Aug 07 Aug 120200400600800Time [PST]Irradiance [W/m2]Aug 02 Aug 07 Aug 12PM#.%[µg/m']01020304050Aug 02 Aug 07 Aug 1201020304050PM2.5 [µgm3]Aug 02 Aug 07 Aug 120200400600800Time [PST]Irradiance [W/m2]Time [PST]Aug 02 Aug 07 Aug 120200400600800Irradiance [W/m#]Figure 4.3: Particulate matter 2.5 concentrations (top) and irradiance values (bottom) measured at VancouverInternational Airport (YVR) for the duration of the smoke event.724.3. Impacts on stabilityThe smoke plume dissipated on August 12, 2017 due to a change insynoptic patterns. The upper level trough moved eastward and changes insea-level pressure resulted in a reversal of ground-level wind direction bring-ing in clean marine air. PM2.5 and irradiance measured at YVR returnedto normal levels on this day.Southwestern British Columbia has been affected by wildfire smoke plumesbefore however the long-lasting nature of this plume makes it a unique eventfor scientific research. The following section will use vertical profile mea-surements of the smoke plume to address the effects this event had on at-mospheric stability and thus boundary layer development.4.3 Impacts on stabilityParticulate matter in smoke plumes has the largest effect on the stabilityof the atmosphere. Particulates are efficient scatterers and black carbonabsorbs radiation. This results in more radiation being absorbed at levelshigher than it would otherwise. Additionally, less direct radiation and morediffuse radiation is observed at the surface. This can alter the lapse rate ofthe atmosphere which is what is usually considered when analyzing stability.Several Windsond’s were launched from the peak of Grouse Mountainduring the smoke event as well as on clear days. As a rough look at thestabilizing effects of wildfire smoke, the potential temperature gradients ofthe atmosphere can be compared between the day before the smoke arrivedand a few days after the smoke was in place. Figure 4.4 shows a comparisonbetween the vertical profiles from a Windsond launched at 1256 PST onJuly 31, 2017 (before the arrival of smoke) and one launched at 1313 PSTon August 09, 2017 (when smoke was present). The first panel in the figure734.3. Impacts on stabilityis a vertical profile of NRB recorded by the LiDAR at 1300 PST for bothdays.744.3.Impactsonstability−0.4 0.0 0.4010002000300040005000NRBAltitude AGL [m]290 310 330010002000300040005000Potential temperature [K]0.000 0.004 0.008010002000300040005000Water vapour mixing ratio [kg/kg]010002000300040005000Altitude AGL [m]010002000300040005000010002000300040005000−0.4 0 0.4010002000300040005000NRBAltitude AGL [m]290 310 33010002000300040005000Potential temperature [K]0 4 0.008010002000300040005000Water vapour m xing ratio [kg/kg]−0.4 0.0 0.4010002000300040005000NRBAltitude AGL [m]290 310 330010002000300040005000Potential temperature [K]0.000 0.004 0.008010002000300040005000Water vapour mixing ratio [kg/kg]− .4 0.0 0.4010002000300040005000NRBAltitude AGL [m]290 310 30010002000300040005000Potenti l temperature [K]0. 0 0. 04 0. 08010002000300040005000Water vapour mixing ratio [kg/kg].000 0.002 0. 4 .006 0.008 0.0100100030005000Water vapour mixing ratio [kg/kg]−5 −4 −3 −2 −1 0020004000NRBAltitude AGL [m]August 09July 31Figure 4.4: Vertical profiles collected by a Windsond weather balloon for July 31, 2017 and August 09, 2017.754.3.Impactsonstability290 310 330010002000300040005000Potential temperature [K]Altitude AGL [m]0.000 0.004 0.008010002000300040005000Water vapour mixing ratio [kg/kg]0.000 0.004 0.008010002000300040005000Water vapour mixing ratio [kg/kg]−0.4 0.0 0.4010002000300040005000NRBAltitude AGL [m]290 310 330010002000300040005000Potential emperature [K]0.000 0.004 0.008010002000300040005000Water vapou  mixin  ratio [kg/kg]−0.4 0.0 0.4010002000300040005000NRBAltitude AGL [m]290 310 330010002000300040005000Potential temperature [K]0.000 0.004 0.008010002000300040005000Water vapour mixing ratio [kg/kg]−0.4 .0 0.4010002000300040005000NRBAltitude AGL [m]290 310 330010002000300040005000Potential t mperatur  [K].000 .004 .008010002000300040005000Water vapour mixing r tio [kg/kg].000 .002 .004 0.006 0.008 0.0100100030005000Water vapour mixing ratio [kg/ ]−5 −4 −3 −2 −1 0020004000NRBAltitude AGL [m]August 09July 31Figure 4.5: Vertical profile output from the WRF model for July 31, 2017 and August 09, 2017.764.3. Impacts on stabilityA near-adiabatic layer is observed on July 31, 2017 extending to roughly500 m AGL before potential temperature begins to increase. This coincideswith a constant water vapour mixing ratio in this layer which also experi-ences a sharp decrease around 500 m. Based on these values, the CBL depthis approximately 500 m AGL. WRF model PBL depth output for this timeis 474 m AGL which matches well with the vertical profile observations. Onthe smoke day, however, no clear adiabatic layer is observed. LiDAR datashows relatively high NRB values on this day throughout the bottom 1200m of the atmosphere indicating the extent of the smoke layer, however, noclear signal is seen in the potential temperature profile at this level. A CBLis not observed on this day.As a rough comparison of the stability of these two days the potentialtemperature gradients of the atmospheres can be compared. Figure 4.6shows this comparison using the change in potential temperature at 100 mincrements for the bottom 2 km of the atmosphere (Equation 4.1). Potentialtemperature gradient is defined in Equation 4.1.Γθ =δθδz(4.1)where Γθ is the potential temperature gradient [K/km], δθ is the changein potential temperature [K], and δz is the change in elevation [km].774.3.Impactsonstability−15 −5 5 150500100015002000Potential temperature lapse rate  [°C/km]Altitude AGL [m]−15 −5 5 150500100015002000Potential temperature lapse rate  [°C/km]−15 −10 −5 0 5 10 150500100015002000Potential temperatureAltitude AGL [m]August 09July 31−15 −10 −5 0 5 10 150500100015002000lapse rate  [°C/km]Altitude AGL [m]A gust 09July 31−1 −10 −5 0 5 10 150500100015002000Potential temperatureAltitude AGL [m]August 09July 31−1 −10 −5 0 5 10 150500100015002000lapse rate  [°C/km]Altitude AGL [m]A gust 09July 31−5 −4 −3 −2 −1 0020004000NRBAltitude AGL [m]August 09July 31Figure 4.6: Potential temperature gradients of the atmosphere on July 31, 2017 (clear day) and August 09, 2017(smoke day) for Windsond data (left) and WRF output (right).784.4. DiscussionSynoptic conditions obviously have large control on atmospheric lapserates however comparing the profiles between a clear day and a smoke daygives a general sense of the radiative effects of the smoke layer. For the first400 m AGL the potential temperature gradient on August 09 varies between5°C/km - 10°C/km. On the clear day however, a large negative potentialtemperature gradient (approximate -15°C/km) is observed in the first 175m and then oscillates around 0°C/km until 500 m, indicating an adiabaticlayer. Between 500 m - 1000 m AGL the potential temperature gradientof the atmosphere is, on average, greater on the clear day than the smokeday. The potential temperature gradient on July 31 reached a maximum of16°C/km in this layer whereas the maximum for this layer on August 09 was6°C/km.When smoke was present in the atmosphere, potential temperature in-creased more rapidly near the surface compared to the clear day and lessrapidly near the top of the smoke layer. An adiabatic layer is not observedin the lower atmosphere on the smoke day. An unstable atmosphere wouldhave a negative or near-zero potential temperature gradient. This is ob-served on July 31, 2017 but not for August 09, 2017. This result makesit clear that the atmosphere was more stable on when the wildfire smokeplume was present in the atmosphere than it was on the clear day.4.4 DiscussionAn increase in the frequency and intensity of wildfires under climate changeputs an impetus on understanding the effects these events have on air quality.In Summer 2017, a thick layer of wildfire smoke advected over SouthwesternBritish Columbia altering radiative forcings in the lower atmosphere leading794.4. Discussionto a stable atmosphere. Wildfire smoke in this area is not a rare occurrence,however, there is still a lack of knowledge on the overall effects this has onpublic and environmental health and climate. A study regarding a smokeplume over the LFV in July 2015 investigated the impacts of a smoke plumewith an AOD of 9. This study showed that increases in diffuse radiationfrom the smoke plume has a large effect on physiological responses fromvarious ecosystems in this region. Of particular interest, when dense smokewas present over the region it was found that the forested site they lookedbecame a source of carbon dioxide, rather than a sink (McKendry et. al.,2018). This has direct implications on climate and must be considered whenanalyzing global and regional radiation budgets.Few studies have looked at the effect of smoke on CBL dynamics andmost that have looked at cases where AOD is less than 2. Taubman et. al.(2004) showed that absorption of solar radiation in a smoke layer with anAOD between 0.42 ± 0.06 - 1.53 ± 0.21 resulted in increased stability inthe lower atmosphere by cooling the surface and heating the air at higherlevels. This created a temperature inversion that remained throughout theafternoon. Wang and Christopher (2006) did a similar analysis using amesoscale model for Southeastern United States and reported similar effectsof smoke stabilizing the atmosphere by altering the radiation budget. Thestabilizing effects of a smoke layer can create a positive feedback loop thathinders vertical mixing and dilution of the smoke leading to an even denserplume that can further amplify the radiative effects. In theory, this feedbackloop would be further amplified under denser layers of smoke.The smoke event analyzed in this chapter had a maximum AOD of 4 andsmoke has been previously observed in Southwestern British Columbia withan AOD of 9. This is the result of an extremely dense smoke layer, which may804.4. Discussionbecome more commonplace as wildfires increase under climate change. Theindirect effects these events have on air quality are due to the radiative effectthey have on the lower atmosphere. This chapter has compared verticalprofile data from a clear day and a smoke day and have shown a more stableatmosphere existing on the smoke day. This is indicated by a larger lapsein potential temperature near the surface due to a potential temperatureinversion. A lower potential temperature gradient was also observed in airaloft during the smoke day. This is because there is no clear CBL presenton the smoke day and therefore no sharp gradient in temperature betweenthe CBL and the free troposphere. A study by Zhuravleva et. al. (2018)showed that the absorption of solar radiation by a smoke layer with an AODbetween 2 and 4 can result in a change in air temperature of approximately2.5 - 5.5 K during daylight hours. This suggests that the longer smoke ispresent in the atmosphere the larger its effect on stability will be. It is clearthat more research should be done in this region the better understand thewide range of effects events such as this have on air quality.81Chapter 5Conclusions5.1 Project outcomeThe two main objectives of this project were to understand the spatial andtemporal variations in CBL height over Southwestern British Columbia andto the assess the impact of a thick layer of wildfire smoke in the loweratmosphere on CBL development. This chapter will summarize the resultsfrom these findings and then provide suggestions for future research.The WRF model was found to be in good agreement with data fromthe Windsond weather balloons which provided confidence that the modelcould be used examine the behaviour of the CBL. Models are inherentlyunrealistic because they are comprised of simplified atmospheric dynamicsand limited by horizontal and vertical grid resolutions. However, agreementbetween model and observed data indicate that the WRF model used in thisstudy is representative of the atmosphere in this region.Using the WRF model, the CBL over Southwestern British Columbiawas found to be the most terrain following during the day and becomesless terrain in the afternoon. Each month in the study period showed asimilar pattern of the CBL closely following terrain until approximately 0800PST where T values begin to increase and r decreases. The least terrainfollowing behaviour for all months was experienced at 1400 PST. August825.1. Project outcomeshowed the least terrain following behaviour at this time with a T of 0.48and r of 0.92. The relation between elevation and CBL height is importantin understanding the distribution of pollutants in the atmosphere becauseit affects the overall morphology of the CBL which determines the locationswhere advective venting may occur.June had the highest mean CBL depth (above ground level) for the entiredomain with a value of 510 m AGL. All proceeding months decreased in valuewith September having the lowest mean CBL depth of 366 m AGL. As CBLdepth is a function of surface sensible heat flux and atmospheric stability,June must have experienced the most unstable conditions. This study was anobservational study to document the characteristics of the CBL, therefore, amore in depth analysis of the day-to-day synoptic conditions and dominantpatterns for each month would help explain these values. Plotting mean CBLdepth AGL over the study region has shown that the CBL tends to growdeeper further up the Fraser Valley than it does near the coast for all months.Although, several locations over the mountains to the north of the FraserValley have mean CBL depths similar to those experienced in the valley.Appian Mountain (49°31’29.48”N, 123°06’26.03”) sticks out as a“hotspot”forhaving a higher mean CBL depth than its immediate surroundings. Again,this indicates the atmosphere in more unstable over this location than itssurroundings. This may be a result of mesoscale circulations increasing CBLdepth or a result of the surface properties of this location.When plotting mean CBL height (mean sea level) over the domain itis clear that, although the CBL tends to grow deeper over the valley, theCBL over any mountain peak is significantly higher than the values seenover the valley. This suggests that advective venting can occur due to anyof the mountains in this region. Grouse Mountain and the other North835.1. Project outcomeShore Mountains have been speculated to be the source of the majorityof elevated layers observed over the valley (McKendry, 1997). This analysisshows that these mountains are capable of advective venting, however, higherelevation areas to the north have a larger difference between the CBL overthese mountains and the CBL over the valley. Specifically, Mount Garibaldi(49°51’02”N, 123°00’17”W) has the highest mean CBL height in the studyregion with a maximum value of over 3000 m MSL achieved in August.Based on this analysis, Mount Garibaldi appears to have a greater potentialfor advective venting than the North Shore Mountains. This mountain islikely the source of high elevation elevated layers over the LFV, given anortherly flow.This study looked at spatial variations in CBL height to gain a betterunderstanding of venting mechanisms in this region. However, when wild-fire smoke is present in the atmosphere, a phenomena that is predicted toincrease in duration and frequency with climate change (IPCC, 2014), theCBL may behave differently than the accepted model for CBL growth underclear sky conditions. The event looked at for this study was a 12 day periodwhere synoptic flow brought wildfire smoke from BC’s interior to Southwest-ern British Columbia. AOD was greater than 1 for the entire period witha maximum of 4 recorded on August 04, 2017. This study has presentedan indirect method for examining the effects of the smoke layer on CBLdevelopment by comparing the potential temperature gradients between aclear sky day and a day during the smoke event. The atmosphere whensmoke was present was much more stable than on the clear day indicatedby a greater potential temperature gradient for the bottom 500 m of theatmosphere. This is due to a potential temperature inversion starting at thesurface. Understanding the impact of wildfire smoke on CBL development845.2. Suggestions for future workis important because the presence of smoke in the atmosphere can inducea positive feedback loop that works to further degrade air quality. By ab-sorbing radiation at upper levels and blocking radiation from reaching thesurface, wildfire smoke works to stabilize the atmosphere by altering thethermodynamic structure of the lower atmosphere. Increased stability inthe atmosphere leads to a lower CBL height, thus less volume available forpollutants to disperse into. Furthermore, increased stability of the atmo-sphere can suppress venting mechanisms that transport pollutants from theboundary layer into the free troposphere. This allows pollutants and smoketo build up in the CBL which can then further suppress these processes byincreasing the concentration and radiative effects of the smoke layer. Un-fortunately, this concept is an understudied area of research and a moremechanistic approach to understanding the impact of wildfire smoke on thedevelopment of a CBL and venting mechanisms would be of great benefit.5.2 Suggestions for future workThe primary goal of this study was to diagnose the characteristics of theCBL over Southwestern British Columbia. Previous studies in this locationhave mostly looked the CBL over the Fraser Valley and have been limitedin space and time. This study used a numerical weather model to gain abetter understanding of variations in the CBL over the region in space andtime. However, models are limited in application as they do not alwayspredict realistic values. The analysis performed here would be well aided byfurther observational data on boundary layer characteristics over mountainpeaks other than Grouse Mountain. As Mount Garibaldi was found to be asource of advective venting, Windsond data from here would help validate855.2. Suggestions for future workthese results and provide a more mechanistic view of venting mechanismsoccurring on this mountain. Additionally, observations of pollutant concen-tration, in addition to thermodynamic observations, would help explain theconnection between the CBL and the height that pollutants released nearthe surface can rise. Ground-based LiDAR’s are convenient because they areable to record changes in time, however, they are limited to one location.Although more expensive, downward-pointing LiDAR attached to a planeor drone would be useful to examine the spatial variations of aerosols in theatmosphere.More observations of the radiative effects of wildfire smoke on CBL de-velopment would also greatly aid the findings from this study. This studyprimarily used data from Windsond flights to examine this concept, however,it is difficult to find causal relationships based on observational data alone.The WRF model used in this does not account for the radiative effects ofaerosols, therefore, the WRF model is not representative of an atmospherewith wildfire smoke present. Various models exist that do account for theradiative effects of aerosols, such as the 1-D libRadTran model (Mayer etal., 2017) or similar 2-D and 3-D models. The application of these modelson this data and for other regions around the world would help to furtherunderstand the suppressing effects of wildfire smoke on CBL development.86BibliographyBanta, R.M., 1984. Daytime boundary-layer evolution over mountainousterrain. Part 1: Observations of the dry circulations. Monthly WeatherReview, 112(2), pp.340-356.Chen, Y., Zhao, C., Zhang, Q., Deng, Z., Huang, M. and Ma, X., 2009.Aircraft study of mountain chimney effect of Beijing, china. Journal ofGeophysical Research: Atmospheres, 114(D8).De Wekker, J., 2002. Structure and morphology of the convective boundarylayer in mountainous terrain (Doctoral dissertation, University of BritishColumbia).Edinger, J.G., 1963. Modification of the marine layer over coastal southernCalifornia. Journal of Applied Meteorology, 2(6), pp.706-712.Gallagher, J.P., McKendry, I.G., Macdonald, A.M. and Leaitch, W.R., 2011.Seasonal and diurnal variations in aerosol concentration on Whistler Moun-tain: Boundary layer influence and synoptic-scale controls. Journal of Ap-plied Meteorology and Climatology, 50(11), pp.2210-2222.87BibliographyHennemuth, B. and Lammert, A., 2006. Determination of the atmosphericboundary layer height from radiosonde and LiDAR backscatter. Boundary-Layer Meteorology, 120(1), pp.181-200.Holzworth, G.C., 1964. Estimates of mean maximum mixing depths in thecontiguous United States. Mon. Weather Rev, 92(5), pp.235-242.Hong, S.Y., Noh, Y. and Dudhia, J., 2006. A new vertical diffusion pack-age with an explicit treatment of entrainment processes. Monthly weatherreview, 134(9), pp.2318-2341.Hu, X.M., Nielsen-Gammon, J.W. and Zhang, F., 2010. Evaluation of threeplanetary boundary layer schemes in the WRF model. Journal of AppliedMeteorology and Climatology, 49(9), pp.1831-1844.IPCC: Climate Change 2014: Impacts, Adaptation, and Vulnerability. PartB: Regional Aspects. Contribution of Working Group II to the Fifth Assess-ment Report of the Intergovernmental Panel on Climate Change [Barros,V.R., C.B. Field, D.J. Dokken, M.D. Mastrandrea, K.J. Mach, T.E. Bilir,M. Chatterjee, K.L. Ebi, Y.O. Estrada, R.C. Genova, B. Girma, E.S. Kissel,A.N. Levy, S. MacCracken, P.R. Mastrandrea, and L.L. White (eds.)]. Cam-bridge University Press, Cambridge, United Kingdom and New York, NY,USA, pp. 688, 2014.Kalthoff, N., H.-J.Binder, M.KoBmann, R. Vdgtlin, U. Corsmeier, F. Fiedler,and H. Schlager, 1998: Temporal evolution and spatial variation of theboundary layer over complex terrain. Atmos. Environ., 32, pp.1179-1194.88BibliographyKossmann, M., Corsmeier, U., De Wekker, S.F.J., Fiedler, F., Vo¨gtlin, R.,Kalthoff, N., Gu¨sten, H. and Neininger, B., 1999. Observations of han-dover processes between the atmospheric boundary layer and the free tro-posphere over mountainous terrain. Beitrage zur Physik der Atmosphare-Contributions to Atmospheric Physics, 72(4), pp.329-350.Lammert, A. and Bo¨senberg, J., 2006. Determination of the convectiveboundary-layer height with laser remote sensing. Boundary-Layer Meteo-rology, 119(1), pp.159-170.Lenschow, D.H., Stankov, B.B. and Mahrt, L., 1979. The rapid morn-ing boundary-layer transition. Journal of the atmospheric sciences, 36(11),pp.2108-2124.Lu, R. and Turco, R.P., 1994. Air pollutant transport in a coastal envi-ronment. Part I: Two-dimensional simulations of sea-breeze and mountaineffects. Journal of the atmospheric sciences, 51(15), pp.2285-2308.Mass, C.: The Weather of the Pacific Northwest, University of WashingtonPress, 2008.McElroy, J.L. and Smith, T.B., 1993. Creation and fate of ozone layers aloftin Southern California. Atmospheric Environment. Part A. General Topics,27(12), pp.1917-1929.McKendry, I.G., Steyn, D.G., Lundgren, J., Hoff, R.M., Strapp, W., Anlauf,89BibliographyK., Froude, F., Martin, J.B., Banta, R.M. and Olivier, L.D., 1997. Elevatedozone layers and vertical down-mixing over the Lower Fraser Valley, BC.Atmospheric Environment, 31(14), pp.2135-2146.McKendry, I.G. and Lundgren, J., 2000. Tropospheric layering of ozone inregions of urbanized complex and/or coastal terrain: a review. Progress inPhysical Geography, 24(3), pp.329-354.McKendry, I., Strawbridge, K., Karumudi, M.L., O’Neill, N., Macdonald,A.M., Leaitch, R., Jaffe, D., Cottle, P., Sharma, S., Sheridan, P. and Ogren,J., 2011. Californian forest fire plumes over Southwestern British Columbia:LiDAR, sunphotometry, and mountaintop chemistry observations. Atmo-spheric Chemistry and Physics, 11(2), pp.465-477.Nieuwstadt, F.T., 1984. The turbulent structure of the stable, nocturnalboundary layer. Journal of the atmospheric sciences, 41(14), pp.2202-2216.Nyeki, S., Kalberer, M., Colbeck, I., De Wekker, S., Furger, M., Ga¨ggeler,H.W., Kossmann, M., Lugauer, M., Steyn, D., Weingartner, E. and Wirth,M., 2000. Convective boundary layer evolution to 4 km asl over high-alpineterrain: Airborne LiDAR observations in the Alps. Geophysical researchletters, 27(5), pp.689-692.Oke, T.R., and Hay, J. 1994. The Climate of Vancouver. 2 edn. BC Geo-graphical Series, no. 50. Vancouver: Department of Geography, Universityof British Columbia.90BibliographyRampanelli, G., Zardi, D. and Rotunno, R., 2004. Mechanisms of up-valleywinds. Journal of the atmospheric sciences, 61(24), pp.3097-3111.Schwartz, J., 1994. Air pollution and daily mortality: a review and metaanalysis. Environmental research, 64(1), pp.36-52.Seibert, P., Beyrich, F., Gryning, S.E., Joffre, S., Rasmussen, A. and Tercier,P., 2000. Review and intercomparison of operational methods for the deter-mination of the mixing height. Atmospheric environment, 34(7), pp.1001-1027.Statistics Canada. 2017. Vancouver [Census metropolitan area], BritishColumbia and British Columbia [Province] (table). Census Profile. 2016Census. Statistics Canada Catalogue no. 98-316-X2016001. Ottawa. Re-leased November 29, 2017. https://www12.statcan.gc.ca/census-recensement/2016/dp-pd/prof/index.cfm?Lang=E (accessed August 11, 2018).Stull, R.B. 1988. An Introduction to Boundary Layer Meteorology. SanDiego: Academic.Stull, R.B., 1998: Studies of mountain climates, and a theory for mixed-layer-top levelness over complex topography. In: Proceedings, symposiumon advances in regional climatology, Karlsruhe, Germany, October 1998.Taubman, B.F., Marufu, L.T., Vant-Hull, B.L., Piety, C.A., Doddridge,B.G., Dickerson, R.R. and Li, Z., 2004. Smoke over haze: Aircraft observa-tions of chemical and optical properties and the effects on heating rates and91Bibliographystability. Journal of Geophysical Research: Atmospheres, 109(D2).van der Kamp, D. and McKendry, I., 2010. Diurnal and seasonal trendsin convective mixed-layer heights estimated from two years of continuousceilometer observations in Vancouver, BC. Boundary-layer meteorology, 137(3),pp.459-475.Wakimoto, R.M. and McElroy, J.L., 1986. LiDAR observation of elevatedpollution layers over Los Angeles. Journal of Climate and Applied Meteo-rology, 25(11), pp.1583-1599.Zhuravleva, T.B., Panchenko, M.V., Kozlov, V.S., Nasrtdinov, I.M., Pol’kin,V.V., Terpugova, S.A. and Chernov, D.G., 2018. Model Estimates of Dy-namics of the Vertical Structure of Solar Absorption and Temperature Ef-fects under Background Conditions and in Extremely Smoke-Laden At-mosphere According to Data of Aircraft Observations. Atmospheric andOceanic Optics, 31(1), pp.25-30.92


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items