"Arts, Faculty of"@en . "Geography, Department of"@en . "DSpace"@en . "UBCV"@en . "Gallagher, John Patrick"@en . "2010-09-29T14:30:45Z"@en . "2010"@en . "Master of Science - MSc"@en . "University of British Columbia"@en . "An observational study was conducted to characterize atmospheric conditions at an air chemistry monitoring site on the summit of Whistler Mountain, British Columbia, Canada. Discrimination of air samples from the observatory as either representative of the free troposphere (FT) or modified by air from the valley-based planetary boundary layer (PBL) is critical to the proper interpretation of air chemistry datasets. Atmospheric data from a one-year study period were used to evaluate indicators and possible driving forces of PBL influence at the Whistler site. Diurnal cycles in water vapour and aerosol concentration were attributed primarily to convective uplift of PBL air during daytime heating hours. Analysis of these variables found that PBL influence was common in the spring and summer months and relatively rare in late fall through early winter. For the one-year period, 37% of the days had diurnal cycles in aerosol concentration that were considered typical of thermally induced vertical transport processes. Patterns of slope and valley winds were also identified for Whistler, and the presence of these diurnal wind systems was associated with enhanced aerosol concentration at the summit. Synoptic classification methods were used to describe the prevailing conditions on days with well-defined indicators of PBL influence. Strong solar insolation and light synoptic scale winds were found to be common on such days. Case studies of particular days confirmed that a deep convective boundary layer (CBL) of well-mixed air can encompass the mountain summits, even during the winter season. During the summer, a tendency for PBL constituents to remain aloft through the night means that the summit observatory can be unrepresentative of the FT for several days at a time. Separation of air chemistry measurements into periods of FT conditions and times of PBL influence requires careful analysis of a variety of datasets on both local and regional scales."@en . "https://circle.library.ubc.ca/rest/handle/2429/28803?expand=metadata"@en . "Patterns of planetary boundary layer influence at the Whistler Mountain air chemistry observatory An observational mountain meteorology study by John Patrick Gallagher B.S., The University of Utah, 1994 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in The Faculty of Graduate Studies (Geography) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) September 2010 c\u00C2\u00A9 John Patrick Gallagher 2010 Abstract An observational study was conducted to characterize atmospheric conditions at an air chemistry monitoring site on the summit of Whistler Mountain, British Columbia, Canada. Discrimination of air samples from the observatory as either representative of the free troposphere (FT) or modified by air from the valley-based planetary boundary layer (PBL) is critical to the proper interpretation of air chem- istry datasets. Atmospheric data from a one-year study period were used to evaluate indicators and possible driving forces of PBL influence at the Whistler site. Diurnal cycles in water vapour and aerosol concentration were attributed primarily to con- vective uplift of PBL air during daytime heating hours. Analysis of these variables found that PBL influence was common in the spring and summer months and rela- tively rare in late fall through early winter. For the one-year period, 37% of the days had diurnal cycles in aerosol concentration that were considered typical of thermally induced vertical transport processes. Patterns of slope and valley winds were also identified for Whistler, and the presence of these diurnal wind systems was asso- ciated with enhanced aerosol concentration at the summit. Synoptic classification methods were used to describe the prevailing conditions on days with well-defined indicators of PBL influence. Strong solar insolation and light synoptic scale winds were found to be common on such days. Case studies of particular days confirmed that a deep convective boundary layer (CBL) of well-mixed air can encompass the mountain summits, even during the winter season. During the summer, a tendency for PBL constituents to remain aloft through the night means that the summit ob- servatory can be unrepresentative of the FT for several days at a time. Separation ii Abstract of air chemistry measurements into periods of FT conditions and times of PBL in- fluence requires careful analysis of a variety of datasets on both local and regional scales. iii Preface Several of the statements included in Chapter 2 concerning weather conditions dur- ing the summer of 2009 were also contributed to a publication listed in the bibliog- raphy of this thesis as McKendry et al. (2010). The time series image of lidar data used in Figure 5.9 was created by Paul Cottle, a member of the technical staff at the UBC Department of Geography. iv Table of Contents Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1 Motivation for study . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Project overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 2 Background and methods . . . . . . . . . . . . . . . . . . . . . . . . . 6 2.1 Long range transport of atmospheric pollution . . . . . . . . . . . . 6 2.1.1 Detection of LRT effects at mountain observatories . . . . . 7 2.2 Air mass discrimination at mountain observatories . . . . . . . . . . 9 2.2.1 Water vapour . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.2.2 Temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.2.3 Aerosol concentration . . . . . . . . . . . . . . . . . . . . . . 14 2.2.4 Slope winds . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 2.2.5 Atmospheric stability . . . . . . . . . . . . . . . . . . . . . . 16 v Table of Contents 2.2.6 Mechanical lift . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.2.7 Other PBL tracers . . . . . . . . . . . . . . . . . . . . . . . . 18 2.2.8 Summary of mountain observatory findings . . . . . . . . . . 19 2.3 Description of study area . . . . . . . . . . . . . . . . . . . . . . . . 20 2.3.1 Climate, weather and emissions of Whistler, BC . . . . . . . 22 2.4 Data analysis methods and datasets used . . . . . . . . . . . . . . . 26 2.4.1 Datasets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 2.4.2 Quality control . . . . . . . . . . . . . . . . . . . . . . . . . . 30 2.4.3 Data analysis methods . . . . . . . . . . . . . . . . . . . . . 31 3 Indicators of PBL influence on Whistler Mountain . . . . . . . . 32 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 3.2 Diurnal cycles of meteorological variables . . . . . . . . . . . . . . . 33 3.2.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 3.2.2 Results and discussion . . . . . . . . . . . . . . . . . . . . . 33 3.3 Comparisons to radiosonde profiles . . . . . . . . . . . . . . . . . . . 38 3.3.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 3.3.2 Results and discussion . . . . . . . . . . . . . . . . . . . . . 39 3.4 Analysis of a CN dataset . . . . . . . . . . . . . . . . . . . . . . . . 47 3.4.1 Details of the dataset and processing methods . . . . . . . . 48 3.4.2 Descriptive statistics . . . . . . . . . . . . . . . . . . . . . . 49 3.4.3 Seasonal patterns . . . . . . . . . . . . . . . . . . . . . . . . 51 3.4.4 Diurnal patterns . . . . . . . . . . . . . . . . . . . . . . . . . 52 3.5 Correlations between water vapour and CN . . . . . . . . . . . . . . 62 3.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 4 Driving forces of PBL influence . . . . . . . . . . . . . . . . . . . . . 70 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 4.2 An example of synoptic scale influence . . . . . . . . . . . . . . . . 71 vi Table of Contents 4.3 Atmospheric stability . . . . . . . . . . . . . . . . . . . . . . . . . . 75 4.4 Slope and valley wind systems . . . . . . . . . . . . . . . . . . . . . 80 4.4.1 Theory and definitions . . . . . . . . . . . . . . . . . . . . . 80 4.4.2 Pattern identification . . . . . . . . . . . . . . . . . . . . . . 81 4.4.3 Diurnal wind case study . . . . . . . . . . . . . . . . . . . . 83 4.4.4 Daily classification . . . . . . . . . . . . . . . . . . . . . . . . 86 4.4.5 Diurnal winds and CN data . . . . . . . . . . . . . . . . . . 88 4.5 Synoptic classification of diurnal cycle days . . . . . . . . . . . . . . 90 4.5.1 Theory and definitions . . . . . . . . . . . . . . . . . . . . . 90 4.5.2 Synoptic weather typing from surface observations . . . . . . 91 4.5.3 Synoptic classification via composite maps . . . . . . . . . . 94 4.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 5 Case studies of vertical mixing . . . . . . . . . . . . . . . . . . . . . 102 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 5.2 Winter PBL-influence case: 20 February 2009 . . . . . . . . . . . . 102 5.3 Summer FT case: 13 July 2009 . . . . . . . . . . . . . . . . . . . . . 106 5.4 Evaluation of BL structure with lidar: 6 \u00E2\u0080\u0093 8 July 2010 . . . . . . . . 110 5.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 6.1 Summary of results . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 6.2 Suggestions for future work . . . . . . . . . . . . . . . . . . . . . . . 123 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124 vii List of Tables 2.1 Mountain observatories and criteria used for air mass discrimination. 19 2.2 Whistler climate normals and observed monthly data from the study period. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 3.1 Summary statistics for the one-year CN dataset . . . . . . . . . . . . 50 3.2 Seasonal median CN concentration values from Whistler Mountain. . 52 3.3 Correlations between water vapour and CN concentration for each month, based on hour-of-day medians. . . . . . . . . . . . . . . . . . 63 6.1 Predominant air mass categories by season for clear and partly cloudy days at the Whistler observatory. . . . . . . . . . . . . . . . . . . . . 119 viii List of Figures 2.1 Venting mechanisms in mountainous terrain. . . . . . . . . . . . . . 10 2.2 Map of the study region showing the location of Whistler Mountain. 20 2.3 Digital terrain map of the local study area. . . . . . . . . . . . . . . 21 2.4 Photo of Whistler Mountain from the north. . . . . . . . . . . . . . . 22 2.5 Photo of the summit of Whistler Mountain. . . . . . . . . . . . . . . 27 2.6 Photo of the mid-mountain OAN weather station (VOL) on Whistler Mountain. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 3.1 Monthly mean values of temperature and water vapour mixing ratio by hour of day for selected months. . . . . . . . . . . . . . . . . . . . 34 3.2 Average amplitudes of diurnal temperature and water vapour variations. 36 3.3 Hourly mean values of temperature and water vapour mixing ratio for selected days in October 2009. . . . . . . . . . . . . . . . . . . . . 37 3.4 Whistler mountaintop temperatures compared to radiosonde profiles from December 2008 and July 2009 . . . . . . . . . . . . . . . . . . . 40 3.5 Whistler mountaintop water vapour mixing ratios compared to ra- diosonde profiles for selected months . . . . . . . . . . . . . . . . . . 42 3.6 Average temperature profiles from the 19-day period in February 2009 when weather balloons were launched from Whistler Valley. . . . . . 44 3.7 Whistler mountaintop temperatures compared to radiosonde profiles from the special observing period in February 2009. . . . . . . . . . 45 ix List of Figures 3.8 Whistler mountaintop water vapour mixing ratios compared to ra- diosonde profiles from the special observing period in February 2009. 46 3.9 A simplified conceptual model of planetary boundary layers in moun- tainous terrain. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 3.10 Effect of automated smoothing scheme on one-minute CN data from April. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 3.11 Monthly box plots of CN concentration data. . . . . . . . . . . . . . 51 3.12 CN diurnal variations by season. . . . . . . . . . . . . . . . . . . . . 53 3.13 Monthly median values of CN concentration by hour of day for se- lected months. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 3.14 Examples of days that meet (\u00E2\u0080\u009CYES\u00E2\u0080\u009D) and do not meet (\u00E2\u0080\u009CNO\u00E2\u0080\u009D) the criteria for a typical diurnal cycle of CN concentration. . . . . . . . . 58 3.15 Number of days per month with a typical diurnal cycle of CN con- centration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 3.16 Median values of CN concentration and water vapour mixing ratio by hour of day for March and June. . . . . . . . . . . . . . . . . . . . . 64 3.17 Plots of hourly CN concentration and water vapour mixing ratio for selected days. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 3.18 Number of days per month with a typical diurnal cycle of CN concen- tration that also showed strong positive correlations between water vapour mixing ratio and CN concentration. . . . . . . . . . . . . . . 67 4.1 Average CN concentration at the Whistler site by wind direction. . . 72 4.2 Model trajectories tracing air parcels backwards 72 hours fromWhistler Mountain. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 4.3 Median stability parameter values for each month of the study period. 77 4.4 Box plots of stability parameter values for days with and without typical diurnal cycles of CN concentration. . . . . . . . . . . . . . . . 78 x List of Figures 4.5 Comparison of potential temperature profiles fromWhistler radiosonde observations and ground-based weather stations for the afternoon of February 9, 2009. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 4.6 Time series of hourly u-wind components from the mid-mountain station VOL for the first week of August 2009. . . . . . . . . . . . . 82 4.7 Wind observations from the evening of August 2, 2009. . . . . . . . . 84 4.8 Wind observations from selected hours of August 3, 2009. . . . . . . 85 4.9 Number of days per month with well-defined diurnal wind patterns. 88 4.10 Comparison of average CN concentrations for days with and without well-defined diurnal wind patterns. . . . . . . . . . . . . . . . . . . . 90 4.11 Composite maps of PBL-influence days represented by synoptic pat- tern A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 4.12 Composite maps of PBL-influence days represented by synoptic pat- tern B. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 4.13 Composite maps of PBL-influence days represented by synoptic pat- tern C. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 5.1 Hourly values of CN concentration and water vapour mixing ratio at the summit of Whistler on 20 February 2009. . . . . . . . . . . . . . 103 5.2 Whistler radiosonde observations from the morning and afternoon soundings on 20 February 2009. . . . . . . . . . . . . . . . . . . . . . 104 5.3 Regional synoptic circulation maps showing averaged 500 hPa and sea level pressure patterns for 20 February 2009. . . . . . . . . . . . 105 5.4 Hourly values of CN concentration and water vapour mixing ratio at the summit of Whistler on 13 July 2009. . . . . . . . . . . . . . . . . 107 5.5 Hourly values of temperature and pressure at the summit of Whistler on 13 July 2009. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 xi List of Figures 5.6 Regional synoptic circulation maps showing averaged 500 hPa and sea level pressure patterns for 13 July 2009. . . . . . . . . . . . . . . 108 5.7 Vertical profiles of temperature and dew point temperature from the Quillayute, WA radiosonde observations of 1600 LST on 13 July 2009. 110 5.8 Regional synoptic circulation maps showing averaged 500 hPa and sea level pressure patterns for 6 \u00E2\u0080\u0093 8 July 2009. . . . . . . . . . . . . . 113 5.9 Time series of backscatter ratio data from the Whistler CORALNet lidar from July 6 \u00E2\u0080\u0093 8, 2010. . . . . . . . . . . . . . . . . . . . . . . . . 114 xii Acknowledgements I would like to thank my research supervisor, Dr. Ian McKendry, for his strong support and unwavering positive attitude throughout my degree program here at UBC. I also appreciate the assistance and feedback provided by the other members of my research committee, Dr. Andreas Christen and Dr. Dan Moore. Staff members at the UBC Department of Geography have been consistently helpful during my studies here, and special thanks must be given to the IT support staff for all their assistance. I am grateful to Wendy Rohrbacher for proofreading drafts of this thesis with her expert eye for grammar and structure. I would also like to acknowledge the camaraderie and support of my fellow graduate students, who have made this department a great place to study. Special mention goes to Lisa Erven, who got me out for fieldwork on the ever-magnificent Whistler Mountain. This project would not have been possible without the support of several other institutions. The Canadian Foundation for Climate and Atmospheric Sciences and the Natural Sciences and Engineering Research Council of Canada provided funding for this research. Most of the datasets used in this study were generously provided by Environment Canada. Of the many divisions and individuals from Environment Canada who have been involved in data collection at Whistler, I would especially like to thank Dr. Paul Joe for his enthusiastic support of my research. This project also benefited from data and fieldwork support given by the Whistler Blackcomb ski area. xiii Chapter 1 Introduction 1.1 Motivation for study Interest in mountain weather and climate likely goes back to the earliest days of hu- man exploration and settlement in mountainous regions. Through basic observation of weather phenomena in mountains, one quickly realizes that topography interacts with atmospheric systems to produce levels of complexity and variability in weather conditions exceeding those of the plains and river valleys where most people reside. However, intensive scientific study of mountain meteorology did not commence until the mid-nineteenth century, and to this day the remoteness and difficulty of access to mountainous areas means that research progress lags behind that of flatter, more populated regions (Barry, 2008). However, as human activity in the mountains steadily increases, there is enhanced impetus to achieve better understanding of mountain weather and climate. The safety and enjoyment of mountain activities depend partly on this knowledge as it is applied to such things as the building of infrastructure and mountain weather forecasting. The general increase of activity in the mountains also presents opportunities for researchers as access becomes easier and meteorological observations proliferate. Air pollution studies constitute another area of meteorological research that has direct implications for human safety and quality of life. While studies of local air quality in settled areas have been ongoing since the Industrial Revolution (Boubel et al., 1994), the phenomenon of long range pollutant transport has only been rec- 1 1.1. Motivation for study ognized and studied over the last few decades. Long range transport (LRT) of air pollution is a process by which air quality on regional and local scales can be influenced by contaminants of distant origin. Particles and trace gases that have long lifetimes in the atmosphere can at times travel intercontinental distances to be eventually entrained into air masses near the surface at population centres on the receiving end. The effect of LRT on air quality in North America has received in- creased research attention in recent years as anthropogenic emissions have expanded dramatically in developing Asian countries. Observation of the large-scale phenomenon of trans-Pacific transport requires a multi-faceted approach to track pollution plumes across the ocean, measure air chemistry parameters at the surface and aloft in North America, and attribute ob- servations of pollutants to Asian sources. An important component of this observing network is the collection of air chemistry data from mountaintop observatories near the west coast of North America, where air streams arriving from over the Pacific can be directly sampled. One such mountaintop observatory is located on Whistler Mountain in southwestern British Columbia, Canada. This study represents an intersection of mountain meteorology and air pollution research. Conceptual models and analysis techniques of mountain meteorology are applied to the problem of determining air mass origin at the Whistler Mountain air chemistry site. Following the establishment of mountain air chemistry observatories around the world, researchers have recognized that interpretation of their datasets needs to be complemented by an understanding of meteorological processes that may affect the measurements (Kleissl et al., 2007). In particular, there needs to be a way to determine whether the constituents being sampled at a given time are of local or distant origin. The influence of complex terrain on meteorological processes makes this sort of analysis different for each mountain observatory. The problem of determining air mass origin has been addressed to various degrees for a number of mountain sites, but thus far the Whistler site has received only very limited 2 1.2. Project overview attention on this topic. This study uses datasets from a one-year period to begin to describe, both qualitatively and quantitatively, the patterns of (local) planetary boundary layer influence at the Whistler mountaintop site. The overall aim of this work is to increase the utility of air chemistry data collected at this site, which is the only mountain observatory for LRT research in Canada. 1.2 Project overview This project is an observational mountain meteorology study. Observational stud- ies rely on analysis of various measurements of the phenomena of interest and can be distinguished from both experiments and modeling projects. Experiments exert some level of control on variables, which is often impractical or impossible in studies of the natural environment. Modeling in this context implies the use of numeri- cal simulations of the environment, run by computer programs. An observational study has both the advantage and disadvantage of using actual, uncontrolled mea- surements: the processes can be studied directly, but the researcher has to contend with the full complexity of the natural world. This makes proving cause-and-effect relationships at best very difficult. However, there is value in simply identifying patterns and using these observations to support, refine or refute conceptual models of how dynamic environmental systems operate. The patterns of interest to this study are those of planetary boundary layer (PBL) influence at the Whistler Mountain air chemistry observatory. The PBL is the lower layer of air that is influenced by the Earth\u00E2\u0080\u0099s surface via friction and exchanges of energy, mass and momentum on timescales of a day or less (Oke, 1987). Turbulent exchange in the PBL makes it a well-mixed layer of air, including any particles and gases that originate at the surface. Above the PBL, the free troposphere (FT) is generally dominated by synoptic scale flows that are free of surface influence. Mountaintop observatories are usually sited to be above the PBL 3 1.2. Project overview for the most part, but even the highest mountains can at times be influenced by boundary layer (BL) air that is transported upward from adjacent valleys (Hindman, 1995). The PBL air can bring with it pollutants of local origin, thus confounding observations meant to sample air masses of more distant origin. Therefore, in order to aid the interpretation of measurements taken at the Whistler mountaintop site, this project focuses primarily on the following two research questions: 1) How much of the time and in what conditions is the Whistler site influenced by PBL air? 2) What routinely available meteorological and/or physico-chemical parameters can be used to distinguish PBL air from free tropospheric conditions? To address these questions, data from numerous sources and several different scales of measurement are analyzed. However, this project would not have been possible without the existence of an observational network on and around Whistler Mountain that is unusually dense for a mountain location. Much of the meteorolog- ical instrumentation was deployed to assist weather forecasting efforts for the 2010 Olympic Winter Games. This wealth of observational data also provides opportu- nities for mountain meteorology studies such as this one. The most important datasets for this project are the meteorological and physico- chemical observations from the summit of Whistler Mountain, for those measure- ments relate to what was happening at the location of interest. It is there that analysis of PBL influence begins; but to elucidate the relevant weather patterns and processes, a host of other observations need to be employed from the local to the synoptic scale. This study uses the available datasets to first identify evidence and patterns of PBL influence, and then examine possible mechanisms and necessary conditions for BL air to reach the mountaintop. This thesis is organized into six chapters. Following this introductory chapter, Chapter 2 contains more detailed background material on the scientific problem addressed and the analysis approach used. Recent literature of LRT studies is 4 1.2. Project overview briefly reviewed, with an emphasis on trans-Pacific transport to North America. Then, a number of other mountain observatories around the world are visited via a review of studies of PBL influence at those locations. This project\u00E2\u0080\u0099s study area is described and an overview of the climate of Whistler is provided. The datasets that have been analyzed are then described, along with some information on quality control and data processing methods. In Chapter 3, indicators of PBL influence at the Whistler observatory are ex- amined. This involves meteorological data from the mountaintop as well as a corre- sponding aerosol dataset representing atmospheric particle concentrations. Patterns on seasonal and diurnal timescales are elucidated, and possible explanations for the patterns are offered. In Chapter 4, meteorological processes that are expected to be conducive to vertical transport of PBL air are analyzed for the Whistler area. Correlations between these processes and the indicators of PBL influence are ex- plored. Also, a synoptic classification approach is employed to depict the favourable weather patterns for PBL air reaching the mountaintop. In Chapter 5, case studies are presented that clarify some details of vertical transport processes in the Whistler area. A winter case with PBL influence and a summer case without PBL influence at the site are presented to demonstrate how the different situations can be distinguished using available measurements. Each of these case studies focuses on one calendar day. Then, a three-day period of lidar data is examined, introducing a new measurement platform that offers great opportunities for mountain meteorology studies at Whistler. A concluding chapter follows, which includes suggestions for future work in this area. 5 Chapter 2 Background and methods 2.1 Long range transport of atmospheric pollution The issue of long range pollution transport may have been first recognized in the aftermath of major volcanic eruptions, such as the Mount Tambora eruption of 1815 in Indonesia, which contributed to an anomalously cold \u00E2\u0080\u009Cyear without a summer\u00E2\u0080\u009D for eastern North America in 1816. Much later, in the 1950s and 1960s, military personnel tracked persistent radioactive debris from nuclear bomb tests via aircraft and ground-based monitoring stations (Boubel et al., 1994). The topic of LRT was more widely recognized as an environmental and political issue in the 1970s when research on the \u00E2\u0080\u009Cacid rain\u00E2\u0080\u009D problem was publicized, introducing many in the public to the idea that anthropogenic emissions from one region can have harmful effects in areas far downwind. After linkages were found between sulphur emissions in continental Europe and the acidification of Scandinavian lakes, the United Nations Economic Commission for Europe established in 1979 the Convention on Long-range Transboundary Air Pollution (UNECE, 2004), setting a precedent for international cooperation on pollution controls. Recently, LRT of pollution across the Pacific Ocean to North America has been recognized as a problem worthy of attention. When hemispheric-scale weather pat- terns are favourable, particulate matter and trace gases originating in Asia that have long lifetimes in the atmosphere are able to make the journey across the Pacific. This trans-Pacific transport can affect North American air quality either through grad- 6 2.1. Long range transport of atmospheric pollution ual increases in background levels of certain pollutants such as ozone (Zhang et al., 2008) or via episodes of enhanced transport that lead to measurable concentration peaks in entities such as carbon monoxide, sulfates, hydrocarbons and crustal dust (Yienger et al., 2000; Heald et al., 2006). One of the first studies to find strong evidence of Asian pollution impacting western North America (Jaffe et al., 1999) used chemistry observations from Cheeka Peak in Washington State to analyze pol- lution episodes from the spring of 1997. In more recent years, modeling efforts and numerous case studies have led to some understanding of the scope and climatology of trans-Pacific transport. Although LRT is a continuous process, it has been found that conditions leading to significant export of Asian emissions to North America are most common in the springtime. Observations and modeling studies have found that synoptic weather conditions in the spring often support production of airborne dust in Asia as well as the rapid transport of dust and industrial emissions in the FT over the Pacific (Cooper et al., 2004; Zhao et al., 2006). This trans-Pacific transport typically occurs at altitudes of 3 to 10 km on timescales of 6 to 14 days (Holzer et al., 2005). Presence of a high pressure ridge aloft over the North American west coast can induce subsidence in the air mass, thereby transporting pollutants down toward the western mountains and the PBL. Subsidence of elevated aerosol layers over southern British Columbia has been observed via ground-based remote sensing during dust transport events (McKendry et al., 2007, 2008). 2.1.1 Detection of LRT effects at mountain observatories Case studies of LRT events necessarily use a variety of observational tools to as- semble the evidence when pollution of distant origin is suspected. In addition to surface-based air chemistry monitoring, remote sensing from both the surface and from satellites has been used to detect pollutants. In cases of intercontinental trans- port, satellite observations are crucial for tracking contaminant plumes over the 7 2.1. Long range transport of atmospheric pollution data-sparse ocean. These measurements are often used to evaluate and calibrate numerical models, which are employed to fill in the remaining gaps in the observing networks. Most air quality monitoring networks are concentrated in urban areas, where a relative abundance of local pollution makes it difficult to discern pollutants from distant sources. However, investigators have had some success in detecting chemical signatures in aerosol data, especially for Asian dust (McKendry et al., 2001; Jaffe et al., 2003). Sampling of the FT from aircraft is one method that can avoid local emissions, but aircraft-based observations are generally limited to infrequent field campaigns. A more permanent and cost-effective method of sampling the FT is the use of mountaintop observatories. Air chemistry instruments similar to those found in urban monitoring networks are placed on mountain peaks that are assumed to be\u00E2\u0080\u0094at least much of the time\u00E2\u0080\u0094above the PBL, i.e. representative of the FT where most of the intercontinental transport occurs. In addition to observing LRT events, these mountain sites are also considered suitable for monitoring background levels of pollutants. Stations for these purposes have been established on mountains and high-altitude plateaus around the world. Many of these stations operate in coop- eration with the World Meteorological Organization\u00E2\u0080\u0099s Global Atmosphere Watch (GAW) program. Studies of trans-Pacific transport benefit from the few mountain observatories that have been established in the region. In addition to the above-mentioned sites on Whistler Mountain and Cheeka Peak, other mountain air chemistry stations in the Pacific region are located on Mauna Loa in Hawaii and Mt. Bachelor, Oregon. The Whistler site is the only high elevation (above 2000 m) air chemistry station in Canada. 8 2.2. Air mass discrimination at mountain observatories 2.2 Air mass discrimination at mountain observatories The PBL is not a static entity; its vertical extent (often referred to as its depth) varies considerably with factors such as radiation fluxes and wind speed, particularly on the diurnal timescale. The concern for mountain observatories is that, at times when air from the valley-based PBL is conveyed up to the height of the measurement station, the site is no longer representative of free tropospheric conditions. During such episodes, the air chemistry equipment is likely to pick up local pollution from the valley air. While the monitoring of locally sourced pollution is important in its own right, it is not the aim of LRT studies. Therefore, it is useful to be able to separate periods of PBL influence from those of free tropospheric conditions at the observing location. The challenge in this air mass discrimination is that there is no known perfect tracer of PBL air. In fact, the definition of the PBL is itself problematic. Con- ceptual models exist that define the depth of the PBL based on vertical profiles of potential temperature, water vapour and wind speed. However, these models tend to be derived from observations and modeling over flat terrain, and conditions can be altered by the presence of complex terrain. Morphology of the daytime PBL, often referred to as the convective boundary layer (CBL), can be influenced by moun- tain ranges. For example, Kossmann et al. (1998) found through an observational study that interactions between the synoptic scale flow and topography can sup- press diurnal CBL growth near the mountains. However, De Wekker et al. (2004) showed that mountain venting mechanisms cause vertical transport of aerosols to greater heights than what would occur over flat terrain. They suggested that the aerosol layer, which extends above the conventional CBL as defined by temperature profiles, is the relevant height for pollution dispersion considerations. The term \u00E2\u0080\u009Cmountain CBL\u00E2\u0080\u009D was suggested to describe this layer that is affected by the surface on timescales of one to several hours. Figure 2.1 depicts the above-mentioned layers 9 2.2. Air mass discrimination at mountain observatories and mechanisms in a conceptual model. 0 10 20 30 40 50 horizontal distance (km) 0 1 2 3 4 5 he ig ht (k m asl ) 4 3 ha h +1 2 Figure 2.1: Venting mechanisms as depicted by De Wekker et al. (2004): 1) moun- tain venting, 2) cloud venting, 3) advective venting, and 4) advection of aerosols from air masses elsewhere. h represents the conventional CBL height and ha is the aerosol layer height. (With kind permission from Springer Science+Business Media: De Wekker et al. (2004) Fig. 10) The concept of an intermediate injection layer between the PBL and the FT was also described by Henne et al. (2005), who found that the layer consisted of up to one third PBL air; that is, the injection layer is a mixture of FT and PBL air. Thus, the presence of such a layer surrounding a mountaintop observing location means that the air sampled can be considered PBL-influenced and not properly representative of the FT. Similarly, at night-time during light wind conditions, a residual layer of BL constituents (water vapour, aerosols, etc.) can remain above the nocturnal BL (Whiteman, 2000). These elevated layers need to be considered in air mass discrimination analysis at mountain sites. The frequency with which a particular mountain observatory is influenced by PBL air from the lowlands depends on the mountain\u00E2\u0080\u0099s size, shape, geographical location and local climate (Kleissl et al., 2007). Therefore, research aimed at distin- guishing air mass types tends to be site-specific. In all cases, however, investigators 10 2.2. Air mass discrimination at mountain observatories can look in the data for indicators of PBL air and/or driving forces of vertical trans- port and subsequent PBL influence at the site (Henne et al., 2008). In the next section, some of the approaches used for several of the well-established mountain observatories are reviewed. Scales of analysis As this project considers processes on a range of atmospheric scales, a few words should be said about the definitions of such scales. LRT can be considered as a global, or at least hemispheric, phenomenon when intercontinental transport is in- volved. The hemispheric (macro) scale is also the largest scale commonly analyzed for atmospheric circulation patterns. Atmospheric systems depicted on weather maps, such as fronts, highs and lows, represent the synoptic scale, covering up to a few thousand kilometres horizontally and lasting for about a week (Stull, 2000). The meso-scale covers phenomena like thunderstorms and local wind systems, op- erating on spatial scales of 10 \u00E2\u0080\u0093 200 km (Oke, 1987). The local scale is smaller still, representing a few kilometres or less. In the context of this study, a mountain slope would be a good example of the local scale. This project is concerned primarily with local and mesoscale processes in the immediate vicinity of Whistler Mountain. However, the different scales of motion do not operate in isolation, and the larger scales always set the stage for what can happen locally. In particular, it will be shown that the synoptic scale weather is crucial to determining the likelihood of PBL influence at a mountain site. 2.2.1 Water vapour A commonly cited indicator of PBL air is water vapour. Because water vapour originates at the surface via evapotranspiration, concentration is usually highest in the PBL with a sharp decrease found in the entrainment zone, which marks the transition to the FT (Oke, 1987). Thus, an increase in water vapour concentration 11 2.2. Air mass discrimination at mountain observatories at a mountain observatory to a level exceeding that of the corresponding height in the FT is cited as evidence of vertical transport of PBL air. Diurnal cycles of water vapour and other variables were analyzed by Obrist et al. (2008) for the Storm Peak Laboratory in the Colorado Rockies. Seasonally averaged data showed diurnal patterns with water vapour concentration peaking in the early afternoon hours and reaching a minimum in early morning. These cycles were attributed to nearly daily transitions between free tropospheric and PBL air masses, driven by slope wind systems. The investigators also correlated water vapour with other entities, finding that up to 75% of the variability in ozone and aerosol concentrations could be explained by the water vapour fluctuations. A sinusoidal diurnal cycle in water vapour concentration was also found byWeiss- Penzias et al. (2006) in a springtime dataset from the Mount Bachelor Observatory in Oregon. They also compared water vapour mixing ratios to moisture profiles from the two nearest radiosonde (weather balloon) sites. Upper air observations from radiosondes are generally considered representative of the free atmosphere. Enhanced water vapour at the top of Mount Bachelor as compared to an equivalent altitude in the free atmosphere was considered to be an indicator of PBL influence. The researchers selected the driest 20% of observations as representative of subsiding free tropospheric air masses. In the above and other studies, water vapour increases have been shown to be at- tributable to infiltration of PBL air. However, one limitation of using water vapour as an indicator is that it is not a fully conservative tracer, i.e. in addition to trans- port, evaporation or condensation within the air mass can change the concentration. Additionally, since water vapour concentrations in the atmosphere are continually changing due to synoptic scale advection, it is not possible to identify a threshold value that represents the PBL. 12 2.2. Air mass discrimination at mountain observatories 2.2.2 Temperature Similar to water vapour, temperature is expected to show regular diurnal cycles in the PBL but not in the FT. These diurnal cycles are driven by the daily course of solar radiation, which follows a sinusoidal pattern with its peak at solar noon. Energy is transported upwards from the heated surface of the Earth in convective thermals that cause the daytime growth of the PBL (Oke, 1987). In this way, heating is felt through the depth of the PBL by day and cooling occurs at night when thermal convection ceases. Thus, the simple presence of a diurnal cycle in temperature may be an indicator of influence from the PBL. For the Mount Washington Observatory in New Hampshire, Grant et al. (2005) used the timing of daily extreme temperatures to identify days when the summit was affected by the PBL. Days with early morning minima and afternoon maxima in temperature were classified as \u00E2\u0080\u009Cradiative\u00E2\u0080\u009D days exhibiting BL behaviour. Days with temperature extremes at the beginning and end of the day were assumed to be dominated by horizontal advection. While this method should find the best-defined examples of PBL growth to the site, there are likely additional days with some PBL air reaching the summit that also have strong horizontal temperature advection, which can overwhelm the diurnal signal. Temperature and water vapour can be expected to increase simultaneously if air is being transported vertically from lower elevations during the daytime. For both of the above-mentioned studies using water vapour (Weiss-Penzias et al., 2006; Obrist et al., 2008), afternoon peaks were also shown for temperature, with timing similar to the diurnal maxima in water vapour. These patterns support the conceptual model of heated, moisture-rich PBL air reaching the mountain sites via convective thermals and/or upslope winds. 13 2.2. Air mass discrimination at mountain observatories 2.2.3 Aerosol concentration An aerosol is a suspension of solid or liquid particles dispersed within a gas, usu- ally air (Glickman, 2000). In practice, particulate matter (PM) datasets are often referred to as aerosol data even though an aerosol technically refers to the whole colloidal system. For the purposes of air pollution studies, the size, composition and concentration of airborne particles are all of concern. PM can come from natural sources such as biogenic emissions, sea salt and desert dust, or from anthropogenic sources such as fossil fuel combustion and biomass burning. Despite large spa- tial and temporal variations, the average vertical profile of aerosol concentration over mid-latitude continental regions shows concentration decreasing with altitude through the PBL, above which a nearly constant background concentration is found (Jaenicke, 1993). However, it should be noted that balloon and aircraft observations have found a secondary maximum of fine particles in the upper troposphere, which has been attributed to gas-to-particle formation (nucleation) processes (Hofmann, 1993; Nyeki et al., 1999). Like water vapour, aerosol concentration can be assumed to decrease sharply at the top of the PBL. This assumption allows one to calculate mixed layer depth based on vertical aerosol profiles, or some proxy thereof such as lidar backscatter data (Steyn et al., 1999; van der Kamp, 2008). Thus, diurnal cycles in aerosol parameters can also be used to identify PBL air at mountaintop observatories. Researchers with the Jungfraujoch research station in Switzerland have identi- fied seasonal and diurnal patterns in aerosol concentration at the site (Baltensperger et al., 1997; Lugauer et al., 1998; Nyeki et al., 1998). They attributed the diur- nal cycles to CBL growth that is enhanced by slope winds acting within a certain catchment area. Baltensperger et al. (1997) found high correlations between aerosol concentration and water vapour, but stated that the aerosol signal is a superior indi- cator of vertical transport due to its larger dynamic range and closer approximation 14 2.2. Air mass discrimination at mountain observatories to conservative behaviour. At the Mauna Loa Observatory on the island of Hawaii, regular diurnal cycles have been observed in aerosol datasets, including number concentration of condensa- tion nuclei (CN) (Bodhaine, 1996). The daily increases in CN corresponded closely to upslope winds that carry polluted island air up to the site from lower eleva- tions. The pattern of daytime upslope and night-time downslope flows at the site is consistent enough that they have used a time of day criterion, selecting data from 0000 \u00E2\u0080\u0093 0800 LST to represent the background (FT) conditions. Bodhaine (1996) considered CN concentration to be an ideal indicator of locally produced pollu- tants\u00E2\u0080\u0094better than humidity, which can signal the presence of PBL air, but not necessarily local pollution. 2.2.4 Slope winds Upslope winds were recognized as the primary mechanism for PBL influence at the Mauna Loa Observatory. Although most sites do not experience these wind systems with the same regularity as Mauna Loa, slope winds have been evaluated as a possible driving force of vertical transport at other locations. Diurnal winds systems, which include slope, valley, and mountain-plain circulations, result from differential heating, which causes pressure gradients to form (Whiteman, 2000). In the case of upslope winds, the air near the mountain slope becomes warmer than air over the valley at the same altitude, creating buoyancy that forces air upward along the slope. After sunset, the slope cools faster than the valley air, leading to downslope (drainage) winds until solar heating commences again the next morning. This twice-daily reversal of wind direction is the characteristic of diurnal wind systems (Whiteman, 2000). Another tropical observatory that has frequent slope wind cycles is Mount Kenya in equatorial East Africa. Henne et al. (2008) used a slope wind pattern as one of two criteria for separating trace gas datasets into different groups for analysis (the 15 2.2. Air mass discrimination at mountain observatories other criterion was based on diurnal variations in specific humidity). A majority of days throughout the year had identifiable slope wind patterns, and those days had larger diurnal cycles in CO2 and O3 than days without slope winds. However, the fact that non-slope wind days still displayed diurnal cycles in the chemistry data led the investigators to conclude that, other than direct vertical transport by slope winds, general CBL growth over the surrounding lowlands can reach the station altitude (3678 m asl). Thus, a time of day criterion was also deemed appropriate for defining FT datasets at Mt. Kenya. 2.2.5 Atmospheric stability The stability of the atmosphere at a given point is determined by the change in temperature with height, known as the lapse rate. A positive lapse rate, which indicates temperature decreasing with height, is considered the normal situation in the troposphere, while an increase of temperature with height is known as a temperature inversion. Atmospheric stability and PBL depth are related in that unstable air leads to thermal convection and CBL growth. This is a situation typical of light wind days with substantial solar heating of the surface. At night, rapid cooling of the surface forms an inversion layer that defines the stable nocturnal BL (Whiteman, 2000). An unstable air mass is characterized by a relatively rapid decrease in temperature with height in the troposphere. Specifically, the air is considered unstable if the environmental lapse rate exceeds the dry adiabatic lapse rate of 10K per km (Aguado and Burt, 2010). For saturated parcels, instability is reached when the lapse rate exceeds the moist adiabatic rate of approximately 5K per km. In the unstable case, a rising thermal will maintain its buoyancy and continue upward until it encounters a stable layer. The relationship between atmospheric stability, convection, and PBL growth means that instability is a possible driving force of valley PBL air reaching moun- taintop locations. Deep instability should encourage more vigorous convection, loft- 16 2.2. Air mass discrimination at mountain observatories ing PBL air to high altitudes. In contrast, a highly stable air mass can act to decouple the mountain peaks from the valley atmosphere. The role of stability was investigated by Parrish et al. (1990) as a meteorological contribution to seasonal and diurnal nitrogen oxide (NOx) cycles at Niwot Ridge, Colorado. They found that the more unstable times of day and year usually corre- sponded to times of higher NOx concentration at the site. The frequent occurrence of daytime temperature inversions below the station elevation in winter was credited with the generally low NOx values observed there during the winter. A more quantitative investigation of stability as it relates to high elevation mea- surements was conducted by Baltensperger et al. (1997) for the Jungfraujoch ob- servatory in the Swiss Alps. They defined a stability parameter based on potential temperature differences between the Jungfraujoch and a valley weather station. This parameter was then compared to results of analysis that identified days with a well-defined diurnal cycle in aerosol concentration. In general, decreased stability showed a tendency toward a greater number of diurnal cycle cases, suggesting that the cycles were caused by convective mixing. However, their results were not con- sistent enough to make stability a reliable predictor of PBL intrusion. In particular, a number of diurnal cycle days with high stability required further explanation. 2.2.6 Mechanical lift In addition to the buoyant forces that cause convective uplift, mechanical orographic lift may also be a mechanism for the transport of PBL air to high elevations. This possibility was investigated by Kleissl et al. (2007) for the Pico Mountain observa- tory in the Azores Islands. They applied a mathematical model to meteorological datasets to calculate how often wind flows impacting the mountain were able to force marine BL air over the top of the mountain. Mechanical uplift was found to be most important during the winter season when synoptic scale winds were strong enough to lift PBL air to the summit up to 59% of the time by month. 17 2.2. Air mass discrimination at mountain observatories The results for Pico, finding that mechanical lift is at least as important as buoyant upslope flow for vertical transport of PBL air, are somewhat unique to relatively small, isolated peaks. Kleissl et al. (2007) noted that for a large massif with relatively low wind speeds such as Mauna Loa, the horizontal winds mostly split around the mountain. As Whistler Mountain is embedded within a very large mountain range, it is expected that mechanical uplift is not important there, at least not for lifting of air all the way from near the valley floor. However, orographic lift may bring PBL air to the summit in cases when the PBL has already reached the upper slopes of the mountain. Due to the lack of sufficient vertical profile data for the Whistler area, this study does not include any calculations of mechanical lifting parameters. 2.2.7 Other PBL tracers At certain sites, trace gases other than those previously mentioned in this section have been used as tracers of local pollution. These tend to be short-lived chemi- cal species that originate in populated areas or lowland surfaces. For example, at the Pico observatory both isoprene and n-butane are routinely measured (Kleissl et al., 2007). Isoprene is emitted from broadleaf vegetation, which is only found more than 500 metres below the station, and n-butane is emitted from heating and cooking in the town centre near sea level. Detection of diurnal cycles in isoprene, but not n-butane, led the investigators to conclude that buoyant upslope flow was not transporting air all the way from sea level to the station. At the Jungfraujoch observatory during special observing periods, black carbon and radon decay prod- ucts 214Pb and 212Pb have been used as tracers of BL air (Baltensperger et al., 1997; Lugauer et al., 2000). NOx has also been used as an indicator of anthropogenic emis- sions from lowland population centres (Reidmiller et al., 2010). The observatory at Whistler has very limited space for instrumentation, and none of these species are currently being measured there. 18 2.2. Air mass discrimination at mountain observatories 2.2.8 Summary of mountain observatory findings The mountain observatories mentioned in the preceding literature review are listed in Table 2.1 along with their geographic parameters and the primary air mass dis- crimination criteria suggested by researchers. Observatory Location/latitude/elev. (asl) Criteria basis Storm Peak Colorado/40.5 \u00E2\u0097\u00A6N/3220 m Diurnal water vapour cycles Mt. Bachelor Oregon/44.0 \u00E2\u0097\u00A6N/2763 m Water vapour threshold Mt. Washington New Hampshire/44.3 \u00E2\u0097\u00A6N/1914 m Timing of daily temperature extremes Jungfraujoch Switzerland/46.6 \u00E2\u0097\u00A6N/3454 m Diurnal aerosol cycles Mauna Loa Hawaii/19.5 \u00E2\u0097\u00A6N/3397 m Time of day, WD, WS Mt. Kenya Kenya/0.1 \u00E2\u0097\u00A6S/3678 m Time of day Niwot Ridge Colorado/40.0 \u00E2\u0097\u00A6N/3050 m WD, stability Pico Azores Islands/38.5 \u00E2\u0097\u00A6N/2225 m Model, diurnal wind cycles Table 2.1: Mountain observatories and criteria used for air mass discrimination. Note: WD=wind direction; WS=wind speed; asl= above sea level. From the above, it is evident that analysis of PBL influence at a mountain site can include BL indicators such as temperature, water vapour and aerosols, as well as driving forces of vertical transport such as slope winds, convective instability and mechanical uplift. This study of the Whistler observatory uses the available datasets to examine several of these parameters with the aim of describing the frequency and conditions of PBL influence at the mountaintop site. From the preceding review of the literature, it is apparent that for most mountain sites, no single parameter is likely to properly characterize the air mass origin at all times. Isolated tropi- cal mountains such as Mauna Loa are perhaps most conformable to simple criteria such as time of day, but Whistler is a mid-latitude peak within a large mountain range. Synoptic scale and seasonal variability, coupled with the complexity of Coast Mountain topography, make development of any broadly applicable air mass dis- 19 2.3. Description of study area crimination criteria challenging. Yet, the identification of relevant patterns and processes of vertical transport on the local scale will inform future investigations of particular time periods and events. 2.3 Description of study area Whistler Mountain is located in the Coast Mountains of British Columbia at 50.06 \u00E2\u0097\u00A6N/ 122.96 \u00E2\u0097\u00A6W. The Coast Mountains cover a large area, extending over 1,500 km north- west from Vancouver, BC to Alaska and the southern Yukon. Whistler is approx- imately 100 km north of Vancouver and less than 35 km from Pacific coastal inlet waters at the head of Howe Sound. See Figure 2.2 for Whistler\u00E2\u0080\u0099s location on a re- gional map. The Resort Municipality of Whistler occupies the Whistler Valley at the base of Whistler and Blackcomb Mountains. The two mountains, separated by Fitzsimmons Creek (see Figure 2.3), comprise the Whistler Blackcomb ski area, and thus include considerable infrastructure in the way of ski lifts, trails and buildings. 0 200 km 135 \u00C2\u00B0 W 130 \u00C2\u00B0 W 125 \u00C2\u00B0 W 120 \u00C2\u00B0 W 115 \u00C2\u00B0 W 50 \u00C2\u00B0 N 55 \u00C2\u00B0 N 45 \u00C2\u00B0 N U.S.A. PACIFIC OCEAN YLW BRITISH COLUMBIA CANADA UIL YZT Figure 2.2: Map of the study region showing the location of Whistler Mountain (blue triangle), Vancouver (red circle) and upper air observation sites (black circles) (YZT=Port Hardy, BC; YLW=Kelowna, BC; UIL=Quillayute, WA). 20 2.3. Description of study area Figure 2.3 provides a view of the study area derived from a digital terrain model. Several of the weather stations used in this study are marked on the map. Whistler Mountain spans 1529 m of elevation from the Creekside base at 653 m to the peak at 2182 m. A network of meteorological sensors has been installed near ski runs that are primarily west to northwest-facing. Trees, mainly conifers, and undergrowth are found to about the 1820 m level on the mountain. The summit area is mostly ex- posed rock and soil with limited alpine vegetation. A summertime photo of Whistler Mountain taken from near Green Lake is provided as Figure 2.4. Figure 2.3: Digital terrain map of the local study area showing several of the automated weather stations. The air chemistry observatory is at the peak of Whistler Mountain. Data source: Canadian Digital Elevation Data from GeoBase (www.geobase.ca). 21 2.3. Description of study area Figure 2.4: Photo of Whistler Mountain from the north. 2.3.1 Climate, weather and emissions of Whistler, BC Whistler\u00E2\u0080\u0099s climate is defined primarily by its midlatitude location near the Pacific Coast and the seasonal shifts in macroscale weather patterns that occur every year. In late fall and winter, a strong westerly jet stream impacts the region with frequent synoptic scale frontal systems, which bring rain to low elevations and snow to the mountains. In the summer, the jet stream weakens and shifts northward as the East Pacific High becomes dominant. The result of these seasonal changes is a Mediterranean climate for southwestern BC, with wet winters and mostly sunny, dry summers (Mass, 2008). The spring and fall tend to be transition periods between the wet and dry ex- tremes of winter and summer. The fall transition is usually marked by an abrupt increase in precipitation sometime during the month of October (Mass, 2008). The spring transition is more gradual; often a cold upper-level low will persist off the southern BC coast, bringing unstable air and showery weather to the region (Oke and Hay, 1994). July and August are the driest and warmest months of the year (see Table 2.2). Due to moderate surface temperatures and stable air associated 22 2.3. Description of study area with the East Pacific High, summer thunderstorms are rare along the West Coast relative to most other parts of the continent (Mass, 2008), but enhanced heating and orographic lift along the slopes can increase shower activity in the mountains. As Whistler lies in a narrow mountain valley, wind flows tend to be topograph- ically channelled. From Squamish at the head of Howe Sound, the valley ascends generally north then northeast to Whistler along the Cheakamus River. Continuing northeast of Whistler, the Green River descends to Pemberton, about 20 km from Whistler. Inflow winds are a southwesterly flow from the coast, whereas outflow winds are northeasterlies from the interior. During the winter, occasional strong flows of cold, dry air from high pressure systems in the interior are known as \u00E2\u0080\u009CArc- tic outflow\u00E2\u0080\u009D events (McKendry, 2000). Otherwise, the valley is often protected from strong winds that can accompany fronts, as those winds tend to be from the south- east ahead of a front, then from the west after the frontal passage. In the warm season, thermally driven mesoscale winds often take over when the synoptic scale flow is weak, resulting in sea breezes, valley winds and slope winds (Oke and Hay, 1994; McKendry, 2000). In the Lower Fraser Valley, where the Vancouver metropolitan area is located, concentrations of pollutants such as O3 and PM tend to be highest during stagnant summertime weather conditions. The synoptic scale pattern most often associated with these events is a high pressure ridge aloft with a surface thermal trough ori- ented along the coast (McKendry, 1994). It is expected that the same pattern causes elevated pollution levels in the Whistler Valley as well. Strong night-time temper- ature inversions in winter can also trap pollutants in a thin layer near the surface, causing high ground-level concentrations in the valley. The Resort Municipality of Whistler has a permanent population of nearly 10,000 people and receives over two million visitors annually (Tourism Whistler statistics). As tourism is the main industry in the valley, emissions of important pollutants such as CO, NOx, SOx and PM are dominated by area sources (mainly space heating) and 23 2.3. Description of study area mobile sources (transportation) (Pitre, 2002). Natural emissions contribute to the PM load in summer via wildfires, and a majority of the volatile organic compound (VOC) emissions are from trees and plants. Weather conditions during the study period The 1971 \u00E2\u0080\u0093 2000 climate normals for Whistler are shown in Table 2.2 along with monthly observations from this project\u00E2\u0080\u0099s study period of December 2008 through November 2009. The Whistler observations are from the Nesters site (station id WAE) in the valley at 659 m elevation, directly across the highway from Whistler Village. Climate statistics were obtained from Environment Canada\u00E2\u0080\u0099s National Cli- mate Archive online (http://www.climate.weatheroffice.gc.ca). Temperatures during the study period were generally cooler than normal during the winter and early spring, but substantially warmer than normal during the sum- mer. Precipitation was below normal in all seasons except the fall. In December 2008, Whistler was dominated by cold air masses from the north and east. While the city of Vancouver was receiving unusually heavy snowfalls, Whistler experienced colder, drier continental air and only light amounts of snow. January 2009 began with a week of considerable snowfall, but starting on the 12th, a blocking high pressure ridge aloft settled over coastal BC, resulting in 15 consecutive days with no precipitation. Strong temperature inversions were common during this period, with temperatures on the upper mountain of Whistler sometimes exceeding 10 \u00E2\u0097\u00A6C. February also featured much more clear, dry weather than usual, with another 15- day dry spell ending on the 22nd. A more typical winter pattern prevailed through March, with numerous frontal systems crossing the area, bringing a mixture of rain and snow. The spring season started off cool, but still drier than normal in April and early May. By the end of May, strong high pressure aloft provided clear skies and the first 25 \u00E2\u0097\u00A6C readings of the season in Whistler, setting the stage for the summer. 24 2 .3 . D escrip tio n o f stu d y a rea Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Temperature (\u00E2\u0097\u00A6C) Normals -3.0 -0.6 2.3 5.9 9.8 13.2 15.9 16.1 12.3 6.6 0.6 -3.2 Study pd -2.6 -1.4 -0.5 5.0 9.9 15.1 19.2 17.1 13.3 5.7 1.7 -5.1 Precipitation (mm) Normals 157.2 119.5 96.1 75.0 66.2 58.1 47.7 47.5 63.7 147.4 188.2 162.3 Study pd 102.3 61.7 75.0 32.2\u00E2\u0088\u0097 37.2 26.2 6.3 30.0 87.0 135.6\u00E2\u0088\u0097 205.1\u00E2\u0088\u0097 70.6 Table 2.2: Whistler climate normals (based on 1971 \u00E2\u0080\u0093 2000 averages) and observed monthly data from the study period. (De- cember values are from 2008; all other months are 2009.) \u00E2\u0088\u0097Denotes monthly precipitation values that may be too low due to significant missing data (more than two days). Data source: Environment Canada. 25 2.4. Data analysis methods and datasets used The persistent high pressure ridge aloft deflected frontal systems to the north for most of the summer. Precipitation for the June \u00E2\u0080\u0093August period was only 41% of normal for Whistler and 43% of normal for Vancouver. At the Nesters observing location in Whistler, July and August only had four days each with measurable rainfall. Temperatures were higher than normal through the summer months, with a peak of the heat wave occurring in late July when four consecutive days had high temperatures exceeding 35 \u00E2\u0097\u00A6C. The rains started to return in September, along with temperatures closer to normal. November was very wet and milder than normal due to a nearly constant influx of Pacific moisture with subtropical origins. A significant consequence of the hot and dry summer was the very active forest fire season that occurred in BC and the northwestern U.S. In addition to numerous large wildfires east of the Coast Mountains, two small fires were reported on the slopes of Blackcomb Mountain. Degraded air quality from smoke was a common occurrence throughout the summer in southern BC. At times, easterly to northerly wind flows advected smoke through the Whistler area and into Metro Vancouver. 2.4 Data analysis methods and datasets used In this section, the primary datasets used for this observational study are intro- duced. Data reduction and quality control methods are briefly outlined here; addi- tional details of data analysis are provided in the subsequent results chapters. Data processing was accomplished using MATLAB version R2009a by The MathWorks and Microsoft Office Excel 2007. Digital terrain maps were produced using ArcMap 9.3.1 by ESRI. 26 2.4. Data analysis methods and datasets used 2.4.1 Datasets Data from the summit of Whistler The Whistler Mountain air chemistry observatory is run by the Air Quality Research Division (AQRD) of Environment Canada (EC). Their instruments are housed in a small building next to the top station of the Peak Express chairlift (Figure 2.5). Meteorological sensors and air intakes for aerosol and gas measurements are mounted on the outside of the building. Data obtained from the AQRD for this study include temperature (T ), pressure (P ), relative humidity (RH ), and CN concentration. Hourly averages were calculated from one-minute data such that each hour\u00E2\u0080\u0099s value represents an average of the preceding 60 minutes. This averaging convention was followed with all other datasets to maintain consistency when comparing variables. Figure 2.5: Photo of the summit of Whistler Mountain. The red circle indicates the location of the AQRD instrumentation. Photo by Dr. Kurt Anlauf of Environment Canada 27 2.4. Data analysis methods and datasets used OAN stations Automated weather stations were installed by EC on Whistler Mountain and else- where to assist meteorologists tasked with providing detailed forecasts for the Olympic venues. These stations comprised the Olympic Autostation Network (OAN). Sev- eral of the OAN stations that were used for this study appear on the terrain map in Figure 2.3. A photo of one of the stations is provided as Figure 2.6. Stations were located at various elevations on Whistler Mountain, mostly along the west to northwest-facing trails where the alpine skiing events were held. The station named VOC was in the valley, co-located with the surface observation point at Nesters. Figure 2.6: Photo of the mid-mountain OAN weather station (VOL) on Whistler Mountain. Monthly files of hourly data were downloaded from the OAN website. Hourly averages of T and P were retained along with RH readings from the top of each hour. Vector average wind speed and direction data from 10-metre towers were also analyzed for several sites. Both hourly averages and 10-minute average winds at the top of each hour were retained. 28 2.4. Data analysis methods and datasets used Regional observations In order to represent the FT and analyze synoptic scale patterns, several upper air datasets were obtained. Radiosonde observations from weather balloon flights provide vertical profiles of T, RH, P and winds through the depth of the troposphere. Twice-daily radiosonde datasets were gathered from the three upper air sites nearest to Whistler: Kelowna, BC, Quillayute, WA and Port Hardy, BC (see Figure 2.3). These datasets were downloaded from the University of Wyoming\u00E2\u0080\u0099s Department of Atmospheric Science website (http://weather.uwyo.edu/upperair/). The data were then linearly interpolated to vertical levels every 25 hectopascals (hPa) from 1000 hPa to 500 hPa. Upper air data were also collected from Whistler during special operations as- sociated with the Olympic Games. Weather balloons were launched from Nesters three times daily during the Olympics in 2010 and during a trial period in February 2009. For this study, data from February 5 \u00E2\u0080\u0093 23, 2009 were obtained from EC and processed in the same manner as the other upper air datasets. Synoptic weather maps were produced by plotting numerical model reanalysis data on regional maps. The model data used in this study consist of NCEP reanalysis products from the Physical Sciences Division website of NOAA\u00E2\u0080\u0099s Earth System Re- search Laboratory: http://www.esrl.noaa.gov/psd/data/gridded/reanalysis. Data files in netCDF format were processed using MATLAB. Grid point data orig- inally on a 2.5\u00E2\u0097\u00A6 x 2.5\u00E2\u0097\u00A6 lat/lon grid were linearly interpolated to every 0.5\u00E2\u0097\u00A6 prior to display. Supplementary observations In addition to the OAN sites, weather stations on Whistler Mountain are also main- tained by the Whistler ski patrol. Temperature, RH and wind data were obtained from Whistler Blackcomb for the station at the peak as well as the \u00E2\u0080\u009Calpine shop\u00E2\u0080\u009D 29 2.4. Data analysis methods and datasets used weather station, which is near the treeline at 1835 m. Remote sensing instruments deployed by EC were also used in this study. A Doppler RADAR was installed 10 km west of Whistler Village in February 2009 and provided reflectivity and velocity profiles over the valley. Additionally, a Vaisala CL31 lidar ceilometer was run in aerosol mode through the spring and summer of 2009 from the Timing Flats location on the lower slopes of Whistler Mountain. Lastly, a high-power aerosol lidar was deployed at the Nesters site in April 2010 as part of Environment Canada\u00E2\u0080\u0099s CORALNet research project. Data displays from this instrument are included in one of the case studies presented in Chapter 5. 2.4.2 Quality control Many of the datasets used in this study came directly from automated stations. Therefore, it was important to check the data for problems before proceeding with analysis. Despite the large volume of data involved, quality control (QC) work was done manually to ensure data integrity. The first step in the QC process was to check for missing data and note any significant gaps. Cases when missing data affected the analysis are mentioned in the appropriate sections of the results chapters. Fortunately, missing data was not a major problem for this project. Typically 97 \u00E2\u0080\u0093 100% of hourly observations per month were available from the weather stations. For upper air soundings, the number of missing flights ranged from 0 to 8 per month for each station. Next, data files of manageable size (usually monthly files) were inspected for potential bad or suspect readings. Time series of each variable were plotted so that outliers and stagnant values were identifiable. Cases of failed sensors were found, and the affected data points were deleted. When feasible, data substitution was performed. This was done in the case of AQRD temperature readings from the summit that remained fixed on the same value for three days in July. For the time period affected, temperatures from the nearby (within 50 m) Whistler Blackcomb 30 2.4. Data analysis methods and datasets used weather station were substituted. Inspection of wind data from the Whistler Blackcomb station on the summit found that the wind directions seemed incorrect starting in May 2009. This discovery was later confirmed by a contact with the Whistler ski patrol. Thus, there was no wind dataset from the peak available for this study. The CN dataset from the air chemistry site required careful inspection for prob- lems as well as semi-automated smoothing, both of which are described in detail in Chapter 3. 2.4.3 Data analysis methods An observational study of this kind has no prescribed method of analysis. Rather, the approach was to proceed using ideas garnered from the relevant literature (re- viewed earlier in this chapter). The importance of diurnal cycles in the data meant that much of the data reduction was aimed at obtaining hour-of-day average and/or median values. Graphical display of results was crucial to pattern identification. Ideas about the relations between variables were then tested via standard statistical methods of regression and hypothesis testing. The following results chapters of this thesis follow the general progression of the analysis that has been performed. First, indicators of PBL influence on Whistler Mountain are evaluated then possible driving forces are examined. Case studies are then used to add details to the analysis by incorporating some of the supplementary datasets. 31 Chapter 3 Indicators of PBL influence on Whistler Mountain 3.1 Introduction In this chapter, datasets from the Whistler observatory are analyzed for signals of PBL air reaching the site. Methods similar to those of studies at other mountain observatories are used to examine meteorological and aerosol data from the peak. First, diurnal cycles of temperature and water vapour are identified and analyzed on different timescales to show both the general patterns and degree of day-to-day vari- ability. Results are interpreted within the framework of existing conceptual models of mountain BL processes and knowledge of weather conditions during the study pe- riod. Next, temperature and water vapour measurements from the mountaintop are compared to free-air observations from weather balloons to evaluate the magnitude of departures from FT conditions at different times of the year. Then, an aerosol dataset is analyzed in detail to reveal seasonal, monthly and diurnal patterns in par- ticle concentration that may be attributable to vertical transport. Lastly, aerosol concentration and water vapour are analyzed together to see how well two different indicators of PBL influence are correlated to each other. 32 3.2. Diurnal cycles of meteorological variables 3.2 Diurnal cycles of meteorological variables 3.2.1 Background The familiar daily heating and cooling cycle near ground level is typically not found in the FT (Stull, 2000), because by definition the FT is above the layer of air directly affected by the surface. Turbulence that causes the PBL to be well mixed in terms of potential temperature and humidity is capped by a stable layer of air at the top of the PBL. Temperature and humidity changes in the FT are mainly due to the dynamics of synoptic scale weather systems. Thus, FT changes tend to occur over longer timescales than the diurnal changes characteristic of the PBL. It follows from this distinction that identification of diurnal cycles in atmospheric variables is one method by which BL behaviour can be discerned. In the only known previous analysis of this kind completed specifically for the Whistler site, Nseir (2007) applied this approach to two springtime periods (2005 and 2006). Diurnal cycles of temperature, water vapour and ozone were found to be common on the peak. In this section, results of similar analysis are presented for the year-long study period (Dec 2008 \u00E2\u0080\u0093Nov 2009). Analysis was based on hourly averaged temperature and moisture readings from the AQRD site on the summit of Whistler. Water vapour mixing ratio (w ) (g/kg) was computed from observations of T, RH and P using Tetens\u00E2\u0080\u0099 formula and related equations in Stull (2000). The mixing ratio is an indication of the actual amount of water vapour in a given air parcel, whereas RH is relative to the temperature. Diurnal cycles were found by calculating hour-of-day means for the time periods in question. 3.2.2 Results and discussion Results from four of the monthly periods are given in Figure 3.1 to show seasonal differences. The results in Figure 3.1 indicate that diurnal cycles in temperature and water 33 3.2. Diurnal cycles of meteorological variables 0 4 8 12 16 20 24\u00E2\u0088\u009212 \u00E2\u0088\u009211 \u00E2\u0088\u009210 Te m pe ra tu re (\u00C2\u00B0C ) Hour of Day (LST) 1.8 2.0 2.2 M ixi ng Ra tio (g /kg ) T w (a) December 2008 0 4 8 12 16 20 24\u00E2\u0088\u009210 \u00E2\u0088\u00925 0 Te m pe ra tu re (\u00C2\u00B0C ) Hour of Day (LST) 2 3 4 M ixi ng Ra tio (g /kg ) T w (b) April 2009 0 4 8 12 16 20 248 10 12 14 16 Te m pe ra tu re (\u00C2\u00B0C ) Hour of Day (LST) 5.4 5.8 6.2 6.6 7.0 M ixi ng Ra tio (g /kg ) T w (c) July 2009 0 4 8 12 16 20 24\u00E2\u0088\u00924 \u00E2\u0088\u00922 0 Te m pe ra tu re (\u00C2\u00B0C ) Hour of Day (LST) 3.0 3.5 4.0 M ixi ng Ra tio (g /kg ) T w (d) October 2009 Figure 3.1: Monthly mean values of temperature (T) and water vapour mixing ratio (w) by hour of day for selected months. 34 3.2. Diurnal cycles of meteorological variables vapour were common during the study period, with minimum values found in the early morning hours and maxima in the afternoon. Diurnal maxima occurred later in the afternoon in summer (Figure 3.1c) as compared to winter (Figure 3.1a), presumably due to the longer daylight hours and subsequent longer period of surface heating. The variables T and w tracked each other fairly closely, indicating that as the water vapour carrying capacity of the air increased with temperature, a source of moisture was available to increase the mixing ratio. The amplitudes of the diurnal cycles show considerable seasonal variability. Note that the y-axis scales differ between the four panels of Figure 3.1. The diurnal pattern for December was relatively weak and poorly defined compared to the other months. Figure 3.2 shows the diurnal cycle amplitudes for each month of the study period. In Figure 3.2a the diurnal temperature amplitudes are shown simply as differences between the maximum and minimum (averaged) hourly values. The amplitudes ranged from 1.0 \u00E2\u0097\u00A6C in December to 5.5 \u00E2\u0097\u00A6C in July, corresponding to the months with the weakest and strongest surface heating, respectively (based on monthly average temperatures). The low average amplitude for December suggests very weak and/or infrequent BL conditions at the summit during that month. The fall months, which were dominated by cloud and precipitation-producing synoptic scale disturbances, had weaker average diurnal cycles than the drier spring months. For water vapour mixing ratio, the amplitudes in Figure 3.2b are given as per- centages of the daily mean, i.e. 100x (max \u00E2\u0088\u0092min)/mean. Rather than showing a summer peak as in the temperature results, the relative amplitude of w peaked in the springtime. There are at least two possible explanations for this observation. One possibility is that the vertical mixing may have been more vigorous and/or the PBL influence was more frequent in spring than summer. As an indicator of air from lower elevations, w should show more of a daytime increase for months with stronger vertical mixing. On the other hand, the smaller diurnal amplitude in w for the summer months may be a reflection of the night-time values remaining 35 3.2. Diurnal cycles of meteorological variables 1 2 3 4 5 6 7 8 9 10 11 120 1 2 3 4 5 6 Month Am pl itu de (\u00C2\u00B0 C ) Temperature (a) Average diurnal temperature amplitude for each month given as a range between maximum and minimun values. 1 2 3 4 5 6 7 8 9 10 11 120 5 10 15 20 25 30 35 Month Am pl itu de (% of m ea n) Water Vapour (b) Average diurnal water vapour mixing ratio amplitude for each month given as a percentage of the daily mean. Figure 3.2: Average amplitudes of diurnal temperature and water vapour variations for each month of the study period. 36 3.2. Diurnal cycles of meteorological variables relatively high due to the presence of a residual layer of moisture. BL constituents that get lofted to high altitudes during the day do not immediately go away when the atmosphere stabilizes in the evening, but rather remain above the nocturnal BL until being replaced by horizontal advection of (usually cleaner and drier) FT air (Lugauer et al., 1998; Stull, 2000). The monthly averaged data shown in Figures 3.1 and 3.2 tend to mask the substantial day-to-day variability present in the diurnal T and w trends. Thus, statements made about the diurnal cycles of a particular month or season cannot be applied to each day of the given period. To provide a sense of the day-to-day variability in the dataset, Figure 3.3 shows four different daily plots of hourly T and w from October 2009. 0 4 8 12 16 20 24\u00E2\u0088\u00925 \u00E2\u0088\u00924 \u00E2\u0088\u00923 \u00E2\u0088\u00922 Te m pe ra tu re (\u00C2\u00B0C ) Hour of Day (LST) 2.5 3.0 3.5 4.0 M ixi ng Ra tio (g /kg ) T w (a) 4 October 0 4 8 12 16 20 24\u00E2\u0088\u00929 \u00E2\u0088\u00928 \u00E2\u0088\u00927 \u00E2\u0088\u00926 \u00E2\u0088\u00925 \u00E2\u0088\u00924 Te m pe ra tu re (\u00C2\u00B0C ) Hour of Day (LST) 0.8 1.2 1.6 2.0 2.4 2.8 M ixi ng Ra tio (g /kg ) T w (b) 10 October 0 4 8 12 16 20 24 \u00E2\u0088\u00928 \u00E2\u0088\u00926 \u00E2\u0088\u00924 \u00E2\u0088\u00922 Te m pe ra tu re (\u00C2\u00B0C ) Hour of Day (LST) 2.0 2.5 3.0 3.5 M ixi ng Ra tio (g /kg ) T w (c) 27 October 0 4 8 12 16 20 24\u00E2\u0088\u00929 \u00E2\u0088\u00926 \u00E2\u0088\u00923 0 3 6 9 Te m pe ra tu re (\u00C2\u00B0C ) Hour of Day (LST) 2.0 3.0 4.0 5.0 6.0 M ixi ng Ra tio (g /kg ) T w (d) 31 October Figure 3.3: Hourly mean values of temperature and water vapour mixing ratio for selected days in October 2009. 37 3.3. Comparisons to radiosonde profiles October 4th (Figure 3.3a) is an example of a day in which both T and w exhibited diurnal cycles with afternoon peaks, but the timing of the maxima were offset by a couple of hours. On October 10th, (Figure 3.3b), T and w were negatively correlated in the afternoon hours: T had a well-defined late afternoon peak, while w decreased through the afternoon. This pattern could reflect the combined effects of local (heating) and synoptic scale (drying) processes acting at the same time. Determining the extent and timing of PBL influence is expected to be difficult for such days. Less ambiguous perhaps is the example of October 27th (Figure 3.3c), when the two variables tracked each other closely and followed what can be considered a typical diurnal course, with an afternoon peak and lowest values in the morning. The close correlation between T and w in this case suggests that both variables were being influenced by the same process(es). The same can be said of the 31st (Figure 3.3d), but in this case, it was probably synoptic scale advection that dominated, causing temperature and moisture to decrease throughout the day. Based on only the T and w data for the 31st, one would conclude that the measurement site was not within the PBL that day. 3.3 Comparisons to radiosonde profiles 3.3.1 Background The diurnal cycles identified in the T and w data from the summit of Whistler seem to indicate considerable influence of PBL air, especially in the spring and summer months of the study period. In this section, readings from the summit are compared to free-air measurements in order to quantify the differences and learn more about the seasonality of PBL influence. Previous work by Nseir (2007) compared Whistler data from several springtime periods in 2005 and 2006 to radiosonde observations from Port Hardy (YZT), Kelowna (YLW) and Quillayute (UIL) (marked on Figure 2.2). For the time periods analyzed, it was found that temperatures and wind speeds 38 3.3. Comparisons to radiosonde profiles on the mountaintop were generally similar to values at the same altitude from the radiosonde profiles, but the water vapour mixing ratios were higher on the mountain. For this analysis, data from the same three upper air sites were used for the whole study period, and data from the February 2009 Whistler radiosonde observations were used for the dates available. Average monthly profiles of temperature and water vapour were computed from the radiosonde data. However, rather than averaging the 0000 and 1200 UTC (0400 and 1600 LST) profiles together as in Nseir (2007), the morning and afternoon sounding times were treated separately to highlight diurnal variations. Average 0400 and 1600 LST T and w values from the AQRD site on the mountaintop were calculated from the quality-controlled hourly values. The average monthly pressure at the AQRD site was used as the representative height (pressure altitude) for comparisons with the radiosonde profiles. It is recognized that the radiosonde profiles themselves are representative of the FT only above the local PBL top at the time of each flight. Thus, the term \u00E2\u0080\u009Cfree air\u00E2\u0080\u009D is used here as a more general term for air that was sampled away from the local terrain, in this case via weather balloon flights. On days when the regional PBL depths were large, comparisons between the radiosonde and mountaintop values will include two different measurements from within the PBL. However, if there are days when PBL air was reaching the mountaintop but not the equivalent height in the free air, then temperature and moisture differences are expected to be apparent in the monthly averaged values. 3.3.2 Results and discussion Examples of the temperature comparisons are given first, with the 0400 and 1600 LST plots for December and July shown in Figure 3.4. The December plots (Figures 3.4a and 3.4b) indicate that there was little diurnal variation between the morning and afternoon observation times. The Whistler temperatures corresponded more closely to the Kelowna profile than to those of the coastal stations, which reflects the 39 3.3. Comparisons to radiosonde profiles fact that during December 2008 Whistler was often under the influence of continental air masses from the interior. \u00E2\u0088\u009230 \u00E2\u0088\u009225 \u00E2\u0088\u009220 \u00E2\u0088\u009215 \u00E2\u0088\u009210 \u00E2\u0088\u00925 0 5 500 600 700 800 900 1000 Temperature (\u00C2\u00B0C) Pr es su re (h Pa ) UIL YZT YLW Whistler (a) December 2008 0400 LST \u00E2\u0088\u009230 \u00E2\u0088\u009225 \u00E2\u0088\u009220 \u00E2\u0088\u009215 \u00E2\u0088\u009210 \u00E2\u0088\u00925 0 5 500 600 700 800 900 1000 Temperature (\u00C2\u00B0C) Pr es su re (h Pa ) UIL YZT YLW Whistler (b) December 2008 1600 LST \u00E2\u0088\u009215 \u00E2\u0088\u009210 \u00E2\u0088\u00925 0 5 10 15 20 500 600 700 800 900 1000 Temperature (\u00C2\u00B0C) Pr es su re (h Pa ) UIL YZT YLW Whistler (c) July 2009 0400 LST \u00E2\u0088\u009220 \u00E2\u0088\u009210 0 10 20 30 500 600 700 800 900 1000 Temperature (\u00C2\u00B0C) Pr es su re (h Pa ) UIL YZT YLW Whistler (d) July 2009 1600 LST Figure 3.4: Whistler mountaintop temperatures compared to radiosonde profiles from December 2008 and July 2009. Data in the left and right panels represent monthly averages for the respective upper air observation times of 1200 UTC/0400 LST and 0000 UTC/1600 LST. Error bars surrounding the Whistler values indicate plus and minus one standard deviation from the averages (based on hourly data). Figures 3.4c and 3.4d provide temperature comparisons for a summer month. In this case, more differences between the morning and afternoon observations are apparent than in the winter example. Lapse rates from the radiosonde data were more stable in the morning hours than in the afternoon, especially for the coastal stations. There was also some diurnal warming of the lower troposphere up to 40 3.3. Comparisons to radiosonde profiles about the level of the Whistler site, indicating that PBL effects\u00E2\u0080\u0094on some days, at least\u00E2\u0080\u0094 reached the height of Coast Mountain ridgetops throughout the region. The average temperature on the peak was on the cool side of the radiosonde profiles in the morning and on the warm side in the afternoon. Although these temperature differences are not large, they show that the average diurnal range in temperature on the peak was larger than that of the radiosonde observations. Water vapour comparisons for three of the months analyzed are given in Figure 3.5. Water vapour (w) is probably a better indicator variable of lower-elevation air than T, because w represents the concentration of a physical entity that can be transported vertically in the atmosphere. The December plots (Figures 3.5a and 3.5b) show that the Whistler water vapour readings were generally close to the free- air values. This suggests that there was not a great deal of vertical motion exclusive to the mountainous area. It is expected that the cold temperatures observed during much of December at Whistler were associated with stable air masses that kept the mountaintop air decoupled from the valley air. The plots for April (Figures 3.5c and 3.5d) show that, on average, the Whistler mixing ratios were notably higher than the same-altitude free-air observations, espe- cially in the afternoon. The greater difference in the afternoon confirms that diurnal cycles of water vapour at that altitude were unique to\u00E2\u0080\u0094or at least better developed in\u00E2\u0080\u0094the mountains. Similar patterns are apparent for August in Figures 3.5e and 3.5f, but with even greater absolute differences between the mountaintop and the free air (note the changing x-axis scale). An interesting feature of the April and August water vapour plots in Figure 3.5 is the enhanced moisture at Kelowna above about 850 hPa as compared to Quillayute and Port Hardy. In a comparison of profiles from radiosonde sites upwind and downwind of the Swiss Alps, Henne et al. (2005) found elevated moisture layers in the afternoon soundings at the downwind site. They attributed this moisture increase to diurnal mountain venting of water vapour in the Alps and subsequent 41 3.3. Comparisons to radiosonde profiles 0 1 2 3 4 5 500 600 700 800 900 1000 Mixing Ratio (g/kg) Pr es su re (h Pa ) UIL YZT YLW Whistler (a) December 2008 0400 LST 0 1 2 3 4 5 500 600 700 800 900 1000 Mixing Ratio (g/kg) Pr es su re (h Pa ) UIL YZT YLW Whistler (b) December 2008 1600 LST 0 1 2 3 4 5 500 600 700 800 900 1000 Mixing Ratio (g/kg) Pr es su re (h Pa ) UIL YZT YLW Whistler (c) April 2009 0400 LST 0 1 2 3 4 5 500 600 700 800 900 1000 Mixing Ratio (g/kg) Pr es su re (h Pa ) UIL YZT YLW Whistler (d) April 2009 1600 LST 0 2 4 6 8 10 500 600 700 800 900 1000 Mixing Ratio (g/kg) Pr es su re (h Pa ) UIL YZT YLW Whistler (e) August 2009 0400 LST 0 2 4 6 8 10 500 600 700 800 900 1000 Mixing Ratio (g/kg) Pr es su re (h Pa ) UIL YZT YLW Whistler (f) August 2009 1600 LST Figure 3.5: Whistler mountaintop water vapour mixing ratios compared to ra- diosonde profiles for selected months. Graphical conventions are the same as for Figure 3.4. 42 3.3. Comparisons to radiosonde profiles horizontal transport to the downwind location. As Kelowna is usually downwind of the Coast Mountains, it may be that similar elevated layers were responsible for the water vapour enhancement observed there during the warm season. Local convection was probably also an important factor in the vertical moisture distribution. The profiles show that the low-altitude mixing ratios at Kelowna decreased diurnally in April and August, which conforms to the conceptual model of convective thermals transporting surface-based constituents upward. Warm season convection is more limited at the coastal sites by the stabilizing factor of cool marine air at the lower levels (Mass, 2008). Despite the enhancement of water vapour over Kelowna between approximately 850 and 600 hPa during the warm season, mixing ratios were still consistently higher at the Whistler site. Analysis of the average morning and afternoon observations from each month of the study period found that mixing ratio differences between the Whistler mountaintop site and the equivalent altitude of the Kelowna soundings were statistically significant for all times except 0400 LST in December and January (based on paired t-tests with \u00CE\u00B1 = 0.05). The most significant differences were found for the 1600 LST observations between May and August, when p-values were on the order of 10\u00E2\u0088\u00929 to 10\u00E2\u0088\u009210. Whistler soundings Weather balloons were launched three times daily from Whistler Valley during the period of 5 \u00E2\u0080\u0093 23 February 2009, providing a valuable dataset of free-air observations in the immediate vicinity of Whistler Mountain. Figure 3.6 shows average temper- ature profiles for each of the three daily observation times. The profile comparison indicates that considerable warming occurred in the low levels from the morning to the afternoon soundings. Slight diurnal warming is evident up to at least 750 hPa, which corresponds approximately to the level of the higher summits in the vicinity of Whistler. Thus, from these temperature profiles alone, one can infer 43 3.3. Comparisons to radiosonde profiles that effects of surface heating were at times discernible throughout the depth of the valley atmosphere. This is a somewhat surprising result for February, considering that other mid-latitude mountain observatories have been considered to be repre- sentative of the FT almost the entire winter (e.g. Baltensperger et al., 1997). The assumption is that the limited solar insolation of winter and high static stability of cold air masses prevent substantial convective mixing from occurring, leaving the mountaintops decoupled from the lowland PBL. A review of the February 2009 weather at Whistler finds that conditions were more typical of springtime for much of the observation period in question. Sunny, dry weather prevailed for most of the February 8 \u00E2\u0080\u0093 21 period, with afternoon tem- peratures in the valley reaching the 6 \u00E2\u0097\u00A6 to 8 \u00E2\u0097\u00A6C range during the 18th \u00E2\u0080\u0093 20th. Thus, it is not unreasonable to assume that convective mixing and possibly slope winds mixed air upwards on those days, growing the CBL to mountaintop height. A case study of February 20th, presented in Chapter 5, supports this interpretation. \u00E2\u0088\u009230 \u00E2\u0088\u009225 \u00E2\u0088\u009220 \u00E2\u0088\u009215 \u00E2\u0088\u009210 \u00E2\u0088\u00925 0 5 500 550 600 650 700 750 800 850 900 950 Temperature (\u00C2\u00B0C) Pr es su re (h Pa ) WAE Temperature Pro!les Feb 5\u00E2\u0088\u009223, 2009 0400 LST 1000 LST 1600 LST Figure 3.6: Average temperature profiles from the 19-day period in February 2009 when weather balloons were launched from Whistler Valley. (WAE is the site id for the Whistler Nesters observing location.) 44 3.3. Comparisons to radiosonde profiles Temperatures on the mountaintop are compared to the radiosonde profiles in Figure 3.7. Profiles from the three more distant upper air sites were averaged to simplify the presentation in this case. From these plots, it can be seen that the temperature on the peak had a slightly larger diurnal range than the adjacent free air. \u00E2\u0088\u009230 \u00E2\u0088\u009225 \u00E2\u0088\u009220 \u00E2\u0088\u009215 \u00E2\u0088\u009210 \u00E2\u0088\u00925 0 5 500 600 700 800 900 1000 Temperature (\u00C2\u00B0C) Pr es su re (h Pa ) Feb 5\u00E2\u0088\u009223 12z Avg UIL,YLW,YZT WAE Whistler Mtn (a) Feb 5 \u00E2\u0080\u0093 23 0400 LST \u00E2\u0088\u009230 \u00E2\u0088\u009225 \u00E2\u0088\u009220 \u00E2\u0088\u009215 \u00E2\u0088\u009210 \u00E2\u0088\u00925 0 5 500 600 700 800 900 1000 Temperature (\u00C2\u00B0C) Pr es su re (h Pa ) Feb 5\u00E2\u0088\u009223 0z Avg UIL,YLW,YZT WAE Whistler Mtn (b) Feb 5 \u00E2\u0080\u0093 23 1600 LST Figure 3.7: Whistler mountaintop temperatures compared to radiosonde profiles from the special observing period in February 2009. Solid lines represent mean temperature profiles from the Whistler soundings (WAE), and dashed lines represent averaged profiles of the three other regional upper air sites. Figure 3.8 contains plots similar to Figure 3.7, but for water vapour. The average value of w on the peak was slightly enhanced compared to the free air values at the early morning observation time, and was more significantly enhanced in the afternoon. The fact that the 1600 LST mountaintop w value exceeded even those of the lowest levels on the WAE profile is not easily explained, but an inspection of RH values indicated that the peak was likely in cloud on several of the afternoons included. During convective conditions, clouds often form around the mountaintops first while the air over the valley remains cloud-free. Converging flows of orographic lift (seen as the mountain venting mechanism in Figure 2.1) can serve to concentrate water vapour around the peaks so that the saturation point is reached and clouds 45 3.3. Comparisons to radiosonde profiles form. The base of convective clouds has often been associated with the top of the PBL (e.g. Oke, 1987). 0 0.5 1 1.5 2 2.5 3 3.5 500 600 700 800 900 1000 Mixing Ratio (g/kg) Pr es su re (h Pa ) Feb 5\u00E2\u0088\u009223 12z Avg UIL,YLW,YZT WAE Whistler Mtn (a) Feb 5 \u00E2\u0080\u0093 23 0400 LST 0 0.5 1 1.5 2 2.5 3 3.5 500 600 700 800 900 1000 Mixing Ratio (g/kg) Pr es su re (h Pa ) Feb 5\u00E2\u0088\u009223 0z Avg UIL,YLW,YZT WAE Whistler Mtn (b) Feb 5 \u00E2\u0080\u0093 23 1600 LST Figure 3.8: Whistler mountaintop water vapour mixing ratios compared to ra- diosonde profiles from the special observing period in February 2009. Solid lines represent mean mixing ratio profiles from the Whistler soundings, and dashed lines represent averaged profiles of the three other regional upper air sites. The preceding analysis of the Whistler sounding dataset has demonstrated that, even during the winter season, BL effects such as diurnal cycles in temperature and moisture variables can be found at the summit. Monthly analysis of the entire one-year study period, which was presented earlier, showed that the strength (and, presumably, frequency) of the diurnal cycles varied seasonally, with pronounced cycles in the spring and summer and weaker average cycles in late fall through early winter. A brief look at the day-to-day variability in diurnal T and w trends served as a reminder that the averaged monthly results cannot be applied to all days that comprise the averages. At this point, it is clear that air parcels on the summit of Whistler quite often do not behave like FT air. However, the presence of diurnal cycles and enhanced water vapour concentration on the mountaintop does not necessarily confirm that air from near the valley bottoms is reaching the peak. It may be that the diurnal 46 3.4. Analysis of a CN dataset temperature cycles can be explained by heating and cooling of surfaces (rocks and soil) immediately surrounding the summit observing site. The diurnal increase in w could likewise be explained by uplift of air from upper slopes of the mountain that was enriched in moisture via evapotranspiration from the soil and vegetation of those slopes. In such a case, it is possible that the PBL and its attendant local pollutants remain below the level of influence for the summit. As depicted in Figure 3.9, a terrain-following BL more local to the mountain ridges may exist on some days, independent of the valley-based PBL. Further analysis is required in order to show if these boundary layers are typically distinct or connected. Polluted valley BL Terrain-following mountain BL ? Measurement site Free Troposphere Figure 3.9: A simplified conceptual model of planetary boundary layers in moun- tainous terrain. 3.4 Analysis of a CN dataset An aerosol dataset from the AQRD site on Whistler is included here as another possible means of detecting PBL effects on the peak. CN number concentration was chosen as the variable for analysis due to its relative simplicity compared to other aerosol measurements and due to data availability considerations. 47 3.4. Analysis of a CN dataset 3.4.1 Details of the dataset and processing methods The aerosol dataset was obtained from Environment Canada and analyzed for the period December 2008 through November 2009. Initially the intended study period was the full calendar year of 2009, but December 2009 could not be included because instrument problems led to bad or missing data for most of the month. The data are from a TSI 3025 Ultrafine Condensation Particle Counter (UCPC), which records counts of particles larger than approximately 3 nm diameter, with near 100% detec- tion efficiency beginning at \u00E2\u0089\u00886 nm (Wiedensohler et al., 1996). The TSI 3025 uses butanol as a condensing fluid to enlarge the particles to a size that can be detected by an optical sensor. The fine aerosol particles that tend to dominate the counts from a UCPC are known as condensation nuclei (CN), and are referred to as such in the following discussion. The raw data files consisted of voltage readings taken every minute from the UCPC. Voltages were converted to particle concentration (number of particles per cubic centimetre, denoted cm\u00E2\u0088\u00923) by the formula 10V , where V is the voltage read- ing. The files were then checked for spurious or missing data. A manual inspection of the minute data was done, and the highest of the spurious spikes in particle con- centration were deleted. The spikes that were deemed spurious were characterized by values that jumped suddenly (within several minutes) an order of magnitude or more and/or were very erratic from minute to minute. Some longer duration (\u00E2\u0088\u00BC30 minutes) peaks were also eliminated if the concentration values were deemed unrealistically high. It is likely that these episodes were from machinery related to ski resort operations, primarily snowmobiles and snow grooming vehicles. However, there is some risk that such editing may remove instances of natural new particle formation (nucleation). But, as will be discussed later in this section, removal of brief nucleation events from the dataset is likely to simplify, rather than hinder, the analysis of PBL influence. 48 3.4. Analysis of a CN dataset After manual editing, each monthly file was further smoothed by two levels of automated filtering designed to remove short-term spikes in the data. The filter compared each minute value to the average one-hour concentration centred on that minute. In the first pass, the filter deleted values that were more than twice the hourly average. In the second pass, values more than 1.5 times the hourly aver- age were deleted. In both passes, the observations immediately before and after the deleted values were also removed. No interpolation was performed; the flagged minutes were simply set to null values. In most cases, a sufficient number of ob- servations remained to obtain robust hourly averages. The number of observations removed from each monthly file by the filtering steps ranged from \u00E2\u0088\u00BC 50 to several hundred. It was noted that the winter months tended to have noisier datasets than the summer months, which may reflect the increased frequency of vehicles on the mountain during the ski season. An example of the effect of the automated smoothing is shown for April 2009 in Figure 3.10. Once the one-minute data were smoothed, hourly averages were calculated for the entire study period. The hourly averages were then used for all subsequent analysis of the particle dataset. To match the practice used for mete- orological datasets, each hour\u00E2\u0080\u0099s average was computed to represent the 60 minutes ending at the top of that hour. 3.4.2 Descriptive statistics The dataset of hourly concentration values that resulted from the previously de- scribed processing steps was 95% complete, with 445 hours of missing data out of a period of 8760 hours (one year). A majority of the missing data (311 hours) are from November, when valid readings were unavailable from the 12th through the 24th. Other periods of missing data include 20 Jan 1500 LST\u00E2\u0080\u0093 21 Jan 1200 LST and 28 Mar 1500 LST\u00E2\u0080\u0093 31 Mar 1000 LST. Summary statistics for the entire Dec 2008 \u00E2\u0080\u0093Nov 2009 analysis period are given 49 3.4. Analysis of a CN dataset 01/00 06/00 11/00 16/00 21/00 26/00 01/000 2000 4000 6000 8000 10000 Day/Hour (LST) Co nc en tra tio n (cm \u00E2\u0088\u0092 3 ) April 2009 pre\u00E2\u0088\u0092filter (a) Particle concentration data before filtering. 01/00 06/00 11/00 16/00 21/00 26/00 01/000 2000 4000 6000 8000 10000 Day/Hour (LST) Co nc en tra tio n (cm \u00E2\u0088\u0092 3 ) April 2009 filtered (b) Particle concentration data after filtering. Figure 3.10: Effect of automated smoothing scheme on one-minute CN data from April. in Table 3.1. The large difference between the mean and median values reflects a skewed dataset containing numerous episodes of high CN concentration. The standard deviation and range show a high degree of variability in measured particle concentration. The lowest 10% of hourly concentrations are represented by values below 95 cm\u00E2\u0088\u00923, while the highest 10% are found above 1883 cm\u00E2\u0088\u00923. Annual mean 872.2 \u00CF\u0083 = 853 Annual median 634.5 Min hourly avg. 1.3 (9 Nov) Max hourly avg. 7849.9 (6 Jun) Table 3.1: Summary statistics for the one-year CN dataset (based on hourly aver- ages). Values are number concentrations in cm\u00E2\u0088\u00923. The following results sections describe the seasonal and diurnal patterns found in the CN data. It should be noted that analysis of interannual variability is not possible with this one-year dataset. 50 3.4. Analysis of a CN dataset 3.4.3 Seasonal patterns In order to elucidate seasonal variation in CN concentration at the Whistler site, median concentrations were computed for each month of the study period. Figure 3.11 shows monthly box plots of the data, with the median value of each month represented by a horizontal line through the relevant box. The monthly medians vary over an order of magnitude from a low of 120.2 cm\u00E2\u0088\u00923 in January to a high of 1600.8 cm\u00E2\u0088\u00923 in July. The intervening months demonstrate a clear trend from low concentrations in the winter to relatively high concentrations during the summer. The only interruption to this trend is manifested by February having a higher median value than March. 0 500 1000 1500 2000 2500 3000 3500 4000 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Whistler Mountain CN Concentration C N C o n c e n tr a ti o n ( c m -3 ) Figure 3.11: Monthly box plots of CN concentration data. The horizontal line through each box represents that month\u00E2\u0080\u0099s median value, while the lower and upper edges of the boxes mark the 25th and 75th percentiles, respectively. The whiskers extending above and below the boxes end at the adjacent values (the most extreme values not considered outliers.) 51 3.4. Analysis of a CN dataset The observed seasonal trend in CN concentration at Whistler is similar to that of the Jungfraujoch in Switzerland. Data reported by Nyeki et al. (1998) show a December minimum and a June peak in aerosol concentration at Jungfraujoch. However, the median concentrations at Whistler are considerably higher for some of the months, particularly in summer (e.g. 1273 cm\u00E2\u0088\u00923 at Whistler vs. 761 cm\u00E2\u0088\u00923 at Jungfraujoch in June). Calculation of median CN concentrations for three-month seasons also reveals the large difference from summer to winter. The seasonal values are shown in Table 3.2, where abbreviations represent the consecutive three-month periods (e.g. MAM=March, April and May). 3-month period conc. (cm\u00E2\u0088\u00923) DJF 201.8 MAM 637.3 JJA 1414.4 SON 608.1 Table 3.2: Seasonal median CN concentration values from Whistler Mountain. 3.4.4 Diurnal patterns On the timescale of a day, trends in aerosol concentration may be brought about by certain meteorological and/or other physical and chemical processes that also operate on diurnal cycles. To identify diurnal patterns in the CN data, median values were calculated for each hour of the day. These hour-of-day medians were computed for each month of the study period as well as each three-month season. The seasonal medians are shown in Figure 3.12. Seasonal differences in aerosol concentration that were mentioned in the previous section are also evident in Figure 3.12. Winter concentrations were generally low and lacking any well-defined diurnal pattern. The other three seasons all had similar 52 3.4. Analysis of a CN dataset 0 4 8 12 16 20 24 0 200 400 600 800 1000 1200 1400 1600 1800 Hour of Day (LST) C o n c e n tr a ti o n ( c m \u00E2\u0088\u0092 3 ) CN Diurnal Variations by Season DJF MAM JJA SON Figure 3.12: Median values of CN concentration for each hour of day computed over each three-month season of the study period. diurnal trends: a morning minimum in CN concentration followed by a late afternoon or early evening peak. The timing of the concentration maxima and minima showed slight seasonal variation. Most notably, the diurnal peak in CN occurred later in the day during summer (1900 LST) than in spring or fall. As the timeframe analyzed decreases, more variability in the median diurnal cycles of CN concentration becomes apparent. Figure 3.13 illustrates this with a selection of four individual months from the study period. The December data lack any discernible diurnal pattern, whereas the April plot shows a relatively smooth sinusoidal trend with a morning minimum and afternoon maximum in concentration. Afternoon peaks are also well-defined in the July and October data. Similar to the previous analysis for water vapour, these diurnal patterns can be interpreted as evidence of vertical transport bringing relatively aerosol-rich air to the summit from lower elevations during the daytime heating period. It is expected that conditions favourable for vertical mixing, such as slope winds and deep CBL development, are found most frequently in the warm season when solar insolation is strong and the 53 3.4. Analysis of a CN dataset synoptic scale forcing is relatively weak. The lack of such conditions in December can therefore explain the absence of a diurnal signal in the aerosol concentration. 0 4 8 12 16 20 24100 120 140 160 180 200 Hour of Day (LST) CN C on ce nt ra tio n (cm \u00E2\u0088\u0092 3 ) (a) December 2008 0 4 8 12 16 20 24400 600 800 1000 1200 1400 Hour of Day (LST) CN C on ce nt ra tio n (cm \u00E2\u0088\u0092 3 ) (b) April 2009 0 4 8 12 16 20 241200 1400 1600 1800 2000 Hour of Day (LST) CN C on ce nt ra tio n (cm \u00E2\u0088\u0092 3 ) (c) July 2009 0 4 8 12 16 20 24300 400 500 600 700 800 Hour of Day (LST) CN C on ce nt ra tio n (cm \u00E2\u0088\u0092 3 ) (d) October 2009 Figure 3.13: Monthly median values of CN concentration by hour of day for selected months. The above analysis shows that large variations occurred in CN concentration at the Whistler site on both seasonal and diurnal timescales. Monthly median values were an order of magnitude higher in summer than in winter. This is similar to findings at other mid-latitude mountain observatories in Colorado and Switzerland (Nyeki et al., 1998). Various reasons for seasonal differences in tropospheric aerosol observations have been suggested, including seasonal rates of gas-to-particle con- version (Hofmann, 1993), greater dust aerosol production in the summertime (Kim et al., 1988), and an increased biogenic component during the summer months (Blif- 54 3.4. Analysis of a CN dataset ford and Ringer, 1969). In addition to these possibilities, researchers associated with the Jungfraujoch observatory have provided evidence that vertical transport of PBL air to the mountaintop site may be the most significant factor influencing seasonal patterns (Baltensperger et al., 1997; Nyeki et al., 1998). This last possibility credits the increased PBL influence of summer with the higher overall aerosol concentrations found at that time of year. While the overall concentrations were highest in the summer months, the abso- lute amplitude of the diurnal cycle was greatest in spring (March \u00E2\u0080\u0093May), varying from near 400 cm\u00E2\u0088\u00923 at 0700 LST to 1000 cm\u00E2\u0088\u00923 at 1700 LST. Considering relative amplitude, as was done for water vapour in Figure 3.2, the average diurnal range for the summer months seems even less impressive, with lower relative amplitudes than all other seasons. It is possible that the climatological tendency for cold air pools aloft in the springtime creates an unstable environment, which enhances the vertical mixing of aerosols and other constituents. A stability analysis in Chapter 4 considers whether the spring period was indeed relatively unstable. The stable high pressure systems of summer may have limited the convective mixing during June through August. However, as suggested previously regarding the water vapour data, the small relative amplitude of summertime diurnal cycles in CN might say more about the night-time situation than the daytime. Frequent presence of a residual layer of aerosol surrounding the mountaintop at night would explain both the overall higher concentrations observed in summer and the dampened diurnal cycles relative to spring and fall. As with T and w, considerable day-to-day variability can be found in CN trends due to the influence of numerous factors affecting concentrations on the timescale of a day. Monthly and seasonal plots are useful for starting to answer the general research question of how often the site is influenced by PBL air, but identification of PBL vs. FT periods is best done on shorter timeframes. To that end, a method for evaluating daily datasets in terms of their fit to the expected pattern of PBL 55 3.4. Analysis of a CN dataset influence is described next. Definition of a typical diurnal cycle In order to explore the idea that diurnal cycles in aerosol concentration at the Whistler observatory may be due to vertical transport, it was necessary to classify each day\u00E2\u0080\u0099s observations based on conformance to a model of a \u00E2\u0080\u009Ctypical\u00E2\u0080\u009D diurnal pattern. The results of this analysis can be compared to meteorological conditions that are expected to influence vertical mixing on a diurnal timescale. The model of a typical diurnal cycle was based on that of Baltensperger et al. (1997) for Jungfrau- joch, and was adjusted to better fit the data from Whistler. The main aspects of the model are that the daily course of CN concentration follows a roughly sinusoidal pattern and that a well-defined peak in concentration occurs in the afternoon or evening hours. Diurnal patterns in meteorological variables tend to be sinusoidal when they are driven by the daily cycle of incoming solar radiation. Thus, if the aerosol concentration is controlled substantially by meteorological factors that act on diurnal cycles, then the CN data should exhibit similar cycles. A quantitative scheme was developed for identification of typical diurnal cycles in CN concentration. The specifics of the scheme were adjusted in a trial-and- error process to ensure good agreement between the scheme\u00E2\u0080\u0099s results and subjective analysis of daily plots. To model the sinusoidal shape of a diurnal cycle, the hourly CN data for each day were fit to the following regression equation: y = b0 + b1 sin ( 2pit T ) + b2 cos ( 2pit T ) + b3 sin ( 4pit T ) + b4 cos ( 4pit T ) (3.1) where t is the hour of day, T = 24, and b0 = 1. The b1 term of the regression equation models a sine wave and the cosine function in the next term allows for a phase shift. The b3 and b4 terms incorporate harmonics, which allow for asymmetry in the wave form. 56 3.4. Analysis of a CN dataset An r2 value was obtained for each day, indicating the proportion of variability in the CN values that could be represented by the regression equation. Comparison of the r2 values with corresponding daily plots of the CN data led to a subjective threshold of 0.70 for days to be considered reasonably representative of a sinusoidal pattern, i.e. days with r2 > 0.70 were retained for further analysis and days with r2 \u00E2\u0089\u00A4 0.70 were classified as not exhibiting typical diurnal cycles. The next step in the scheme was to find days when the diurnal peak in CN concentration occurred in the afternoon or evening hours. This criterion fits with the conceptual model of daytime CBL and/or slope flow development that gradually brings valley air to higher altitudes as the day progresses. As with the diurnal course of temperature, the maximum values are expected to lag behind the maximum solar input, which occurs at solar noon on clear days. To define this criterion, running three-hour averages were computed for each day, and the highest of these means had to occur in the afternoon or evening. For the months of March through October, the hours of 1400 \u00E2\u0080\u0093 2300 LST were used as the allowable window for the highest three- hour average (ending at the given hour). For the \u00E2\u0080\u009Cwinter\u00E2\u0080\u009D months of November through February, a narrower window of 1300 \u00E2\u0080\u0093 2000 LST was used to reflect the shorter daylight period, which tends to produce earlier peaks in diurnal signals. Lastly, days when the CN concentration remained below 300 cm\u00E2\u0088\u00923 were not included even if the other conditions were met. It is assumed that concentrations this low reflect an absence of lowland PBL air. The three above-mentioned criteria were combined to designate each day of the study period as either representative of a typical diurnal cycle in aerosol concen- tration or not. Examples of \u00E2\u0080\u009Cyes\u00E2\u0080\u009D days and \u00E2\u0080\u009Cno\u00E2\u0080\u009D days are given in Figure 3.14 to illustrate characteristic results of the scheme. Figure 3.15 shows how many days per month met the criteria. Overall, 127 out of 344 days analyzed (37%) were classified as exhibiting typical diurnal variations of CN concentration. There are clear sea- sonal differences in the results, with relatively few days qualifying in late fall and 57 3.4. Analysis of a CN dataset early winter. In the April through September period, all months but August had slightly over 50% of the days meet the criteria. It should be noted again that CN data were unavailable from 12 \u00E2\u0080\u0093 24 November, so there were just 17 days from that month included in the analysis. 0 4 8 12 16 20 240 200 400 600 800 1000 Hour of Day (LST) CN C on ce nt ra tio n (cm \u00E2\u0088\u0092 3 ) NO (a) 11 April 2009 0 4 8 12 16 20 240 1000 2000 3000 4000 5000 Hour of Day (LST) CN C on ce nt ra tio n (cm \u00E2\u0088\u0092 3 ) YES (b) 28 April 2009 0 4 8 12 16 20 240 200 400 600 800 1000 Hour of Day (LST) CN C on ce nt ra tio n (cm \u00E2\u0088\u0092 3 ) NO (c) 22 February 2009 0 4 8 12 16 20 24200 400 600 800 1000 1200 Hour of Day (LST) CN C on ce nt ra tio n (cm \u00E2\u0088\u0092 3 ) YES (d) 13 February 2009 Figure 3.14: Examples of days that meet (\u00E2\u0080\u009CYES\u00E2\u0080\u009D) and do not meet (\u00E2\u0080\u009CNO\u00E2\u0080\u009D) the criteria for a typical diurnal cycle of CN concentration. The scheme for identifying diurnal patterns generally agrees with subjective analysis of the data, but it does have some limitations. The criteria are somewhat arbitrary, and thus different thresholds would produce different results. However, comparison of results from this scheme with an earlier version that used a more complicated set of criteria to define the shape of the diurnal cycle (rather than 58 3.4. Analysis of a CN dataset Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 0 5 10 15 20 25 30 N um be r o f d ay s Month Number of Days with \"Typical\" Diurnal Cycle Figure 3.15: Number of days per month that qualified as displaying a typical diurnal cycle of CN concentration for the period December 2008 \u00E2\u0080\u0093November 2009. using a sine wave model) showed similar outcomes, e.g. 130 days qualified under the previous version vs. 127 with the final scheme. A specific limitation noted with the scheme is that it can disqualify days that have relatively sharp peaks in CN concentration. These sharp peaks could at times be due to BL influence if the PBL air only reached the mountaintop for a brief period of the day. It is also possible that PBL air occasionally affects the summit at atypical times, e.g. such that a peak in CN concentration occurs before noon. Therefore, the scheme used here can be considered conservative, as it was designed in accordance with an idealized conceptual model of diurnal PBL growth. The results in Figure 3.15 show that well-defined diurnal variations in CN con- centration generally occurred more often in the warm season than during winter. It is expected that this is because conditions conducive to vertical mixing are found more frequently during the warm season. However, some inconsistencies in the observed seasonal trends are worth noting. First, February had a rather high fre- quency of diurnal cycles for a winter month, with 11 of 28 days (39%) meeting the 59 3.4. Analysis of a CN dataset criteria. While this may not be generally expected for a winter month, earlier dis- cussion in this chapter has suggested that February 2009 had weather more typical of a spring month in Whistler. The large number of dry, sunny days with light synoptic gradients in February may have facilitated convective mixing capable of transporting lowland air to the summits. In a \u00E2\u0080\u009Cnormal\u00E2\u0080\u009D February, frequent strat- iform precipitation events would tend to limit convective mixing and also reduce aerosol concentrations via washout of particles. August is the other month that appears to be in conflict with seasonal trends found in Figure 3.15. In this case, the number of days with diurnal cycles was anomalously low compared to the other summer months. The statistics in Chapter 2 indicate that August 2009 was generally a hot and dry month at Whistler, which is normally thought to be favourable for convective mixing. However, it may be that strong high pressure aloft induced strong atmospheric stability, suppressing the convection on some of the days. Additionally, closer inspection of the summer weather reports revealed another possible explanation for the small number of days with typical cycles in the CN data: smoke was frequently reported in the Whistler area during August, particularly in the early and late portions of the month. It is believed that the abundance of forest fire smoke in the atmosphere during parts of the summer profoundly influenced aerosol concentrations in the region. Elevated smoke layers that moved through the Whistler area at various times caused peaks in the measured concentrations that often did not fit the timing expected if aerosol movement was dominated by vertical transport. Days with peaks in CN concentra- tion which did not occur in the afternoon or evening hours did not qualify as typical cycle days, even though there may have been PBL influence at the mountaintop. This scenario alludes to the complexity of air mass discrimination at a mountain site, where motions of different dimensions and scales interact to form a complex three-dimensional transport field. 60 3.4. Analysis of a CN dataset Consideration of nucleation events CN data from the UCPC instrument are thought to be generally representative of aerosol concentrations at the site, although the particle counts are dominated by very small aerosol particles. Consequently, it is possible that, on some days, nucleation events significantly affect the measured concentrations. Nucleation in the atmosphere occurs when materials in the vapour phase convert into a lower energy state by undergoing a phase change to liquid or solid (Glickman, 2000). Nucleation, also known as gas-to-particle conversion, usually forms particles smaller than 3 nm diameter. The process of nucleation and subsequent particle growth is often referred to as new particle formation (NPF) (Nishita et al., 2008). Nucleation requires the presence of condensable gases of low volatility such as sulfuric acid, ammonia, or iodide species and may be initiated by atmospheric mixing processes (Kulmala, 2003). For interpretation of the CN data, the concern is that NPF events could be the cause of diurnal peaks rather than vertical transport of PBL air. NPF events have been observed at some mountain observatories via analysis of particle size distribution data. For example, at Storm Peak Lab in Colorado, nucle- ation events were found during 62% of 400 measurement days, consistently occurring in mid-afternoon (Hallar, 2010). As the Storm Peak site is thought to experience a nearly daily transition from FT to PBL air near mid-day (Lowenthal et al., 2002), it is possible that mixing of condensable vapours from the PBL provides some of the necessary conditions for the nucleation events. At a mountain site in Japan, Nishita et al. (2008) found that concentrations of nucleation mode particles were strongly correlated with water vapour mixing ratio. Rapid concentration increases of small particles were concurrent with increases in w and periods of upslope valley winds, leading the investigators to conclude that nucleation had occurred in PBL air, which was subsequently transported to the site. Identification of NPF events requires particle size distribution data, which are 61 3.5. Correlations between water vapour and CN unavailable for large portions of the study period considered here. For the purposes of this study, it will be assumed that nucleation at the Whistler site is usually dependent on the presence of condensable gases from the PBL. Thus, the resulting diurnal increase in CN concentration can still be considered a signature of PBL influence. It appears to be an open scientific question whether or not nucleation commonly occurs in the lower FT (A.G.Hallar, personal communication), and full analysis of the nucleation topic is beyond the scope of this study. Therefore, the above assumption is made while recognizing that it may not be valid in some cases. 3.5 Correlations between water vapour and CN If diurnal increases in both water vapour and particle concentration are indicators of PBL air arriving at the mountain site, then it is expected that w and CN concen- tration will be well correlated on days with PBL influence. To examine the relation between these two variables, linear correlation coefficients were calculated first on a monthly basis and then on a daily basis. For the monthly analysis, hour-of-day medians of w and CN concentration were calculated and the Pearson correlation coefficient (r) was computed for each month as an indicator of the linear dependence between the two medians. Results are given in Table 3.3. The r values in Table 3.3 show that correlations between water vapour and CN concentrations were high in the spring and summer, with the exception of March. Significant positive correlations were also found for January, February and October. These results suggest that water vapour and aerosol concentrations are often af- fected by the same mechanism(s). The lack of correlation between the two variables in November is not surprising, as November was the wettest month of the study period. Any positive correlation between w and CN from PBL-influenced days was likely cancelled out by the negative correlation of humid days with precipitation 62 3.5. Correlations between water vapour and CN Month Pearson r p -value Jan 0.49 0.016 Feb 0.60 1.77x10\u00E2\u0088\u00923 Mar 0.12 0.572 Apr 0.78 7.56x10\u00E2\u0088\u00926 May 0.81 1.57x10\u00E2\u0088\u00926 Jun 0.85 1.80x10\u00E2\u0088\u00927 Jul 0.81 1.90x10\u00E2\u0088\u00926 Aug 0.83 7.02x10\u00E2\u0088\u00927 Sep 0.34 0.108 Oct 0.57 3.79x10\u00E2\u0088\u00923 Nov -0.06 0.780 Dec 0.21 0.321 Table 3.3: Correlations between water vapour and CN concentration for each month, based on hour-of-day medians. (December values are from 2008; all other months are 2009.) and resulting washout of aerosol particles. Examples of months with low and high correlations are given in Figures 3.16a and 3.16b, respectively. Median w and CN values tracked each other closely in June, with the only deviations from a near-perfect correlation being due to the changes in w occurring 1 \u00E2\u0080\u0093 2 hours earlier than those of CN concentration. In the case of March, it can be seen that while the diurnal cycle was well-defined for w, the median CN values remained high through the evening, resulting in a poor overall correlation for the month. A review of daily CN plots for March revealed that several days with high late-evening values affected the monthly medians. In particular, March 2nd had generally rising CN concentrations through the day to a value of nearly 2000 cm\u00E2\u0088\u00923 at midnight. And on the 6th, a rather dramatic peak concentration of over 5000 cm\u00E2\u0088\u00923 occurred in the early evening. These very high values may have been associated with events other than diurnal PBL influence, e.g. regional or long-range pollution layers being advected through the site in the synoptic scale flow. 63 3.5. Correlations between water vapour and CN 0 4 8 12 16 20 24100 200 300 400 500 CN C on ce nt ra tio n (cm \u00E2\u0088\u0092 3 ) Hour of day (LST) CN and WV March 2009 2.0 2.2 2.4 2.6 2.8 M ix in g Ra tio (g /kg ) CN w (a) March 2009 (r = 0.12) 0 4 8 12 16 20 24800 1000 1200 1400 1600 CN C on ce nt ra tio n (cm \u00E2\u0088\u0092 3 ) Hour of day (LST) CN and WV June 2009 4.8 5.0 5.2 5.4 5.6 M ix in g Ra tio (g /kg ) CN w (b) June 2009 (r = 0.85) Figure 3.16: Median values of CN concentration and water vapour mixing ratio by hour of day for March and June. In order to further investigate the relation between water vapour and aerosol concentration, correlations between the two variables were analyzed on a daily basis. In this case, Spearman\u00E2\u0080\u0099s rank correlation coefficient (rs) was used, because the daily data consist of small sample sizes and cannot be assumed to be normally distributed. A great deal of variety was found in the daily trends, with various ways in which both positive and negative correlations were manifested. Examples of four different scenarios are given in Figure 3.17. Proposed explanations for each scenario, based on the synoptic and mesoscale weather conditions observed on the relevant days, are given below. On August 26th (Figure 3.17a), there was a strong positive correlation between w and CN concentration, with an rs value of 0.80. However, due to the sharp decline in concentration during the late afternoon and evening, the day did not qualify as having a typical diurnal CN cycle. Based on a review of synoptic weather maps from that day (not shown here), it appears as though the passage of a high pressure ridge caused a wind direction change in coastal BC from onshore westerly to offshore northerly flow. The northerly winds ushered in drier air, which in this case was also cleaner in terms of particulate matter. The morning minimum and afternoon peak in both variables suggest that PBL influence may have been felt at the mountaintop 64 3.5. Correlations between water vapour and CN 0 4 8 12 16 20 24500 1000 1500 CN C on ce nt ra tio n (cm \u00E2\u0088\u0092 3 ) Hour of day (LST) CN and WV August 26 2.0 3.0 4.0 5.0 6.0 M ix in g Ra tio (g /kg ) CN w (a) 26 August 2009 (rs = 0.80) 0 4 8 12 16 20 240 500 1000 1500 CN C on ce nt ra tio n (cm \u00E2\u0088\u0092 3 ) Hour of day (LST) CN and WV April 19 2.0 3.0 4.0 5.0 M ix in g Ra tio (g /kg ) CN w (b) 19 April 2009 (rs = \u00E2\u0088\u00920.82) 0 4 8 12 16 20 240 500 1000 1500 2000 2500 CN C on ce nt ra tio n (cm \u00E2\u0088\u0092 3 ) Hour of day (LST) CN and WV April 7 1.0 1.5 2.0 2.5 3.0 3.5 M ix in g Ra tio (g /kg ) CN w (c) 7 April 2009 (rs = 0.89) 0 4 8 12 16 20 240 500 1000 1500 2000 CN C on ce nt ra tio n (cm \u00E2\u0088\u0092 3 ) Hour of day (LST) CN and WV March 2 2.5 3.0 3.5 4.0 4.5 M ix in g Ra tio (g /kg ) CN w (d) 2 March 2009 (rs = \u00E2\u0088\u00920.95) Figure 3.17: Plots of hourly CN concentration and water vapour mixing ratio for selected days. 65 3.5. Correlations between water vapour and CN that day, but the departure from a classic sinusoidal shape meant that the criteria for a typical diurnal cycle were not met. An example of a strong negative correlation between w and CN can be seen in the case of April 19th (Figure 3.17b). Weather observations from Whistler Nesters indicated steady light rain through the day. Thus, water vapour increased to the saturation point while aerosol particles were washed out of the local air mass by precipitation. It is expected that, on days such as this, stable stratification of the atmosphere and limited solar insolation at the surface usually act to limit CBL growth to below the level of the mountaintop. The next example of a highly significant positive correlation between water vapour and CN comes from April 7th (Figure 3.17c). Surface observation sites in the area reported clear skies and light winds, i.e. a quiet synoptic scale pattern. It is assumed that mesoscale processes of CBL growth and/or slope winds brought PBL air to the summit on this day, producing the \u00E2\u0080\u009Cclassic\u00E2\u0080\u009D diurnal pattern in both variables. This day exemplifies the pattern expected from these two variables on days with PBL influence. Lastly, the case of March 2nd (Figure 3.17d) shows the two variables almost perfectly anti-correlated. Water vapour decreased substantially after about 0800 LST as the CN concentration increased sharply. A review of synoptic maps and surface observations revealed that these changes began as a strong cold front passed through the area. The synoptic scale subsidence and advection of colder, drier air typically found behind a cold front explains the decreasing mixing ratio. However, in this case, the air mass change was not to a cleaner one in terms of PM. This appears to be a case of an elevated pollution layer of some kind, possibly from LRT, impacting the mountain site by way of a subsiding air mass. Rapidly decreasing w, especially when coincident with increasing temperature, is a signature of subsidence, which can be useful for identification of FT conditions. Days with strong correlations between w and CN were identified by way of t-tests 66 3.5. Correlations between water vapour and CN on whether the correlations were different from zero. Figure 3.18 compiles the days on which the resulting p-values for positive correlations were less than 0.01 and the CN data exhibited typical diurnal cycles. Note that these results are a subset of the days represented by Figure 3.15. It is possible that some days had afternoon peaks and morning minima in aerosol concentration by way of processes other than PBL influence, e.g. horizontal advection of pollutants in the FT or NPF events not associated with the PBL. By including only days that also had strong positive correlations between aerosol and water vapour, Figure 3.18 represents the days that best fit the conceptual model of convective vertical transport of lowland air to the mountaintop. While the months of November, December and January were nearly free of this pattern, the remainder of the months all had at least a few such days, and the highest frequency of the pattern was found in the April \u00E2\u0080\u0093 July period. Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 0 5 10 15 Month N um be r o f d ay s Typical Diurnal Cycle and High r s Value Figure 3.18: Number of days per month with a typical diurnal cycle of CN concen- tration that also showed strong positive correlations between water vapour mixing ratio and CN concentration for the period December 2008 \u00E2\u0080\u0093November 2009. 67 3.6. Conclusions 3.6 Conclusions Datasets from the Whistler Mountain observatory have been analyzed for the full one-year study period in order to assess possible indicators of PBL influence that have been used in previous studies. Analysis of diurnal cycles has been emphasized, with the idea that PBL influence at Whistler most often comes from thermally induced upward motions, which are normally limited to the daytime hours. Lift of lowland air by fronts, low pressure systems or via the mechanical forcing of topography\u00E2\u0080\u0094 i.e. synoptic scale lift\u00E2\u0080\u0094 is not expected to operate on any regular diurnal timescale, and thus is not included for now in the analysis. Similar preliminary conclusions can be made from the analysis of temperature, water vapour and aerosol observations. The high frequency of diurnal cycles found in all the variables for most of the year provides considerable evidence for PBL influence on the mountaintop, particularly during afternoon and evening hours in the warm season. Only the months of November, December and January could be characterized as rarely exhibiting well-defined diurnal cycles (e.g. less than five days per month for CN data). However, analysis based solely on observations from the summit does not fully elucidate the nature of the BL condition typically encountered there. From the mountaintop data, it cannot be known if there is a deep CBL extending from the valley floor to the mountaintops and beyond, or if the diurnal cycles are merely a reflection of more local upslope flows which have their origins above the valley BL. This uncertainty arises because none of the entities measured at the site are perfect tracers of valley air. Comparisons between Whistler mountaintop readings and regional free-air obser- vations from radiosondes confirm that the mountain site is often not representative of free-air conditions. However, the large distance (250+ km) separating Whistler from the three nearest routine sounding sites limits the conclusions one can draw from these datasets. The supplementary upper air dataset from Whistler-based ra- 68 3.6. Conclusions diosonde observations, which was introduced in this chapter, does offer information that may reduce the uncertainty regarding the nature of the mountain boundary layers. Results from analysis of temperature profiles from this dataset suggest that diurnal heating effects do at times get distributed through the depth of the valley atmosphere, which gives support to the deep CBL scenario. One of the case studies presented in Chapter 5 uses this upper air dataset to evaluate the vertical structure of the Whistler BL in greater detail. In the next chapter, some of the expected driving forces of PBL growth and vertical transport are examined to see how often they were present in the Whistler area during the study period. Correlations are then explored between driving forces and indicators of PBL influence. 69 Chapter 4 Driving forces of PBL influence 4.1 Introduction The PBL indicators analyzed in the preceding chapter provide a sense of how often PBL air affects the Whistler site. Including a full year of data in the analysis provides insight into seasonal differences, but it should be kept in mind that interannual variability is expected to be considerable due to the vagaries of macro and synoptic scale weather patterns. Previous studies have shown that the frequency of PBL air reaching mountain sites depends on the synoptic circulation type (e.g. Lugauer et al., 1998; Collaud Coen et al., 2010). In additional to affecting vertical transport, the synoptic scale pattern can directly affect pollutant loads at a given location via horizontal transport and through removal by precipitation. Therefore, interpretation of air chemistry data collected from a mountain site should consider synoptic scale effects as well as PBL influence to properly characterize the sampled air parcels. In this chapter possible mechanisms for the lifting of valley-based PBL air to the summit of Whistler Mountain are investigated. Although mesoscale processes related to convective mixing are expected to be most important in this regard, this chapter begins and ends with synoptic scale analyses that highlight the importance of considering the different scales of motion and their interactions. 70 4.2. An example of synoptic scale influence 4.2 An example of synoptic scale influence Horizontal wind direction is an obvious variable to consider when characterizing the air stream being sampled at any air chemistry site. Knowledge of upstream sources can then be applied during data interpretation. Ideally, one would like to know the wind flow not just at the measurement point but also for a considerable distance upstream. For this purpose, trajectory models have often been applied to source apportionment problems. Several of the previously mentioned studies from other mountain observatories have used back-trajectory modeling to aid their data interpretation (e.g. Weiss-Penzias et al., 2006; Henne et al., 2008; Obrist et al., 2008). In this section, wind direction over Whistler Mountain is analyzed for effects on CN concentration and then modeled back-trajectories are employed to add information about source regions. For this analysis, synoptic scale wind flow is represented by daily mean winds at the 700 hPa level, which is the first standard pressure level encountered above Whistler Mountain. As the upper air observations from Whistler are only available for a brief part of the study period, NCEP reanalysis datasets were used to create the daily wind averages. For each LST day, u and v wind components from the four available synoptic times were acquired and used to calculate vector-average wind speed and direction. Data from the grid point nearest Whistler were used: 50.0 \u00E2\u0097\u00A6N/122.5 \u00E2\u0097\u00A6W, which is approximately 34 km east of Whistler Mountain. Synoptic wind analysis for the Jungfraujoch by Lugauer (1998) found that both the mean aerosol concentration and character of the diurnal signal depended on wind direction. Here, only the mean concentrations are evaluated by wind direction to demonstrate the importance of synoptic scale flow. Figure 4.1 shows the mean CN concentration by cardinal wind direction for the entire study period. It is evident from Figure 4.1 that CN concentrations tended to be highest when the 700 hPa wind was from the east and lowest when the wind was from the west. 71 4.2. An example of synoptic scale influence North East South West 400 600 800 1000 1200 1400 Wind Direction CN C on ce nt ra tio n (cm \u00E2\u0088\u0092 3 ) CN Concentration by Wind Direction Figure 4.1: Average CN concentration at the Whistler site by wind direction for the Dec 2008 \u00E2\u0080\u0093Nov 2009 study period, based on daily means and winds from 700 hPa NCEP reanalysis datasets. Analysis of the data by season was done to see if these wind direction dependencies may simply reflect concurrent seasonal differences in aerosol load and wind direction. It was found that west wind days had the lowest average concentration in all seasons and easterly days had the highest concentration in all seasons except autumn (Sept \u00E2\u0080\u0093 Nov), when only two days had easterly winds. The differences were most pronounced in the spring (Mar \u00E2\u0080\u0093May), when the mean concentration for easterly days was more than double that of westerly days. Additionally, the proportion of days that fell into each wind direction category did not vary a great deal by season. In each season, westerly winds were dominant on 50 \u00E2\u0080\u0093 62% of the days and easterly winds accounted for less than 10% of the days. For the whole study period, statistical t-tests (without consideration of possible temporal autocorrelation) found the mean aerosol concentrations for the four wind directions to be significantly different from each other (at \u00CE\u00B1 = 0.05) for every possible comparison except north vs. south. In order to learn more about the flow patterns associated with the different 700 hPa wind directions, back-trajectories were plotted for a number of the individual 72 4.2. An example of synoptic scale influence days in the study period. The back-trajectories were produced using an on-line version of the NOAA HYSPLIT model (Draxler and Rolph, 2010). NCEP\u00E2\u0080\u0099s Global Data Assimilation System (GDAS) archive was selected as the meteorological input, and three-day back-trajectories were run from the latitude, longitude and elevation of Whistler Mountain. Two typical examples each of west wind and east wind days are given in Figure 4.2. Figures 4.2a and 4.2b show two very different parcel trajectories that both arrived at Whistler as flow from the westerly sector. Inspection of a number of westerly days indicated that, although the individual paths varied a great deal, the parcels usually spent most of their time over the Pacific Ocean before taking a fairly direct route from the ocean to Whistler. In contrast, the easterly days (Figures 4.2c and 4.2d) involved parcels that had traveled over the North American continent for the entire three-day period. Trajectories for nearly all of the easterly days from the study period showed pathways over North America. In some cases, such as the 18 December example in Figure 4.2c, the air spent considerable time close to the surface (i.e. in the PBL) before being lifted to the level of Whistler Mountain. Thus, pollutants from anthropogenic and natural sources in the United States and/or Canada were likely carried to the observing site, enhancing the CN concentrations on the east wind days. The westerly cases may have included episodes of trans-Pacific pollution transport, but the long journey over the ocean would have given time for dispersion processes to act on such plumes, lowering the particle concentrations. These examples provide a sense of the variety of three-dimensional pathways by which air parcels can arrive at the Whistler site. Synoptic scale transport de- termines which source regions influence the air being sampled at the mountaintop. Back-trajectory modeling is a useful tool for source apportionment studies. How- ever, the global and continental scale weather models used as input for trajectory models typically are not able to simulate the local and mesoscale processes that are crucial to determining vertical transport. In the next two sections, attention is 73 4.2. An example of synoptic scale influence (a) 20 April 2009 (b) 26 September 2009 (c) 18 December 2008 (d) 29 April 2009 Figure 4.2: Model trajectories tracing air parcels backwards 72 hours from Whistler Mountain for two days with westerly flow (a and b) and two days with easterly flow (c and d). Model output from Draxler and Rolph (2010). 74 4.3. Atmospheric stability again focused on these smaller scales of motion to investigate some possible driving forces of vertical transport in the immediate vicinity of Whistler Mountain. 4.3 Atmospheric stability While mechanical turbulence can cause mixing within a thin layer near the surface, convective thermals are the primary means by which the daytime BL (the CBL) in- creases its depth (Oke, 1987). Convection commences when surface heating creates an unstable lapse rate near the ground, making air parcels buoyant. The height to which thermals will rise depends on the environmental lapse rate present through the rest of the troposphere. Deep convection occurs when an unstable tempera- ture profile extends high into the troposphere, allowing buoyant parcels to continue accelerating upward. Nseir (2007) applied stability analysis to estimate the timing of PBL influence at the Whistler site. For the springtime periods considered, the transition times from FT to PBL in the morning, and back to FT in the evening, were estimated based on the onset of unstable and stable lapse rates, respectively. Lapse rates in this case were determined from temperature sensors located approximately every 300 m vertically on the mountainside. For this study, a static stability parameter was created based on lapse rates be- tween the valley floor and the summit of Whistler Mountain. Following the practice of Baltensperger et al. (1997), an \u00E2\u0080\u009Ceffective\u00E2\u0080\u009D potential temperature difference was calculated between the VOC station at Nesters in Whistler Valley and the AQRD meteorological sensors on the peak. For this parameter, the potential temperature difference (\u00E2\u0088\u0086\u00CE\u00B8) was used whenever RH on the summit was \u00E2\u0089\u00A4 95%, and the equivalent potential temperature difference (\u00E2\u0088\u0086\u00CE\u00B8e) was used when RH exceeded 95%. Potential temperature is the temperature a parcel would have if it were moved adiabatically up or down to a standard reference pressure of 1000 hPa. Thus, an unstable profile 75 4.3. Atmospheric stability would be one with \u00CE\u00B8 decreasing with height (Oke, 1987). For saturated parcels, effects of latent heating are accounted for in the calculation of \u00CE\u00B8e. Hourly readings of T, RH and P from VOC and the peak were used to calculate the effective potential temperature difference for each hour of the study period. \u00CE\u00B8 was calculated from the equation given in Stull (2000) and \u00CE\u00B8e was computed from equations in Bolton (1980). A daily stability parameter was then obtained by tak- ing the lowest effective potential temperature difference for each day. Because the difference was calculated as summit \u00E2\u0088\u0092 valley, negative numbers indicate unstable conditions in the layer and positive numbers indicate stable conditions. This pa- rameter simplifies the actual situation by considering only one layer; at times, there could be stable and unstable sub-layers between the valley floor and the summit. Stability parameter values for the study period ranged from \u00E2\u0089\u0088\u00E2\u0088\u00925 to +22. Figure 4.3 gives the median values for each month. As expected, the most stable conditions were found during the winter. In particular, January had numerous days with very high static stability during the period of strong high pressure aloft that dominated Whistler\u00E2\u0080\u0099s weather for two weeks. Only the months of May, June and October had median stability parameter values below zero, with the lowest median value occurring in June. It is expected that values were higher in the late summer period (July \u00E2\u0080\u0093 Sept) because a persistent high pressure ridge pattern effectively blocked any disturbances with cold air aloft from transiting the area. The results in Figure 4.3 broadly agree with some of the seasonal patterns re- garding diurnal cycles that were presented in Chapter 3. The relatively unstable conditions of May were consistent with springtime peaks in diurnal cycle amplitudes of water vapour (Figure 3.2b) and CN (Figures 3.15 and 3.18). However, in the case of water vapour, the highest relative diurnal amplitude occurred in April, not May. Likewise, the low stability parameter value for October does not match the timing of the secondary seasonal peak in diurnal cycle strength that was found for Septem- ber. The high median stability parameter value for February also conflicts with the 76 4.3. Atmospheric stability Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec \u00E2\u0088\u00924 \u00E2\u0088\u00922 0 2 4 6 8 Month \u00E2\u0088\u0086\u00CE\u00B8 e ff (K ) Median Stability Parameter by Month Figure 4.3: Median stability parameter values for each month of the study period. Parameter values are daily minimum effective potential temperature differences be- tween the summit and the valley. relatively high frequency of diurnal cycles found in the T, w and CN data for that month. If diurnal cycles in aerosol concentration can be attributed to convective uplift of PBL air, then such cycles may be expected to occur on days with low static stability. To test this idea, daily stability parameter values were grouped according to results of the CN analysis that identified days with typical diurnal cycles (Section 3.4). Figure 4.4 shows the data as box plots representing the entire study period. The \u00E2\u0080\u009CNO\u00E2\u0080\u009D box plot gives stability parameter values for days that did not have typical CN cycles and the \u00E2\u0080\u009CYES\u00E2\u0080\u009D box plot shows values for days that did fit the typical diurnal cycle criteria. Figure 4.4 indicates that the subset of days lacking well-defined diurnal CN cy- cles had a higher median stability parameter value and included more of the very stable (\u00E2\u0088\u0086\u00CE\u00B8eff > 10) cases than the subset of days with diurnal CN cycles. However, numerous days with diurnal CN cycles occurred when the static stability was high, and there were also days without diurnal cycles that had unstable (\u00E2\u0088\u0086\u00CE\u00B8eff < 0) con- 77 4.3. Atmospheric stability NO YES \u00E2\u0088\u00925 0 5 10 15 20 \u00E2\u0088\u0086\u00CE\u00B8 e ff(K ) CN Diurnal Cycles and Stability Figure 4.4: Box plots of stability parameter values for days with (\u00E2\u0080\u009CYES\u00E2\u0080\u009D) and without (\u00E2\u0080\u009CNO\u00E2\u0080\u009D) typical diurnal cycles of CN concentration. The horizontal line through each box represents the median value, while the lower and upper edges of the boxes mark the 25th and 75th percentiles, respectively. The whiskers extending above and below the boxes end at the adjacent values (the most extreme values not considered outliers.) Outliers are shown as open circles. ditions. This substantial overlap in the datasets means that the stability parameter used here does not reliably predict the presence of diurnal aerosol cycles. Inspection of the data by season yielded results qualitatively similar to the annual box plots presented here: a majority of the very stable days did not have typical diurnal cycles in the CN data, but a wide range of stability parameter values was found for both the \u00E2\u0080\u009CYES\u00E2\u0080\u009D and \u00E2\u0080\u009CNO\u00E2\u0080\u009D days. This outcome agrees with the findings of Baltensperger et al. (1997), who stated that synoptic scale analysis was required to explain the aerosol data in terms of diabatic transport processes. The failure of a stability parameter to predict diurnal cycle days could mean that stability and diurnal cycles are not physically related or it could reflect in- adequacies in the definitions of the stability parameter and/or the typical diurnal cycle. An issue with the stability parameter that is likely significant on some days is that temperatures measured from ground stations on the mountainside may not 78 4.3. Atmospheric stability be representative of the free-air temperatures over the valley. This issue has been investigated for Whistler Mountain by Erven and McKendry (2009), who found that temperature profiles from ground stations on the mountain often differed from those of the local upper air soundings, particularly under clear, high pressure conditions. As a result, lapse rates and any related static stability parameters derived from ground stations at times do not properly characterize the potential for convective mixing. An example of differing slope-wise and free-air temperature profiles is given in Figure 4.5, where potential temperatures from the 9 February afternoon radiosonde observation are compared to a profile of \u00CE\u00B8 values from OAN and AQRD weather stations on the mountain. In this case, the slope-wise profile shows much greater stability than the free-air profile, largely due to the relatively high temperature measured on the peak. It should noted that temperature sensors at all the ground stations had radiation shields but were not aspirated. Thus, for sunny, light wind days a warm bias can be expected. 278 279 280 281 282 283 284 600 650 700 750 800 850 900 950 \u00CE\u00B8 (K) Pr es su re (h Pa ) Feb 9 1600 LST Raob OAN Figure 4.5: Comparison of potential temperature profiles from Whistler radiosonde observations (Raob) and ground-based weather stations (OAN) for the afternoon of February 9, 2009. 79 4.4. Slope and valley wind systems From the example in Figure 4.5, it is apparent that evaluation of the potential for convection and CBL growth based on slope-side temperature observations is problematic. It is possible that a dynamic stability parameter such as the Richardson number or the Obukhov Length would be more useful in this context, but calculation of such parameters requires observations that are not available from Whistler (e.g. sensible heat flux for the Obukhov Length). Another approach\u00E2\u0080\u0094to identify the synoptic scale conditions conducive to strong convective mixing\u00E2\u0080\u0094 is considered in Section 4.5. 4.4 Slope and valley wind systems 4.4.1 Theory and definitions Diurnal wind systems, which are driven by differential heating and cooling in com- plex topography, are an important aspect of mountain climates (Barry, 2008). As discussed in Chapter 2, slope winds have been analyzed at other mountain obser- vatories as mechanisms for upward transport of PBL constituents by day and the return of FT air to alpine sites at night. Valley winds operate similarly to slope winds but on a larger scale. A pressure gradient forms when the upper portion of a mountain valley heats or cools faster than the same-altitude air over the lower val- ley, resulting in a wind directed toward the lower pressure (Whiteman, 2000). Thus, the valley-scale diurnal winds blow up-valley during the daytime heating period and down-valley overnight. Up-valley winds can bring local and regional pollutants to mountain sites by facilitating vertical mixing and via advection of pollutants from sources in the lower valleys and/or adjacent plains. The frequency of thermally driven diurnal wind systems varies depending on the topography and climate of different mountain locations (Whiteman, 2000). For the Whistler area, a lack of wind observations in the Cheakamus, Green, and Fitzsim- mons drainages has precluded any detailed study of valley winds. Also, prior to in- 80 4.4. Slope and valley wind systems stallation of the OAN weather stations in the years leading up to the 2010 Olympics, there were not sufficient wind observations on the mountain to analyze slope wind patterns. For this study, select OAN stations are used for initial investigations of diurnal wind patterns in the Whistler area in order to assess such winds as potential contributors to PBL influence at the mountaintop. Diurnal mountain winds are most likely to develop during clear weather periods with weak synoptic scale forcing, i.e. light winds aloft (Whiteman, 2000; Barry, 2008). Upslope winds typically begin soon after the slopes become sunlit in the morning, but can become overpowered by the stronger valley winds, which develop more gradually (Whiteman, 2000). The timing of wind reversals is not uniform along a mountain slope, and often the period of upslope and/or up-valley flow occurs later in the day on the upper mountain than on the lower slopes. A fully developed valley wind will often be found through the depth of the valley, i.e. up to the ridgetops. However, Reuten et al. (2005) found that upslope flows may fill as little as half the CBL, while return flow occupies the upper portion of the CBL. Thus, observations of upslope flows on the upper half of a mountain may be sufficient to assume that some mixing of PBL air is occurring at the summit. 4.4.2 Pattern identification Observations from OAN stations on and around Whistler Mountain were analyzed for the characteristic twice-daily reversal of wind direction. Initially, fair weather summertime periods were selected for review. The topography of Whistler Valley is such that the along-valley direction is not exactly perpendicular to the slopes of Whistler Mountain. This is because the valley orientation changes at Whistler from approximately northeast\u00E2\u0080\u0093southwest to north\u00E2\u0080\u0093south (see Figure 2.3). There- fore, determination of a valley wind signature can be difficult, but at the locations of the mountainside OAN stations, it is expected that an up-valley wind is from the southwest and down-valley flow is from the northeast. Slope winds should be 81 4.4. Slope and valley wind systems normal to the slope orientation, which, in the case of OAN stations on Whistler, is west to northwest. Thus, both upslope and up-valley winds should have a westerly component, and downslope/down-valley winds should have an easterly component. Numerous examples of twice-daily wind reversals were found in the OAN data. Winds from the summit were not analyzed due to instrument problems; also, right at the summit any wind direction could be interpreted as an upslope flow. There- fore, this analysis relies on stations from approximately the lower two-thirds of the mountain. An example of a period with regular diurnal wind shifts is given in Figure 4.6. 01/00 02/00 03/00 04/00 05/00 06/00 07/00 08/00 \u00E2\u0088\u00922 \u00E2\u0088\u00921.5 \u00E2\u0088\u00921 \u00E2\u0088\u00920.5 0 0.5 1 1.5 2 2.5 Day/Hour (LST) u \u00E2\u0088\u0092 w in d (m s\u00E2\u0088\u0092 1 ) VOL 1\u00E2\u0088\u00927 Aug 2009 Figure 4.6: Time series of hourly u-wind components from the mid-mountain station VOL for the first week of August 2009. The u-wind component is shown in Figure 4.6 to isolate the east-west portion of the winds. Positive u-wind values indicate westerly components and negative values indicate easterly components. The data plotted are from the first week of August 2009, which was generally sunny and warm with daytime convective clouds observed at Whistler. The data show a clear diurnal trend, with the westerly component prevalent during the daytime, shifting to an easterly flow overnight. The wind 82 4.4. Slope and valley wind systems shifts typically occurred between 0800 and 1000 LST (easterly to westerly) and then between 2000 and 2200 LST (back to easterly). Variation in the actual wind directions from hour to hour makes it difficult to distinguish slope winds from valley wind regimes. It is expected that the diurnal winds observed at this mid-mountain elevation are a combination of both slope and valley winds, with some influence from the synoptic scale pressure gradients as well. 4.4.3 Diurnal wind case study A more detailed wind analysis was performed on data from the first four days of August 2009. The 10-minute average winds at the top of each hour were obtained from five of the OAN stations: VOT, VOL and VOH on the slopes of Whistler Mountain; VOI on the lower slope of Blackcomb Mountain; and VOC in the valley. Whistler Blackcomb\u00E2\u0080\u0099s Alpine Shop (ALP) wind observations from near the treeline at 1836 m were also included to represent the area immediately below the peak. Figures 4.7 and 4.8 show a typical example of the diurnal cycle found in the wind data. In these figures, wind barbs are plotted on a terrain map to show wind speed and direction from four different times on August 2nd and 3rd. The wind barbs are displayed in knots, which are the conventional units for synoptic weather maps (1 kt = 0.5144ms\u00E2\u0088\u00921). For display purposes, wind speeds greater than 0.5 kt were rounded up to the next 5 kt category; otherwise too many of the observations would appear as calm. Wind barbs representing the 1900 LST observations on August 2nd (Figure 4.7a) show that all stations had up-valley and/or upslope wind directions. A similar flow pattern had been present throughout the afternoon. Two hours later at 2100 LST (Figure 4.7b), winds on the lower-mountain slopes (VOT, VOL and VOI) had turned to downslope directions, and VOC had shifted nearly 180\u00E2\u0097\u00A6 to a down-valley direction. Meanwhile, the upper mountain stations VOH and ALP, which receive sunshine later in the evening than the lower slopes, continued to show up-valley flow. 83 4.4. Slope and valley wind systems (a) 2 August 1900 LST (b) 2 August 2100 LST Figure 4.7: Wind observations from the evening of August 2, 2009. 84 4.4. Slope and valley wind systems (a) 3 August 0200 LST (b) 3 August 1200 LST Figure 4.8: Wind observations from selected hours of August 3, 2009. 85 4.4. Slope and valley wind systems By 0200 LST on the 3rd (Figure 4.8a), fully developed drainage flow was present, with downslope winds observed at all mountainside stations. Near the valley bottom at VOC, the wind was calm, presumably due to the decoupling effect of a nocturnal temperature inversion. Another transition period occurred in the morning hours following sunrise, and by noon (Figure 4.8b) all six stations were again showing up- valley and/or upslope wind directions. This example illustrates the twice-daily wind shifts observed on a typical fair weather summer day at Whistler and confirms that the timing of diurnal wind direction changes can vary amongst different elevations on the mountain. Furthermore, the diurnal changes observed at the Alpine Shop indicate that valley wind flows do, at times, reach the upper mountain of Whistler. 4.4.4 Daily classification After identifying the pattern of diurnal wind flows at Whistler, the next step in the analysis was to evaluate how often during the one-year study period the pattern could be detected. For this purpose, diurnal wind criteria were defined for two of the mountainside OAN stations: VOL at mid-mountain (1320 m) and VOH on the upper mountain (1643 m). The criteria were conceived following Kleissl et al. (2007), but with adjustments for the local patterns at Whistler. For VOL, a day was considered to be a diurnal wind day if at least five out of six hours between 1300 and 1800 LST (inclusive) had wind directions between 210 and 350\u00E2\u0097\u00A6 and at least five out of six hours between 0400 \u00E2\u0080\u0093 0600 and 0000 \u00E2\u0080\u0093 0200 LST (next day) had wind directions between 30 and 170\u00E2\u0097\u00A6. For VOH, the same criterion was used for the daytime winds, but it was found that the night-time downslope flows were often not as well defined as at VOL. Thus, the requirements were adjusted to include any wind directions with an easterly component (0 \u00E2\u0080\u0093 180\u00E2\u0097\u00A6) for at least four of the six designated overnight hours. Additionally, it was found that on some warm-season days the upslope flow on the upper mountain was delayed until quite late in the afternoon. This finding led to a manual inspection of VOH data for days that met 86 4.4. Slope and valley wind systems the night-time criterion but did not have enough hours of upslope flow between 1300 and 1800 LST to meet the daytime criterion. From these cases, days with at least four hours of upslope flow during any afternoon/evening five-hour period were also accepted. This resulted in acceptance of seven additional days, all in the May \u00E2\u0080\u0093August timeframe. The directional ranges were not centred exactly on west (270\u00E2\u0097\u00A6) and east (90\u00E2\u0097\u00A6) because, for both VOL and VOH, the local terrain contours appear to sometimes de- flect the upslope (downslope) flow to a more northwesterly (southeasterly) direction. The downslope hours considered in the criteria were separated into two periods to ensure that the wind reversal occurred twice-daily, which fits the conceptual model of diurnal winds and likely excludes changes that were due solely to synoptic scale wind shifts. Results from the wind analysis are shown in Figure 4.9. A greater number of days qualified as exhibiting the diurnal pattern at VOL than at VOH, indicating that on some days the slope and/or valley winds did not reach the upper mountain. The days for which both stations met the criteria are considered to have had well- developed diurnal wind flows. The monthly breakdown in Figure 4.9 shows that July had the most such days (13), followed by August (10 days), coinciding with the warmest months of the year. However, the fact that May had more days with well- developed diurnal winds than June suggests that synoptic scale variability in the weather conditions plays an important role. Development of diurnal wind systems will be curtailed or eliminated if the synoptic scale flow is too strong and/or the solar insolation is too weak. No days during December or January exhibited well-defined diurnal winds. Re- sults are unavailable for November due to considerable missing data from VOL. Similarly, April results are incomplete because wind data were missing from VOH for the first half of the month. As expected, diurnal wind cycles are mostly a warm-season phenomenon, with relatively few days meeting the criteria between 87 4.4. Slope and valley wind systems Jan Feb Mar Apr* May Jun Jul Aug Sep Oct Nov* Dec0 5 10 15 20 Month N um be r o f d ay s Diurnal Wind Days VOL VOL and VOH Figure 4.9: Number of days per month with well-defined diurnal wind patterns for the period December 2008 \u00E2\u0080\u0093November 2009. \u00E2\u0088\u0097Denotes months with insufficient data. September and April. However, the finding that two days in February exhibited diurnal wind cycles is further evidence that thermally driven mixing processes can occur at Whistler in the winter months, given the right conditions. Overall, of the 318 days included in the analysis, 46 days (14.5%) met the criteria for diurnal winds, with up to 13 days (42%) in given month (July) qualifying. Note that the criteria were meant to find days with well-defined slope and/or valley flows. Thus, some of the excluded days had weaker diurnal patterns, which usually meant the prescribed flows were of shorter duration than the criteria required. 4.4.5 Diurnal winds and CN data Although the seasonal results for diurnal winds and diurnal CN cycles are qualita- tively similar, comparison of Figures 3.15 and 4.9 readily reveals that the relation between the two phenomena is at best limited. For the datasets analyzed, the pres- ence of diurnal winds does not significantly increase the probability that a day will qualify as having a typical diurnal CN cycle. There are several possible reasons for 88 4.4. Slope and valley wind systems this weak correlation in the results. First, there are likely days when slope winds and other mixing processes reach the height of VOH, but not the peak. This is a limitation of the observing network; installation of a 3-D anemometer at the peak would greatly enhance the network by providing vertical velocity readings. Second, there are likely other days when the synoptic scale gradient is strong enough to sup- press one of the diurnal wind shifts, yet ample solar insolation and instability still lead to deep CBL development. Such days would have diurnal CN cycles without well-defined slope or valley wind cycles. Lastly, the timing constraints imposed by the criteria for diurnal winds and diurnal CN cycles were designed to capture the best-defined cases of each, which inevitably leaves out days with weaker expressions of the phenomena. In order to ascertain if the presence of diurnal winds has any effect on the aerosol data, mean daily CN concentrations for each month were grouped by days with and without diurnal wind signatures. The results are given in Figure 4.10. For each of the months that had diurnal wind cycles, the mean CN concentration was higher for the diurnal wind days than for days without diurnal winds. Although the monthly difference in means was statistically significant only for May (based on right-tailed t-tests with \u00CE\u00B1 = 0.05), the consistency of the results suggests that diurnal wind systems act to enhance aerosol concentration at the summit. It is expected that this is because slope and valley winds can bring PBL air to the mountaintop by direct transport and/or by facilitating vertical mixing and CBL growth in the immediate area. Like atmospheric instability, slope and valley winds appear to be contributing factors in the upward influence of PBL air and consequent afternoon maxima in CN concentration on Whistler Mountain. Yet, neither stability nor diurnal winds correlate strongly to results from the CN diurnal cycle analysis. This finding reflects the circumstance that diurnal CBL growth in the mountains is a complex 3-D process that cannot be easily characterized by parameters derived from a limited observing 89 4.5. Synoptic classification of diurnal cycle days Feb Mar Apr May Jun Jul Aug Sep Oct0 500 1000 1500 2000 Month CN C on ce nt ra tio n (cm \u00E2\u0088\u0092 3 ) Diurnal Wind Days and CN \"No\" days \"Yes\" days Figure 4.10: Comparison of average CN concentrations for days with (\u00E2\u0080\u009CYes\u00E2\u0080\u009D) and without (\u00E2\u0080\u009CNo\u00E2\u0080\u009D) well-defined diurnal wind patterns. network. While a site such as Mauna Loa may have very regular slope winds that correlate well to observed cycles in air chemistry datasets, the mid-latitude coastal setting of Whistler means that synoptic scale influences are important to processes at the smaller scales\u00E2\u0080\u0094and the synoptic scale pattern changes frequently. With this in mind, attention is now focused again on the synoptic scale in an attempt to define what kind of day is favourable for vertical transport of PBL to the summit of Whistler. 4.5 Synoptic classification of diurnal cycle days 4.5.1 Theory and definitions Although diurnal CBL growth is a consequence of local and mesoscale mechanisms, the process can be difficult to assess or predict based on limited local observations. However, the synoptic scale conditions known to be favourable for vigorous CBL development\u00E2\u0080\u0094strong solar insolation coupled with weak regional pressure gradi- 90 4.5. Synoptic classification of diurnal cycle days ents\u00E2\u0080\u0094are simpler to diagnose or forecast. If the diurnal cycles found in measure- ments on the summit of Whistler are indeed primarily caused by convective uplift of PBL air to the mountaintops, then days with well-defined diurnal cycles should exhibit these synoptic weather conditions. The fact that mean diurnal cycles in CN concentration were much less pronounced in winter than in summer supports the idea that convective uplift plays an important role in transport of lower-elevation constituents to the site, because the limited solar heating of winter makes the en- vironment less supportive of deep convection. Dynamic lifting mechanisms such as frontal lift may also be important at times, but would need to be assessed separately, as these mechanisms do not operate on any regular diurnal schedule. From the above, it appears as though identification of conditions conducive to PBL influence at the mountaintop may be amenable to a synoptic classification approach. The field of synoptic climatology involves finding relations between the atmospheric circulation and the surface environment of a region (Yarnal, 1993). Synoptic climatology has the advantage of being able to synthesize complex interac- tions into recognizable patterns; it can provide a depiction of the pattern associated with a particular environmental effect without having to render the details of all the processes involved. In this section, days with diurnal cycles in CN concentration are examined to determine if a common set of synoptic conditions describes them. This is known as the environment-to-circulation approach to synoptic classification: the environmental effect (diurnal CN cycles) determines which portions of the synoptic dataset are included for analysis (Yarnal, 1993). 4.5.2 Synoptic weather typing from surface observations For a first look at what meteorological conditions might be associated with CBL growth in the Whistler area, a simplified version of synoptic climatology was em- ployed. Synoptic weather typing, which is the classification of weather conditions or patterns into categories (Sheridan, 2002), was performed based solely on the weather 91 4.5. Synoptic classification of diurnal cycle days conditions observed from the surface. This is a departure from the more traditional synoptic analysis procedures, which are based on regional circulation patterns, i.e. synoptic weather maps. In this case, observations from the Whistler Valley site at Nesters were examined for each day of the study period. Hourly surface observa- tions in the METAR format are reported from the site daily between 0600 and 1800 local time by a human observer. A manual classification scheme was created to categorize each day based on the reported conditions. Days that were mainly sunny were divided into convective (type 1) and non-convective (type 2) categories based on the observer\u00E2\u0080\u0099s cloud type codes in the remarks section of the METARs. Days that were primarily cloudy, without precipitation, were designated as type 3. The type 4 category included cloudy days with at least four hours of precipitation, and type 5 was the \u00E2\u0080\u009Cmixed\u00E2\u0080\u009D category for days with periods of both precipitation and sunshine. For the entire study period (344 days after accounting for missing CN data), 57% of the days were \u00E2\u0080\u009Csunny\u00E2\u0080\u009D (type 1 or 2), 24% were precipitation days (type 4), 10% were \u00E2\u0080\u009Ccloudy\u00E2\u0080\u009D (type 3) and 9% were \u00E2\u0080\u009Cmixed\u00E2\u0080\u009D (type 5). For days with a typical diurnal cycle in CN concentration (127 days, see Section 3.4), 72% were sunny and only 8% were precipitation days. Including the mixed days, 83% of the typical cycle days were sunny or partly sunny (types 1, 2 and 5). These results support the concept of PBL influence arising primarily from convective transport on days with strong solar insolation. When the subset of typical cycle days also having strong correlations between water vapour and CN concentration was considered (59 days, see Section 3.5), 86% of the days were sunny and 92% were sunny or partly sunny. Only one day within this subset was categorized as a precipitation day (type 4). The presence of convective clouds on the sunny days (type 1) increased the likelihood of diurnal CN patterns, but a substantial portion (36%) of the 127 diurnal cycle days coincided with \u00E2\u0080\u009Cnon-convective\u00E2\u0080\u009D sunny days (type 2). Indeed, generation of thermals and CBL growth can occur on clear days with little or no visible evidence 92 4.5. Synoptic classification of diurnal cycle days of the convective activity; the term \u00E2\u0080\u009Cnon-convective\u00E2\u0080\u009D here only refers to the lack of convective clouds reported in the observations. A more robust weather typing scheme based on surface observations was devel- oped by Sheridan (2002). This scheme, known as the Spatial Synoptic Classification (SSC), is considered a hybrid classification scheme, with both manual and automated elements. For each location, \u00E2\u0080\u009Cseed days\u00E2\u0080\u009D were identified to represent typical cases of each weather type, and then each day\u00E2\u0080\u0099s actual conditions were classified accord- ing to which seed day they most closely resembled (Sheridan, 2002). Sheridan\u00E2\u0080\u0099s SSC consists of six types\u00E2\u0080\u0094several of which are analogous to traditional air mass types\u00E2\u0080\u0094plus a seventh \u00E2\u0080\u009Ctransitional\u00E2\u0080\u009D category for days when conditions change from one weather type to another. Use of the SSC in this study is limited by the fact that, due to insufficient historical data, the scheme has not been run for Whistler. However, SSC types are available for Vancouver (YVR airport). During the one-year study period, the dominant SSC types for Vancouver were \u00E2\u0080\u009Cdry moderate\u00E2\u0080\u009D (46% of days with complete data), \u00E2\u0080\u009Cmoist moderate\u00E2\u0080\u009D (14%), \u00E2\u0080\u009Cmoist polar\u00E2\u0080\u009D (22%) and \u00E2\u0080\u009Ctransitional\u00E2\u0080\u009D (12%). No days were classified as \u00E2\u0080\u009Cdry tropical\u00E2\u0080\u009D and only four days were considered \u00E2\u0080\u009Cmoist tropical,\u00E2\u0080\u009D as these categories are more typical of conditions found at low latitudes. Despite the issue of coastal Vancouver weather not always being representative of conditions at Whistler, analysis of SSC designations for days with diurnal CN cycles yielded results qualitatively similar to those of the above analysis based on Whistler observations. For days with typical CN cycles and strong correlations between CN and water vapour (which, for simplicity, are referred to here as \u00E2\u0080\u009CPBL- influence\u00E2\u0080\u009D days), 67% fell into the \u00E2\u0080\u009Cdry moderate\u00E2\u0080\u009D category and another 10% were \u00E2\u0080\u009Cdry polar.\u00E2\u0080\u009D Thus, for PBL-influence days, a considerably higher percentage fell into the \u00E2\u0080\u009Cdry\u00E2\u0080\u009D categories than did all days in the study period (77% vs. 52%). Likewise, the percentage of \u00E2\u0080\u009Cmoist\u00E2\u0080\u009D days was lower for the PBL-influence subset than for the whole study period (13% vs. 37%). This is analogous to saying that the 93 4.5. Synoptic classification of diurnal cycle days PBL-influence days were dominated by sunny conditions, particularly the relatively warm days of the \u00E2\u0080\u009Cdry moderate\u00E2\u0080\u009D type. It is interesting to note that two of the four PBL-influence days in February were of the \u00E2\u0080\u009Cdry moderate\u00E2\u0080\u009D type, even though this type was generally much more common in summer than in winter. Sheridan\u00E2\u0080\u0099s SSC takes into consideration seasonal changes, so that the \u00E2\u0080\u009Cdry moderate\u00E2\u0080\u009D days are relatively mild for the time of year compared to the \u00E2\u0080\u009Cdry polar\u00E2\u0080\u009D days. One of the \u00E2\u0080\u009Cdry moderate\u00E2\u0080\u009D days in February (the 20th) was also one of two days that month with well-developed diurnal winds. This case provides a good example of a day during the winter season when the synoptic scale weather conditions provided an environment favourable for considerable vertical mixing in the Whistler area. Based on the above analysis, a generalization can be made that days with in- dicators of PBL influence on Whistler Mountain were more often associated with sunny, mild conditions than any other weather situation. However, not all sunny days at Whistler during the study period had well-defined diurnal cycles in the vari- ables. In order to learn more about the synoptic scale conditions conducive to PBL influence on the mountaintop, the following analysis uses a common environment- to-circulation classification approach to incorporate data that describe regional cir- culation patterns. 4.5.3 Synoptic classification via composite maps For days that had typical diurnal CN cycles, 500 hPa and sea level pressure (SLP) NCEP reanalysis grid point values were obtained and plotted on regional maps. As with the 700 hPa wind analysis in Section 4.2, daily averages were based on the four synoptic times available for each LST day. The analysis presented in this section is limited to the 59 PBL-influence days when diurnal CN cycles were coincident with strong correlations between water vapour and CN concentration, thereby focusing on a manageable number of days that best fit the typical pattern of diurnal PBL influence. 94 4.5. Synoptic classification of diurnal cycle days A day-by-day review of the 59 PBL-influence cases revealed considerable variety in the synoptic weather patterns. Thus, it would not be appropriate to produce composite (average) maps that included data from all 59 days, as that would be a case of averaging unlike circulation patterns, which does not yield meaningful results (Yarnal, 1993). Rather, the days were first subjectively categorized into six circulation pattern types (A \u00E2\u0080\u0093F) and then composite maps were produced by type. Most of the PBL-influence days (51 out of 59 cases) fit the first three pattern types, which are depicted and described below. Pattern A is characterized by a high pressure ridge aloft (500 hPa) and a weak surface (SLP) gradient in the Whistler region (see Figure 4.11). Pattern A was by far the most common synoptic pattern identified in the subset, representing 32 of the 59 days. Several of the summertime pattern-A days featured weak surface thermal troughs along the BC coast. It is interesting to note that the ridge aloft in most pattern-A cases was not particularly strong; this pattern type appears to capture days with enough ridging aloft to support clear or partly cloudy skies, but not such a strong upper high that CBL growth was severely restricted by dynamic subsidence. At the surface, weak pressure gradients allowed for mesoscale transport processes such as convection and diurnal wind development. Pattern B is essentially a wintertime version of pattern A, but is defined more by the surface pattern than the 500 hPa circulation. On these days, low pressure to the south coupled with high pressure in the BC interior created moderate offshore pressure gradients through the Coast Mountains (see Figure 4.12b). This low-level offshore flow and the accompanying northerly flow aloft produced clear to partly cloudy skies in Whistler on the six winter days that fit the pattern. The scenario defined as pattern C features either a trough or zonal flow aloft, rather than a ridge (Figure 4.13a). But similar to pattern A, the surface pressure gradient was weak on pattern C days, usually with a weak high pressure ridge over the region (Figure 4.13b). Review of surface observations and additional synoptic 95 4.5. Synoptic classification of diurnal cycle days 55 50 5550 5580 5610 5640 5670 5700 5730 5760 140 \u00C2\u00B0 W 135 \u00C2\u00B0 W 130 \u00C2\u00B0 W 125 \u00C2\u00B0 W 120 \u00C2\u00B0 W 115 \u00C2\u00B0 W 110 \u00C2\u00B0 W 40 \u00C2\u00B0 N 45 \u00C2\u00B0 N 50 \u00C2\u00B0 N 55 \u00C2\u00B0 N 60 \u00C2\u00B0 N (a) 500 hPa Pattern A 10 16 1016 1016 102 0 140 \u00C2\u00B0 W 135 \u00C2\u00B0 W 130 \u00C2\u00B0 W 125 \u00C2\u00B0 W 120 \u00C2\u00B0 W 115 \u00C2\u00B0 W 110 \u00C2\u00B0 W 40 \u00C2\u00B0 N 45 \u00C2\u00B0 N 50 \u00C2\u00B0 N 55 \u00C2\u00B0 N 60 \u00C2\u00B0 N (b) SLP Pattern A Figure 4.11: Composite maps of PBL-influence days represented by synoptic pattern A. The triangle symbol marks the location of Whistler Mountain. 5250 5280 5310 5340 5370 5400 5430 5460 5490 5490 5520 5520 5550 5580 5610 5640 140 \u00C2\u00B0 W 135 \u00C2\u00B0 W 130 \u00C2\u00B0 W 125 \u00C2\u00B0 W 120 \u00C2\u00B0 W 115 \u00C2\u00B0 W 110 \u00C2\u00B0 W 40 \u00C2\u00B0 N 45 \u00C2\u00B0 N 50 \u00C2\u00B0 N 55 \u00C2\u00B0 N 60 \u00C2\u00B0 N (a) 500 hPa Pattern B 10 16 1020 10 24 1024 1024 1028 1028 1032 1032 1036 1036 140 \u00C2\u00B0 W 135 \u00C2\u00B0 W 130 \u00C2\u00B0 W 125 \u00C2\u00B0 W 120 \u00C2\u00B0 W 115 \u00C2\u00B0 W 110 \u00C2\u00B0 W 40 \u00C2\u00B0 N 45 \u00C2\u00B0 N 50 \u00C2\u00B0 N 55 \u00C2\u00B0 N 60 \u00C2\u00B0 N (b) SLP Pattern B Figure 4.12: Composite maps of PBL-influence days represented by synoptic pattern B. 96 4.5. Synoptic classification of diurnal cycle days maps for the relevant days found that, on most of the pattern C days, the Whistler area was in a convective post-frontal environment. There were 13 days that fit pattern C, all in the spring and summer months. On these days, a mix of sunshine and cloud was observed at Whistler. On nearly all of the pattern C days, convective clouds were reported by the observer, with towering cumulus (TCU) reported on six of the days. Surface heating may not have been as intense on these days as for the pattern A days, but the destabilizing effect of cold air aloft enhanced convective mixing, likely resulting in a deep CBL. 554 0 5540 5570 5600 5630 5660 5690 5720 5750 5780 140 \u00C2\u00B0 W 135 \u00C2\u00B0 W 130 \u00C2\u00B0 W 125 \u00C2\u00B0 W 120 \u00C2\u00B0 W 115 \u00C2\u00B0 W 110 \u00C2\u00B0 W 40 \u00C2\u00B0 N 45 \u00C2\u00B0 N 50 \u00C2\u00B0 N 55 \u00C2\u00B0 N 60 \u00C2\u00B0 N (a) 500 hPa Pattern C 1012 1016 1016 101 6 1020 140 \u00C2\u00B0 W 135 \u00C2\u00B0 W 130 \u00C2\u00B0 W 125 \u00C2\u00B0 W 120 \u00C2\u00B0 W 115 \u00C2\u00B0 W 110 \u00C2\u00B0 W 40 \u00C2\u00B0 N 45 \u00C2\u00B0 N 50 \u00C2\u00B0 N 55 \u00C2\u00B0 N 60 \u00C2\u00B0 N (b) SLP Pattern C Figure 4.13: Composite maps of PBL-influence days represented by synoptic pattern C. The remaining eight days not conforming to any of the patterns A \u00E2\u0080\u0093C can be described by three additional scenarios. Three of these days featured an upper low to the south of BC, which gave the Whistler area dry easterly flow (pattern D). Another three days were considered to be in a pre-frontal environment with offshore SLP gradients and considerable sunshine (pattern E). Finally, two PBL-influence days (17 March and 27 June) appear to have had frontal systems passing through Whistler (pattern F). This last category is of interest because frontal passages are common at Whistler, particularly in the wet fall \u00E2\u0080\u0093 spring months, but they appear to account for very few of the PBL-influence days. It is expected that, on most such 97 4.6. Conclusions days, low solar insolation and strong pressure gradients limit convective mixing to a shallow layer in the valleys. However, dynamic lift associated with fronts is another mechanism which may force PBL air upwards. Because this analysis uses aerosol concentration as the primary means of identifying days with possible PBL influence at the site, cases of dynamic lift bringing valley air to the summit may elude detection due to the washout of PM by precipitation. A more thorough investigation of dynamic lift effects would require use of a gaseous boundary layer tracer that is not significantly reduced by clouds or precipitation. Overall, the composite maps corroborate the understanding of PBL-influence days derived from the preceding analysis of surface observations. Days with well- defined diurnal cycles in CN concentration tend to be those with abundant solar insolation and, with the exception of the pattern-B cases, weak surface pressure gradients. The synoptic circulation analysis has shown that there are numerous ways in which the atmosphere can produce such a day at Whistler. The two most common scenarios are represented by patterns A and C. Pattern A can be consid- ered a classic fair weather scenario for the Coast Mountains, dominated by a high pressure ridge aloft. Pattern C is more representative of a break in a synoptically active situation: following the passage of a front or trough, an unstable atmosphere promotes convective mixing, which is often enhanced by orographic lift in the moun- tains. This analysis has found that most cases of diurnal CN cycles at Whistler can be explained by coincident synoptic weather patterns that support the mesoscale processes associated with vertical mixing. 4.6 Conclusions In this chapter, meteorological analysis on different scales has been employed to attain insight about the driving forces of PBL influence on Whistler Mountain. Mesoscale considerations of atmospheric stability and diurnal wind systems showed 98 4.6. Conclusions seasonal variations that were broadly in accordance with the observed seasonality of PBL indicators, i.e. unstable conditions and diurnal wind cycles were much more common in the warm season than in winter. But, like the PBL indicators examined in Chapter 3, these conditions were not entirely absent during the winter. In partic- ular, identification of well-defined diurnal wind patterns in February demonstrates that thermally driven processes can occur at Whistler in the winter season. This lends support to the idea that some days in winter are subject to convective mixing and subsequent CBL growth to a height that encompasses the mountaintop. While other mid-latitude mountain sites have been assumed to be decoupled from the PBL throughout the winter season (e.g. Lugauer et al., 1998), the relatively low altitude of Whistler and its susceptibility to mild air masses can support considerable verti- cal mixing in winter, particularly late in the season when increasing daylight hours enhance the daily solar insolation. A static stability parameter based on potential temperature differences between the valley and the peak of Whistler was found to be a poor predictor of diurnal CN cycles on the peak. While this may suggest that stability is not of great importance to vertical transport processes, it appears more likely that a stability parameter based on slope-wise lapse rates does not accurately represent the convective po- tential of the adjacent free air. Indeed, examples could be found of days with high stability parameter values during which cumulus and TCU clouds were observed over Whistler. Lacking routine upper air observations, evaluating stability via an obser- vational network is challenging. Stability parameters based on mesoscale numerical models may provide better information than surface-based observations regarding the potential for diurnal CBL growth. Patterns of slope and valley wind systems were found in data from weather stations on and near Whistler Mountain. Well-defined diurnal winds reaching the upper portions of the mountain were common during the warmest summer months and rare during the cool season. Although the days with well-defined diurnal winds 99 4.6. Conclusions did not correlate well with days having typical diurnal cycles in CN concentration, it was found that average CN concentrations were enhanced for days with diurnal wind signatures. Thus, it can be assumed that slope and valley winds either directly increase CN concentrations on the peak via vertical transport, or that days with diurnal wind cycles coincide with conditions that support general CBL growth. The location and climate of Whistler means that frequently changing synoptic scale weather conditions play an important role in determining the sources of air parcels sampled from the mountaintop. The synoptic scale pattern directly affects pollutant loads via horizontal transport and dispersion processes. Equally important are the indirect effects of creating an environment either favourable or restrictive to- wards mesoscale vertical transport processes. Using the CN diurnal cycle signature as the primary indicator of PBL influence on the mountaintop, a synoptic classifi- cation approach was applied to datasets covering the study period. It was found that most of the \u00E2\u0080\u009Ctypical diurnal cycle\u00E2\u0080\u009D cases corresponded to days with abundant sunshine and weak synoptic scale pressure gradients. Analysis of 500 hPa and SLP circulation patterns on the diurnal cycle days led to identification of six charac- teristic situations that produced diurnal cycle days. Two of the six pattern types appeared to be most common, accounting for a majority of PBL-influence days: 1) a high pressure ridge of moderate strength aloft accompanied by a weak surface pattern, and 2) a post-frontal convective situation with a weak high at the surface and cold air aloft. These patterns are consistent with established conceptual models of CBL growth in complex terrain. Convective mixing appears to be the primary means of PBL influence at Whistler. However, identification of two PBL-influence days that likely involved frontal pas- sages at Whistler serve as a reminder that dynamic lift may also play a role in the upward transport of valley air. Forrer et al. (2000) found evidence for uplift of trace gases from the lowlands to Jungfraujoch due to both fo\u00CC\u0088hn wind events and fronts. Their conclusion that transport processes on different spatial and timescales 100 4.6. Conclusions are important to the interpretation of air chemistry measurements at Jungfraujoch also applies to Whistler. A full analysis of the role of dynamic lift is beyond the scope of this study, as it would require the use of a gaseous tracer of PBL air other than water vapour. Also, due to considerable variability in the strength and specific nature of synoptic scale disturbances, it seems that evaluation of dynamic vertical transport in the Whistler area would need to be done on a case-by-case basis. This has already been done to a limited extent for specific cases of dynamic subsidence that brought pollutants of distant origin to Whistler after trans-Pacific transport in the FT (Nseir, 2007). In the next chapter, indicators of PBL influence and possible driving forces are examined together in case studies of different vertical mixing scenarios. Additional datasets are incorporated to help determine whether diurnal cycles detected on the peak are the result of boundary layer effects very local to the mountaintop or due to a more thorough mixing of valley air into a deep CBL. 101 Chapter 5 Case studies of vertical mixing 5.1 Introduction The following case studies build upon analysis from the previous chapters, which has provided some answers to the question of how often and in what conditions the Whistler site is exposed to PBL air. The first case study focuses on a day with PBL influence on the peak that occurred in a manner assumed to be common, but at a time of year when mesoscale thermal forcing is not expected to be prevalent. This case corroborates earlier evidence for wintertime PBL influence at the site. A second case study describes a summertime day when the valley-based PBL remained below the mountaintop. Synoptic weather conditions are examined to determine which limiting factors restricted BL growth on that day. The last case study considers an important new data source in the Whistler area, which was deployed after the primary study period of this thesis. An example using three days of data from the Whistler CORALNet lidar provides a preview of the potential benefits of this ground-based remote sensing instrument to mountain BL studies. 5.2 Winter PBL-influence case: 20 February 2009 February 20th, 2009 exemplifies a day during the winter season when well-defined diurnal cycles could be found in meteorological data from Whistler Mountain. The criteria for a typical diurnal cycle in CN concentration (Section 3.4) were met and 102 5.2. Winter PBL-influence case: 20 February 2009 the CN values were well correlated with water vapour mixing ratio (see Figure 5.1). The Spearman\u00E2\u0080\u0099s rs value based on the day\u00E2\u0080\u0099s 24 hourly CN and w values was 0.66 with a corresponding p-value of 5.0x 10\u00E2\u0088\u00924. Thus, the day qualifies as one of the 59 \u00E2\u0080\u009CPBL-influence\u00E2\u0080\u009D days included in Figure 3.18, and one of four such days from February. 0 4 8 12 16 20 240 500 1000 1500 CN C on ce nt ra tio n (cm \u00E2\u0088\u0092 3 ) Hour of day (LST) CN and WV February 20 0 1 2 3 M ix in g Ra tio (g /kg ) CN w Figure 5.1: Hourly values of CN concentration and water vapour mixing ratio at the summit of Whistler on 20 February 2009. As this day coincided with the period of upper air observations at Whistler, free- air vertical profile data are available to help determine the structure of the lower atmosphere. Figure 5.2 shows profiles of temperature and potential temperature from both the morning and afternoon soundings. From the temperature profiles it is apparent that a strong surface-based inversion was in place in the morning, with a surface temperature in the valley of about \u00E2\u0088\u00927 \u00E2\u0097\u00A6C. Surface heating later destroyed the inversion, and by 1600 LST diurnal heating effects were manifest to the mountaintop level, which is marked with an \u00E2\u0080\u0098X\u00E2\u0080\u0099 in Figure 5.2a. In addition to showing the same diurnally warmed layer, the potential temperature profiles in Figure 5.2b give an indication of the atmospheric stability. The atmosphere was highly stable at 0400 LST, but at 1600 LST the air between the surface and the mountaintop level was only slightly stable, indicating a relatively well-mixed layer. The inflection point 103 5.2. Winter PBL-influence case: 20 February 2009 in the afternoon \u00CE\u00B8 profile at the approximate height of the mountaintop can be interpreted as the top of the PBL. From these profiles, it appears as though surface heating effects were distributed throughout the depth of the valley. \u00E2\u0088\u009220 \u00E2\u0088\u009210 0 10 500 600 700 800 900 1000 Temperature (\u00C2\u00B0C) Pr es su re (h Pa ) Whistler Soundings 20 Feb 0400 LST 1600 LST (a) Temperature profiles 270 280 290 300 310 320 500 600 700 800 900 1000 Potential Temperature (K) Pr es su re (h Pa ) WAE \u00CE\u00B8 Profiles 20 Feb 0400 LST 1600 LST (b) Potential temperature profiles Figure 5.2: Whistler radiosonde observations from the morning (0400 LST) and afternoon (1600 LST) soundings on 20 February 2009. The \u00E2\u0080\u0098X\u00E2\u0080\u0099 in the left panel represents the pressure and temperature values observed at the AQRD site on the peak at 1600 LST. METAR observations from the Nesters site indicated that February 20th had abundant sunshine with just scattered cirrus clouds. Winds were light and the surface temperature reached 5 \u00E2\u0097\u00A6C by the time of the afternoon balloon launch. No convective clouds were reported. The stability parameter value for this day was rather high at 8.2. Thus, it is expected that this day was not a case of the post- frontal convective situation (Pattern B) described in Section 4.5. Instead, this day best fits Pattern A, as seen in the synoptic circulation maps of Figure 5.3. The 500 hPa pattern (Figure 5.3a) shows a stronger (higher-amplitude) high pressure ridge than average for Pattern A. As Whistler was right under the upper ridge, it can be assumed that the synoptic scale vertical motion was downward, which gave the region nearly clear skies. At the surface, high pressure was centred over southeastern BC, providing a weak offshore pressure gradient to the Whistler area. This quiescent pattern on the synoptic scale allowed the daytime heating to 104 5.2. Winter PBL-influence case: 20 February 2009 produce local and mesoscale thermal gradients even though snow cover must have reduced solar absorption at the surface. Convective thermals and upslope flows mixed air upwards from the valley, re-distributing heat as evident in the profiles of Figure 5.2. Therefore, on Pattern A days the thermally driven vertical transport and the synoptic scale vertical motion are of opposite direction, which is what Lugauer (1998) found to be the most common situation for PBL influence at Jungfraujoch. 5440547 0 5470 5500 550 0 5500 5530 5530 5560 5560 5590 5590 5620 5620 5650 5680 140 \u00C2\u00B0 W 135 \u00C2\u00B0 W 130 \u00C2\u00B0 W 125 \u00C2\u00B0 W 120 \u00C2\u00B0 W 115 \u00C2\u00B0 W 110 \u00C2\u00B0 W 40 \u00C2\u00B0 N 45 \u00C2\u00B0 N 50 \u00C2\u00B0 N 55 \u00C2\u00B0 N 60 \u00C2\u00B0 N (a) 500 hPa 1008 1012 1016 1020 10 201024 10 24 1024 1024 10 24 1028 1028 140 \u00C2\u00B0 W 135 \u00C2\u00B0 W 130 \u00C2\u00B0 W 125 \u00C2\u00B0 W 120 \u00C2\u00B0 W 115 \u00C2\u00B0 W 110 \u00C2\u00B0 W 40 \u00C2\u00B0 N 45 \u00C2\u00B0 N 50 \u00C2\u00B0 N 55 \u00C2\u00B0 N 60 \u00C2\u00B0 N (b) SLP Figure 5.3: Regional synoptic circulation maps showing averaged 500 hPa and sea level pressure patterns for 20 February 2009. Evidence for the presence of thermally driven transport can be found in the wind data from the mountainside stations VOL and VOH, which fit the criteria for diurnal winds on February 20th. At mid-mountain, VOL had easterly downslope winds during the preceding night and until late morning on the 20th. At mid-day the wind was nearly calm, and then by 1300 LST a light upslope flow from the northwest had developed. Winds with a westerly component continued through the afternoon, switching to southeasterly in the evening. Winds at the higher-elevation VOH station followed a very similar pattern, but with the switch to westerly com- ponent flow occurring earlier (around 1100 LST) than at VOL. This day was one of only two during the winter that met the criteria for well-developed diurnal winds. Several other days during the February dry spell also had diurnal cycles in measured 105 5.3. Summer FT case: 13 July 2009 CN concentration without having well-defined diurnal winds. The weaker diurnal wind signatures on such days may be due to less intense surface heating and/or slightly stronger synoptic scale flows that affected the wind directions. Slope and valley winds are not necessary for CBL growth, but they are expected to expedite upward mixing of valley air. In this case, it appears as though thermally driven vertical transport was just strong enough to work against synoptic scale subsidence, bringing PBL air to the summits. The temperature profiles in Figure 5.2 indicate that the PBL top was very near the summit of Whistler Mountain at 1600 LST. This observation, along with the timing of the upslope winds, is in accordance with the somewhat sharp peak in CN concentration that occurred between 1600 and 1900 LST (see Figure 5.1). Although an afternoon peak in aerosol concentration could conceivably be caused by horizontal advection of a pollution plume or by new particle growth, the strong correlation between CN and water vapour in this case and the preceding evidence for diabatic processes support the notion that convective uplift resulted in intrusions of PBL air at the mountaintop site during the afternoon. The measurement site was then representative of a mixture of PBL and FT air until several hours after mesoscale uplift had ceased, allowing FT air from upstream to displace the PBL constituents via horizontal (synoptic scale) advection. 5.3 Summer FT case: 13 July 2009 Examples of PBL influence during the warm season are numerous from the study period, with most such days having patterns in the data that are qualitatively similar to those of the preceding winter case. Therefore, it is more instructive at this point to examine a summer day when it appears as though the PBL remained below the summit of Whistler. On July 13th no rainfall was reported at Whistler, but CN concentrations on the peak generally declined through the day (Figure 5.4). Water vapour mixing ratio 106 5.3. Summer FT case: 13 July 2009 was unsteady through the morning, with some increase between 0800 and 1200 LST, followed by decreasing values through the afternoon and evening. Neither of these variables showed a typical afternoon or evening peak as expected for a PBL-influence day, nor were they well correlated to each other. 0 4 8 12 16 20 240 500 1000 1500 2000 2500 CN C on ce nt ra tio n (cm \u00E2\u0088\u0092 3 ) Hour of day (LST) CN and WV July 13 6.0 6.5 7.0 7.5 8.0 8.5 M ix in g Ra tio (g /kg ) CN w Figure 5.4: Hourly values of CN concentration and water vapour mixing ratio at the summit of Whistler on 13 July 2009. Figure 5.5 shows the day\u00E2\u0080\u0099s temperature and pressure trends. There appears to be a slight diurnal cycle in the temperature trend, but with the daily high coming earlier than usual for the time of year (1300 LST). The temperature then fell through the afternoon and evening, ending the day about 5 \u00E2\u0097\u00A6C cooler than it began. Meanwhile, the atmospheric pressure gradually increased by 3 hPa through the day. These changes in the meteorological variables in the afternoon hours suggest that an air mass change was taking place. However, rather than intrusion of humid, aerosol- rich air from the lowlands, this case appears more characteristic of a synoptic scale change to cooler, drier air. The synoptic scale pattern is summarized by the maps in Figure 5.6. At 500 hPa, a trough had recently passed through southwestern BC and a high pressure ridge was approaching the area from the eastern Pacific. Inspection of 500 hPa maps from the days before and after July 13th confirmed that this was a progressive rather 107 5.3. Summer FT case: 13 July 2009 0 4 8 12 16 20 24 2 4 6 8 10 12 Te m pe ra tu re (\u00C2\u00B0C ) Hour of Day (LST) Temperature and Pressure July 13 784 785 786 787 Pr es su re (h Pa ) T P Figure 5.5: Hourly values of temperature and pressure at the summit of Whistler on 13 July 2009. than a stagnant pattern. Therefore, the rising pressure observed at the Whistler observatory can be attributed to the building high pressure ridge. The stability parameter value for this day was 7.4; only one day in July had a higher value. It can be expected that dynamic subsidence was present in this upper level pattern, but in this case Whistler did not have clear skies. Broken to overcast cloud cover was reported throughout the day with ceilings of 1500 to 3000 feet. 5620 5650 5680 5710 5740 5770 5770 5800 5830 5860 140 \u00C2\u00B0 W 135 \u00C2\u00B0 W 130 \u00C2\u00B0 W 125 \u00C2\u00B0 W 120 \u00C2\u00B0 W 115 \u00C2\u00B0 W 110 \u00C2\u00B0 W 40 \u00C2\u00B0 N 45 \u00C2\u00B0 N 50 \u00C2\u00B0 N 55 \u00C2\u00B0 N 60 \u00C2\u00B0 N (a) 500 hPa 10 12 10161020 102 0 10 24 140 \u00C2\u00B0 W 135 \u00C2\u00B0 W 130 \u00C2\u00B0 W 125 \u00C2\u00B0 W 120 \u00C2\u00B0 W 115 \u00C2\u00B0 W 110 \u00C2\u00B0 W 40 \u00C2\u00B0 N 45 \u00C2\u00B0 N 50 \u00C2\u00B0 N 55 \u00C2\u00B0 N 60 \u00C2\u00B0 N (b) SLP Figure 5.6: Regional synoptic circulation maps showing averaged 500 hPa and sea level pressure patterns for 13 July 2009. 108 5.3. Summer FT case: 13 July 2009 Regional surface observations and the sea level pressure pattern (Figure 5.6b) for the day in question provide an explanation for the cloud cover at Whistler. A moderate onshore pressure gradient was present across coastal BC. Brisk southerly winds (10 \u00E2\u0080\u0093 20 kt at Squamish and 8 \u00E2\u0080\u0093 12 kt at Whistler) brought moist marine air up Howe Sound and the Cheakamus Valley. This low-level moisture advection combined with orographic lift to produce a stratocumulus deck of clouds in the region. Although radiosonde observations are not available from Whistler for this case, the nearest coastal upper air sites provide a representative picture of the vertical temperature and moisture structure responsible for the observed conditions. The afternoon observations from Quillayute (Figure 5.7) show a clearly defined marine BL capped by a subsidence inversion at about 880 hPa, or approximately 1400 m asl. Saturated conditions near the top of the BL indicate cloud layers, whereas the FT air above was much drier. Sounding data from Port Hardy showed a very similar scenario with a slightly lower BL top, indicating that the moist marine BL was found throughout a broad region on this day. Cloud bases at Whistler were observed to be 3000 feet (900 m) above ground level (agl) at 1600 LST. This corresponds to a height of 1560 m asl, slightly higher than the BL height at Quillayute, but well below the summit of the mountain. The pressure level at the summit is marked on Figure 5.7, showing that the site was well above the marine BL. Additionally, the mountaintop temperature corresponded well to the FT temperature profile from Quillayute. Winds above the BL at Quillayute showed backing with height, which corre- sponds to cold air advection. Thus, the cooling, drying and rising pressure observed at the Whistler site on this day can all be explained by the synoptic scale changes that were occurring in the FT. The strong subsidence inversion capping the PBL effectively kept the air at the mountaintop level decoupled from the valley air. Cloud cover limited surface heating such that convection was not vigorous enough to force the PBL top to the height of the peak. Therefore, for this day, the AQRD site can 109 5.4. Evaluation of BL structure with lidar: 6 \u00E2\u0080\u0093 8 July 2010 \u00E2\u0088\u009240 \u00E2\u0088\u009230 \u00E2\u0088\u009220 \u00E2\u0088\u009210 0 10 20 500 600 700 800 900 1000 Temperature (\u00C2\u00B0C) Pr es su re (h Pa ) UIL 13 July 1600 LST T T d Whistler Figure 5.7: Vertical profiles of temperature and dew point temperature from the Quillayute, WA radiosonde observations of 1600 LST on 13 July 2009. The + symbol represents the pressure and temperature values observed at the AQRD site on the peak at the same time. be considered representative of the FT. These inland intrusions of marine air are an occasional feature of the summertime weather at Whistler, and may represent an easily-recognizable pattern that maintains FT conditions at the measurement site. However, it is best to use the available datasets to estimate the PBL height on a case-by-case basis. 5.4 Evaluation of BL structure with lidar: 6 \u00E2\u0080\u0093 8 July 2010 Lidar backscatter datasets have been used for boundary layer studies in a variety of environments (e.g. Endlich et al., 1979; Wakimoto and McElroy, 1986; De Wekker et al., 2004). Time series images of backscatter intensities can be used to evaluate mixed layer depth based on the marked drop in signal strength at the top of the PBL (Boers et al., 1984). Several automated methods using lidar to calculate mixed 110 5.4. Evaluation of BL structure with lidar: 6 \u00E2\u0080\u0093 8 July 2010 layer depth have been developed, all of which depend on the assumption that aerosol concentration is substantially higher in the mixed layer (PBL) than in the lower FT (Steyn et al., 1999). During the spring and summer of 2009, a Vaisala CL31 ceilometer operated at the Timing Flats site near the Creekside base of Whistler Mountain. Laser ceilometers are low-power lidars that are used primarily to measure cloud base heights, but have been shown to be useful for aerosol detection and BL observations during dry weather (Zephoris et al., 2005; Mu\u00CC\u0088nkel et al., 2007). A CL31 ceilometer situated in urban Vancouver was found to be effective for observation of aerosol layers and for computation of mixed layer depth (van der Kamp, 2008; McKendry et al., 2009). However, a review of backscatter data from the Whistler ceilometer found that on most clear days the aerosol concentration was too low to provide an unambiguous depiction of the PBL. Exceptions to this situation included days with smoke in the air from regional biomass burning. Smoke plumes were detectable with the CL31, and provided episodic opportunities to analyze vertical motion and mixing processes through ground-based remote sensing (e.g. McKendry et al., 2010). The ceilometer was removed from the site in September 2009. In April 2010, a more powerful aerosol lidar was installed at the Nesters site in Whistler Valley. As part of the Canadian Operational Research Aerosol Lidar Network (CORALNet) operated by Environment Canada, this lidar was deployed for aerosol research and LRT studies. Operating continuously except during pre- cipitation and aircraft overflights, the CORALNet lidar produces time series of backscatter ratio values, which can be used to create visual representations of BL evolution. The higher power of the laser as compared to the CL31 ceilometer means that more backscattered energy is received, resulting in better definition of relative aerosol concentrations. The lidar is expected to be in place at Whistler for at least one year, providing a unique opportunity for BL studies in a mountainous setting through all seasons. Such a dataset is a expected to be highly valuable for assessing 111 5.4. Evaluation of BL structure with lidar: 6 \u00E2\u0080\u0093 8 July 2010 periods of PBL influence at the mountaintop: while meteorological and physico- chemical data can provide evidence for PBL influence as demonstrated herein, lidar imagery allows one to \u00E2\u0080\u009Csee\u00E2\u0080\u009D the PBL and other layers of enhanced aerosol concen- tration. An example is illustrated and discussed below. Although the lidar dataset does not overlap with other datasets used in this study, this case serves to confirm some of the conjectures made thus far concerning the nature of PBL influence at Whistler while showing the potential value of this instrument to interpretation of the chemistry data. For this example, a three-day period was selected from the first extended dry spell of the 2010 summer season. The period of 6 \u00E2\u0080\u0093 8 July featured clear to partly cloudy skies at Whistler with increasingly warm temperatures, reaching a high of 33.5 \u00E2\u0097\u00A6C on the 8th. The 500 hPa and SLP patterns, averaged over the three days, are given in Figure 5.8. The 500 hPa map shows a large high pressure ridge over the area. This ridge moved slowly toward the southeast during the period; the map in Figure 5.8a represents an average position. At the surface (Figure 5.8b), the East Pacific High was situated offshore of BC with a ridge extending southeastward through the province. A slight thermal trough was in place along the coast. Similar to the 20 February case study, the synoptic situation for this case was like that of the Pattern A composite maps (Section 4.5), but with a stronger ridge aloft. The synoptic scale pressure gradient was fairly weak, both at the surface and aloft. Although the AQRD and OAN datasets are not available for this period, it can be expected that the synoptic scale conditions supported thermally driven convection and diurnal wind systems, resulting in diurnal cycles of T, w and CN concentration on the peak. Surface observations from the Nesters site support this assumption in that a valley wind signature was present on the 7th and 8th: light northerly winds switched to southerly in late afternoon on both days. Also, a few cumuliform clouds appeared each day by early afternoon, a sign of upward mesoscale motion occurring in an environment of synoptic scale subsidence. Inspection of Kelowna 112 5.4. Evaluation of BL structure with lidar: 6 \u00E2\u0080\u0093 8 July 2010 5650 5680 5710 57 40 5740 5770 5800 5830 5830 5860 140 \u00C2\u00B0 W 135 \u00C2\u00B0 W 130 \u00C2\u00B0 W 125 \u00C2\u00B0 W 120 \u00C2\u00B0 W 115 \u00C2\u00B0 W 110 \u00C2\u00B0 W 40 \u00C2\u00B0 N 45 \u00C2\u00B0 N 50 \u00C2\u00B0 N 55 \u00C2\u00B0 N 60 \u00C2\u00B0 N (a) 500 hPa 1012 1012 10121 016 1016 1020 1020 10 24 1028 1032 140 \u00C2\u00B0 W 135 \u00C2\u00B0 W 130 \u00C2\u00B0 W 125 \u00C2\u00B0 W 120 \u00C2\u00B0 W 115 \u00C2\u00B0 W 110 \u00C2\u00B0 W 40 \u00C2\u00B0 N 45 \u00C2\u00B0 N 50 \u00C2\u00B0 N 55 \u00C2\u00B0 N 60 \u00C2\u00B0 N (b) SLP Figure 5.8: Regional synoptic circulation maps showing averaged 500 hPa and sea level pressure patterns for 6 \u00E2\u0080\u0093 8 July 2009. upper air soundings from the period (not shown here) revealed that subsidence inversions were present at approximately 700 hPa, below which the afternoon lapse rates were nearly dry adiabatic. Thus, if the Kelowna soundings were representative of conditions at Whistler, a well-mixed layer of air extended from the surface to above the mountaintop on each of the three afternoons. A time series of backscatter data from the Whistler lidar is given in Figure 5.9 for the three-day period of interest. The higher returns found in the lowest 3 km of the atmosphere reflect the greater concentration of aerosol in the PBL as compared to the FT. As expected from previous observations of aerosol profiles (e.g. Jaenicke, 1993), concentrations were generally highest near the ground, decreasing gradually through the mixed layer to a fairly distinct PBL top, which varied between approximately 2400 and 3200 m agl during the period. A horizontal line across the image at 1517 m agl marks the summit height of Whistler Mountain. The backscatter data from July 6th through early afternoon on the 7th give the appearance of a more layered structure than what is found later in the period. It may be that the region of lower concentrations between approximately 1500 and 2500 m (as compared to data closer to the surface) is representative of the injection and/or 113 5.4. Evaluation of BL structure with lidar: 6 \u00E2\u0080\u0093 8 July 2010 Figure 5.9: Time series of backscatter ratio data from the Whistler CORALNet lidar from July 6 \u00E2\u0080\u0093 8, 2010. Labels A and B point to features mentioned in the text. residual layer as described in Chapter 2. This area is labelled \u00E2\u0080\u0098A\u00E2\u0080\u0099 in the figure. A free-air temperature profile, which is not available for this timeframe, would be needed to determine how this image best fits the conceptual models. An injection layer would appear to be slightly stable in terms of potential temperature, whereas the conventional CBL is defined by a nearly constant potential temperature profile (De Wekker, 2002). Regardless of the details of its inner structure, it seems clear from the image that an aerosol layer (AL) extended beyond the height of Whistler Mountain for the entire three-day period. The time series of backscatter data shows how the depth of the AL changed over time. On each of the three days, CBL growth began before noon and the diur- nal maximum depth occurred in late afternoon or evening. The ragged appearance of the AL top is likely due to updrafts and downdrafts associated with convective thermals and mountain venting. On the afternoon of the 8th, a small area of very high returns in the data (appearing as white in the image and labelled as \u00E2\u0080\u0098B\u00E2\u0080\u0099) can be seen at the top of the AL. The height of these returns, just above 3000 m agl, corresponds well to the altocumulus clouds reported by the Whistler observer that afternoon at 10,000 feet agl. The image at that time shows relatively high returns right up to the cloud base, suggesting that there was a well-mixed PBL between the surface and 3000 m agl, which is also in agreement with the height of the tem- 114 5.4. Evaluation of BL structure with lidar: 6 \u00E2\u0080\u0093 8 July 2010 perature inversion in the Kelowna sounding data. This is much higher than typical summertime mixed layer heights for Vancouver as determined by van der Kamp (2008) from ceilometer backscatter data. Data from the Whistler lidar support the idea that enhanced thermal circulations and venting mechanisms produce deep boundary layers in mountainous areas. Although the summit of Whistler was within the AL throughout the three-day period, it is still expected that diurnal cycles occurred in the meteorological and aerosol variables, with maximum values in late afternoon or early evening as the period of most vigorous upward motion came to an end. Overnight, the absence of updrafts allows for a lowering of concentrations as synoptic scale advection prevails at the mountaintop. However, BL constituents which were lofted above ridgetops in upstream mountainous areas during the day would pass through the site at night, maintaining a residual layer that is not representative of FT conditions (Lugauer, 1998). The lidar image presented here supports the explanation offered in Chapter 3 for the dampened diurnal cycles observed in summer for CN and water vapour as compared to springtime. The presence of a residual layer at night means that daily average concentrations at the site will be relatively high and the night-time concentrations may not recover to those of the FT. It is surmised that night-time residual layers are more common and persistent in the summer season than during other times of the year. By operating for an entire year, the CORALNet lidar at Whistler will reduce the uncertainty in these conceptual models. Additionally, comparisons to data from the CORALNet lidar in coastal Vancouver will help resolve differences in BL structure produced by the different physical environments. Despite the limited data available, this case study provides strong evidence that, during synoptic weather conditions conducive to convective mixing, a deep CBL can develop in the Whistler area to heights well above the ridgetops. Updrafts and downdrafts mix the air and its surface-based entities during the daytime heating period such that the observatory at the summit of Whistler samples a mixture of FT 115 5.5. Conclusions and valley air. On some nights, this mixture will be replaced by FT air as downslope flows and synoptic advection become dominant. However, on nights such as those in this case study, the site remains within a residual layer of enhanced water vapour and PM. Presumably, trace gases from local pollution sources in nearby valleys would also linger in the residual layer overnight. Thus, for the purposes of LRT and background air chemistry studies, some days will contain no representative data. Information obtained from the lidar simplifies the identification of such days. 5.5 Conclusions The three case studies of this chapter have shown that air mass discrimination can be performed for the Whistler observatory by using available datasets to focus on the conditions and meteorological mechanisms of a particular short (e.g. one day) time period. However, not all days will be amenable to easy interpretation, as the complexity of 3-D atmospheric motions on various scales can lead to results that seem ambiguous. These case studies have reiterated the importance of the synoptic scale setting to outcomes concerning PBL depth in the mountains. It was shown in the 20 February case that the synoptic situation provided an environment comparable to conditions more commonly found in spring or summer. The 13 July example showed that not all summertime cases of high pressure aloft result in deep CBL development. The three case studies also demonstrated the value of vertical profiles amongst the datasets. For evaluation of BL structure, ideally one would always have profiles of tempera- ture, moisture, wind and aerosol concentration available from the area of interest. As upper air soundings are not normally conducted anywhere in southwestern BC, the deployment of the CORALNet lidar constitutes an important contribution to the observational network in the Whistler area. 116 Chapter 6 Conclusions 6.1 Summary of results Datasets from Whistler Mountain and its region were analyzed in order to investi- gate the frequency and nature of PBL influence at the mountaintop air chemistry observatory on seasonal, monthly and diurnal timescales. The seasonal and monthly analyses enabled general conclusions about PBL influence at the site, while analysis of daily data highlighted considerable variability found in the results from day to day, especially during times of changing synoptic scale weather. To summarize the results of this study, the two primary research questions posed at the outset are now re-visited to assess the extent to which they have been answered. 1) How much of the time and in what conditions is the Whistler site influenced by PBL air? Analysis of measurements from the summit of Whistler Mountain formed the basis of assumptions made about PBL influence at the site. Meteorological and aerosol variables were used as indicators of air from lower elevations arriving at the moun- taintop. From all indicators, it is apparent that PBL influence is most common at Whistler in the spring and summer months and least common in late fall through early winter. Criteria were developed using CN concentration to define a typical diurnal cycle that would be expected from thermally induced vertical motions. It was found that 37% of all days in the one-year study period exhibited typical diurnal cycles. 117 6.1. Summary of results Monthly results ranged from 3 days in December that met the criteria to 17 days each in May and September. The scheme for finding days with typical diurnal cycles can be considered conservative, as it likely missed days with PBL influence that had maximum daily concentrations at atypical times due to synoptic scale factors. Also, it is not expected that the scheme would reliably include days with PBL influence via mechanisms unrelated to thermal forcing, e.g. dynamic lift. Results from February 2009, when 11 days had typical diurnal cycles in the CN data, suggest that the Whistler site is susceptible to vertical mixing and subse- quent PBL influence during the winter. A case study from February showed diurnal changes in local temperature profiles that were consistent with CBL growth to the mountaintop level. However, a review of weather conditions over the study pe- riod found that February 2009 was not characteristic of a normal winter month at Whistler. The connection between synoptic weather conditions and local vertical transport means that considerable interannual variability can be expected in the patterns of PBL influence at the site. Investigation into the possible driving forces of PBL influence found that synoptic scale analysis was more successful at explaining the patterns than analysis on more local scales. The synoptic scale weather situation can either facilitate or inhibit local vertical motions in mountainous terrain. Days with abundant solar insolation and weak synoptic scale pressure gradients usually support convective uplift and diurnal wind systems. From this knowledge, synoptic classification was used to define the weather types mostly frequently associated with PBL influence during the study period. The most common synoptic pattern found on PBL-influence days featured a high pressure ridge aloft and a weak, often offshore, surface pressure gradient. Another common pattern for PBL-influence days involved a more unstable air mass, often in a post-frontal environment, but also with a weak surface gradient and considerable sunshine. In order to make some generalizations about the seasonal nature of PBL influence 118 6.1. Summary of results at Whistler, results have been synthesized into predominant air mass categories for each season as given in Table 6.1. This categorization is considered appropriate only for days with considerable sunshine and little or no precipitation. Cloudy days and those with precipitation are more likely to maintain FT conditions at the site, but further study is needed to assess the possible role of dynamic lift in PBL influence. Hours (LST) Winter Spring Summer Fall Day (0900 \u00E2\u0080\u0093 2400) FT PBL PBL PBL Night (0000 \u00E2\u0080\u0093 0900) FT FT RL FT Table 6.1: Predominant air mass categories by season for clear and partly cloudy days at the Whistler observatory. Note: RL= residual layer. The seasonal categories in Table 6.1 are further separated by time of day, with the hours of midnight to 0900 LST given as representative of \u00E2\u0080\u009Cnight.\u00E2\u0080\u009D This range of hours was estimated based on average diurnal cycle graphs from Chapter 3, which show water vapour and CN concentration reaching their daily minimum values early in the morning before the effects of daytime upward motions reach the site. The remaining hours of the day are those which are more clearly prone to BL effects on days with typical diurnal cycles. It must be emphasized that these seasonal categories are broad generalizations meant only to show the predominant condition expected for each period. As seen in the preceding chapters, day-to-day variation in the results was substantial through- out the study period. Whistler\u00E2\u0080\u0099s location in a synoptically active mid-latitude coastal zone precludes the use of simple criteria, such as time of day, for routine air mass discrimination. Based on the results summarized in Table 6.1, it can be said that the Whistler site is representative of the FT most of the time in winter, but it is subject to the influence of PBL air a majority of the time in summer, both day and night. The persistence of a residual layer of BL constituents at night is assumed common in 119 6.1. Summary of results summer when convective mixing is generally most vigorous and the daily period of solar insolation is longest. Thus, the night-time period is not long enough for FT air to completely replace the mixed PBL/FT air lingering around the summit on nights when the synoptic scale flow is weak. Lidar backscatter data from July 2010 provided an example of a multi-day period when the peak of Whistler remained in PBL or RL layers at all hours. The spring and fall periods are prone to PBL influence on clear to partly cloudy days, but it is expected that the longer nights and slightly stronger synoptic scale winds typical of these transition seasons promote a return to FT conditions at the summit by early morning. 2) What routinely available meteorological and/or physico-chemical parameters can be used to distinguish PBL air from free tropospheric conditions? This study has used the diurnal trends in several variables as indicators of air from lower elevations affecting the mountaintop site. Following the methods of previous studies from other mountain observatories, diurnal increases in temperature, water vapour and aerosol concentration were interpreted as signals of PBL air infiltrating the site. Of these three indicators, temperature is likely the least robust and aerosol concentration is likely the best indicator of PBL influence. Temperature cycles can be expected from heating and cooling of surfaces very local to the mountain summit, whereas water vapour and aerosols are physical entities that have known typical profiles in the atmosphere. Aerosol concentration is expected to be generally superior to water vapour mixing ratio in this context, as it has a larger dynamic range, comes closer to conservative behaviour, and is more closely associated with local pollution sources (Bodhaine, 1996; Baltensperger et al., 1997). The diurnal and seasonal cycles found in these variables fit well with conceptual models of mesoscale vertical motion in complex terrain, but none of the indicators used are ideal tracers of valley air; there are always other possible explanations 120 6.1. Summary of results for an observed cycle of temperature, water vapour or aerosol concentration. For instance, a diurnal maximum in w could occur during cloud formation in the FT and/or horizontal moisture advection that happens to occur in the afternoon hours. Examining correlations between two indicator variables, as was done for w and CN in this study, can increase confidence in the vertical motion/CBL growth explanation for diurnal cycles when the two variables are highly correlated. In these cases, it becomes more difficult to explain the patterns in terms of coincidental FT processes. Rather, the upward movement of humid, aerosol-rich valley air is the simpler and more likely explanation for daytime increases in these variables at the mountaintop. However, the actual complexity of atmospheric motions often prevents the measured variables from conforming neatly to models. A day with diurnal cycles in w and CN that are temporally offset from each other may fail to exhibit a high degree of correlation between the two variables, but the vertical transport of locally sourced pollutants is still likely on such a day. Case studies using vertical profiles of meteorological variables showed that a mixed layer developed on PBL-influence days from the valley floor to altitudes be- yond the mountain summits. It appears that growth of the mountain CBL via convective thermals is the primary mechanism for PBL influence at the Whistler site. On some days the upward mixing of valley air is enhanced by slope and valley wind systems. Well-defined diurnal wind patterns were found to occur on 14.5% of all days in the study period and up to 42% of days in the summer months. Monthly average CN concentrations were higher for days with slope and valley winds than for days without diurnal wind cycles. Orographic and dynamic lift associated with synoptic weather systems may also play a role in transporting valley air upward. Two of the PBL-influence days during this study appeared to involve frontal passages. Additional data would be needed to fully assess the role of dynamic lift in PBL influence. This study has relied on CN concentration as the primary indicator of lower-elevation air. Aerosols are not 121 6.1. Summary of results suitable PBL tracers during precipitation, because many of the particles get washed out of the air mass. This study has used the available data to describe patterns of PBL influence at the Whistler observatory for a full year, but criteria have not been developed for air mass discrimination. The lack of a definitive tracer of PBL air and the above-mentioned complexity of atmospheric processes on different scales preclude the routine use of threshold values or simple decision paths to characterize the sampled air as PBL-influenced or representative of the FT. This outcome is not unique to this study or the Whistler site. For, even at the Mauna Loa Observatory, where the consistency of the island\u00E2\u0080\u0099s tropical weather has encouraged the use of a simple time of day criterion, Bodhaine (1996) cautioned that a manual inspection of datasets should still be performed to identify cases that do not fit the usual pattern. Therefore, a challenge exists in selecting periods to represent background FT conditions for a particular site. For short-term cases (up to a few days), analysis of a variety of datasets, as was done for the case studies in Chapter 5, can be used to make judgements about the representativeness of the measurements. However, this approach becomes less practical as the timeframe of interest increases. The method of Weiss-Penzias et al. (2006)\u00E2\u0080\u0094 to select the driest 20% of observations for a particular period as representative of subsiding free tropospheric air masses\u00E2\u0080\u0094has appeal for its simplicity, but could still allow periods of PBL influence to be included. However, if the threshold is conservative and the timeframe relatively short (e.g. not longer than one month), then this type of approach may be the best available for differentiating large datasets. Analysis of short time periods should emphasize signatures of vertical motion at the site. Well-correlated diurnal cycles in w and CN with afternoon maxima strongly suggest PBL influence, whereas increasing T concurrent with decreasing w is a signature of subsidence. Synoptic weather maps and vertical profile data (from regional upper air soundings, lidar data, etc.) should then be used to confirm that 122 6.2. Suggestions for future work the synoptic environment was supportive of the inferred processes. Back-trajectory models can provide additional information about the provenance of sampled air streams. Although more local observations would be useful, the observational net- work surrounding Whistler is sufficient in most cases to distinguish episodes of LRT from local or regional pollution by using the above-outlined course of analysis. 6.2 Suggestions for future work Future research of this kind would benefit most from availability of a better tracer of local pollution from the valley-based BL. A gaseous tracer that is not substantially reduced by clouds or precipitation would be best. Observations of NOx or some other short-lived pollutant known to have sources in the Whistler area should be added to the measurement program at the AQRD observatory. The CORALNet lidar at Whistler will provide a rich dataset for aerosol and mountain BL studies. Development of an algorithm for computing mixed layer heights from the lidar data could provide an automated, objective method of de- termining PBL depth. If such an algorithm were validated with the meteorological and physico-chemical data, then it could be used for air mass discrimination and subsequent separation of air chemistry datasets. The scientific value of having a li- dar operating continuously in a mountainous setting is substantial, and it would be beneficial to keep the instrument in Whistler for more than the proposed one-year period. Finally, it would be instructive to extend the analysis of this study to include data from several consecutive years. The robustness of conclusions made here about the frequency and seasonal patterns of PBL influence could be assessed and interannual variability could be quantified. Such analysis would bring the study closer to an actual climatology of air mass characterization for the Whistler observatory. 123 Bibliography Aguado, E. and Burt, J.: Understanding Weather and Climate, 5th edition, Prentice Hall, 2010. Baltensperger, U., Ga\u00CC\u0088ggeler, H., Jost, D., Lugauer, M., Schwikowski, M., Weingart- ner, E., and Seibert, P.: Aerosol climatology at the high-alpine site Jungfraujoch, Switzerland, Journal Of Geophysical Research, 102, 19,707\u00E2\u0080\u009319,715, 1997. Barry, R.: Mountain Weather and Climate, 3rd edition, Cambridge University Press, 2008. Blifford, I. and Ringer, L.: The size and number distribution of aerosols in the continental troposphere, Journal of the Atmospheric Sciences, 26, 716\u00E2\u0080\u0093726, 1969. Bodhaine, B.: Aerosol measurements during the Mauna Loa Photochemistry Ex- periment 2, Journal of Geophysical Research, 101, 14,757\u00E2\u0080\u009314,765, 1996. Boers, R., Eloranta, E., and Coulter, R.: Lidar observations of mixed layer dy- namics: tests of parameterized entrainment models of mixed layer growth rate, Journal of Climate and Applied Meteorology, 23, 247\u00E2\u0080\u0093266, 1984. Bolton, D.: The computation of equivalent potential temperature., Monthly Weather Review, 108, 1046\u00E2\u0080\u00931053, 1980. Boubel, R. W., Fox, D. L., Turner, D. B., and Stern, A. C.: Fundamentals of Air Pollution, 3rd edition, Academic Press, 1994. 124 Bibliography Collaud Coen, M., Weingartner, E., Nyeki, S., Steinbacher, M., and Baltensperger, U.: Impact of synoptic weather types on the planetary boundary layer influence at the Jungfraujoch, in: Symposium on Atmospheric Chemistry and Physics at Mountain Sites, Interlaken, Switzerland, 2010. Cooper, O., Forster, C., Parrish, D., Trainer, M., Dunlea, E., Ryerson, Y., Huber, G., Fehsenfeld, F., Nicks, D., Holloway, J., de Gouw, J., Warneke, C., Roberts, J., Flocke, F., and Moody, J.: A case study transpacific warm conveyor belt transport: Influence of merging airstreams on trace gas import to North America, Journal of Geophysical Research D: Atmospheres, 109, 1\u00E2\u0080\u009317, 2004. De Wekker, S.: Structure and morphology of the convective boundary layer in mountainous terrain, Ph.D. thesis, University of British Columbia, 2002. De Wekker, S. F., Steyn, D., and Nyeki, S.: A comparison of aerosol-layer and convective boundary-layer structure over a mountain range during STAAARTE \u00E2\u0080\u009997, Boundary-Layer Meteorology, 113, 249\u00E2\u0080\u0093271, 2004. Draxler, R. and Rolph, G.: HYSPLIT (HYbrid Single-Particle Lagrangian Integrated Trajectory) model access via NOAA ARL READY website (http://ready.arl.noaa.gov/HYSPLIT.php), NOAA Air Resources Laboratory, Silver Spring, MD., 2010. Endlich, R., Ludwig, F., and Uthe, E.: An automatic method for determining the mixing depth from lidar observations, Atmospheric Environment, 13, 1051\u00E2\u0080\u00931056, 1979. Erven, L. and McKendry, I.: Slope air temperatures to aid in forecasting the 2010 Winter Olympic and Paralympic alpine events, Poster at: 13th Conference on Mesoscale Processes, Salt Lake City, Utah, U.S.A., 2009. Forrer, J., Ruttimann, R., Schneiter, D., Fischer, A., Buchmann, B., and Hofer, P.: 125 Bibliography Variability of trace gases at the high-Alpine site Jungfraujoch caused by meteo- rological transport processes, Journal of Geophysical Research D: Atmospheres, 105, 12,241\u00E2\u0080\u009312,251, 2000. Glickman, T. S., ed.: Glossary of Meteorology, 2nd edition, American Meteorological Society, 2000. Grant, A. N., Pszenny, A. A., and Fischer, E. V.: The 1935-2003 air tempera- ture record from the summit of Mount Washington, New Hampshire, Journal of Climate, 18, 4445\u00E2\u0080\u00934453, 2005. Hallar, A. G.: Persistent daily aerosol nucleation events at mountain-top location, in: Symposium on Atmospheric Chemistry and Physics at Mountain Sites, Interlaken, Switzerland, 2010. Heald, C. L., Jacob, D. J., Park, R. J., Alexander, B., Fairlie, T. D., Yantosca, R. M., and Chu, D. A.: Transpacific transport of Asian anthropogenic aerosols and its impact on surface air quality in the United States, Journal of Geophysical Research D: Atmospheres, 111, 2006. Henne, S., Furger, M., and Prevot, A. S.: Climatology of mountain venting-induced elevated moisture layers in the lee of the Alps, Journal of Applied Meteorology, 44, 620\u00E2\u0080\u0093633, 2005. Henne, S., Klausen, J., Junkermann, W., Kariuki, J., Aseyo, J., and Buchmann, B.: Representativeness and climatology of carbon monoxide and ozone at the global GAW station Mt. Kenya in equatorial Africa, Atmospheric Chemistry and Physics, 8, 3119\u00E2\u0080\u00933139, 2008. Hindman, E. E.: Himalayan banner clouds and thermals, Proceedings of the 7th Conference on Mountain Meteorology, pp. 115\u00E2\u0080\u0093120, 1995. 126 Bibliography Hofmann, D.: Twenty years of balloon-borne tropospheric aerosol measurements at Laramie, Wyoming, Journal Of Geophysical Research, 98, 12,753\u00E2\u0080\u009312,766, 1993. Holzer, M., Hall, T. M., and Stull, R. B.: Seasonality and weather-driven variability of transpacific transport, Journal of Geophysical Research D: Atmospheres, 110, 1\u00E2\u0080\u009322, 2005. Jaenicke, R.: Tropospheric aerosols, in: Aerosol-Cloud-Climate Interactions, edited by Hobbs, P. V., pp. 1\u00E2\u0080\u009331, Academic Press, 1993. Jaffe, D., Anderson, T., Covert, D., Kotchenruther, R., Trost, B., Danielson, J., Simpson, W., Berntsen, T., Karlsdottir, S., Blake, D., Harris, J., Carmichael, G., and Itsushi, U.: Transport of Asian air pollution to North America, Geophysical Research Letters, 26, 711\u00E2\u0080\u0093714, 1999. Jaffe, D., McKendry, I., Anderson, T., and Price, H.: Six \u00E2\u0080\u0098new\u00E2\u0080\u0099 episodes of trans- Pacific transport of air pollutants, Atmospheric Environment, 37, 391\u00E2\u0080\u0093404, 2003. Kim, Y., Sievering, H., and Boatman, J.: Airborne measurement of atmospheric aerosol particles in the lower troposphere over the central United States, Journal of Geophysical Research, 93, 12,631\u00E2\u0080\u009312,644, 1988. Kleissl, J., Honrath, R., Dziobak, M., Tanner, D., Val Martin, M., Owen, R., and Helmig, D.: Occurrence of upslope flows at the Pico mountaintop observatory: A case study of orographic flows on a small, volcanic island, Journal of Geophysical Research D: Atmospheres, 112, 2007. Kossmann, M., Vogtlin, R., Corsmeier, U., Vogel, B., Fiedler, F., Binder, H. J., Kalthoff, N., and Beyrich, F.: Aspects of the convective boundary layer structure over complex terrain, Atmospheric Environment, 32, 1323\u00E2\u0080\u00931348, 1998. Kulmala, M.: How particles nucleate and grow, Science, 302, 1000\u00E2\u0080\u00931001, 2003. 127 Bibliography Lowenthal, D. H., Borys, R. D., and Wetzel, M. A.: Aerosol distributions and cloud interactions at a mountaintop laboratory, Journal of Geophysical Research D: Atmospheres, 107, 1\u00E2\u0080\u00931 \u00E2\u0080\u0093 1\u00E2\u0080\u00936, 2002. Lugauer, M.: Vertical transport of atmospheric trace species in the Alps, Ph.D. thesis, University of Bern, 1998. Lugauer, M., Baltensperger, U., Furger, M., Ga\u00CC\u0088ggeler, H., Jost, D., Schwikowski, M., and Wanner, H.: Aerosol transport to the high Alpine sites Jungfraujoch (3454 m asl) and Colle Gnifetti (4452 m asl), Tellus, Series B, 50B, 76\u00E2\u0080\u009392, 1998. Lugauer, M., Baltensperger, U., Furger, M., Ga\u00CC\u0088ggeler, H., Jost, D., Nyeki, S., and Schwikowski, M.: Influences of vertical transport and scavenging on aerosol par- ticle surface area and radon decay product concentrations at the Jungfraujoch (3454 m above sea level), Journal of Geophysical Research D: Atmospheres, 105, 19,869\u00E2\u0080\u009319,879, 2000. Mass, C.: The Weather of the Pacific Northwest, University of Washington Press, 2008. McKendry, I.: Synoptic circulation and summertime ground-level ozone concen- trations at Vancouver, British Columbia, Journal of Applied Meteorology, 33, 627\u00E2\u0080\u0093641, 1994. McKendry, I.: PM10 levels in the Lower Fraser Valley, British Columbia, Canada: An overview of spatiotemporal variations and meteorological controls, Journal of the Air and Waste Management Association, 50, 443\u00E2\u0080\u0093452, 2000. McKendry, I., Hacker, J., Stull, R., Sakiyama, S., Mignacca, D., and Reid, K.: Long-range transport of Asian dust to the Lower Fraser Valley, British Columbia, Canada, Journal of Geophysical Research D: Atmospheres, 106, 18,361\u00E2\u0080\u009318,370, 2001. 128 Bibliography McKendry, I., MacDonald, A., Leaitch, W., Van Donkelaar, A., Zhang, Q., Duck, T., and Martin, R.: Trans-Pacific dust events observed at Whistler, British Columbia during INTEX-B, Atmospheric Chemistry and Physics, 8, 6297\u00E2\u0080\u00936307, 2008. McKendry, I., van der Kamp, D., Strawbridge, K., Christen, A., and Crawford, B.: Simultaneous observations of boundary-layer aerosol layers with CL31 ceilometer and 1064/532nm lidar, Atmospheric Environment, 43, 5847\u00E2\u0080\u00935852, 2009. McKendry, I., Gallagher, J., Campuzano Jost, P., Bertram, A., Strawbridge, K., Leaitch, W., and MacDonald, A.: A subsiding regional forest fire aerosol layer at Whistler, BC: implications for interpretation of mountaintop chemistry ob- servations, Atmospheric Chemistry and Physics Discussions, 10, 20,303\u00E2\u0080\u009320,327, 2010. McKendry, I. G., Strawbridge, K. B., O\u00E2\u0080\u0099Neill, N. T., Macdonald, A. M., Liu, P. S., Leaitch, W. R., Anlauf, K. G., Jaegle, L., Fairlie, T. D., and Westphal, D. L.: Trans-pacific transport of Saharan dust to western North America: A case study, Journal of Geophysical Research D: Atmospheres, 112, 2007. Mu\u00CC\u0088nkel, C., Eresmaa, N., Rasanen, J., and Karppinen, A.: Retrieval of mixing height and dust concentration with lidar ceilometer, Boundary-Layer Meteorology, 124, 117\u00E2\u0080\u0093128, 2007. Nishita, C., Osada, K., Kido, M., Matsunaga, K., and Iwasaka, Y.: Nucleation mode particles in upslope valley winds at Mount Norikura, Japan: implications for the vertical extent of new particle formation events in the lower troposphere, Journal of Geophysical Research D: Atmospheres, 113, 2008. Nseir, R.: Technical report for meteorological data at Whistler Peak, Tech. rep., Air Quality Research Division, Science and Technology Branch, Environment Canada, 2007. 129 Bibliography Nyeki, S., Baltensperger, U., Colbeck, I., Jost, D., Weingartner, E., and Ga\u00CC\u0088ggeler, H.: The Jungfraujoch high-alpine research station (3454 m) as a background clean continental site for the measurement of aerosol parameters, Journal of Geophysical Research D: Atmospheres, 103, 6097\u00E2\u0080\u00936107, 1998. Nyeki, S., Kalberer, M., Lugauer, M., Weingartner, E., Petzold, A., Schroder, F., Colbeck, I., and Baltensperger, U.: Condensation nuclei (CN) and ultrafine CN in the free troposphere to 12 km: A case study over the Jungfraujoch high-alpine re- search station, Geophysical Research Letters, 26, 2195\u00E2\u0080\u00932198, condensation nuclei;, 1999. Obrist, D., Hallar, A. G., McCubbin, I., Stephens, B. B., and Rahn, T.: Atmospheric mercury concentrations at Storm Peak Laboratory in the Rocky Mountains: Evi- dence for long-range transport from Asia, boundary layer contributions, and plant mercury uptake, Atmospheric Environment, 42, 7579\u00E2\u0080\u00937589, 2008. Oke, T.: Boundary Layer Climates, 2nd edition, Routledge, 1987. Oke, T. and Hay, J. E.: The Climate of Vancouver, 2nd edition, BC Geographical Series, no. 50, Department of Geography, University of British Columbia, 1994. Parrish, D., Hahn, C., Fahey, D., Williams, E., Bollinger, M., Hubler, G., Buhr, M., Murphy, P., Trainer, M., Hsie, E., Liu, S., and Fehsenfeld, F.: Systematic variations in the concentration of NOx (NO Plus NO2) at Niwot Ridge, Colorado, Journal of Geophysical Research, 95, 1817\u00E2\u0080\u00931836, 1990. Pitre, T.: 1995 Sea-to-Sky airshed emissions inventory of common air contaminants, summary report, Tech. rep., British Columbia Ministry of Water, Land & Air Protection, 2002. Reidmiller, D. R., Jaffe, D. A., Fischer, E. V., Finley, B., and Kangas, R.: NOx observations in the free troposphere at the Mt. Bachelor Observatory, Poster at: 130 Bibliography Symposium on Atmospheric Chemistry and Physics at Mountain Sites, Interlaken, Switzerland, 2010. Reuten, C., Steyn, D., Strawbridge, K., and Bovis, P.: Observations of the rela- tion between upslope flows and the convective boundary layer in steep terrain, Boundary-Layer Meteorology, 116, 37\u00E2\u0080\u009361, 2005. Sheridan, S. C.: The redevelopment of a weather-type classification scheme for North America, International Journal of Climatology, 22, 51\u00E2\u0080\u009368, 2002. Steyn, D., Baldi, M., and Hoff, R.: The detection of mixed layer depth and entrain- ment zone thickness from lidar backscatter profiles, Journal of Atmospheric and Oceanic Technology, 16, 953\u00E2\u0080\u0093959, 1999. Stull, R. B.: Meteorology for Scientists and Engineers, 2nd edition, Brooks/Cole, 2000. UNECE: Clearing the air: 25 years of the Convention on Long-range Transboundary Air Pollution, Tech. rep., United Nations, 2004. van der Kamp, D.: Ceilometer observations of Vancouver\u00E2\u0080\u0099s urban boundary layer, Master\u00E2\u0080\u0099s thesis, University of British Columbia, 2008. Wakimoto, R. and McElroy, J.: Lidar observation of elevated pollution layer over Los Angeles, Journal of Climate and Applied Meteorology, 25, 1583\u00E2\u0080\u00931599, 1986. Weiss-Penzias, P., Jaffe, D. A., Swartzendruber, P., Dennison, J. B., Chand, D., Hafner, W., and Prestbo, E.: Observations of Asian air pollution in the free troposphere at Mount Bachelor Observatory during the spring of 2004, Journal of Geophysical Research, 111, 2006. Whiteman, C. D.: Mountain Meteorology: Fundamentals and Applications, Oxford University Press, 2000. 131 Bibliography Wiedensohler, A., Orsini, D., Covert, D. S., Coffmann, D., Cantrell, W., Havlicek, M., Brechtel, F., Russel, L., Weber, R. J., Gras, J., Hudson, J. G., and Litchy, M.: Performance of the counting efficiency of the TSI-3025 UCPC, the TSI-3010 CPC, and the TSI-3760/7610 CPC for different operating conditions, Journal of Aerosol Science, 27, S337\u00E2\u0080\u0093S338, 1996. Yarnal, B.: Synoptic climatology in environmental analysis, Belhaven Press, 1993. Yienger, J. J., Galanter, M., Holloway, T. A., Phadnis, M. J., Guttikunda, S. K., Carmichael, G. R., Moxim, W. J., and Levy II, H.: The episodic nature of air pollution transport from Asia to North America, Journal of Geophysical Research D: Atmospheres, 105, 26,931\u00E2\u0080\u009326,945, 2000. Zephoris, M., Holin, H., Lavie, F., Cenac, N., Cluzeau, M., Delas, O., Eideliman, F., Gagneux, J., Gander, A., and Thibord, C.: Ceilometer observations of aerosol layer structure above the Petit Luberon during ESCOMPTE\u00E2\u0080\u0099s IOP 2, Atmo- spheric Research, 74, 581\u00E2\u0080\u0093595, 2005. Zhang, L., Jacob, D., Boersma, K., Jaffe, D., Olson, J., Bowman, K., Worden, J., Thompson, A., Avery, M., Cohen, R., Dibb, J., Flocke, F., Fuelberg, H., Huey, L., McMillan, W., Singh, H., and Weinheimer, A.: Transpacific transport of ozone pollution and the effect of recent Asian emission increases on air quality in North America: An integrated analysis using satellite, aircraft, ozonesonde, and surface observations, Atmospheric Chemistry and Physics Discussions, 8, 8143\u00E2\u0080\u0093 8191, 2008. Zhao, T., Gong, S., Zhang, X., Blanchet, J., McKendry, I., and Zhou, Z.: A sim- ulated climatology of Asian dust aerosol and its trans-Pacific transport. Part I: Mean climate and validation, Journal of Climate, 19, 88\u00E2\u0080\u0093103, 2006. 132"@en . "Thesis/Dissertation"@en . "2010-11"@en . "10.14288/1.0071327"@en . "eng"@en . "Geography"@en . "Vancouver : University of British Columbia Library"@en . "University of British Columbia"@en . "Attribution-NonCommercial-NoDerivatives 4.0 International"@en . "http://creativecommons.org/licenses/by-nc-nd/4.0/"@en . "Graduate"@en . "Patterns of planetary boundary layer influence at the Whistler Mountain air chemistry observatory : an observational mountain meteorology study"@en . "Text"@en . "http://hdl.handle.net/2429/28803"@en .