Strong Along-Channel Winds on the Coast of British Columbia:
Synoptic Climatology and Case Studies

by

Talaat Bakri
B.A., Damascus University, 2008

THESIS SUBMITTED IN PARTIAL FULFILLMENT OF
THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE
IN
NATURAL RESOURCES AND ENVIRONMENTAL STUDIES

UNIVERSITY OF NORTHERN BRITISH COLUMBIA
AUGUST 2014

© TALAAT BAKRI, 2014

UMI Number: 1526514

All rights reserved
INFORMATION TO ALL USERS
The quality of this reproduction is dependent upon the quality of the copy submitted.
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
a note will indicate the deletion.

Di!ss0?t&iori Publishing
UMI 1526514
Published by ProQuest LLC 2015. Copyright in the Dissertation held by the Author.
Microform Edition © ProQuest LLC.
All rights reserved. This work is protected against
unauthorized copying under Title 17, United States Code.

ProQuest LLC
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, Ml 48106-1346

Abstract
Strong along-channel winds on the coast o f British Columbia have been investigated at several
locations. Large-scale (synoptic-scale) atmospheric situations have been determined for different
types of the along-channel winds that blow up and down channels throughout different seasons.
Wintertime outflow winds are associated with a sea-level high-pressure area over the interior o f
British Columbia caused by an arctic air mass. Wintertime inflows are caused by sea level lowpressure systems that advance toward the coast of British Columbia. Summertime inflows are
initiated by a frontal passage or an inland thermal trough. Case studies at Howe Sound are
provided utilizing a period of intensive observations. Inflows (outflows) are accompanied by
higher (lower) temperature and dew point and caused by an along-channel pressure gradient
pointed inland (toward the coast). An outflow case shows potential hydraulic features, and the
vertical structures o f two cases indicate a lower gap flow layer within 1-2 km height.

TABLE OF CONTENTS
Abstract...................................................................................................................................................ii
Table of Contents.................................................................................................................................. in
List of Tables.......................................................................................................................................... v
List of Figures........................................................................................................................................ vi
Glossary................................................................................................................................................. ix
Acknowledgement............................................................................................................................... xiii
1.

2.

Study Rationale, Objectives and Research Questions.................................................. 1
1.1. Introduction..........................................................................................................1
1.2. Rationale of the Study.......................................................................................... 1
1.3. Objectives and Research Questions......................................................................3
1.4. Structure of Thesis...............................................................................................4
Literature Review and Methodology............................................................................ 5
2.1.
Literature Review................................................................................................5
2.1.1. Gap Winds: Definition and Theory........................................................ 5
2.1.2. Forces behind Gap Winds..................................................................... 11
2.1.2.1. Synoptic-scale Pressure Gradient.............................................11
2.1.2.2. Hydrostatically / Thermally Induced Pressure Gradient............ 15
2.1.2.3. Other Factors Affecting Gap Winds......................................... 17
2.1.2.3.1.
Venturi Effect..................................................17
2.1.2.3.2.
The Coriolis Force........................................... 18
2.1.2.3.3.
Friction.............................................................19
2.1.2.3.4.
Inertia and Entrainment of Air from Aloft ...... 20
2.1.3. Gap Winds, Wakes, and Downslope Windstorms.................................21
2.1.4. Gap Winds and Other Along-Channel Winds.......................................22
2.2. Methodology......................................................................................................23
2.2.1. Review of the Methods Used in Gap Wind Studies.............................. 23
2.2.2. Methods Employed in the Study of the Synoptic Climatology of Strong
Along-Channel Winds on the Coast of British Columbia...................... 26
2.2.2.1. Selection of the Observations.................................................. 26
2.2.2.2. Investigation of the Wind Behaviour and Selection of the Events
28
2.2.2.3. Constructing the Composites................................................... 31
2.2.2.4. K-means Clustering of the Composites....................................32
2.2.2.5. Frontal Investigation................................................................ 35
2.2.3. Methods Employed in the Study of Strong Along-channel Winds in
Howe Sound...................................
35
2.2.3.1. Data Description......................................................................36
2.2.3.2. Climatological Study...............................................................40
2.2.3.3. Case Studies............................................................................ 41

3.

A Synoptic Climatology of Strong Along-Channel Winds on the Coast of British
Columbia...................................................................................................................... 44
3.1.
Abstract............................................................................................................. 44
3.2.
Introduction....................................................................................................... 45
3.3.
Data, Methods, and Study Locations................................................................. 47
3.3.1. Study Locations.................................................................................... 47
3.3.2. Data.......................................................................................................49
3.3.3. Methods................................................................................................ 50
3.3.4. Validation of the Ability of MeteorologicalStations to Detect Outflows
and Inflows.......................................................................................... 53
3.4.
Annual and Diurnal Distribution of Outflow/Inflow Events...............................54
3.5.
Composite Analysis...........................................................................................56
3.5.1. Wintertime Outflow.............................................................................. 56
3.5.2. Wintertime Inflow................................................................................ 60
3.5.3. Summertime Outflow............................................................................61
3.5.4. Summertime Inflow...............................................................................63
3.6.
Manual Investigation of Hand-drawn Surface Analysis Charts........................... 65
3.7.
Clustering Analysis............................................................................................70
3.8.
Conclusion........................................................................................................ 73
4.
Case Studies of Gap Winds in Howe Sound........................
75
4.1.
Abstract............................................................................................................. 75
4.2.
Introduction....................................................................................................... 76
4.3.
Data and Methods..............................................................................................81
4.4.
Climatology of Outflows/Inflows.......................................................................83
4.4.1. Wind Climatology................................................................................ 84
4.4.2. Wind Speed - Pressure Gradient Correlations........................................87
4.4.3. Air Temperature and Humidity as a Function of Gap Winds..................89
4.5.
Case Studies...................................................................................................... 90
4.5.1. Synoptic Description.............................................................................91
4.5.2. Observational Description.....................................................................93
4.5.3. The Outflow Events in the SAR Images.................................................99
4.5.4. Vertical Observations.......................................................................... 102
4.6.
Conclusion.......................................................................................................107
5.
Conclusions, Further Researches and Recommendations........................................ 109
5.1.
Introduction......................................................................................................109
5.2.
Summary of Results and Conclusions.............................................................. 109
5.3.
Recommendations and Suggesting for Further Research.................................. 112
Bibliography........................................................................................................................................115

List of Tables
Table 2.1 Weather stations employed in the synoptic climatology. Station name, identifier, start
date o f available data, and location information are indicated for each weather station.. 27
Table 2.2 Weather stations employed in the study o f along-channel winds in Howe Sound.
Meteorological stations are provided along with their ID, locations and period of
available d ata.......................................................................................................................... 39

List of Figures
Figure 2.1. A topographic map o f the coast of British Columbia showing different inlets and
channels that dissect it. Channels that have been used in this study are circled and
labelled in r e d ............................................................................................................................7
Figure 2.2. A diagram illustrates the hydraulic effects on creating gap winds and the contribution
o f the synoptic scale pressure gradient (2014). Department o f Atmospheric Sciences,
University o f Washington.......................................................................................................15
Figure 3.1. A topographic map of western BC showing the locations o f the gaps and a
meteorological station (red triangle, circle or square) for each gap. Arrows show
directions o f the wind that a meteorological station can detect as channelized winds
(directions are extracted from wind roses). WEK: Portland Inlet. WME: Burke Channel.
WAS: Howe Sound. WGT: Qualicum. WQK: Juan de Fuca Strait.................................... 48
Figure 3.2. Wind roses and detailed topography for each weather station representing a gap wind.
Note the visual agreement between the topographic conditions o f each gap and the yearly
wind rose observed at the selected meteorological station o f each gap Speeds > 10 m s-1
approximately match Beaufort scale 6 ...................................................................................51
Figure 3.3. Boxplots showing the average time period (hours) o f an event for each type of gap
wind at different meteorological stations. The upper edge o f boxes indicates the 75 th
percentile o f the data set, and the lower edge indicates the 25th percentile. The horizontal
red lines insides boxes represent the median value. The ends of the vertical lines indicate
the 5th and 95th percentile of data values, while plus signs represent outliers. ID-h: highwind-speed events, ID-1: low-wind-speed events. Stations that do not detect events in
each category are not shown.................................................................................................. 52
Figure 3.4. (a) Annual, and (b) diurnal distributions of outflows (blue) and inflows (brown) for
hourly observations (percentages) that meet the defined criteria of gap winds, (L) is the
local standard time (PST)....................................................................................................... 55
Figure 3.5. Climatology of wintertime (a) and summertime (b) for MSLP and 500 hPa heights. 56
Figure 3.6. Composites of wintertime outflow events. MSLP and 500 hPa geopotential height
maps have the standard deviation in light contours, and the anomaly maps use shading to
indicate when the anomaly is statistically significant (p < 0.05) using the student’s t test,
(a) Portland Inlet high-wind-speed, (b) Portland Inlet low-wind-speed, (c) Burke Channel
high-wind-speed, (d) Burke Channel low-wind-speed, (e) Howe Sound, and (f) Juan de
Fuca Strait. Intervals: MSLP composites (4 hPa), MSLP anomalies (2 hPa), 500 hPa
composites (60 m), and 500 hPa anomalies (30 m). The green dots represent the weather
station locations...................................................................................................................... 59
Figure 3.7. Composites o f wintertime inflow events. MSLP and 500 hPa geopotential height
maps have the standard deviation in light contours, and the anomaly maps use shading to
indicate when the anomaly is statistically significant (p < 0.05) using the student’s t test,
(a) Burke Channel, (b) Howe Sound, (c) Qualicum, and (d) Juan de Fuca. Intervals:

MSLP composites (4 hPa), MSLP anomalies (2 hPa), 500 hPa composites (60 m), and
500 hPa anomalies (30 m). The green dots represent the weather station locations
61
Figure 3.8. Composites of summertime outflow events. MSLP and 500 hPa geopotential height
maps have the standard deviation in light contours, and the anomaly maps use shading to
indicate when the anomaly is statistically significant (p < 0.05) using the student’s t test,
(a) Portland Inlet, (b) Burke Channel, and (c) Howe Sound. Intervals: MSLP composites
(4 hPa), MSLP anomalies (2 hPa), 500 hPa composites (60 m), and 500 hPa anomalies
(30 m). The green dots represent the weather station locations...........................................63
Figure 3.9. Composites of summertime inflow events. MSLP and 500 hPa geopotential height
maps have the standard deviation in light contours, and the anomaly maps use shading to
indicate when the anomaly is statistically significant (p < 0.05) using the student’s t test,
(a) Burke Channel, (b) Qualicum, and (c) Juan de Fuca. Intervals: MSLP composites (4
hPa), MSLP anomalies (2 hPa), 500 hPa composites (60 m), and 500 hPa anomalies (30
m). The green dots represent the weather station locations................................................65
Figure 3.10. Surface analysis charts of the synoptic evolution of a summertime inflow Qualicum
wind event. The event started at 0300 LST 11 Sep 2004 and end at 1200 LST 11 Sep
2004. The brown dot identifies Sisters Island weather station (WGT). Intervals: 4 hPa, L
represents a low centre, the curved line with triangles represents a cold front, and the
curved line with half circles represents warm fronts. Meteorological service of
Environment Canada - Pacific Storm Prediction Center.....................................................68
Figure 3.11. Surface analysis charts of the synoptic evolution o f a summertime inflow Qualicum
wind event. The event started at 1300 LST 03 May 1999 and ended at 0400 LST 04 May
1999. The brown dot identifies Sisters Island weather station (WGT). Intervals: 4 hPa, L
represents a low centre. Meteorological service of Environment Canada - Pacific Storm
Prediction C enter....................................................................................................................69
Figure 3.12. Mean and k-means clusters of Burke Channel summertime inflows for MSLP with
the variance of each cluster. Intervals: 4 hPa........................................................................ 71
Figure 3.13. Mean and k-means clusters o f Qualicum summertime inflows for MSLP with the
variance o f each cluster. Intervals: 4 hPa.............................................................................. 72
Figure 4.1. A topographic map o f Howe Sound - Cheakamus Valley showing the locations of the
meteorological stations that are used in this study................................................................80
Figure 4.2. Wind roses of a 3-years period for summer (JJA), winter (DJF) and transition seasons
(TRANS). Concentric rings are frequency in 5% increments. Wind directions are divided
into 16 groups......................................................................................................................... 86
Figure 4.3. Scatterplots showing the relationship between PG and hourly wind speed o f (a)
outflows at Pam Rocks, and (b) inflows at Squamish........................................................ .88
Figure 4.4. Seasonal averages of temperature and dew point at stations along Howe Sound and
Cheakamus Valley as a function of wind speed and direction. Solid lines show
temperature and dashed lines show dew point. Green lines with cross markers show the
overall means, red lines with stars show the inflows and black lines with squares show

the outflows. PM is Pemberton, SQ is Squamish Air Quality, PR is Pam Rocks, and El is
Entrance Island........................................................................................................................ 90
Figure 4.5. (a) January Event surface analysis chart and the associated 500 hPa geopotential
height level on 2010-Jan-06 (0600 UTC), (b) February Event: surface analysis chart and
the associated 500 hPa geopotential height level on 2010-Feb-26 (1200 U T C )............... 93
Figure 4.6. (a) Wind speed during the January Event at Pam Rocks (blue) and Squamish (red),
(b) PG between Pemberton and Pam Rocks during the January Event, (c) Potential
temperature time series during the January Event at Pam Rocks (blue), Squamish (black)
and Pemberton (dashed red), (d) Dew point time series similar to (c). The two vertical
lines represent the start and the end of the event. The spike in dew point temperature at
Pam Rocks at 2000 LST on 5th January indicates a suspicious observation since it was
accompanied with lower wind speed, change in wind direction and higher relative
humidity than the previous and subsequent observations.................................................... 95
Figur 4.7. (a) Wind speed during the February Event at Pam Rocks (blue) and Squamish (red),
(b) PG between Pemberton and Pam Rocks during the February Event, (c) Potential
temperature time series during the February Event at Pam Rocks (blue), Squamish
(black) and Pemberton (dashed red), (d) Dew point time series similar to (c). The two
vertical lines represent the start and the end o f the event.................................................... 97
Figure 4.8. MSLP during the (a) January Event, and the (b) February Event at Pam Rocks
(blue), Squamish Airport (dashed black), Whistler (dashed green), Whistler Mountain
High-level (pink), and Pemberton (dashed orange). The two vertical lines represent the
start and the end of the event................................................................................................. 99
Figure 4.9. SAR images during the (a) January Event, and the (b) February Event. Zoomed in
images of the Howe Sound region are depicted on the left with the larger scene to the
right........................................................................................................................................ 101
Figure 4.10. (a) Wind components at Squamish Airport during the January Event within the first
2 km. (b) The along-channel wind component during the January Event. Areas in the plot
that are blank indicate a lack of reliable profiler data........................................................ 104
Figure 4.11. (a) Wind components at Squamish Airport during the February Event within the
first 2 km. (b) The along-channel wind component during the February Event, (c)
Potential temperature profile during the February Event. Areas in the plot that are blank
indicate a lack o f reliable profiler data................................................................................ 105
Figure 4.12. Vertical potential temperature profiles during the February Event. The red line at the
start o f the event (25/02/2010 - 2100 LST), the blue line 14 hours after the start o f the
event (26/02/2010 - 1100 LST), and the black line at the end o f the event (27/02/2010 0600 LST).............................................................................................................................. 106

Glossary

Abbreviations:
ECM W F - The European Centre for Medium-Range Weather Forecasting. It provides
reanalysis data of the global atmosphere from three projects: ERA-15 (December 1978 February 1994), ERA-40 (September 1957 - August 2002), and ERA-interim (1979 - present)
each with different resolution and time scale.
hPa - Hectopascal. A metric measurement unit of pressure. It is commonly used to measure
atmospheric pressure.
LST - Local Standard Time.
MSLP - Mean Sea Level Pressure. The atmospheric pressure at sea level.
M SLPG - Mean Sea Level Pressure Gradient. The atmospheric pressure gradient at sea level.
NARR - North American Regional Reanalysis. Reanalysis data provided by the National
Oceanic and Atmospheric Administration (NOAA), Boulder, Colorado, USA. It covers the North
American Region with horizontal resolution of 32 km and 45 vertical layers at 8 times per day.
NCEP-NCAR - National Centers for Environmental Prediction - National Center of
Atmospheric Research. Reanalysis data provided by the National Oceanic and Atmospheric
Administration (NOAA), Boulder, Colorado, USA. They cover the globe with resolution o f 2.5°
at 4 times per day.

ix

PCA - Principal Component Analysis. A statistical approach that aims to convert a set of
possibly correlated variables into a set o f linearly uncorrelated variables called principal
components.
PG - Pressure Gradient. The change of the atmospheric pressure over a given distance.
R2 - Coefficient of determination for a linear regression. An indicator from 0 tol that reveals
how strong a linear relationship is between two variables and how close estimated values from a
linear equation will correspond to actual data.
RASS - Radio Acoustic Sounding System. The profiles o f the virtual temperature o f the lower
atmosphere provided by a five-beam boundary layer wind profiler.
SAR —Synthetic Aperture Radar wind satellite images. Wind fields that provide wind data over
the water surface through using SAR sensors to provide active microwave imaging of the Earth’s
surface.
G eneral definitions:
Anticyclone - An area of high atmospheric pressure around which the wind blows clockwise in
the Northern Hemisphere and counterclockwise in the southern Hemisphere. Also called a
“high”.
B arrier je t - An elevated wind maximum on the windward side of a mountain barrier, blowing
parallel to the barrier.
B eaufort scale - An empirical measure of wind speed that relates the wind speed to observed
conditions at sea or on land. It divides the wind speed into 12 scales.

Composite analysis - A method that depends on considering a sufficient number o f atmospheric
events to create a composite which represents the average conditions o f a given event.
Coriolis force - An apparent force observed on any free moving object in a rotating system. On
Earth, this deflective force results from Earth’s rotation and causes moving particles (including
the wind) to deflect to the right in the Northern Hemisphere and to the left in the Southern
Hemisphere.
Cyclone - An area o f low pressure around which the winds blow counterclockwise in the
Northern Hemisphere and clockwise in the southern Hemisphere.
Downslope windstorm - A type of large-amplitude mountain wave that can occur downwind of
mountain barriers resulting in strong, often gusty low-level flow that accelerates down the lee
slopes of a mountain.
Front - The transition zone between two distinct air masses.
Geopotential height - The actual height o f a pressure surface above mean sea level. A
geopotential height observation represents the height o f the pressure surface on which the
observation was taken.
Geostrophic winds - A theoretical horizontal wind blowing in a straight path, parallel to the
isobars or contours, at a constant speed, and assuming there is no friction. It results when
Coriolis force balances exactly the horizontal pressure gradient force.
Inversion layer - A layer in the troposphere where the air temperature increases with height.

Lee wave - or mountain wave - vertical undulation of airstreams on the lee side o f a mountain.
The first wave occurs above the mountain that causes it, with a series of waves o f equal
horizontal wavelength extending downstream.
Mesoscale - The scale of meteorological phenomena that range in size from a few kilometers to
about 100 km. It includes local winds, thunderstorms, and tornados.

Ridge - An elongated area of high atmospheric pressure.
Rotor - A turbulent eddy that forms downwind o f a mountain chain, creating hazardous flying
conditions.
Synoptic climatology - The study o f weather conditions over the synoptic scale that are
associated with particular environmental conditions.
Synoptic scale - The typical weather map scale that shows features such as high- and lowpressure areas and fronts over a distance spanning a continent.
Trough - An elongated area o f low atmospheric pressure.
Venturi effect - The reduction in fluid pressure that results when a fluid flows through a
constricted section of a pipe.
Wind Rose - A diagram that shows the percentage of time that the wind blows from each
direction by speed category, over a given time, at a wind monitoring site.

Acknowledgement
It is hard to find enough words to express my deep thanks to my supervisor Professor Peter
Jackson for his unlimited support, continuous encouragement, and dependable guidance. My
thanks to Mr. Ford Doherty for his help in accessing different types o f data; and to other
supervisory committee members, Prof. Elie Korkmaz and Dr. Stephen Dery. In addition, thanks
to Wilma Huneke and Ayham Rezk for their assistance. My special and warm gratitude to my
caring, and loving ones, my mother and my wife for their constant emotional support, and to all
others who are not mentioned here but supported me during my studies.
This research was made possible by graduate scholarships provided by the University of
Northern British Columbia. Funding was also provided by Damascus University and Prof. Peter
Jackson’s NSERC Discovery Grant.
Finally, I dedicate this thesis to the one who was absent in his body but always present in his
works and love, to the soul of my father.

1. Study Rationale, Objectives and Research Questions
1.1. Introduction
Winds are dynamically affected by mountains, with the topography of the mountain
barrier playing a crucial role in the behaviour of the airflow. However, different conditions of
the lower atmosphere lead to different patterns of winds affected by topographic barriers.
When a statically stable air mass in the lower atmosphere encounters a mountain barrier in
the presence of a horizontal pressure gradient (PG), the air will overcome the mountain
barrier if it has enough kinetic energy, and low enough stability. Such an air mass that flows
over a mountain barrier may lead to the creation of lee waves, rotors or downslope
windstorms. If the air mass is blocked by the barrier, it may detour around it as a barrier jet.
In the case of a barrier that is dissected by mountain passes, inlets and channels, the air may
flow through the barrier. Flows o f this kind are called gap winds (Jackson et al., 2013). Gap
winds are considered as a subset o f a larger category o f channelized winds, which will be
called from now on along-channel winds. Along-channel winds through fjords and channels
that perpendicularly dissect the British Columbia (BC) coastal mountains blow in two
directions parallel to the channel orientation, as outflows: from inland toward the coast, and
as inflows: from the coast inland.
1.2. Rationale of the Study
The importance of studying the along-channel winds arises not just from their hazards
but also from their potential benefits. While, strong but relatively rare, outflow events can be
dramatic, the weaker cases are more common and may still cause significant hazards. An
example o f consequences o f a strong outflow case is one that hit the lower Fraser River
Valley and Howe Sound (Jackson, 1996) between 30 January and 2 February 1989. The

effects include knocked down trees, power outages, temporary cancellation of ferry services,
the highest recorded gas and electricity consumption in Metropolitan Vancouver, as well as
damages to domestic water reservoirs and to boats. Outflow events may be accompanied by
blizzard conditions, low visibility and the accumulation of snow and ice (Jackson and Steyn,
1994a). Strong outflows can be dangerous for marine service in the channels including
fishing and transportation. In addition, they can be hazardous to aircrafts, especially when an
airport is located within the potential path o f the flow, such as Squamish Airport in
northeastern Howe Sound, BC. They can also transport, build-up, or flush-out pollutants.
However, flows have some potential benefits, such as producing wind power. This is the case
in the Columbia Gorge, Oregon, where several wind farms have been constructed (Sharp and
Mass, 2004). Inflows during the summer can make channels attractive for windsurfing sports
such as at Howe Sound near Squamish and the Columbia Gorge. Channelized winds are an
important feature o f the local climatology of channels and fjords. Therefore, studying them is
an important step in identifying the local climatology o f these regions.
The main goal of the current study is to better understand and forecast these alongchannel winds both in previously studied and unstudied regions of the BC coast. It is
motivated by the consequences and benefits o f the along-channel winds discussed above.
The study is conducted on five different flows on the BC coast, three o f which have not been
systematically studied as of yet. These occur at Portland Inlet, Burke Channel, and Qualicum
on southeastern Vancouver Island. In addition, the study considers two other winds that have
been studied intensively before: gap winds in Howe Sound and in the Strait o f Juan de Fuca.
This study also investigates inflows in some o f these gaps; this is the first investigation of
inflows in this region (see Figures 2.1 and 3.1 for locations).

1.3. Objectives and Research Questions
The primary objective of this research is to investigate the synoptic-scale situations that
are associated with strong along-channel winds on the BC coast. Determining these synopticscale patterns may improve the forecasting of this phenomenon. In addition, the study aims
to initiate research on inflows. The case-study method has been carried out for one of the
mentioned gaps to improve our understanding o f these winds. This study will achieve these
goals through the following objectives:
Objective #1 - Determine the synoptic-scale patterns that are associated with
outflows and inflows during different seasons and throughout several gaps.
Objective #2 - Identify other weather phenomena, which do not appear in the
synoptic composites, but are potentially important for some o f the flow types.
Objective #3 - Determine the local climatology o f the along-channel flows in one of
the gaps - Howe Sound.
Objective #4 - Investigate the different characteristics o f gap flows in Howe Sound
through studying the horizontal and vertical structure of the flow using the case-study
method.
The objectives and rationale o f this study result in the following research questions:
Research question #1 - What are the synoptic patterns associated with different
outflows/inflows on the BC coast?

3

Research question #2 - What are the effects of the channelized winds on the
climatology o f Howe Sound-Cheakamus Valley region? Moreover, what are the different
characteristics o f the Howe Sound gap flow?

1.4. Structure of Thesis
This research involves two main components: a synoptic climatology o f along-channel
winds on the BC coast is provided in chapter 3, and a climatological study of along-channel
winds in Howe Sound along with case studies are provided in chapter 4. The synoptic
climatology aims to address objectives 1 and 2 of the current study, and the climatological
study with case studies aim to address objectives 3 and 4. Each o f the two main components
is written in the format of manuscripts that will be submitted to journals. Each manuscript
contains an abstract, an introduction, and a conclusion along with presenting data and
methods involved in each component and providing results with discussions for each section
separately.
Chapter 2 provides a literature review of gap winds and then differentiates it from other
along-channel winds. The literature review is followed by a detailed discussion of all
methods involved in the current research. While chapter 3 and 4 contain the two main
components o f this study, chapter 5 provides a summary o f results and conclusions and
suggestions for further research. The author o f the present thesis is the principal investigator
and the first writer of the main two chapters as well as all other parts of the thesis.

4

2. Literature Review and Methodology
2.1. Literature Review
2.1.1. Gap Winds: Definition and Theory
The current chapter mainly describes gap winds, and then it distinguishes these from
other along-channel winds. Depending on the gap geometry, gaps are classified as short gaps
and long gaps (Stull, 2000), or as level gaps and elevated gaps (Jackson, 2013). In long gaps,
the gap width is much less than the gap length. Long gaps have typical lengths o f 100 km or
more, and the gap width is 2-20% of the gap length. Due to the transit time for air to flow
through long gaps, the Coriolis force must be considered; however it is negligible for short
gaps. An example o f a long gap on the BC coast is the Strait o f Juan de Fuca, which has a
length of about 150 km and a width of 20 - 30 km.
Gap winds can be level gaps, in which the flows are restricted laterally by the
topography, or elevated gaps, where the flow is restricted both laterally and vertically by the
topography. Most channels on the BC coast are level gaps where they are water-covered by
the ocean but considered elevated further inland. Examples of such gaps on the BC coast that
are partially level are: Howe Sound, Jervis Inlet, Toba Inlet, Desolation Sound, Bute Inlet,
Dean Channel, Burke Channel, Douglas Channel and Portland Inlet (Figure 2.1). All these
gaps are fjords dissecting the coastal mountains of BC in a semi-perpendicular direction,
formed by glaciations in past geological eras. They have NE-SW to E-W orientations,
making them function as connection paths for airflows between the interior of BC and the
coast. The Strait of Juan de Fuca is a level gap. It separates southern Vancouver Island from
Washington State, forming the southwestern extension of the mountain gap o f the Fraser
River Valley.

Gap winds are mostly observed during winter, when the contrast in the air masses
between both sides of the barrier is largest. In the cold season, cold, dense, and statically
stratified air forms over the continent and is blocked by the coastal mountain barrier, creating
a region of high surface pressure. In most cases, the blocked air is relatively shallow, with
height less than the mountaintop and is overlaid by warmer air. This situation leads to a
stable layer (an inversion layer) at the interface between the cold layer near the surface and
warmer air aloft (Sharp and Mass, 2004; Koletsis et al., 2009; Lackmann and Overland,
1989). The base of the inversion layer is often located below the mountaintop. Consequently,
the gap flow is constricted vertically from above by the inversion layer, which prevents it
from overcoming the mountaintop, but allows it to blow through mountain gaps toward the
coast. The past description of gap wind formation reveals their three-dimensional nature,
where specific conditions of the airflow’s vertical structure are crucial for the formation of
these flows. Therefore, in addition to the two horizontal dimensions o f the flow, a vertical
dimension is observed in gap flows.

6

Figure 2.1. A topographic map of the coast of British Columbia showing different inlets
and channels that dissect it. Channels that have been used in this study are circled and
labelled in red.
The behaviour of a gap flow is similar to that of water flow in an open channel.
Therefore, the single-layer hydraulic theory in open channels, which is a three-dimensional
nonlinear theory, has been considered in several studies to describe the behaviour o f gap
winds and to simulate it (e.g. Jackson and Steyn, 1994b; Finnigan et al., 1994; Schar and

Smith, 1993; Pan and Smith, 1999; Finnigan et al., 1998; Jackson, 1996). In this theory, the
flow is controlled by hydraulic control points, which can be horizontal contractions or
changes in surface elevation. The hydraulic controls determine the relationship between the
flow depth and the flow discharge (Finnigan et al., 1998). A non-dimensional number called
a Froude number (F) is used to investigate this relationship. The Froude number, in this case
applied to water, relates the single-layer flow velocity u to the surface wave speed c= (gh)

1/2

,

where g is the gravitational acceleration, and h is the fluid depth:

F = u /J g h

(1.1)

When applied to gap winds, two stratified layers should be considered with a cold layer
flowing beneath a thick, warmer layer. For two-layer flows, Armi (1986) defines the
composite Froude number:
G2 = F ,2 + F / - ( 1 - r ) F 2 F22
where

Fn2 = u„2 / (g' hn),

n = 1,2

(1.2)
(1.3)

are the Froude numbers for each layer, g ' - (1 - r) g is the reduced gravity, r = pi / p2, p is
the density, h„ are the individual layer depths, and subscript 1 refers to the upper layer, while
2 to the lower layer (Finnigan et al., 1998; Armi, 1986). In gap winds, when hi »
» I// and the stability Froude number

h2 and u2

0: {F 2 = (u2 - uj)2 / g'h), (h - h/ + h2) which

is the case during outflows, the Froude number is very similar to the single-layer Froude
number with g replaced by g' :
Fj = u2/ ( g ' h2) 1/2.

(1.4)

8

When F/ < 1 the flow velocity is less than the phase velocity and the flow is
subcritical. The flow may become supercritical (Fj > 1) after passing a control point such as a
contraction. In this case, at a control point the flow is critical (F/ = 1). Another transition
from supercritical to subcritical flow occurs through a hydraulic jump. In the hydraulic jump,
the depth o f the flow increases and the speed decreases abruptly because o f energy loss and
increased turbulence (Jackson and Steyen, 1994b; Gohm and Mayr, 2004; Pan and Smith,
1999). The transition from subcritical to supercritical creates a mesoscale, hydrostaticallyinduced pressure gradient. More details will be given in the next section when discussing the
forces behind gap winds. The single-layer hydraulic theory can represent the lower layer of
the outflow wind system.
The shallow water equations of a single layer for a free surface height o f the fluid h(x,y) are
written as (Jackson et al., 2013; Armi, 1986):

(1.5)

( 1.6 )

dh

d(hu)

dt

dx

d(tiv) __

(1.7)

dy

where / i s the Coriolis force, and u,v are the velocity components on the x and y directions,
respectively. For simplicity, the steady-state flow in a channel o f width b(x) over topography
H(x) can be written without friction, external pressure gradient and the Coriolis force as:

( 1.8)

9

In the equation (1.8), the pressure gradient has been partitioned into a component resulting
from the change of the height of the free surface and a component resulting from topographic
changes. In a channel with vertical sidewalls, the continuity equation can be written as
(Jackson et al., 2013):

. du ,

dh

uhdb

dx

dx

b dx

h— +u— =

,,

—

(1.9)

If both internal Froude numbers are high, F 2 and F ? »

1, so u 2 ~ u i , the equations (1.8)

and (1.9) can be re-written to relate derivatives o f the dependent flow variables (u, h) to the
derivatives o f the independent topographic variables (H b ), with the result:
1 du _
u dx

Idh
h dx

_

f

1

1

1 db

( 1. 10)

—

L l - F 2J b d x

. 1 - F 2l h d x

1 db
f—
b
L l- F 2. d x

-1 —F 2J “h d x

V

41

'

cv i . i i )/

This pair of equations governs the flow of two layers with a free surface between them. F 2 is
now just the external Froude number (Armi, 1986).
Jackson and Steyn (1994b) simulate the outflow winds at Howe Sound using a hydraulic
model called hydmod. The model includes a number o f parameters such as along-channel
elevations, channel cross section, variations of the gap wind-layer height, friction and the
external pressure gradient. However, they consider a number o f assumptions such as the flow
being steady in time and gradually varying in space. Hydmod appears to be successful in
simulating spatial variations of gap wind patterns at Howe Sound, showing regions of
subcritical, supercritical flows and hydraulic jumps. Hydraulic physical modeling has also
been used to simulate outflows in Howe Sound by Finnigan et al. (1994 and 1998).

10

Improvements in the geometry o f the physical model of the channel are developed between
these two studies. The physical model appears to be capable o f simulating spatial
complexities of the outflows in Howe Sound.

2.1.2. Forces behind Gap Winds
This section addresses the main forces that produce gap winds and provides observational
results related to each force from several gap wind studies. While early studies explain gap
winds by the funnel effect or “Venturi effect”, researchers later find that the along-channel
pressure gradient is the main force behind gap winds. A classification o f the forces that cause
gap winds is provided in the following sections.
2.I.2.I. Synoptic-scale Pressure Gradient
Specific synoptic-scale features can lead to suitable conditions that create mesoscale gap
winds. Identifying the synoptic features is an essential step to understand the evolution of a
gap wind event. In this section, a description of the synoptic features associated with outflow
events is provided for gap winds in different locations.
Three outflow events in Juan de Fuca Strait have been studied by Overland and Walter
(1981). They explain the different synoptic features that were responsible for the outflow
cases. One o f the events was caused by a sea level continental anticyclone located over the
interior o f BC that was accompanied by an upper-level (500 hPa) trough over southern BC.
Outflows for this event were observed within the Fraser River Valley and the southern Strait
of Georgia out to the Strait o f Juan de Fuca. The boundaries o f the flow were sharp and
defined by large horizontal wind shears and the difference in dew point between the air in the
jet and that outside of it. The pressure gradient was oriented NNE-SSW and pointed from

11

inland toward the coast. In another event, a centre of sea level low pressure was located
offshore o f Washington State and was accompanied by a 500 hPa ridge over the coast. In the
third case, a similar low pressure centre was located west o f Haida Gwaii with a 500 hPa low
situated over the surface low. The last two cases formed coast-parallel isobars and therefore a
channel-parallel pressure gradient that pointed from inland toward the coast creating a flow
blowing out o f Juan de Fuca Strait. Winds were light over the Fraser River Valley and
southern Strait of Georgia in contrast to the first case. Mass et al. (1995) describes an outflow
event blowing out o f the Fraser River Valley gap toward the Strait o f Juan de Fuca and Puget
Sound. The synoptic situation was characterized by a 500 hPa trough located over eastern
Washington State, an 850 hPa trough west of the 500 hPa trough, and a centre o f sea level
high pressure located over southern BC near the international border that extends offshore
along the BC coast. Colle and Mass (2000) studied the synoptic features o f an outflow event
in Juan de Fuca Strait. A sea level centre of low pressure located in the Pacific Northwest is
accompanied by an eastward advancing occluded front. Over the interior of BC, an arctic
high pressure area with low temperatures o f -35°C and sea level pressure o f 1040 hPa
resulted in a pressure difference of 20 hPa between the interior and the coast. At the 500 hPa
level, a ridge was situated along the west coast as a trough moved toward the Aleutians.
In Howe Sound, the synoptic situation o f several outflow events is described by Jackson
and Steyn (1994a), Jackson (1996), and Finnigan et al. (1994). At the surface, a similar
situation was found in all cases that are similar to the first case described earlier in Overland
and Walter (1981). An arctic high pressure advanced from the Yukon Territory toward
southern BC, blocked by the coastal mountains, and an associated arctic front moved across
Howe Sound. Aloft, a trough intensified over interior BC and moved southward toward the

12

coast and a ridge built over the Aleutians, resulting in a north to northeasterly airflow aloft
over the coast.
The synoptic situation of an outflow event in the Columbia Gorge, Oregon, is described
by Sharp (2002). Similar to the previously mentioned studies, an upper-level ridge along the
coast caused a northerly flow aloft and was associated with an anticyclone at the surface over
interior BC and Washington State. The strongest flow was accompanied by the passage o f an
arctic front across the gorge. Other North American west coast studies showed similar
synoptic situations. Gap winds at the Shelikof Strait, Alaska, showed an extension of the
Siberian high over Alaska at the surface level with a shallow centre o f low pressure south of
the Gulf o f Alaska (Lackmann and Overland, 1989). The upper level situation showed a
northerly flow with cold air advection over Alaska. Another study by Liu et al. (2008)
considered gap winds in the Prince William Sound, Alaska. The strong outflow event was
accompanied by a sea level low pressure centre o f 972 hPa located over the Gulf o f Alaska,
causing a pressure difference of 40 hPa between the interior high pressure and the coast.
Most the above-mentioned gap winds are induced by a synoptic scale pressure gradient that
points from the interior o f the continent toward the coast, resulting in outflow gap winds.
This is caused by the damming o f cold air behind the coastal mountains of western North
America and / or the movement of a low pressure area toward the coast. However, similar
synoptic situations have been identified for other regions o f the world.
An outflow event over the Gulf o f Tehuantepec, Mexico is discussed by Steenburgh et al.
(1998). The event was caused by a surge o f cold air from the north into Mexico along the
Sierra Madre, causing low temperatures and high wind speeds. In Japan, the Soya Strait gap
winds are associated with the Okhotsk high, which causes the damming o f cold air to the
13

eastern side o f Hokkaido Island (Shimada and Kawamura, 2011). A similar situation has
been reported for the Tsugaru Strait, Japan (Shimada et al., 2010). The synoptic situation of a
northerly flow through the main mountain gap o f Crete Island was characterized by an upperlevel ridge over the Balkans and a trough over eastern Turkey. At the surface level, the
Turkish thermal low was located over Cyprus and a high pressure appeared over continental
Greece (Koletsis et al., 2009). Several other studies have investigated the synoptic conditions
that accompany particular flows (e.g. Chenoli et al., 2012); such research is called synoptic
climatology studies.
Synoptic climatology is the study o f synoptic weather conditions that are associated with
particular environmental conditions. One way o f studying the synoptic-scale conditions that
are associated with gap winds is using the composite analysis method o f synoptic
climatology. This method considers a sufficient number of events to create a composite that
represents the average conditions of a gap wind event. A similar study has been conducted to
understand the synoptic conditions associated with the Columbia Gorge gap wind (Sharp and
Mass, 2004). Composites o f the Columbia Gorge gap winds showed an amplified upper-level
ridge located along the west coast o f North America and a trough extending to the east o f the
upper-level ridge. This situation led to southward cold air advection, which enhances the
surface-level anticyclone over southern BC. Deviations of different wind speed groups were
calculated from the Columbia Gorge composites. It was found that the larger the magnitude
o f the upper ridge deviation, the stronger the outflow wind speed. The methods used in this
paper will be discussed in detail in section 2.2.1. All of these previous studies indicate the
importance o f the synoptic scale pressure gradient as the main driving force for gap winds.

14

Consequently the synoptic-scale pressure gradient must be investigated to understand gap
winds.
2.I.2.2. Hydrostatically / Thermally Induced Pressure Gradient
A cool air mass forming on one side of a mountain barrier can form a mesoscale pressure
gradient pointing toward the other side o f the barrier, which enhances the synoptic-scale
pressure gradient. In the case o f the coastal mountains of BC, the denser, cold air
accumulates on the eastern side o f the mountain range producing higher pressure at low
levels. After the cold air moves through the mountain gap, it spreads out at the region where
the gap widens near the gap exit. The spreading layer o f the cold air near the gap exit makes
the depth shallower than upstream. Decreases in the depth o f the cold air layer result in lower
surface pressures at the gap exit region. This is a simplified picture of the hydrostatic
pressure effects on creating gap winds. Figure 2.2 illustrates this behaviour.
Figure 2.2. A diagram illustrates
the hydraulic effects on creating
gap winds and the contribution
of the synoptic scale pressure
gradient.
Department o f Atmospheric
Sciences, University o f
Washington (2014).
The earlier-mentioned hydraulic theory explains this behaviour. Once the flow becomes
shallow, a supercritical region is formed and the speeds increase downstream o f the flow.
Upstream of the supercritical region the blocking effect makes the flow deeper creating a
subcritical region where the winds are light. In general, the complexity of the flow increases
as more layers of different density are considered in modelling the flow. In addition,
topographic changes and the effects of friction and rotation increase the complexity (Mayr et
15

al., 2004). The basic hydraulic theory applied to gap winds is analogous with the flow of
water in an open channel; there is only one layer that flows (the lower layer). In reality, the
atmosphere, even in the case o f a lower layer o f cold air, is not a shallow fluid like water in a
channel as it is continuously stratified. The linear theory with continuous stratification has
been shown to fail in representing gap flows (Pan and Smith, 1999); thus the non-linear
hydraulic theory is the most efficient theoretical approach for gap flows that are caused by
the cold air damming effect. However, the hydraulic theory will not be adequate when the
gap flow layer is not separated from the flow aloft (Schar and Smith, 1993).
There are situations when gap flows are not caused by a continental anticyclone. Thermal
contrasts between both sides of the barrier create differences in surface pressure. Several
situations can lead to the thermal contrast. A passing warm (cold) frontal system across the
barrier can bring warm (cold) air to one side of the barrier. A similar situation initiated gap
flows and foehn winds in the Alps, most notably downwind o f the Brenner Pass (Mayr et al.,
2004). Warm air associated with a warm frontal system can raise the temperature north of the
Alps and cause the surface pressure on the southern side to be on order of 5 hPa higher than
on the northern side.
Good examples of gap flows that are induced by thermal contrasts are seen across the
coastal mountain chains o f the Red Sea. The Red Sea is surrounded by desert-lands that
create large cross-shore, diurnal temperature gradients. The most famous gap wind observed
on the Red Sea coast is the Tokar gap wind on the Sudanese coast (Jiang et al., 2009).
Diurnal outflows are observed during nighttime at the Tokar gap because o f the nighttime
radiative cooling over the desert-land. The Santa Ana gap winds in Southern California have
similar conditions of the synoptic forcing and the local thermodynamic forcing (Hughes and
16

Hall, 2010). In the Santa Ana winds, a large temperature gradient between the cold desert
during nighttime and the warm ocean leads to an offshore pressure gradient, which causes
strong shallow offshore flows. This thermal forcing is the main driving force of the Santa
Ana winds. Katabatic winds observed in Antarctica can be significantly affected by the
topography when the flow blows through valleys and passes. At the McMurdo dry valleys,
winds blow mainly in directions parallel to the valley orientation, either up valley as sea
breezes or down valley from the plateau as katabatic winds (Nylen et al., 2004).
During the warm season, a thermal trough can develop over the western United States
and extend northward into Washington State, southern BC and the Pacific Northwest,
causing easterly winds across the Cascades (Brewer et al., 2012). The thermal trough can
move eastward across the Cascades to bring a shift to westerly winds (inflows) and increase
the wind speed and cloud cover causing a drop in temperature, resulting in the end of a warm
period. The westerly winds have moderate speeds and large diurnal variability. They are
usually strongest during the afternoon and evening when the thermal low intensifies (Sharp,
2002). Such changes in the wind speed and direction, which are caused by the changes of the
thermal trough location, have crucial impacts on the wind energy generation and wildfire
behaviour (Brewer et al., 2012).
2.I.2.3. Other Factors Affecting Gap Winds
2.I.2.3.I. Venturi Effect
It was thought in the early works o f gap winds that the Venturi effect (funnel effect) is
the main factor that causes strong flows within gaps (Reed, 1931). The simple explanation of
the Venturi effect is that the constricted regions o f the channel play a role similar to the
Venturi tube by increasing the speed of the flow at the constriction. In other words, the
17

strongest winds occur at the narrowest parts o f the gap. According to the principles o f fluid
dynamics, the pressure must drop in the area where the width of the cross-section decreases
causing the strongest speeds (Sharp and Mass, 2004). The Venturi effect would explain the
flow in the case where the depth o f the gap flow could not change; however a gap flow is
more like the flow in an open channel, in which the depth of flow is high in the subcritical
region and it decreases toward the critical and supercritical regions. Many observational
studies o f gap flows find that the strongest wind speeds are observed at the gap exit region,
not at the narrowest section of the gap. This indicates that the main driving force o f gap
flows is the acceleration down the pressure gradient (Overland and Walter, 1981; Mass et al.,
1995; Colie and Mass, 2000; Sharp and Mass, 2004; Koletsis et al., 2009).
2.I.2.3.2. The Coriolis Force
The Coriolis force directs moving objects to the right (left) o f their motion in the
Northern (Southern) Hemisphere. With the absence o f surface friction and acceleration,
winds away from the equator are affected by two main forces: the pressure gradient force and
the Coriolis force. In the case when the pressure gradient force is balanced by the Coriolis
force, the winds are in a geostrophic balance blowing parallel to isobars and are called
geostrophic winds. In gap winds, however, the flow is restricted from turning due to the
Coriolis force by the topography and friction is not negligible. Therefore, the winds cannot
be in a geostrophic balance. The Rossby number (Ro = U/fL; U is the wind speed, f is the
Coriolis parameter, L is the length scale) is the ratio of inertial to Coriolis forces and defines
conditions when the Coriolis force is important (Ro < 1) or can be neglected (Ro > 1). Ro
equals 1 with winds of 10 m s '1, f of 10'4 s'1 and lengths of 100 km, therefore the Coriolis
force can be neglected when the gap length in less than about 100 km long (Jackson et al.,

18

2013). Steenburgh et al. (1998) show the importance o f the Coriolis force on the behaviour
o f the outflows over the Gulf of Tehuantepec, Mexico. Similar results are found for a gap
wind event in Shelikof Strait, Alaska (Lackmann and Overland, 1989) where the time
required for an air parcel to transit the strait is almost double the time scale for the
geostrophic adjustment at the latitude of Shelikof Strait, which gives the Coriolis force
sufficient time to divert the flow. However, in the cases where the time scale of the
geostrophic adjustment is larger than the time required for air parcels to transit the gap, the
Coriolis force would be negligible.
2.I.2.3.3. Friction
When friction (drag) is added to the net force balance that affects winds, a cross-isobaric
component of the wind will appear. The influence o f the surface friction depends mainly on
the roughness of the surface and on the vertical stability o f the gap flow layer. More surface
roughness leads to a larger cross-isobaric component and vice versa. On the other hand, more
stable air leads to surface drag that is confined to a shallower layer which consequently slows
the winds. The friction effect can be ignored above the boundary layer when winds are in a
geostrophic balance (Lackmann, 2012). In an outflow event that struck northwest
Washington State (Mass et al., 1995), the frictional drag approximately balances the pressure
gradient force in the lee o f the Cascade Mountains resulting in minimal acceleration o f the
flow. The role o f surface friction explains the stronger winds recorded over the water surface
than over the land.
Through considering the previous forces, the momentum equation of gap flows can be
written as follows (Colie and Mass, 2000):

19

^ = --A P +F r-fV
dt
p
J
A

B

( 1. 12)

C D

where term A is the total velocity vector, term B is the pressure gradient (p is density of air),
and terms C and D represent the frictional and the Coriolis forces, respectively.

2.L2.3.4. Inertia and Entrainment of Air from Aloft
Other controlling factors considered important for gap flows are inertia and entrainment
of air from the layer above the gap flow. They are both retarding forces similar to the surface
drag force. Inertia plays crucial role in determining whether the air mass would overcome the
barrier or not, and it can balance, along with friction, the pressure gradient force (Jackson et
al., 2013). Mass et al. (1995) use a three-way balance to explain the down-flow acceleration
within the Fraser River Valley gap. They include the pressure gradient along the gap Px, the
initial wind speed w(0), and the friction coefficient K, which is a function o f surface
roughness, stability, and boundary layer depth:

u2

(1.13)

where u(x) is the wind speed some distance x from the starting point along the gap. Applying
this three-way balance, which includes the pressure gradient force, friction and inertia, gave
results consistent with observations. Entrainment is the vertical mixing of momentum into
the boundary layer (Stevens and Duan, 2002). Lackmann and Overland (1989) have
conducted work similar to Mass et al. (1995), but they include the entrainment. They find
that the acceleration o f gap winds in Shelikof Strait can be explained by the balance between
the pressure gradient, surface friction and the entrainment in addition to the Coriolis
acceleration. The entrainment of air from aloft into the boundary layer is the largest retarding
20

force. After explaining the different factors that affect gap flows, it is important to illustrate
the similarities with other types of dynamically driven winds.

2.1.3. Gap winds, Wakes and Downslope Windstorms
As mentioned earlier, gap winds are dynamically driven flows produced by the
interaction o f specific orographic conditions and particular atmospheric states. However,
some circumstances o f topography and atmosphere can lead to the creation o f gap winds
along with other dynamically driven flows. Most notably, gap winds can be observed with
mountain wakes (Pan and Smith, 1999), or with downslope windstorms (Mayr et al., 2004).
By definition, wakes accompany gap winds, as they are formed in the lee o f high
mountains as slower air when a gap flow occurs downstream o f the mountain. Pan and Smith
(1999) explore the hydraulic features o f wakes and gap winds that form in the region of
Unimak Island, in Alaska’s Aleutian Island chain, by comparing the results from a shallow
water model with Synthetic Aperture Radar (SAR) wind images. The SAR images show, in
four cases, wake patterns that formed in the lee of several mountains with gap flows blowing
through two different gaps. One of the important hydraulic features found in this paper that
relates wakes formation to the gap wind behaviour, is that strong, supercritical flows over the
mountain peaks lead to strong hydraulic jumps and weak flows in the wake region. However,
subcritical flows cause strong flows in the wake region with the absence of hydraulic jumps.
Downslope windstorms are observed down the lee slopes of the mountains resulting in
strong, gusty, shallow winds. They are caused by a large amplitude mountain wave that can
occur when there is a stable layer near the mountain top (Jackson et al., 2013). Downslope
windstorms can accompany gap flows. In this case, the inversion layer above the gap flow is
21

located above the mountain crest; consequently, the gap flow is very deep and not limited to
the gap topography. In such a situation, a downslope windstorm is superimposed on the gap
flow. This can be observed in elevated gaps such as the Brenner Pass in the Alps (Mayr et
al., 2004). In this elevated gap, when a downslope windstorm accompanies a gap flow, it is
called a deep foehn. However, a “pure” gap flow has been referred to as a shallow foehn. In
this case the stable layer is above the altitude o f the mountain pass, but below the altitude of
the mountain peaks. The hybrid gap flow-downslope windstorm in the Brenner Pass - Wipp
valley is studied through observations and model simulations by Flamant et al. (2002). The
gap flow was accompanied by a mountain wave at the entrance o f the gap (Brenner Pass),
and downstream o f the gap a downslope windstorm was observed and extended toward the
city of Innsbruck, Austria.
2.1.4. Gap Winds and Other Along-channel Winds
The previous discussions have placed gap flows into context by providing theoretical
background, discussing the main forces that cause them, and illustrating important results
from previous studies. As discussed, gap winds are caused mainly by the along-channel
pressure gradient, in which isobars are oriented perpendicular or semi-perpendicular to the
channel orientation. Therefore, when channelized winds are not caused mainly by the alongchannel pressure gradient they are likely not considered to be gap winds.
Gap winds are characterized by a lower layer separated from the air aloft by a stable
layer. The presence of both a stable layer near the mountain top and an along-channel
pressure gradient distinguishes gap winds from other channelized flows. Therefore, a
channelized flow that does not show one o f the past two characteristics cannot be considered

22

a gap flow. In the current research, several along-channel winds at the BC coast are
examined by focusing on specific aspects o f these winds. In particular, the synoptic-scale
pressure gradient is investigated and determined. In addition, case studies are conducted
utilizing a period of intensive observations.

2.2. Methodology
2.2.1. Review of the Methods Used in Gap Wind Studies
Gap flows have been studied through theoretical works, observations, and simulations.
Observational analyses have been conducted using different sources of data. In situ
observations from weather stations have been used to characterize surface flows. A weather
(meteorological) station may contain a wind monitor, which should be mounted at a 10 m
height above ground level in an unobstructed area for wind studies, a barometer, temperature
and humidity sensors, and a rain gauge. The wind monitor gives the wind observations and
averages at particular temporal intervals. A weather station can have different or similar
observation and averaging periods. Another wind monitor might be mounted at the same
weather station but at different height. The barometer gives surface pressure measurements,
while the temperature and humidity sensors and the rain gauge provide the necessary
information about the weather that accompanies gap flows. Such weather stations have been
used in many gap wind studies (e.g. Koletsis et al., 2009; Sharp and Mass, 2004; de Foy et
al., 2006; Jackson and Steyn, 1994a). Satellite observations have also been utilized for
recording the surface wind measurements using the Quick Scatterometer data or
“QuikSCAT” (Shimada et al., 2010; Shimada and Kawamura, 2011; Koletsis et al., 2009).
SAR wind images are also used for observations over water surfaces (Liu et al., 2008).
Reanalysis data are utilized for determining the synoptic conditions o f the flows using the
23

North American Regional Reanalysis (NARR) data (Hughes and Hall, 2010) or the National
Centers for Environmental Prediction - National Centre of Atmospheric Research (NCEPNCAR) data (Sharp and Mass, 2004). The vertical structure o f the flows have been observed
using aircraft measurements (Overland and Walter, 1981; Lackmann and Overland, 1989), or
by vertical soundings (Jackson and Steyn, 1994a). To simulate gap flows, shallow water
models have been mostly applied (Pan and Smith, 1999; Jackson and Steyn, 1994b; Ghom
and Mayr, 2004; Schar and Smith, 1993). Physical hydraulic models are created to simulate
gap winds in Howe Sound (Finnigan et al., 1994; Finnigan et al., 1998). Jackson and Steyn
(1994a) develop a 3-D numerical model to simulate the same flows.
In observational studies, wind roses have been created for locations along gaps to
investigate the wind climatology (Sharp and Mass, 2004; Jackson and Steyn 1994a).
Correlations between the surface pressure gradient along the gap and the wind speed have
been examined, while temperature and humidity gradients have been investigated during gap
flow events (Colle and Mass, 2000; Koletsis et al., 2009; Sharp, 2002; Sharp and Mass,
2004). The synoptic conditions that accompanied gap flow events are discussed in most of
the gap wind studies for both the sea level pressure and the 500 hPa geopotential height
(Overland and Walter, 1981; Mass and Albright, 1985; Lackmann and Overland, 1989;
Jackson and Steyn, 1994a; Finnigan et al., 1994; Mass et al., 1995; Colle and Mass, 2000;
Sharp, 2002; Flamant et al., 2002; de Foy et al., 2006).
In the current study, two o f the objectives aim to investigate the synoptic-scale
climatology o f several along-channel winds on the BC coast. As mentioned in section
2.1.2.1, Sharp and Mass (2004) conducted a synoptic composite analysis for gap winds in the
Columbia Gorge, Oregon. They created composites for sea level and 500 hPa geopotential
24

height by averaging daily NCEP data for all days where flow events occur, and then they
calculated deviations of composites from the climatology. The composites were created for
different wind speed groups. However, synoptic studies are also conducted for wind
phenomena other than gap flows (e.g. Chenoli et al., 2012; Mercer et al., 2012; Moore and
Renfrew, 2005; Harden et al., 2011; Goyette, 2010). For example, a synoptic composite was
built for severe wind events at McMurdo station, Antarctica (Chenoli et al., 2012), following
procedures similar to Sharp and Mass (2004). As another example, synoptic composites have
been made for tomadic and nontomadic outbreaks within the United States (Mercer et al.,
2012). In another case, satellite data were used instead o f the surface-based observations to
determine barrier wind events around Greenland (Moore and Renfrew, 2005) and then
synoptic averages were created depending on six years of available satellite data. Harden et
al. (2011) used reanalysis data from the European Centre for Medium-Range Weather
Forecasting (ECMWF) to obtain the barrier wind events around southern Greenland for 20
years, and then to create synoptic composites.
The long history of the synoptic system studies start with the classical work, which
depends on manual analysis. Synoptic climatology studies usually relate atmospheric
circulations to the surface environment (Halios et al., 2012). Two approaches have been used
in synoptic climatology studies (Yamal, 1993), environment-to-circulation and circulationto-environment. The first approach assesses the environmental data first and then sets criteria
to build the synoptic classes based on the environmental data. The second approach classifies
the synoptic patterns first and then seeks a relationship between the synoptic patterns and
environmental data that fit those patterns. An example o f the first approach is Murphy et al.
(2004). They studied synoptic patterns that are associated with the emergence and flight of

25

the mountain pine beetle in B.C. They determined the possible days o f flights and then they
built a synoptic composite for the flight days. An example o f the second approach is
employed in Stahl et al. (2006), where a number of synoptic-scale circulation types are
classified first, and then the types are correlated to specific patterns of precipitation and
temperature anomalies across BC. The first approach (environment-to-circulation) has been
used in the present study by determining criteria from the observational data, particularly the
wind speed and direction, and then constructing the associated synoptic classes. This
approach is explained in detail in the following section.
2.2.2. Methods Employed in the Study of the Synoptic-scale Climatology of Strong
Along-channel Winds on the Coast of British Columbia
2.2.2.I. Selection of the Observations
Investigating the synoptic conditions o f the along-channel winds requires determining a
sufficient number of flow events to produce a reliable composite of the events. At least one
weather station that is suitably located to detect the along-channel winds in both directions is
a minimum for each gap. Ideally, the location o f the weather station would be near the
oceanic exit o f the channel for outflows, and inland near the channel entrance for inflows. A
water-based or a near-water location is preferable since the surface friction is less over water
surfaces. Therefore, a central, clear location within the flooded part o f the channel would
also be effective for measuring the flow in both directions. Table 2.1 lists the selected
weather stations for each gap.

26

Table 2.1. Weather stations employed in the synoptic climatology. Station name, identifier,
start date o f available data, location information are indicated for each weather station, and
percentage o f wind speed > 11.3 m s'1 (refer to Figures 2.1 and 3.1 for locations). * Note that
at Sisters Island the number between brackets represents Qualicums, and the other number
represents all winds over 11.3 m s'1 that are mostly blow in directions parallel to the
orenation o f the Strait o f Georgia and thus not considred as Qualicum Winds.

Gap

Longitude

Elev.

(°N)

(°W)

(m)

Percentage
o f wind
speed
observations
> Beaufort
scale 6 (11.3
m s’1 (%)

54.58

130.70

8.2

33.39

52.19

127.47

31.0

16.42

49.49

123.30

4.9

11.71

48.30

123.53

3.0

23.85

49.49

124.43

20

20.36 (01.23)

Station

ID

Start

Location

Name

(Enviro

date

of Latitude

-nment

available

Canada

data

)
Portland Inlet

Grey

WEK

1994

Islet
Burke

Cathedr

Channel

-al Point

Howe Sound

Pam

WME

Race

February
1994

WAS

February
1994

Rocks
Juan de Fuca

February

WQK

February
1994

Rocks
CS
Qualicum*

Sisters
Island

WGT

March
1995

In Portland Inlet, the location of Grey Islet near the oceanic exit of the inlet, allows
detection o f outflows but not inflows, especially since Grey Islet is located on the inland side
o f Dundas Island. Unfortunately, there is no weather station that can detect potential inflows
in Portland Inlet. In Burke Channel, Cathedral Point records winds in both directions in the
27

channel as it has a clear, central location. In Howe Sound, Pam Rocks detects outflows but it
is not an ideal location for inflows since it is close to the mouth of the channel. Another
weather station that can detect inflows better than Pam Rocks is located at Squamish, BC.
This station does not have sufficient observations to create composites; therefore it is not
considered in the synoptic analysis. This will be discussed in more detail in the next section.
In the Strait of Juan de Fuca, Race Rocks is selected as it observes both inflows and outflows
similar to Cathedral Point in Burke Channel. The Qualicum Wind is a special case, in which
gap winds blow in only one direction (from the southwest) and are observed mainly over the
Strait of Georgia south o f Hornby Island and over Home Lake inland. The only available
station for detecting the Qualicum Wind is Sisters Island (refer to figures 3.1 and 3.2 for
locations; and table 2.1 for available observations).
All weather stations are automated and operated by Environment Canada. They provide
in situ hourly observations with available data starting from 1994/1995. The observations in
some of the selected stations include: air temperature, humidity, dew point, air pressure and
tendency, precipitation amount and wind speed and direction at 10 m height. However, not
all o f these observations are available in some stations or for specific periods. For selecting
wind events, observations o f wind speed and direction are only considered. The next step is
to verify the ability o f these weather stations to detect the flows.
2.2.2.2. Investigation of the Wind Behaviour and Selection of the Events
Selecting wind events is a main procedure in creating synoptic composites. A wind event
is determined depending on criteria for wind direction, wind speed and seasonality. A
number o f successive observations are considered as an event if winds blow from a particular

28

direction and over a predefined wind speed threshold. Yearly and seasonal wind roses have
been created for each of the selected weather stations. A wind rose gives information about
the distribution of wind speeds by direction at a particular location (Sharp and Mass, 2004).
This includes the prevalent wind directions in different seasons and the wind percentage of
each direction. Therefore, it shows whether the station can detect outflow and/or inflow
winds. Creating these wind roses is the first step in setting the criteria to define the wind
events and dividing them into different groups depending on season, direction and speed.
As the synoptic conditions vary considerably between the seasons, partitioning between
the warm and cold seasons is a crucial step in identifying different synoptic situations
throughout the year. In this study, the warm season (henceforth referred to as summertime) is
considered as the months o f May through September, and the cold season (henceforth
referred to as wintertime) as the months of November through March. The monthly
composites between the months o f November and March are found to be similar, as are the
composites between May and September. The months of April and October are ignored for
both seasons since they have characteristics o f both warm and cold events. Consequently,
including them in either o f the seasons will add variance to the composites. The direction is
determined mainly by inspecting the wind roses, as inflows have winds that blow inland and
vice versa for outflows. The wind roses allow the accurate determination o f the range of
directions that correspond to each flow type in each gap. As an example, the outflows in the
Burke channel blow from between 45°N - 70°N, and the inflows blow from between 210° 240°. A wind speed threshold is selected to define wind events. The selected threshold
should be low enough so that there is a sufficient number of events as well as high enough so
that there is not too much variance in the composite pattern. The Beaufort scale 6, which is

29

11.3 m s '1, is found to be suitable to meet both needs. The same threshold is selected in other
similar wind studies (e.g. Chenoli et al., 2012). The second threshold necessary to define an
event is duration. As in the wind speed threshold, the duration threshold is set to balance
between the number of events and producing a low variance composite. After investigating
the observations, the length o f four successive hours is selected as a threshold time.
Therefore, any consecutive four hours that meet the wind speed and direction criteria are
considered as an event. Consequently, applying the previous criteria produced four different
categories of strong along-channel winds: wintertime outflows, summertime outflows,
wintertime inflows, and summertime inflows. Every category has a separate composite in
each gap.
The number of events that need to be included to achieve a reliable composite
depends on the variance o f the events, and on the available observations that meet the event
criteria. If the events result in a composite with large variance, then more events are needed
for a reliable estimate o f the composite. A trial-and-error procedure is applied on a
progressively increasing number of events. It is found that between 50 and 75 events is
sufficient to build reliable composites. Very strong outflow events are observed in both
Portland Inlet and Burke Channel, with wind speeds in some events reaching the Beaufort
Violent Storm speed category for at least one hour (>28.6 m s '1). Therefore, another
partitioning is considered to separate the high-speed wintertime outflow events as a separate
group for those two channels. An event is classified in this group when the maximum wind
speed exceeded the Beaufort scale 9, which is 20 m s '1. The next step is to construct the
synoptic composite for each flow type.

30

2.2.2.3. Constructing the Composites
NCEP-NCAR data are used to build the synoptic composites (Kalnay et al., 1996).
NCEP-NCAR reanalysis data are globe-covering data that have a spatial resolution o f 2.5°
latitude by 2.5° longitude. These data are used for constructing similar composites in
previous studies (Sharp and Mass, 2004; Chenoli et al., 2012), and are found to be suitable
for producing the targeted composites of this study. The data are available every six hours
and also as daily averages. They are produced using a numerical model that assimilates
observations from many different sources. NCEP-NCAR reanalysis data are extracted from
the archives provided by the National Oceanic and Atmospheric Administration / Oceanic
Atmospheric Research / Earth System Research Laboratory - Physical Sciences Division
(NOAA/OAR/ESRL PSD), Boulder, Colorado, USA, for the period o f the available wind
observations (1994-2012), and for both the mean sea level pressure and 500 hPa geopotential
heights. The synoptic domain follows that o f work by Stahl et al. (2006) who also built
synoptic composites of BC for different purposes. The domain coordinates are 40 - 62.5°N
and 1 1 0 - 157.5°W.
A climatology is created for the defined wintertime and summertime seasons by
averaging all days o f each season for the study period. Then, a composite is created for each
flow type in each gap for mean sea level pressure and 500 hPa height. Composites are
constructed for times corresponding to the events by averaging the data from the NCEPNCAR gridpoint dataset o f the selected synoptic domain for all times coinciding with the
events. Wintertime outflow events have an average duration ranging between 37 and 132
hours. Therefore, the daily average of NCEP-NCAR data are used for wintertime outflows.
An NCEP-NCAR day should observe 13 hours o f the event or more to be included in the
31

composite. For example, if an event starts at 10 am January 1st and ends at 9 am January 3rd:
January 1st is included in the composite as it observes 14 hours o f the event (from 10 am to
00 am), January 2nd is also included as the whole day observes the event; however the last
day (January 3rd) observes 9 hours o f the event (from 00 am until 9 am) thus it is not
included in the composite. For all other flow types, the 6-hourly NCEP-NCAR data are used.
Similar to the NCEP-NCAR daily averages, the 6-hourly time step should observe four hours
at least of the event to be considered in the composite.
An anomaly is calculated for each composite by subtracting the composite synoptic map
from the climatology over the summer or winter period. Anomalies are useful since they can
explain some of the synoptic features that are not obvious in the composites. Standard
deviations of the composites are calculated and included in the composite maps to show the
variability o f the composite pattern. Finally, the statistical significance of the anomalies is
calculated using the Student’s t test at the 95% significance level (Wilks, 2011) and included
in the anomaly maps as shaded areas.
2.2.2.4. K-means Clustering of the Composites
Events of some flow types have large variability. As composites are averages of the
events, large variability resulting from the presence of several synoptic patterns may obscure
the interpretation. Therefore, another automated approach has been considered to determine
whether the variability within some o f the composites can be reduced by further partitioning
the composites into clusters. This is particularly an issue for the summertime inflows, which
have composites with large variance.

32

A non-hierarchical clustering approach called k-means clustering is applied. K-means
clustering is an unsupervised technique (data do not originate from clearly defined groups or
prior groups) that aims to find the natural grouping of a set of patterns. Therefore, the n
patterns or “data points” are divided into k groups or “clusters” based on a measure of
similarity (Jain, 2010). The similarity between the patterns within the same cluster is the
highest, and it is lowest between patterns in different clusters. In other words, the approach
minimizes the variance within each cluster and maximizes it between different clusters. Kmeans clustering has been applied in many synoptic studies (e.g. Jiang et al., 2012, Jiang,
2011, Stahl et al., 2006 and Burlando, 2009). In most of these studies, k-means clustering is
preceded by the application of principal component analysis (PCA), which helps identify an
initial set o f the synoptic types. In these studies, the data set contains a large number o f
variables, which usually includes all days for several decades. Therefore, applying PCA is
useful to reduce noise in the data as well as to obtain the initial types.
Determining the number o f clusters has been always considered a difficult problem in
this type o f unsupervised clustering. This issue becomes more complex with an increasing
number of data points (Jain, 2010). The number of clusters must be determined prior to
performing the k-mean cluster analysis. The ideal target of the k-means clustering would be
to reach to the best homogeneity within each group by increasing the number o f clusters until
the within-cluster variance does not decrease much. At the same time, the clusters should be
limited to a reasonable number. Therefore, a balance between the cluster homogeneity and
the number of clusters should be achieved. Stahl et al. (2006) in their classifications of the
daily synoptic circulation types over BC, accomplished the previous balance by considering

33

a local minimum between the number of clusters and the mean distance from the cluster
centroid. The local minimum in their study occurred at 13 clusters.
K-means analysis starts by placing a centroid for each cluster randomly, so all data will
be assigned for the nearest centroid. The centroid positions are then re-estimated and located
in the centre o f the cluster they belong to (Kassomenos et al., 2010). This procedure is
repeated until the centroids are fixed. The initial placement o f the centroids affects the
production of the final clusters; therefore, the procedure should be repeated several times
(Wilks, 2011). The similarity (or the distance) between the clusters is measured using the
Squared Euclidean distance method (Jiang, 2011). This method considers each centroid as a
mean o f the cluster.
In the present study, k-means clustering is performed on each flow type that has about 70
events (data points). All mean sea level pressure values at each NCEP-NCAR grid point are
put into one vector for each wind case and this is used as a variable. In fact, the small number
of variables limits the use of many automated methods such as the method applied by Stahl
et al. (2006), who use nearly 20,000 variables in their analysis. Therefore, a manual
classification is applied initially by looking at each event alone, which provides an
opportunity to investigate the potential number o f clusters for each flow type. Although
manual classification is more subjective, may result in bias, and requires an experienced
classifier, it is still considered as an initial step for determining the number o f clusters. The
manual verification is combined and followed by a trial-and-error procedure (Kassomenos et
al., 2010), until the most efficient number o f clusters is reached.

34

2 2 2 .5. Frontal Investigation
In addition to the automated clustering analysis, a manual investigation is conducted to
detect potential weather phenomena that might be associated with some of the flow types,
particularly, for composites that do not show important differences from the climatology. In
the description provided by Lange (1998) about winds on the southern BC coast, inflow
winds can be initiated by the passage of a frontal system. Therefore, fronts are investigated
for summertime inflow winds in Burke Channel, Qualicum, and Juan de Fuca Strait, using
the 6-hourly hand-drawn surface analysis charts produced at the Meteorological Service of
Environment Canada’s Pacific Storm Prediction Centre. The charts are available for 13
summer seasons from 1999 to 2011. The investigation is conducted for 104 cases from these
three channels.
The manual investigation considers weather maps that occur at the time o f the events.
The investigation aims to detect fronts, their location, type, orientation, and frontal
movement, in addition to any other potentially important synoptic condition, such as the
presence o f low and/or high-pressure systems. This analysis is followed by examples of a
synoptic evolution of Qualicum wind events. The examples aim to describe the evolution of
synoptic conditions associated with inflow case studies.
2.2.3. Methods Employed in the Study of Strong Along-channel Winds in Howe
Sound
The third and fourth objectives of this research are achieved by detailed study o f gap
winds in a selected gap along the BC coast. This study is conducted both climatologically
and through the case study method. Howe Sound - Cheakamus Valley is chosen as a
35

meteorologically well-instrumented gap on the BC coast. Objectives 3 and 4 are met through
applying the following procedures.

2.2.3.I. Data Description
Conducting an observational study of gap winds requires both a well-distributed surfacemeteorological network and vertical observations of the lower atmosphere within a suitable
channel. An enhanced set of observations that are made for the Vancouver 2010 Winter
Olympics provided the opportunity for an observational gap wind study in the Howe Sound Cheakamus valley. However, the surface weather stations are placed mainly at three
locations and most o f them are outside the main gap (Joe et al., 2014). The locations include
Cypress Mountain north of Vancouver, Whistler Mountain, and Callaghan Valley west of
Whistler (Figure 4.1). The wind profiler, which records the vertical wind in the lower
atmosphere, is installed within the gap at Squamish Airport. These observational platforms
are in operation between nine months to three years during and prior to the Olympics and
Paralympics, and they are available either hourly (such as the surface pressure), or every 15
minutes. Other surface observations in Howe Sound - Cheakamus Valley, available prior to
and after the Olympics, are well distributed along the valley and are available for longer
periods. These observations include weather stations operated by Environment Canada or the
BC Ministry o f Environment. Since they are available for longer periods, they are used to
carry out the climatological study, hence to meet the third objective o f this research.
All the in situ observations used are provided by automated weather stations. A typical
station, whether it is installed during the Olympic period or before it, includes a wind
monitor at 10 m above ground level height, a barometer, air temperature and humidity

36

sensors, a rain gauge and snow depth sensor. However, most of the stations contain only
some o f these instruments. The hourly surface pressure observations are extrapolated to sea
level pressure ( P Ms l ) t0 remove the influence of the station elevation using the sea level
reduction equations provided by Stull (2000):

Pun = P * n .« tp ( f j f r )

(1.14)

where: Pstnis the reported station pressure, Zstn is the station elevation, a = R J |g| = 29.3 m
K"1, Rd is the gas constant for dry air, g is the average gravitational acceleration on Earth at
sea level. T*v is the average air virtual temperature in Kelvins:
T \ = 0.5 * [Tv(t0) + Tv(t0- 12h) + Tsa* Zstn]

(1.15)

where to is the time of the observations at the weather stations, and Ysa = 0.0065 K m '1 is the
standard-atmosphere moist adiabatic lapse rate for the troposphere.
The wind profiler (refer to Figure 4.1 for location) data are provided by a five-beam
boundary layer wind profile (915 MHz by Vaisala), and available half-hourly for the wind
data and hourly for RASS (Radio Acoustic Sounding System) profiles o f virtual temperature
(Joe et al., 2014; Isaac et al., 2014; Gultepe et al., 2014; and Joe et al., 2010). The surface
observations are extracted from the website o f the Historical Climate Data Archive of
Environment Canada. Observing stations that are run by the BC Ministry of Environment are
extracted from the website o f the BC Air Data Archive. The wind profile data are provided
by Environment Canada through personal communication (Ford Doherty, George Isaac and
Ivan Heckman). All the available surface observations are investigated to select useful
locations for conducting this study. Thus, a number of weather stations are selected

37

depending on the location o f the station and the available observations. The weather stations
employed along with their period of available data are provided in Table 2.2.
The hand-drawn surface analysis charts (mentioned earlier) are used to depict the sea
level synoptic situation for the case studies, and the NCEP-NCAR reanalysis data are used
for illustrating the synoptic conditions at the 500 hPa geopotential height. SAR wind images
are used to depict the horizontal distribution of the flow over the water surface in the two
case studies. SAR wind images are high-resolution wind fields that can provide wind data
over the water surface. They have been developed from the Canadian RADARSAT Satellite
data, which uses SAR sensors to provide active microwave imaging o f Earth's surface at a
frequency o f 5.3 GHz in the C-band. The satellites are in sun-synchronous orbits at 798 km
above the Earth’s surface at periods o f 100 minutes (Chan et al., 2010). SAR sensors can
image the water surface through clouds and precipitation.

38

Table 2.2. Weather stations employed in the study o f along-channel winds in Howe Sound.
Weather stations are provided along with their ID, locations and period of available data.

Data source

Meteorological Station
(ID)

Location
Latidude
(°N)

Period of
Available
Longitude Elevation data
above
(°W)
sea level
(m)

Pemberton
(WGP)

Airport 50.30

122.74

204.3

19 8 8 present

Pemberton
Wind (WPN)

Airport-

50.30

122.74

203.0

2007present

Squamish
(WSK)

Airport 49.78

123.16

52.1

1982 present

Pam Rocks (WAS)

49.49

123.30

4.9

19 9 4 present

Point Atkinson (WSB)

49.33

123.26

14.0

19 9 6 present

Entrance Island (WEL)

49.21

123.81

5.0

19 9 4 present

49.70

123.15

20.0

2009
(November) present

Ferry 49.43

123.46

15.0

2005present

Whistler (VOC)
Monitoring
Network o f the
Vancouver 2010
Mountain
Winter Olympics Whistler
High Level (VOA)

50.13

122.95

659.0

2005present

50.08

122.95

1640.0

2 0 0 5 -2 0 1 0
(May)

Whistler
Mountain
High Level - Remote
Wind (VOH)

50.07

122.95

1690.0

20 0 5 -2 0 1 0
(July)

Wind Profile / RASS

49.78

123.16

52.0

2 0 0 7 -2 0 1 0
(March)

Environment
Canada

BC Ministry of Squamish
Environment
Langdale
Terminal_60

39

2.2.3.2. Climatological Study
The climatological study that is conducted within Howe Sound - Cheakamus Valley aims
to illustrate and quantify the climatological gradients that exist within and near the gap. The
effects of the inflows/outflows on these gradients are explored, focusing on winds, pressure
gradients and air temperature/humidity gradients.
A wind climatology is conducted by creating seasonal wind roses at different locations,
showing the wind behaviour within the gap and the surrounding regions, and illustrating the
channeling effect o f the gap. Most o f the selected locations do not have more than three years
of available wind observations. A comparison between a 3-year wind rose and a 10-year one
shows similar results at three locations that have long time series o f available observations.
Therefore, a 3-year wind rose is constructed for all locations since this is the available period
of data for some stations.
The sea level pressure gradients are examined along the gap and outside it as a function
o f wind speed and direction. Three weather stations are selected along the gap (Pam Rocks,
Pemberton, Squamish Airport) and one outside it (Entrance Island). The best weather station
that can detect inflows is the Squamish air quality weather station operated by the BC
Ministry of Environment. This station does not have long time series of meteorological
observations since it started in November 2009. Consequently, correlations o f inflows are
limited to a 4-year period (2010-2013). The outflows are detected by Pam Rocks and they are
examined for longer time series (-12 years). Four correlations of wind speed/pressure
difference are illustrated for outflows as well as for inflows. Pressure differences are
calculated by deducting sea level pressure observations of southwestern stations from the

40

northeastern ones for outflows, and vice versa for inflows. Surface pressure measurements of
the Squamish airport station are used for inflows rather than those from the Squamish air
quality station as it does not measure surface pressure. The coefficient of determination, R2,
is calculated for each correlation, and then the one-tailed Student’s t test is calculated to
determine whether R2 is significantly different from zero at the 95% significance level with
n-2 degrees o f freedom (Koletsis et al., 2009; Devore, 2010): t =

Temperature and humidity gradients are calculated seasonally as a function o f wind
speed and direction (the gradients are correlated to the wind speed and direction) for the
previously-mentioned stations. Along-channel wind speeds o f 3 m s'1or less are excluded to
reduce the noise that could be caused by including lighter winds that may result from local
effects. Temperature and humidity gradients show the effects o f the outflows and inflows on
the air temperature and humidity along the gap and throughout different seasons.
2.2.3.3. Case Studies
An investigation o f specific gap wind events has been limited to the period of the
available wind profile data (from late 2007 to early 2010). The preferred period is from late
2009 until the end o f the wind profile record as this was the period of intensive observations
for the 2010 Olympics. Accordingly, the winter of 2009/2010 is investigated for outflow
events. It is found that this winter had six outflow cases, two o f which are selected for the
present study as they have different synoptic conditions and good wind profile data. Another
target of this research is to examine inflows, particularly the summertime ones. As there is
not any common summer season between the wind profiler data and the observations of the

41

Squamish air quality weather station (the station that can detect inflows), no inflow event
case studies can be done.
The selected outflow cases are studied synoptically using hand-drawn mean sea level
pressure charts produced four times a day at the Meteorological Service of Canada’s Pacific
Storm Prediction Centre, and contoured NCEP-NCAR maps for the 500 hPa geopotential
heights. Then winds, air temperature, humidity, and pressure gradients are examined using
different locations along the gap to show the gradients along the channel. The mean sea level
pressure changes are depicted at several locations throughout both cases, which demonstrate
the effects o f different synoptic conditions on the mean sea level pressure. One SAR wind
image is used for each event to show the horizontal distribution of the flow over the water
surface. Those are the only images that have been captured during the two cases. Finally, the
vertical structure o f the flows during both events is examined using the available wind
profiler data. The RASS profiles o f the virtual temperature are available for just one o f the
selected wind cases. The wind profiler data are filtered using values of the Signal-to-Noise
Ratio (SNR) as recommended by Riddle et al. (2012) who determined a minimum threshold
o f SNR for the wind profiler data to be considered reliable. The SNR threshold aims to
distinguish and then omit the wind profiler data that are produced by weak atmospheric
signals resulting from the fluctuations or noise in the data. Riddle et al. (2012) determined
the SNR threshold as -12 dB as a result o f theoretical considerations and empirical
modifications. Consequently, the wind profiler data that has a SNR o f less than -12 dB have
not been considered in this study. The valid wind profiler wind directions and speeds are
depicted jointly as vectors. Furthermore, the along channel wind component (Vac) is
calculated to clarify the wind speeds along the channel: Vac = M cos (Wd - Aid), where M is

42

the reported wind speed, Wd is the reported wind direction, and Aid is the along-channel
direction at the wind profile site (340°).

43

3. A Synoptic Climatology of Strong Along-channel Winds on the Coast of British
Columbia
3.1. Abstract
A synoptic climatology using mean sea level pressure and 500-hPa geopotential height
composites is conducted for five strong along-channel winds that occur through the channels
and passes dissecting British Columbia’s coastal mountains. Seasonal (winter, summer) and
directional (inflow: air moving from the coast inland; outflow: air moving from inland
toward the coast) partitioning of the winds results in four distinct along-channel winds that
occur: summertime inflow, summertime outflow, wintertime inflow and wintertime outflow.
Composite analyses using NCEP-NCAR reanalysis data and in-situ observations are used to
examine each wind type at all five locations. Wintertime composites produce distinctive
patterns from the overall winter climatology, in which outflows occur when cold-air
damming associated with an arctic surface high pressure area on the inland side of the coastal
mountains is accompanied by the presence of an area of low surface pressure in the
northeastern Pacific. Inflow composites in the winter indicate low pressure areas associated
with mid-latitude cyclones over the Gulf of Alaska. In this winter inflow situation, winds
blow semi-parallel to isobars in near a geostrophic balance, whereas the wind crosses isobars
down the pressure gradient in the wintertime outflows. Summertime composites are similar
to the overall summer climatology; therefore, other approaches are applied to explain these
winds. We analyze 104 summertime inflow events at three locations using surface analysis
charts for 13 summer seasons. The analysis reveals the importance o f fronts at the time of
gap wind events, with fronts associated with almost half of the events at the three gap
locations. Non-hierarchical clustering analysis is applied for the summertime inflow events.

The clustering analysis gives a better explanation o f the summertime inflows than the
composites. This is achieved by grouping events into different clusters, allowing for better
illustration of the variance between the clusters and showing the importance o f the role of
pressure gradients across the coast in initiating the flows.

3.2. Introduction
Specific orography in the presence of particular atmospheric conditions can lead to
gap winds. The synoptic-scale pressure gradient (PG) has a profound role in creating these
atmospheric conditions (Hughes and Hall, 2010). A coast-parallel mountain barrier that often
separates air of continental origin from air o f maritime origin creates a typical environment
for gap winds. Channels and passes, that are perpendicular to the mountain chain, function as
connection paths for airflows between the ocean and the interior o f the continent leading to
gap winds (Jackson et al., 2013), which at times might reach the speed of violent storms
(Mass et al., 1995). The typical atmospheric conditions for gap winds occur more frequently
during the winter (Jackson and Steyn, 1994a), in which stronger pressure gradients occur
across the coastal mountains. Therefore, gap winds have been widely studied in the literature
as a wintertime wind phenomenon and are often called outflows. However, winds can be
observed blowing through channels but are likely not defined as gap winds because o f the
absence of some gap wind characteristics, such as a stable layer that separates the gap flow
layer from the air aloft, and an along-channel PG. Consequently, gap winds are considered as
a subset of a larger category called along-channel winds, which include all flows that are
channelized by the topography of the gap and blowing up and down the channel. Alongchannel winds can be blowing into the continent and called “inflows” or out of the continent
as “outflows”, which at times can be considered as gap winds.
45

Similar wind phenomena have been studied in many geographical regions, such as
Soya Strait north of the Japanese island o f Hokkaido (Shimdada and Kawamura, 2011),
Crete Island (Koletsis et al., 2009), Shelikof Strait in Alaska (Lackmann and Overland,
1989), Red Sea (Jiang et al., 2009) and Mexico City (de Foy et al., 2006). The synoptic-scale
weather pattern has been investigated previously in many outflow studies, emphasizing the
importance of such an analysis for understanding the evolution of these winds (Mass et al.,
1995, Overland and Walter, 1981, and Sharp, 2002). The coast of British Columbia (BC) has
a topographic configuration conducive to strong along-channel winds. Two of these winds
have been studied systematically before. These are the Strait of Juan de Fuca gap wind (Reed
1931; Overland and Walter, 1981 and Mass et al., 1995), and the Howe Sound gap wind
(Jackson and Steyn, 1994a; Jackson and Steyn, 1994b; and Jackson, 1996). Most of these
studies identified the synoptic forcing o f outflow cases using both the mean sea level
pressure (MSLP) and 500 hPa level synoptic patterns. Other channelized flows along the BC
coast have not been previously studied.
However, most of the previous studies are concerned with the synoptic evolution o f a
case study. Synoptic analyses based on composites of a number of events have been
performed for many other wind studies (Chenoli et al., 2012; Harden et al., 2011; Mercer et
al., 2012), as well as in the present study. Two general approaches have been used in
synoptic climatology studies (Yamal, 1993), environment-to-circulation (e.g. Chenoli et al.,
2012) and circulation-to-environment (e.g. Stahl et al., 2006). The present study uses the first
approach, which assesses the environmental data first, particularly the wind speed and
direction, and then sets criteria to build the associated synoptic classes.

46

The objectives of this study are to better understand the atmospheric conditions that are
associated with strong along-channel winds along the BC coast, and in addition, to
characterize some of the previously unstudied flows as well as to conduct a synoptic
climatology for several outflows/inflows on the BC coast. First, a climatology o f these winds
based on in-situ observations and reanalysis o f existing data is presented. In section 3.3, we
discuss data and methods used to build these composites. Annual and diurnal variability of
events are discussed in section 3.4. Section 3.5 is devoted to the composite analysis o f the
different wind types. In sections 3.6 and 3.7, we present some additional manual and
automated tools to explain some o f the composites that do not show important differences
from the average climatology, while conclusions are drawn in section 3.8.
3.3. Data, Methods and Study Locations
3.3.1. Study Locations
Since the geographic conditions o f many gaps on the BC coast are similar, a sample
of them have been chosen in this study to represent the variation among strong along-channel
wind cases on the BC coast. Figure 3.1 shows the locations o f the selected gaps and a
proximal weather station that is used to represent the flow at each gap location. The selected
gaps were intended to be distributed from the north to the south coast of BC and to include a
gap within an island. The availability of a weather station and its location within the gap
plays an important role in the selection procedure. Those gaps are (from north to south):
Portland Inlet to represent the northern coast, Burke Channel to represent the central coast,
Howe Sound and Juan de Fuca Strait to represent the southern coast and Qualicum to
represent a gap wind blowing over Vancouver Island.

47

INFLOW

■

OUTFLOW

opography/m
Wue
2909
2329.2
M | 1749.4
™
1169.6
568.8

Figure 3.1. A topographic map of western BC showing the locations of the gaps
and a meteorological station (red triangle, circle or square) for each gap. Arrows
show directions of the wind that a meteorological station can detect as
channelized winds (directions are extracted from wind roses). WEK: Portland
Inlet. WME: Burke Channel. WAS: Howe Sound. WGT: Qualicum. WQK: Juan
de Fuca Strait.

48

3.3.2. Data
To perform the synoptic assessment of targeted winds, two types of data are utilized.
First, hourly wind observations are used to define the wind events. Table 2.1 (section 2.2.2.1)
shows the selected weather station for each gap, along with the period of available data (see
Figure 3.1 for locations). All meteorological stations are automated, land-based, and operated
by Environment Canada. A station can contain a wind monitor, which is deployed at a 10 m
height, a barometer, air temperature and humidity sensors, and a rain gauge. The data provide
in-situ surface observations. They are extracted from the Environment Canada’s National
Climate Data and Information Archive website (Environment Canada, 2014).
The second type of data is NCEP-NCAR reanalysis I project data that are used to
construct the synoptic composites of the wind events. MSLP and 500 hPa geopotential
heights from the NCEP-NCAR reanalysis data for the period o f 1994-2012 are extracted
from the archives provided by the National Oceanic and Atmospheric Administration /
Oceanic Atmospheric Research / Earth System Research Laboratory - Physical Sciences
Division (NOAA/OAR/ESRL - PSD), Boulder, Colorado, USA (Kalnay et al., 1996). NCEPNCAR reanalysis data are available at 6-hourly temporal resolution as well as daily averages.
The spatial resolution of 2.5° x 2.5° latitude/longitude of the data is found to be sufficient to
build composites of gap winds at the synoptic-scale. NCEP-NCAR reanalysis data are
extracted for a synoptic domain suitable for establishing the synoptic climatology o f BC:
between 40°- 62.5°N and 110°- 157.5°W, following the work o f Stahl et al. (2006) who
conduct a synoptic climatology of the region for different purposes.

49

3.3.3. Methods
Seasonal and yearly wind roses are created for winds at each meteorological station
to determine whether the station can detect outflow and/or inflow winds. The along-channel
wind directions are determined by assessing wind roses and comparing the most common
wind direction with the orientation of gaps from topographic maps (Figure 3.2). Then the
seasonality is considered to distribute the observations between the cold season “wintertime”
and the warm season “summertime”. Finally, the minimum wind speed threshold to define an
event is chosen as 11.3 m s'1 that corresponds to the Beaufort scale 6 (Strong Breeze)
(Chenoli et al., 2012). A wind event is defined as any consecutive four hours that meet the
wind speed and direction criteria. The time and wind speed thresholds are chosen to consider
a sufficient number o f events as well as to produce significant results. In each o f Portland
Inlet and Burke Channel, many events are detected with winds much higher than 11.3 m s '1.
Therefore, another wind-speed partitioning is performed to define high-speed events as the
peak wind speed exceeds the Beaufort scale 9, which is 20.0 m s'1. Consequently, the winds
have been separated into wintertime outflow, which is a subset o f high-speed events in
Portland Inlet and Burke Channel, wintertime inflow, summertime outflow, and summertime
inflow. The duration o f each category varied considerably. Wintertime outflows have the
longest average period, and Juan de Fuca Strait gap winds have the longest inflow period
(Figure 3.3).

50

form
m,
P o r tla n d In le t
Weatherstation
G rey Islet-

war

tnsr

%

W EK

SC
U
TH
B urke C h an ­
n el
Weatherstation
Cathedral P oint
-W M E

H ow e Sound
Weatherstation
Pam R ock sW AS

J u a n d e Fuca
Weather station
R a c e R ock sW QK

Q u a lic u m
Weatherstation
S isters IslandWGT
Outflow

a

Inflow

^

Speed m/i

E32o-»
Qi#-20

llutinflow #

Figure 3.2. Wind roses and detailed topography for each weather station
representing along-channel winds. Note the visual agreement between the
topographic conditions of each gap and the yearly wind rose observed at the
selected meteorological station of each gap. Speeds > 10 m s'1 approximately
match Beaufort scale 6.

51

50
C

3

250

£

2

200

43

e

150

.2

s3 100

Q

50

3

O 40
B

30

g

20

3
O

10

O
T
WEK-h WEK-I WME-h WME-I WAS

Wintertime outflows

WQK

WME

WAS

WGT

T

nr
WEK

WME

Summertime outflows

WQK

Wintertime inflows

WME

WAS

WGT

WQK

Summertime Inflows

Figure 3.3. Boxplots showing the average time period (hours) of an event for
each type of gap wind at different meteorological stations. The upper edge of
boxes indicates the 75th percentile of the data set, and the lower edge indicates
the 25th percentile. The horizontal red lines inside boxes represent the median
value. The ends of the vertical lines indicate the 5th and 95th percentile of data
values, while plus signs represent outliers. ID-h: high-wind-speed events, ID-1:
low-wind-speed events. Stations that do not detect events in each category are
not shown.
To seasonally partition the winds more accurately, the months o f April and October
are ignored for the study period since these two months have characteristics of both summer
and winter wind cases. Consequently, including them in the summer and/or winter would add
variance to the composites and anomalies. In fact, both seasons are determined after
investigations that examined event composites from each month. A climatology of MSLP
and 500 hPa heights for the period 1994/1995 to 2012 is created by averaging the NCEPNCAR reanalysis data for the months November through March as the winter season, and for
May through September as the summer season over the BC synoptic domain. Then MSLP
and 500 hPa geopotential height event composites are created by averaging the NCEPNCAR reanalysis data over the synoptic domain o f BC for all events in each category. The
52

anomaly is then calculated for each composite by subtracting the composite from the
climatology. Standard deviations o f the composites are calculated and included in the
composite maps to show the variability o f the isobars. Finally, the statistical significance of
the anomalies is calculated using the Student’s t test at the 95% significance level (Wilks,
2011) and included in the anomaly maps as shaded areas.
3.3.4. Validation

of

the

Ability

of

Meteorological

Stations

to

Detect

Outflows/Inflows
The ability o f detecting along-channel winds varies considerably between the five gaps.
This depends mainly on the location o f the meteorological station within the gap. The
validation o f the capability o f each station to detect along-channel flows is achieved by the
coincidence o f the most frequent wind directions from a yearly wind rose with the channel
orientation from topographic maps. Both WME and WQK are ideal for detecting along
channel winds in Burke Channel and in the Strait o f Juan de Fuca, respectively. This is not
the case for WEK (Portland Inlet) and WAS (Howe Sound) since both are blocked by islands
on the ocean side and located near the termination of the gaps (see Figure 3.2). Qualicum
(WGT) is a special case. This station observes mainly northeast and southwest winds
blowing along the axis of the Strait o f Georgia. However, these winds are not the focus of
this study. The Qualicum wind is a relatively infrequent plume o f southwesterly wind that
crosses Vancouver Island through the Albemi Inlet, passing over the crest of the land and
Home Lake toward the Strait of Georgia north o f Qualicum Beach as a southwesterly wind
(personal investigation with local people, 2012).

53

3.4. Annual and Diurnal Distribution of Outflow/Inflow Events
The annual frequencies of outflows and inflows, as previously defined, are given in
Figure 3.4, where inflows (outflows) are not detected at WEK-Portland (WGT-Qualicum).
The annual distribution of hours that observed outflows (Figure 3.4a-blue) shows a similar
shape for all stations with a peak from November until March. The winter peak is related to
the normal mean sea level pressure gradient (MSLPG) between the inland Canadian High
and the offshore Aleutian Low, resulting in a high frequency of outflows.
The annual distribution o f hours that observed inflows (Figure 3.4a-brown) shows a peak
frequency at the southern gaps (Qualicum and Juan de Fuca Strait) in summer, and a
minimum between September and February. This is likely a result o f the normal MSLPG
between the Pacific High and a thermal low-pressure centre over Washington State and
southern BC that commonly develops over the summer (Brewer et al., 2012). The locations
for both WAS-Howe Sound and WME-Burke Channel are not ideal to detect inflows that are
generally largest at the end of the channel. The middle location of these two stations will not
easily detect strong inflows, except those caused by the passage of a surface low. Such a
situation is more frequent in winter.

54

Annual Distribution

Diurnal Distribution

%

e

1.1 ,
1

2

3

4

5

8

7

8

. I l f l r ’j 1 1 1 n 1 1 1

9 10 11 12

'

01

40

i
g

%

20

3L

6L

9L

12L 15L 18L 21L

40

.

- WME

i L

h j J J —I —■ J

1

3

2

4

5

6

7

8

4

2o

|

IllililiUuiIll

Ik l i Channel) o

8 10 11 12

0L

3L

6L

8L

121 151 181 21L

ll

h bI

A

i

d I

0L

3L

9L 12L

15L

18L 21L

20,

| Jl

Is
2

i ________ . 1 1

u
1

0L.

2»

2

3

4

S 6 7

8 9

a
10 11 12

j IhkL h A d i

Ll... u 11 ..lL
1 2

3

4

5

6

7

8

6L

Qi

9 10 11 12

OL

3L

6L

9L

12L 15L 181 211

40

1

(*)

2

3

4

5

6

7

8

1

WGT
(Qualicum)

9 10 11 12

20

Jl I I a ■ I I L
01

Months

<b )

3L

6U

91

12L 15L 18L 21L

Hours

Figure 3.4. (a) Annual, and (b) diurnal distributions of outflows (blue) and inflows
(brown) for hourly observations (percentages) that meet the defined criteria of gap winds,
(L) is the local standard time (PST).
The diurnal distribution o f outflows (Figure 3.4b-blue) shows a common minimum at the
wannest portion o f the day indicating the role o f daily MSLP variations in which warm air
temperatures decrease sea level pressure over the continent, reducing the ocean-ward PG.
The maximum frequency appears between 3 Local Standard Time (LST - Pacific Standard
Time) and 9 LST - the time o f day with coldest temperatures which act to increase the sea
level pressure over land and increase the ocean-ward PG. Inflows (Figure 3.4b-brown) show
more variability. WGT-Qualicum and WQK-Juan de Fuca have similar distributions with
peaks near midnight and minima before noon. Daily inflow distributions at WME-Burke
Channel and WAS-Howe Sound are more uniform, probably because these stations are less
ideal for detecting inflows.

55

3.5. Composite Analysis
Following the criteria, noted earlier, for defining events and categorizing them in
different groups, a composite is created for each category. The composites are subtracted
from the climatology to obtain anomalies. (See Figure 3.5 for climatology).

M SLP

500 hPa

1004

Figure 3.5. Climatology of wintertime (a) and summertime (b)
for MSLP and 500 hPa heights.

3.5.1. Wintertime Outflow
Figure 3.6 shows the wintertime outflow MSLP and 500 hPa geopotential height
composites. The MSLPG is the main difference between the high-speed (3.6a and 3.6c) and
the low-speed (3.6b and 3.6d) composites and anomalies of Portland Inlet and Burke
Channel. The MSLPG of the high-speed group has a NNE-SSW orientation. The surface
pressure difference has values o f approximately 20-18 hPa between the high and low
pressure centres, very similar to the southern gaps (3.6e and 3.6f). However, MSLPG in the
low-speed group has a NE-SW orientation, and the surface pressure difference values are 10-

56

12 hPa. The similarities between the high-speed group and the southern gaps lead to the
conclusion that same synoptic conditions that cause strong outflow events in the northern
gaps of the BC coast can produce lower-wind-speed outflows in the southern gaps. To
investigate the correlation in timing between the outflows in the northern and southern gaps,
hourly wintertime outflow observations that occurred at the same time in two different gaps
(a northern gap and a southern gap) are determined. It is found that between 60% and 77% of
wintertime outflows in southern channels (Howe Sound and Juan de Fuca Strait) occurred at
the same time as wintertime outflows in northern channels (Portland Inlet and Burke
Channel).
Isobars are oriented perpendicular to the channels, suggesting that the MSLPG is the
main driving force of these outflows. The Coriolis force is negligible when the channel
length is less than approximately 100 km (Jackson et al., 2013). However, this is not the case
in Juan de Fuca Strait where the channel is 150 km long (Overland and Walter, 1981), and
the isobars here are more parallel to the strait.
The northern continental positive MSLP anomalies in all groups can be explained by
the presence o f an enhanced arctic anticyclone (Canadian High), which is similar to synoptic
type 6 provided by Stahl et al. (2006). During winter, the shallow cold air o f an arctic
anticyclone can advance southward and be blocked by the coastal mountains o f BC. This
phenomenon is known as cold-air damming (Lackman, 2012). The depth o f this stable cold
air is probably not high enough to overcome the barrier, which separates the interior cold,
dry, stably stratified air from the relatively warm, moist air over the coast (Jackson, 1996). In
this case, wind uses channels and fjords to blow from the inland High toward the offshore

57

Low-pressure area as an outflow wind, which in this case is also a gap wind (Overland and
Walter, 1981).
500 hPa geopotential height composites and anomalies also show similarities between
the high-speed group (3.6a and 3.6c) and the southern gaps (3.6e and 3.6f). Both groups have
larger anomalies than the low-speed group (3.6b and 3.6d), and show a trough in the
anomalies extending across the southern BC coast. The eastern Pacific 500 hPa ridge is
amplified in all composites, and extends across Alaska and northern BC with positive
anomalies in these areas. The enhanced 500 hPa ridge is associated with a strengthened
surface anticyclone over the interior of BC and Yukon that is indicated in the MSLP
anomalies in Figure 3.6.

58

MSLP

MSLP Anomaly

500 hPa

(*)

500 hPa Anomaly

—E»*00
2

100*
%

(b)
1012

5

<d)
1008

150M
O

to
1

(0
!3*!r^tEv^i5rNr^i»v
Figure 3.6. Composites of wintertime outflow events. MSLP and 500 hPa geopotential height
maps have the standard deviation in light contours, and the anomaly maps use shading to
indicate when the anomaly is statistically significant (p < 0.05) using the student’s t test, (a)
Portland Inlet high-wind-speed, (b) Portland Inlet low-wind-speed, (c) Burke Channel highwind-speed, (d) Burke Channel low-wind-speed, (e) Howe Sound, and (f) Juan de Fuca Strait.
Intervals: MSLP composites (4 hPa), MSLP anomalies (2 hPa), 500 hPa composites (60 m),
and 500 hPa anomalies (30 m). The green dots represent the weather station locations.
59

3.5.2. Wintertime Inflow
All MSLP composites show a low-pressure centre over the northern Gulf of Alaska (<
992 hPa in most cases) with a trough extending south-southeastward (Figure 3.7). There is a
strong MSLPG over the BC coast to the south and southeast o f the trough from S-N or from
SE-NW. This is very similar to the synoptic type 9 provided by Stahl et al. (2006), which
occurs mostly in the winter season and causes westerly to southwesterly winds over BC. The
surface low over the Gulf of Alaska is a quasi-permanent wintertime feature (the Aleutian
Low) associated with mid-latitude cyclones that track eastward across the Pacific Ocean. The
zone of strong MSLPG near each o f the stations in Figure 3.7 has isobars oriented parallel to
the channel, suggesting that the inflow winds are likely in near geostrophic balance at the
synoptic scale, at least above the surface, with the PG force balanced by the Coriolis force.
The relationship between the surface and upper-level pattern in Figure 3.7 is consistent,
the surface low-pressure centre and trough is coincident with or just east o f a trough at 500
hPa. The nearly coincident locations of the upper-level trough and the surface trough suggest
a “cold low”: the closed circulation of an occluded mid-latitude cyclone at the 500 hPa level
(Parker et al., 1989). It is important to note that winds that accompany cold lows (or mid­
latitude cyclones) are not limited to the mountain gaps, and cannot be considered as gap
flows; however, they are channelized winds. The Juan de Fuca winter inflow pattern (3.7d)
shows a shallower surface low-pressure centre and a less pronounced upper-level trough
compared with other inflow locations. This may be related to the fact that the flow in Juan de
Fuca Strait has a distance > 100 km to accelerate over the water surface before reaching
WQK (Mass et al., 1995), while the other inflows have < 100 km to blow over the water.

60

Therefore, similar wind speeds, which are produced by a deep cyclone in the short gaps, can
be induced by a shallower cyclone in long gaps such as Juan de Fuca Strait.

M SLP

M SLP Anom aly

500 hPa

500 hPa A nom aly

355

I

•M l

IT

✓ ^0*2

_ \

" - f--

Figure 3.7. Composites of wintertime inflow events. MSLP and 500 hPa geopotential height
maps have the standard deviation in light contours, and the anomaly maps use shading to
indicate when the anomaly is statistically significant (p < 0.05) using the student’s t test, (a)
Burke Channel, (b) Howe Sound, (c) Qualicum, and (d) Juan de Fuca. Intervals: MSLP
composites (4 hPa), MSLP anomalies (2 hPa), 500 hPa composites (60 m), and 500 hPa
anomalies (30 m). The green dots represent the weather station locations.
3.5.3. Summertime Outflow
Summertime outflow events have been detected in three of the gaps considered in this
study (Portland Inlet-WEK, Burke Channel-WME, and Howe Sound-WAS). The average
61

maximum wind speed o f these events is much less than for the wintertime outflows, but it
remains above 14 m s'1. These events are infrequent; the occurrence o f such cases is 0.3%,
0.28% and 0.48% of observations for WEK, WME and WAS, respectively. Figure 3.8 shows
the composites of summer outflows. A weak MSLPG compared with winter outflows is
common between all composites, which show similarities with the climatology. The
anomalies give a better explanation o f the synoptic conditions resulting in summer outflows.
The climatology (Figure 3.5b) shows that the Pacific High migrates northeast in the summer
with a ridge extending northeastward across the BC central coast. The MSLP anomalies
(Figure 3.8) are negative in the location of the Pacific High and ridge and positive over the
continent and northern BC resulting in an anomalous PG oriented NE-SW parallel to each
channel and perpendicular to the coast.
The 500 hPa geopotential height composites (Figure 3.8) are also similar to the
climatology (Figure 3.5b). In the Portland Channel composite (Figure 3.8a), there are
negative anomalies across the whole domain lowering 500 hPa geopotential heights
everywhere. The anomalous PG o f the MSLP composites leads to the conclusion that the
synoptic factor causing these infrequent events is a summertime stormy period in which a
low approaches BC from the SW. In this case, the flow pattern is more similar to that in the
winter, with the Pacific High displaced to the south, lower than normal pressures offshore o f
the BC coast and higher than normal pressures over the interior o f BC. This reverses the
normal summer PG that points from ocean inland, so that it now points from land toward the
ocean.

62

MSLP

MSLP Anomaly

500 hPa

500 hPa Anomaly

SMS

<»)

A

e 1014

.. ....... '»—

< J

a ?
■
I*
I -’5
’*»nn»

io ia s

P

uSt

»

\

IS *
. js f e t o

*

S

if

w*

u rw

urw

ur*

ttrw

*4*

hAi

iM
>

Figure 3.8. Composites of summertime outflow events. MSLP and 500 hPa geopotential
height maps have the standard deviation in light contours, and the anomaly maps use
shading to indicate when the anomaly is statistically significant (p < 0.05) using the student’s
t test, (a) Portland Inlet, (b) Burke Channel, and (c) Howe Sound. Intervals: MSLP
composites (4 hPa), MSLP anomalies (2 hPa), 500 hPa composites (60 m), and 500 hPa
anomalies (30 m). The green dots represent the weather station locations.
3.5.4. Summertime Inflow
It is important to note that summertime inflows cannot just be explained by the diurnal
land-water thermal circulation, although mesoscale land-water thermal contrasts, which
cause a land/sea breeze, can be enhanced by synoptic conditions to produce an intensive sea
breeze (Segal and Pielke, 1985). The sea breeze most frequently occurs in the afternoon
hours and is substituted by the land breeze during the nighttime (Abbs and Physick, 1992).
However, inflow winds are mostly recorded during the nighttime at a number of the gaps
considered in this study (see Figure 3.4), indicating the importance of investigating synoptic
factors that are probably the cause o f such flows.
63

Three o f the gaps investigated in this study (Burke Channel, Qualicum, Juan de Fuca
Strait) frequently report summertime inflows (Figure 3.9). As in the summertime outflows,
the anomalies suggest a clearer explanation o f these flows. However, Burke Channel
anomalies do not show important differences from the climatology (Figure 3.9a).

At

Qualicum, the anomaly shows negative MSLP of -6 hPa on the northwest BC coast (Figure
3.9b). Both MSLP anomalies of Burke Channel and Qualicum imply an anomalous
geostrophic contribution to the inflow in both locations. In Juan de Fuca, positive anomalies
of 5 hPa are located west o f the domain creating W-E PG, which shows the contribution of
the PG pointed inland to the inflow (Figure 3.9c) at this location. In general, the summertime
inflow composites are not very different from the summertime climatology, indicating the
importance o f seeking other meteorological features that are not apparent in the composites.
Composites are averages o f the events, so they are not able to display the variability o f the
events. Therefore, another automated approach has been considered to explain the synoptic
conditions o f the summertime inflows. This approach is discussed in section 3.7.

64

SLP

SOOhPa

SLP Anomaly

SOO hPa Anomaly

(a)

a

102011

72

(b)

dfr

*

0

M
Vi t*sV
i i»»wn.V
Figure 3.9. Composites of summertime inflow events. MSLP and 500 hPa geopotential
height maps have the standard deviation in light contours, and the anomaly maps use shading
to indicate when the anomaly is statistically significant (p < 0.05) using the student’s t test,
(a) Burke Channel, (b) Qualicum, and (c) Juan de Fuca. Intervals: MSLP composites (4
hPa), MSLP anomalies (2 hPa), 500 hPa composites (60 m), and 500 hPa anomalies (30 m).
The green dots represent the weather station locations.

Lange (1998) reports that summertime inflows are usually accompanied by the passage
o f a front over the coast coming from the west or northwest. Therefore, to better understand
the possible role o f fronts in initiating summertime inflows, a manual assessment of MSLP
charts is applied in the following section.
3.6. Manual Investigation of Hand-drawn Surface Analysis Charts
A manual investigation of the 6-hourly hand-drawn surface analysis charts produced at
the Meteorological Service of Environment Canada’s Pacific Storm Prediction Centre, is

65

conducted. The main purpose of this investigation is to better understand the synoptic
conditions responsible for summertime inflows because the composite analysis conducted in
the previous section reveals small and largely insignificant differences from the climatology.
The hand analyzed MSLP charts are available for 13-summer seasons (1999 to 2011). The
investigation is conducted for 104 cases for the three gaps with 33, 35, and 36 cases for
Burke Channel, Qualicum, and Juan de Fuca Strait, respectively. The investigation focuses
mainly on detecting fronts associated with the events, and determining their location and type
in addition to any other potentially important synoptic conditions, such as the presence of
low and/or high-pressure systems. The following description considers charts associated with
the time of the wind events.
NE-SW oriented cold and/or occluded fronts over the BC coast accompanied with an
MSLP low to the west or northwest of BC are detected in 20 out o f 33 Burke Channel
summertime inflow events. Six other cases show an MSLP low over Yukon, and two cases
show a N-S oriented warm front over the BC coast. The remaining five events do not appear
to be associated with surface weather systems.
The investigation of the summertime Qualicum inflows shows a cold front, a surface
trough, and a stationary front in 16 cases, nine cases, and four cases, respectively (out o f the
35 cases examined). The remaining six events do not show any nearby frontal activity or
surface trough situations. Cold fronts are located over the Vancouver Island region and
mostly oriented N-S with eastward or southeastward frontal movement. The surface trough is
located over the interior of BC, west o f Haida Gwaii or Vancouver Island.

66

The analysis of Strait of Juan de Fuca summertime inflows shows an MSLP low in 19
out o f 36 cases, a cold front accompanied by an advanced mid-latitude cyclone to the west or
northwest in nine cases, and a frontal system without the presence of a mid-latitude cyclone
in three cases. Meanwhile five cases show the Pacific High located west o f Washington State
or Vancouver Island. The MSLP low is located over Washington State or southern BC and
sometimes appears as a northward trough extending from a larger low-pressure system to the
southeast. The frontal system detected is oriented NE-SW or N-S, sometimes with eastward
to southeastward movement.
An example of the synoptic-scale evolution of a Qualicum Wind event that is caused
by a frontal system is given in Figure 3.10. Seventeen to eleven hours prior to the event, a
mid-latitude cyclone with a MSLP low centre approaches Vancouver Island from the
southwest (Figure 3.10a and 3.10b). Five hours prior to the event, the MSLP low advances to
the north of Vancouver Island and the warm front passes the coast, while the cold front is
still over Vancouver Island (Figure 3.10c). At the onset time of the Qualicum event (0300
LST 11 September), a shift in the wind direction from 120° to 200° begins between 0100
LST and 0300 LST at the Sisters Island weather station, indicating a frontal passage and
initiating a Qualicum Wind event. One hour after the start o f the event (Figure 3.10d), the
wind direction becomes 220° and the SW-NE oriented cold front has passed Vancouver
Island toward eastern BC. By 1000 LST 11 September, the cold front is located over Alberta
and the Qualicum event is still observed (Figure 3.10e). At 1200 LST 11 September, the
wind becomes weaker and the event ends. Four hours after the end of the event (Figure
3.1 Of), the system becomes weaker and the cold front advances further eastward; however,
the winds are still from the southwest with speeds lowered to around 5 m s '1. A similar

67

synoptic evolution is found in several summertime inflow events in the other two locations,
particularly in Burke Channel. In most cases, a frontal system that has passed the channel
initiates the inflows, and the wind event ends once the system becomes weaker or disappears.

(a )

17 h o u r s p r io r to th e e v e n t (1 0 L S T 10 S e p )

(d

1 h o u rs a f te r th e sta r t o f th e e v e n t (4 L S T 11 S ep )

(b)

11 h o u rs p r io r to t h e e v e n t (1 6 L S T 10 S ep )

(e) 7 h o u r s a fte r th e sta r t o f t h e e v e n t (1 0 L S T 11 S ep )

(c)

S h o u r s p r io r to t h e e v e n t (2 2 L S T 10 S e p )

(f

4 h o u r s a fte r th e e n d o f t h e e v e n t (1 6 L S T 11 S ep )

Figure 3.10. Surface analysis charts of the synoptic evolution of a summertime inflow
Qualicum wind event. The event started at 0300 LST 11 Sep 2004 and end at 1200 LST 11
Sep 2004. The brown dot identifies Sisters Island weather station (WGT). Intervals: 4 hPa, L
represents a low centre, the curved line with triangles represents a coldfront, and the curved
line with half circles represents warmfronts. Meteorological service o f Environment Canada
—Pacific Storm Prediction Centre.

68

The other common case of Qualicums involves an interior surface trough which, with
a ridge to the south o f the BC coast, creates a SW-NE PG. Such a case represents a typical
summertime event of Qualicum winds. An example o f this pattern is given in Figure 3.11.

a)

3 hours prior to the event (10 LST03 May 1999)

(c) 9 hours after the start of the event (22 LST 03 M*v 1999)

(b) 3 hours after the start of the event (16 LST03M«y 1999) (d)

The end of the event (04 LST 04 M»> 1999)

I

Figure 3.11. Surface analysis charts of the synoptic evolution of a summertime inflow Qualicum
wind event. The event started at 1300 LST 03 May 1999 and ended at 0400 LST 04 May 1999.
The brown dot identifies Sisters Island weather station (WGT). Intervals: 4 hPa, L represents a
low centre. Meteorological service o f Environment Canada - Pacific Storm Prediction Centre.
Before the start of the event (Figure 3.1 la), a surface ridge dominates to the southwest of
Vancouver Island and a surface low is located over Saskatchewan with a trough through the
interior of BC. At the time of the event (Figure 3.11 b &c), the surface ridge advances
northeastward over the southern coast of BC and the interior trough is located over
northeastern BC and Alberta, creating a SW-NE PG. Such a synoptic scale PG leads to
Qualicum Winds, similar to the pattern shown in the composite analysis (Figure 3.9b). At the

69

end of the event (Figure 3.1 Id), the surface ridge advances northeastward and dominates
over southern BC. At this time, the winds become weaker (around 5 m s '1), although it is still
blowing from the same direction for the next few hours.

3.7. Clustering Analysis
Non-hierarchical clustering approaches have been widely used in synoptic type
classifications, particularly, the partitioning k-means approach (Jiang et al., 2012, Jiang,
2011, Stahl et al., 2006 and Burlando, 2009). The k-means approach provides an easily
interpretable visual explanation o f the variance. Accordingly, it is applied in this study on the
MSLP summertime inflow composites, which do not show important differences from the
summertime climatology. The number o f clusters in k-means clustering should be specified
in advance o f the analysis. One of the statistical approaches used to determine the number of
clusters is to consider a local minimum between the number o f groups and the homogeneity
of a cluster (Stahl et al., 2006).
In this study, k-means clustering is applied to a limited number o f variables, where each
composite has about 70 cases and each case is considered as one variable. Specifically, all
MSLP values at each NCEP grid point are put into one vector for each wind case and this is
used as a variable. The small number o f variables puts limits on using many automated
methods to determine the number o f clusters. Therefore, a manual classification is conducted
initially by looking at each event alone, which gives the opportunity to investigate the
potential number of clusters in each group o f events. The manual verification is combined
and followed by a trial-and-error procedure (Kassomenos et al., 2010), which allows
determining the most efficient number o f clusters for each composite. The k-means

70

clustering divides n patterns or “data points” into k groups or “clusters” based on a measure
of similarity (Jain, 2010). The similarity (or the distance) between clusters is measured using
the Squared Euclidean distance method (Jiang, 2011). The clustering procedure is repeated
up to 2000 times as suggested by Wilks (2011), each with a new set o f initial centroid
positions.
In Burke Channel (Figure 3.12), the main cluster (54.90% o f the variance) shows a
situation similar to the wintertime inflow composites but with a weaker PG, and a weaker
low-pressure centre over the northern Gulf o f Alaska. This cluster suggests that a frontal
system is involved. The other two clusters (~ 45% of the variance) show a PG pointing
inland suggesting the down gradient contribution in the inflows. The Pacific High extends
northeast toward the BC coast, and a thermal low located inland. This synoptic situation
results in a PG parallel to the channel orientation.

M ean

C luster 1 (54.90% )

Cluster 2 (23.52% )

C luster 3 (2 1 .5 6 % )

Figure 3.12. Mean and k-means clusters of Burke Channel
summertime inflows for MSLP with the variance of each
cluster. Intervals: 4 hPa.
71

In Qualicum (Figure 3.13), clusters 1 and 3 (accounting for 53.33% of the variance) show
a centre of low pressure over the northern Gulf o f Alaska. Cluster 4 (13.33% of the variance)
shows similar positions o f the surface pressure fields but with larger PG, and Cluster 2 shows
a low pressure located over Haida Gwaii. All these clusters are similar to the wintertime
inflow composites but with weaker systems. All Qualicum clusters, except cluster 2, have a
SW-NE ridge over Vancouver Island and Washington State with the high to the SW. Cluster
2 can be considered as a post-frontal situation although cluster 5 might also have a
dissipating front near the Washington State coast. Clusters of both Burke Channel and
Qualicum show a variety of synoptic situations including a low-pressure centre over the Gulf
of Alaska, a high-pressure centre over northeastern Pacific, and an inland thermal low.

V*

'

Now

ibQB
012........ ......W12
1018

1016

1^12

'
|

C l u s t e r 2 ( 2 1 .6 6 % )

Cluster 1 (33.33%)

Mean

1012.

163*»

Cluster 3(2(1%)

144**

1SS*W

126*W

C lu s te r 4 (1 3 3 3 % )

!17*W

isj*w

U4*w

if**

C l u s t e r 5 ( 1 1 .6 6 % )

Figure 3.13. Mean and k-means clusters of Qualicum summertime inflows for MSLP with the
variance of each cluster. Intervals: 4 hPa.

72

K-means clusters for the summertime inflow in Juan de Fuca Strait produce a very
similar situation to the composite (not shown). K-means clustering can better show the PG
orientation and values corresponding to summertime inflow events than the composites in
both Burke Channel and Qualicum. However, the Juan de Fuca Strait summertime inflow
composite cannot be divided into different distinct patterns due to the small variance within
it.
3.8. Conclusion
In this chapter, the synoptic climatology of inflows and outflows for a number of
gaps on the BC coast were examined using MSLP and 500 hPa geopotential height maps.
The wintertime composites showed less variance as well as larger anomalies than the
summertime composites. The cold-air-damming phenomenon, which is caused by a cold
arctic air mass advancing from Alaska and Yukon toward southern BC, initiates the
wintertime outflows which create a strong ocean-ward PG. The composites of the wintertime
inflows showed a low-pressure area over the northern Gulf of Alaska, in which the isobars
are parallel to the orientation o f the gaps. These composites do not indicate that the
wintertime inflows are gap winds; however, they are still considered as strong along-channel
winds. Summertime outflows have a weak ocean-ward PG that is the reverse o f the normal
summer PG situation. They probably occur during stormy periods when a MSLP low
approaches from the southwest. MSLP composites of the summertime inflows showed a SWNE PG; however, the Burke Channel summertime inflow composite showed a pattern very
similar to the climatology. Therefore, other investigations are applied to the summertime
inflows. A manual analysis is conducted using MSLP hand-drawn charts, revealing the
importance of a frontal passage in initiating these inflows. The fronts are mostly oriented N73

S or NW-SE and move eastward or southeastward across the coast. In addition, nonhierarchical k-means clustering was conducted on the summertime inflows. The clustering
analysis showed little variance in the Juan de Fuca Strait composite, but can subdivide the
Burke Channel and Qualicum composites into distinct clusters. These clusters provide a
better explanation for the synoptic conditions that are responsible for the inflows in both
gaps. Some of the clusters showed similarities to the wintertime inflow composites, while
other clusters showed a PG parallel to the channel orientation and caused by an inland
thermal low. In future studies, a similar cluster analysis could be conducted on summertime
outflows particularly in the Strait of Juan de Fuca which will help to identify patterns that
produce the reverse winds in the strait. The results of this study indicate the importance of
more detailed investigations such as using a case study approach and clustering for situations
in which the composite has high variance such as for the summertime inflows.

74

4. Case Studies of Gap Winds in Howe Sound
4.1. Abstract
This chapter examines along-channel winds within Howe Sound, British Columbia, Canada
that occur in both directions, from the interior plateau out toward the coast as outflows, and
from the coast inland as inflows. First, a climatological study o f the along-channel winds in
Howe Sound-Cheakamus Valley is conducted for wind, pressure, air temperature and
humidity as a function of wind speed and direction. The Pam Rocks station (Squamish
station) has the most appropriate location to detect outflows (inflows). The pressure gradient
between Pam Rocks-Squamish and Pam Rocks-Pemberton has the strongest correlation with
outflow strength, and Pam Rocks-Squamish with inflow strength. Outflows (inflows) are
accompanied by lower (higher) temperature and dew point, except for the inflows in the
summer, which bring lower dew point than the overall mean. Second, two case studies of
outflow events are presented and described during the period o f intensive observations prior
to, and during, the Vancouver 2010 Winter Olympics. The January 2010 outflow event is
caused by a zone of strong across-barrier mean sea level pressure gradient. The pressure
gradient is formed behind an arctic front that moved southward across Howe Sound. The
February 2010 outflow event is caused by an advancing centre of sea level low pressure from
the Pacific that formed a NE-SW mean sea level pressure gradient across southern British
Columbia (BC). Synthetic Aperture Radar (SAR) wind images show the spatial distribution
of the flows within Howe sound. Hydraulic behaviours are detected during the January event.
The vertical profile of the events shows an along-channel layer at a height of around 1.5 km
above the surface in the January case, while the height is less than 1.5 km in the other case.

75

The potential temperature of the outflow winds layer in February case is lower than that of
the overlying layer by 4°C indicating the potential for hydraulic behaviour of the flow.

4.2. Introduction
The complex topography of the coastal mountains of British Columbia (BC) profoundly
affects the wind regimes along the coast. The coast-parallel mountain barrier separates two
different air masses: air of continental characteristics inland and air o f maritime
characteristics over the Pacific Ocean. This air mass contrast appears in temperature and
pressure fields between both sides of the coastal mountains, particularly in the lower
atmospheric levels, leading to the creation of a cross-barrier pressure gradient (PG). Such a
PG causes wind to blow through the mountain barrier from high to low pressure using
channels and passes as “gap winds”. In northwestern North America, gap winds are usually
observed during the winter and referred to as outflows. However, other flows that are
channelized by the topography o f the gap blowing up and down the channel are likely not
defined as gap winds because of the absence of the principal conditions that characterize gap
winds. Such conditions include a stable layer that separates the gap flow layer from the air
aloft and an along-channel PG. Channelized winds can blow from the coast inland as inflows
and from the interior plateau out toward the coast as outflows. In this chapter, the terms o f
“inflows” and “outflows” are used rather than “gap winds”.
Outflows along the BC Coast during winter are usually associated with Arctic
outbreaks. Cold, dense, statically stable air advances from the Yukon and Alaska southward
across the interior plateau o f BC creating a pool of cold air that is blocked by the coastal
mountains forming a cross-barrier PG. Outflows observed through the channels dissecting

76

the coastal mountains can reach storm-force (28.5 m s'1) and bring extreme wind chill
(Jackson, 1996). They can cause serious hazards for mariners, aircraft and the public (Mass
and Albright, 1985). The importance o f studying these winds does not just arise from their
hazards, but also because they have high spatial variability making them a challenge to
forecast. Inflows are much less extreme than outflows, and are mostly observed during the
summer, as found in chapter 3 in the study o f a synoptic climatology o f inflows on the BC
coast. They can be caused by a thermal low over the BC interior or accompanied by a
passing front. Inflows have a recreational and tourist importance during the summer,
particularly for windsurfing sports.
Similar wind phenomena have been intensively studied in the literature. A synopticscale PG is found to be the main driving force of outflows in many studies (Lackmann and
Overland, 1989; Mass et al., 1995; Overland and Walter, 1981; Liu et al., 2008).
Temperature profile differences between the two sides of a barrier can create an acrossbarrier PG (Shaip, 2002; and Sharp and Mass, 2004). The temperature contrast reverses
between summer and winter creating an opposite PG (Jiang et al., 2009). The Coriolis force
becomes important in channels longer than about 100 km (Jackson et al., 2013). Some
studies have examined the importance o f the “Venturi effect”; they did not find this to be a
factor for gap wind behaviour (Koletsis et al., 2009). Other factors that contribute to
characterizing outflow events are inertia, surface friction and entrainment o f air from above
the flow layer (Sharp and Mass, 2004). However, the down-pressure gradient acceleration is
mainly caused by the along-channel PG. The down-gradient acceleration is considered as the
main characteristic of air movements in many observational studies of outflow events
(Overland and Walter, 1981; Sharp, 2002; Colie and Mass, 2000; Koletsis et al., 2009), in

77

which the highest wind speed is often observed near the gap exit. Hydraulic theory explains
the mechanisms of gap-type outflows (Jackson and Steyn, 1994b; Gohm and Mayr, 2004;
and Finnigan et al., 1994). It explains the continuous acceleration of the flow across the gap
as a result of a transition from a subcritical to a supercritical regime by passing through a
critical point, and then ending in a hydraulic jump. The single-layer hydraulic theory in open
channels is used to simulate the behaviour o f outflow winds in several studies (Jackson and
Steyn, 1994b; Finnigan et al., 1994; Schar and Smith, 1993; Pan and Smith, 1999; Finnigan
et al., 1998; Jackson, 1996). In outflow events, the presence of a stable layer that separates
the colder more dense air on the bottom from the less dense air above is similar to the
water/air interface in hydraulic flows.
Outflow winds have a three-dimensional structure (Jackson and Steyn, 1994); therefore,
vertical observations are critical when examining case studies o f them. The lack o f such
observations has put limitations on conducting observational case studies at many locations.
A well-instrumented pass on the BC coast is Howe Sound-Cheakamus Valley (Howe Sound
is a glacial fjord formed in the past glaciation eras). The location of Howe Sound and some
of the weather stations are provided in Figure 4.1. This channel has a number of public
meteorological stations distributed along the gap, starting from its entrance near the Strait of
Georgia and ending at Pemberton to the northeast. This observational network is augmented
with additional meteorological stations during and prior to the Vancouver 2010 Winter
Olympics and Paralympics (Joe et al., 2014). The Olympics stations are installed at new sites
and at different elevations. In addition, a five-beam boundary layer wind profiler (915 MHz
by Vaisala) is installed at the Squamish Airport. The location of the wind profiler allows the
detection of along-channel winds near Squamish, ~10 km northeast o f Howe Sound. This

78

enhanced meteorological monitoring network provides sufficient data to carry out an
observational case-study o f outflow events along the Howe Sound-Cheakamus Valley
channel during the period of intensive observations in the winter season of 2009/2010.
Several outflow events occurred during the period of intensive observations, allowing
analysis o f two outflow cases that have different initial conditions, as well as good quality
wind-profile data.
Outflows in Howe Sound, which are known locally as Squamish Winds or
“Squamishes” (Lange, 1998), have been an object o f several studies before. Jackson and
Steyn (1994a) provide a comparison between observations o f an outflow event at Howe
Sound and a three-dimensional numerical model o f gap winds. The comparison shows a
good agreement, and the model outputs indicate a hydraulic behaviour o f the flow. They
apply a one-dimensional hydraulic model in a subsequent paper (Jackson and Steyn, 1994b),
which appears successful in predicting the gap flow compared with observations. Similar
results are found in Jackson (1996) regarding the hydraulic behaviour of an outflow event in
Howe Sound. Another small-scale hydraulic-physical model that simulates outflows in Howe
Sound shows a good agreement with observations and confirmed the hydraulic behaviour of
the flow (Finnigan et al., 1994). However, all these papers study outflow events in Howe
Sound using observations mainly for comparing them with model outputs, and all the
considered events have similar synoptic-scale situations. In addition, there are very limited
observations o f the flow above the surface to support these previous studies in Howe Sound.
However, flows within Howe Sound have not been characterized, and their climatological
effects have not been determined.

79

*ttiTt*tt

\

B c b 'M d A A

Figure 4.1. A topographic map of Howe Sound - Cheakamus Valley showing the
locations of the meteorological stations that are used in this study.
This study is motivated by the previously mentioned aspects of outflows as well as
inflows and aims to improve the capabilities of forecasting these winds in Howe Sound. A

80

climatological assessment o f outflows and inflows is conducted in Howe Sound. In addition,
an observational analysis is presented for two outflow case-studies that have different
synoptic-scale conditions. The case-study analysis utilized the period o f intensive
observations associated with the Vancouver 2010 Winter Olympics. In section 4.3 a
description of the data used as well as methods is provided. The climatological study o f the
gap winds is presented in section 4.4. Section 4.5 is devoted to the case-study analysis, and a
conclusion is drawn in section 4.6.
4.3. Data and Methods
In situ meteorological observations are distributed along the gap as well as outside its
topographic limitations. This allows examining both the along-channel meteorological
gradients, as well as the funneling effects o f the gap on the wind behaviour outside the
channel. Table 2.2 shows the different sources o f observations used in this study (see Figure
4.1 for locations). In situ hourly observations at six stations along the gap operated by
Environment Canada are extracted from the Historical Climate Data Archive (Environment
Canada, 2014). Meteorological observations from two Air Quality stations operated by the
BC Ministry of Environment are extracted from the BC Air Data Archive (BC Environment,
2014). Intensive in situ meteorological observations operated by Environment Canada prior
to and during the Vancouver 2010 Winter Olympics are utilized at four locations: the wind
profiler data at Squamish Airport and three meteorological stations near Whistler Mountain
(Joe et al., 2014; Isaac et al., 2014; Gultepe et al., 2014; and Joe et al., 2010). The half-hourly
wind profiles are used along with hourly RASS (Radio Acoustic Sounding System) profiles
of virtual temperature to depict the vertical structure o f the wind flows for two outflow wind
events. A Signal to Noise Ratio (SNR) threshold is used to omit the wind profiler data that
81

occur below a pre-specified SNR value and cannot be considered as valid data. The threshold
is determined following the work o f Riddle et al. (2012) as -12 dB.
SAR wind images are used to visualize the gap flow over open water surfaces for the two
outflow cases. SAR wind images are operated by the National SAR Winds Project (NSWP)
of the Meteorological Service of Environment Canada using the Canadian RADARSAT
satellites. SAR sensors image the water surface through clouds and precipitation, providing
clear information on location of surface features over water (Chan et al., 2010). The synoptic
situation o f the two outflow events is presented using hand-drawn surface-level analysis
charts over western Canada. The hand-drawn surface charts are provided by Meteorological
Service of Environment Canada’s Pacific Storm Prediction Centre. To identify the 500 hPa
atmospheric level during the outflow cases, 6-hourly NCEP-NCAR (National Centers for
Environmental Prediction - National Center for Atmospheric Research) reanalysis I project
data o f 500 hPa geopotential height are used to produce synoptic maps of the two outflow
cases. The NCEP-NCAR reanalysis data are extracted from the archives provided by the
National Oceanic and Atmospheric Administration / Oceanic Atmospheric Research / Earth
System Research Laboratory - Physical Sciences Division (NOAA/OAR/ESRL PSD),
Boulder, Colorado, USA (Kalnay et al., 1996).
A climatological study is conducted for the winds and pressure along the channel to
identify the best locations to detect inflows and outflows, and to assess the along-channel
PG. The coefficient of determination is calculated for the wind speed - pressure gradient
correlations, and then the one-tailed Student’s t test is calculated for these correlations at p <
0.05. Seasonal gradients o f the along-channel air temperature and humidity profiles are
created as a function of wind speed and direction to obtain a clearer picture o f climatic
82

gradients along the gap. Then, two outflow events are selected during the period of intensive
observations to conduct an observational case-study. This includes synoptic investigations
for both cases for the MSLP and 500 hPa geopotential heights, and pressure, temperature,
and humidity gradients are determined at different locations along the channel. The
horizontal and vertical structures of both cases are investigated using the SAR images and
the wind profiles, respectively.
4.4. Climatology of Outflows/Inflows
Wind regimes within Howe Sound are largely determined by the tunneling effects of the
fjord. The topography is characterized by high mountains on both sides o f the channel. This
topographic system persists northeastward o f Howe Sound through the Cheakamus Valley
until Pemberton but with less restricted mountain walls. Howe Sound widens southwestward
before it opens onto the Strait o f Georgia, where a number of islands are located within the
fjord (Figure 4.1). The complex topography o f Howe Sound affects the spatial distribution of
the wind flows over the water surface (Jackson and Steyn, 1994a; Finnigan et al., 1998). Air
movement within this complex topographic system is highly channeled and observed mainly
in the two directions that are parallel to the channel orientation. The wind that flows up or
down the channel is largely determined by the orientation o f the along-channel PG, which
can be induced by specific synoptic-scale conditions, or by regional thermal differences
between both sides o f the barrier. To quantitatively assess the channelized wind climatology
in Howe Sound, the following analyses have been conducted to illustrate the climatological
gradients within the gap.

83

4.4.1. Wind Climatology
Seasonal wind roses for summer, winter and transition (spring and autumn) seasons are
created for a number o f weather stations within the gap and outside it (Figure 4.2). Wind
observations are available for 3-5 years at some weather stations and for longer periods at
others. A comparison between a 3-year wind rose and a 10-year one shows very similar
results at three locations; therefore, a 3-year wind rose is used for all weather stations. The
correspondence between wind frequencies and channel orientation shown in the wind roses
in Figure 4.2 illustrate the importance of topographic effects on the wind behaviour within
the channel.
The bidirectionality is shown in ail roses, even for those outside the gap such as at
Entrance Island and Point Atkinson where winds are channeled by the wider Strait of
Georgia. Pemberton recorded light winds, and the directions reflect the topographic location
of the station. Whistler Remote-wind is located nearly outside the channel at 1690 m above
sea level and around 1000 m above the valley bottom. The bidirectionality is less pronounced
here, but cannot be explained by only the channeling effects because of the high topographic
complexity of the station location. Katabatic winds account for Whistler Remote-wind flows,
which blow down different slopes surrounding the station. Each o f these stations (Entrance
Island, Point Atkinson, Pemberton and Whistler Remote-wind) is outside the tunneling
effects of Howe Sound-Cheakamus Valley. The location o f Langdale, near the termination of
the fjord, does not allow it to detect inflows since there is not enough distance for inflows to
accelerate into the channel. Therefore, southwesterly winds at Langdale during the summer
are due to sea breezes. Furthermore, Langdale cannot detect outflows because it is located on

84

the lee side of Gambier Island. The light northwesterly winds during the winter and transition
seasons are likely due to the katabatic flows blowing from the nearby slopes to the west.
Each o f Pam Rocks, Squamish and Whistler are located within the gap. They all show
high bidirectionality but with speeds becoming stronger from Whistler to Pam Rocks. The
high surface roughness surrounding the Whistler location would reduce the wind speed there,
preventing that station from detecting high wind speeds. In addition, the Whistler station is
near the channel pass, and if the flow behaves as a hydraulic flow, then it is expected that the
strongest winds would be observed downstream. At Squamish, surface roughness is less than
at Whistler, the station is downwind from the channel pass, and outflows are detected during
the winter and transition seasons, but the speeds do not exceed 9 m s '1. However, downchannel winds frequently exceed 15 m s'1 at Pam Rocks. This is due both to the acceleration
o f outflows before reaching Pam Rocks, as well as to the lower roughness over the water
surface. Inflows are observed at both Squamish and Pam Rocks during summer with similar
wind speeds. Pam Rocks (similar to Langdale) is located near the channel terminus and
blocked by Gambier Island to the south, and as such it is not an ideal place to detect inflows.
This analysis suggests that Pam Rocks (Squamish) is the most appropriate location for
detecting outflows (inflows). The following climatological analysis therefore uses data from
Squamish for inflows and Pam Rocks for outflows.

85

JJA

DIF

DJF

JiA

TRANS

DIF

1

t

$ 0

TRANS

JJA

TRANS

€3 *•>*■« jjp
11.0 161

DIF

9.0 -12.0
t.O> 8.0

I

MI

^gJT

30- *0
0 5 - 3.0

m
JJA
DIF

h

TRANS>

JJA

i

i
TRANS
DIF

JJA

TRANS

OIF

TRANS

DIF

JJA

^ _ ■Sfcjta

TRANS

JJA

_. iSyte

Figure 4.2. Wind roses of a 3-years period for summer (JJA), winter (DJF) and transition
seasons (TRANS). Concentric rings are frequency in 5% increments. Wind directions are
divided into 16 groups.

86

4.4.2. Wind Speed - Pressure Gradient Correlations
The along-channel PG is used in many studies to explain and quantify the wind speed at
the gap exit (Overland and Walter, 1981; Colle and Mass, 2000; Sharp and Mass, 2004;
Koletsis et al., 2009). Hourly pressure observations corrected to mean sea level, are available
for a number o f weather stations employed in this study. These observations are used to
calculate pressure gradients between Pam Rocks, Pemberton, Squamish Airport and Entrance
Island. The Squamish air quality station does not measure surface pressure; therefore
pressure data from Squamish Airport, which is ~9 km northeast of Squamish, are used. It is
noteworthy that Squamish Airport does not detect inflows or outflows. Reported wind speeds
are used for all years, and wind speeds are chosen from specific sectors for outflows
(inflows) using the yearly wind rose at Pam Rocks (Squamish).
Figure 4.3a shows the relationship between various pressure gradients and outflow winds
at Pam Rocks over a 10-year period. Strong relationships are found between stations along
the gap for both Pemberton - Pam Rocks and Squamish Airport - Pam Rocks, where the
coefficient of determination R2 = 0.65, 0.66 respectively. These high R2 values indicate that
pressure gradients between Pemberton - Pam Rocks and Squamish Airport - Pam Rocks can
both be correlated to the outflow strength at Pam Rocks. To assess the statistical significance
o f R2, the one-tailed Student’s i test is calculated at p < 0.05 with n-2 degrees o f freedom
(Koletsis et al., 2009; Devore, 2010). The Statistical test shows whether the R2 value is
significantly different from zero.
The calculated t values for Pemberton - Pam Rocks and Squamish Airport - Pam Rocks
indicate that the null hypothesis o f R2 = 0 cannot be accepted and the correlation is

87

statistically significant. Further from the gap at Entrance Island the relationships do not show
good values o f the coefficient of determination, in which R2 = 0.04, 0.01 at Pam Rocks Entrance Island, Pemberton - Entrance Island, respectively.

Pemberton - Pam Rocks

Squamish airport - Pam Rocks

(#- 066)

0

OIK

0

Pressure gradtsnl (hPa)

Pemberton - Entrance island

Pam Rocks - Entrance Island

005

0

[ # - 0. 015)

o

8

I f f ' - 0. 0*1

0 05

Pressure gredteni (hPadtm)

Prassurt gredfrnt (hPe)

Pam Rocks - Squamish airport

Entrance island - Squamish airport
0. 37)0

y-8T **31

O09

003

OEM

0 06

-0XJ2

0CB

0

002

0(M

006

Pressure gradient (hPeAoti)

Prassura gradient (ftPafcm)

Entrance - Pemberton

Squamish airport - Pemberton

I f f * - 0.46)

y« W* ♦2.2

? ^ !a>
-0®

006

034

-002

Pressure gradient (hPa/km)

006 0

0Oo o8

002

oo*

Pressure gradient (hPafan)

Figure 4.3. Scatterplots showing the relationship between PG and hourly wind speed
of (a) outflows at Pam Rocks, and (b) inflows at Squamish.
88

PG - wind speed correlations for inflows at Squamish are shown in Figure 4.3b. These
correlations are calculated for ~4 years, which is the period of available wind data from the
Squamish air quality station. The strongest along-channel correlation o f the inflows is
between Pam Rocks - Squamish Airport with R2 = 0.51. Entrance Island - Pemberton also
has a relatively strong correlation (R2 = 0.46), indicating the importance of the larger scale
PG. Squamish Airport - Pemberton correlation is weak (R = 0.2) where the thermal
differences might not be large enough, similar to Entrance Island - Squamish Airport but
with a better correlation (R2 - 0.37); however, all the inflow correlations are statically
significant
4.4.3. Air Temperature and Humidity As a Function of Gap Flows
Along-channel air temperature and humidity gradients are examined for the winter,
transition and summer seasons as a function of wind speed and direction. Figure 4.4 shows
the gradients o f the overall averages (green), outflows (black) and inflows (red) for wind
speed over 3 m s'1. This speed is selected to avoid winds resulting from local effects. On
average, outflows (inflows) have lower (higher) air temperatures. Also, outflows are
associated with lower dew points in the winter and transition seasons. The average air
temperature (dew point) gradients of the outflows between Pemberton and Pam Rocks is
6.5°C (4°C) in the winter and 4°C (3°C) in the transition season. The cooler air temperature
and dew point associated with outflows at Pemberton confirms the cold-air damming effect
that causes these northeasterly winds.
The average temperature gradient of the summertime inflows is the opposite of that for
outflows, where Pemberton is 5°C warmer than Pam Rocks. This is likely responsible for a

89

mesoscale PG that induces inflows, making the area near Squamish attractive for
windsurfing. Wintertime inflows have a similar air temperature gradient as the outflows,
indicating that wintertime inflows are caused by a reason other than the thermal gradient
such as a passage of a mid-latitude cyclone. This can be inferred from the higher humidity
values at Pemberton, Squamish and Pam Rocks during wintertime inflows. The transition
seasons always has moderate gradients that occur between the winter and summer gradients.

W in te r

T ran sitio n

Sum m er

21

21

18

18

® 15

15

15

£ 12

12

12

9

9

6

e

3

3

0

0

-3

•3

.M'
a>

PM

SQ

PR

„,J§

El

Figure 4.4. Seasonal averages of temperature and dew point at stations
along Howe Sound and Cheakamus Valley as a function of wind speed
and direction. Solid lines show temperature and dashed lines show dew
point. Green lines with cross markers show the overall means, red lines
with stars show the inflows and black lines with squares show the
outflows. PM is Pemberton, SQ is Squamish Air Quality, PR is Pam
Rocks, and El is Entrance Island.

4.5. Case Studies
During the period o f intensive observations in the 2009/2010 winter, only six outflow
events are observed at Pam Rocks. The Beaufort scale 6 (11.3 m s'1) is used as a wind speed
threshold for determining a wind event. Two of these events are chosen for a detailed case
90

study analysis. The events selected have different synoptic conditions and both have reliable
wind profile data. The events are determined depending on the wind speed and direction. The
directions should be located within the outflow sector at Pam Rocks. The initiation of an
event is chosen as speeds exceeding 11.3 m s '1 (Beaufort scale 6), and ending when winds
moderate to less than 7.5 m s'1 (less than Beaufort scale 5). The first event begins on 5
January 2010 at 1400 LST and lasts for three days, and the second event starts on 25
February 2010 at 0600 LST and lasts for one day and ten hours. No inflow events can be
analyzed in this study, since there is not a common summer season between the wind profiler
data and the observations of Squamish air quality station, which can detect inflows.
4.5.1. Synoptic Description
a) January Event: (Figure 4.5a)
A N-S ridge at the 500 hPa level is located across Vancouver Island throughout the event
with a trough to the east over the continental interior. The upper-level situation results in a
north to northeasterly flow aloft over western North America, which helps push arctic air
southward through BC. This is similar to the situation described by Sharp (2002) o f an
outflow in the Columbia Gorge, and Jackson (1996) for an outflow event in Howe Sound and
the Fraser River Valley. Associated with the ridge aloft, a mean sea level pressure (MSLP)
anticyclone is located over northeastern BC with a centre of 1048 hPa over the southern
Northwest Territories. A strong across-barrier MSLP gradient (MSLPG) of ~ 0.03 hPa km '1
formed over the southern BC coast between the MSLP anticyclone and a MSLP cyclone over
the Pacific Ocean. An arctic front separates the cold air of the interior anticyclone and the
warmer air over the Pacific Ocean. The passage of this front indicates the onset o f the

91

outflow event (Jackson and Steyn, 1994a). Eight hours before the end o f the event, the
anticyclone weakens and moves southeastward.
b) February Event: (Figure 4.5b)
The synoptic situation of the February Event is different from the January Event. In the
February Event, there is not a continental anticyclone present in the vicinity o f the study
area. The main feature o f this event is a MSLP cyclone offshore o f Washington State and
southern BC that is approaching the region. The MSLPG across the southern BC coast is
weaker than in the January Event with values o f - 0.02 hPa km '1. The surface cyclone is
accompanied by a trough at 500 hPa level indicating the presence o f an occluded mid­
latitude cyclone. The surface cyclone advances eastward throughout the event until it
weakens and passes the coast toward eastern BC and Alberta pronouncing the end o f the
event.

92

5520
54j

' “U. *

-9340

\

\

5460
5520

\s

5580
i220

5340
5280
5640

January Event 400 hP» level

(*)

January Event Surface analysis chart

5400

5340
■9521

5340

5400

February Event 600 hP*

(•»)

February Event Surface analysis chart

Figure 4.5. (a) January Event surface analysis chart and the associated 500 hPa geopotential
height level on 2010-Jan-06 (0600 UTC), (b) February Event: surface analysis chart and the
associated 500 hPa geopotential height level on 2010-Feb-26 (1200 UTC).

4.5.2. Observational Description
To examine the gradients along the channel, each of the wind speed, air temperature,
humidity and PG are depicted at several locations including Pam Rocks, Squamish and
Pemberton during both wind cases.

93

Figure 4.6a shows the wind speed at Pam Rocks and Squamish during the January Event.
The first 24 hours o f the event observe the highest wind speed at both locations. The speed at
Pam Rocks is higher than Squamish by 7-13 m s '1 likely due to the change o f surface
roughness conditions and to the acceleration o f the flow between Squamish and Pam Rocks.
A daily variation can be seen in the wind speed distribution, in which the highest speed
during a day is recorded at around 0600 LST matching the higher PG values between Pam
Rocks and Pemberton (Figure 4.6b). This suggests that the diurnal heating cycle may play a
role, due to the lowest air temperatures in the early morning. The cooling at Pemberton is
higher than at Pam Rocks due to the radiational cooling at Pemberton and the marine effects
at Pam Rocks (Figure 4.6c). This thermal difference between Pemberton and Pam Rocks
creates a mesoscale PG that enhanced the synoptic-scale one. In general, the January Event
is accompanied by lower air temperatures and humidity levels along the channel (Figure 4.6c
and 4.6d) with values lower at Pemberton and higher at Pam Rocks.

94

0000

0*01/10

1200

0000

0*01/10

1200

0000

1200

orm/io

0000

0*01/10

1200

0000

i:

0*01/10

Thm (LST)

Figure 4.6. (a) Wind speed during the January Event at Pam Rocks (blue) and Squamish
(red), (b) PG between Pemberton and Pam Rocks during the January Event, (c) Potential
temperature time series during the January Event at Pam Rocks (blue), Squamish
(black) and Pemberton (dashed red), (d) Dew point time series similar to (c). The two
vertical lines represent the start and the end of the event. The spike in dew point
temperature at Pam Rocks at 2000 LST on 5th January indicates a suspicious observation
since it was accompanied with lower wind speed, change in wind direction and higher
relative humidity than the previous and subsequent observations.
95

Figure 4.7 shows the gradients observed during the February Event. Many
characteristics o f the February Event are different from the January Event. First, PG values
are less than those of the January Event and they do not exceed 0.04 hPa km '1 except for four
hours at the beginning o f the event. The PG is always more than 0.05 hPa km'1 during the
January Event. This is reflected in the wind speed, which shows lower values during the
February Event (Figure 4.7a and 4.7b). Second, the differences in the air temperature and
humidity during and prior to the event in the February Event is less than the January Event,
particularly at Pam Rocks and Squamish; however, they still show lower values during the
event (Figure 4.7c and 4.7d). The most significant characteristic is that the February Event
shows higher air temperatures than the January Event. Air temperature and dew point are
always above 4°C at Pam Rocks and Squamish and above 0°C at Pemberton during the
February Event, however, they are mostly below these values during the January Event. This
can be explained by the different synoptic situations for both events. The mid-latitude
cyclone that created an across-barrier PG during the February Event is not accompanied by
an arctic front. Consequently, air temperature and dew point do not record subfreezing values
along the channel during the February Event. Finally, Pemberton - Pam Rocks thermal
differences during the February Event are less than those o f the January Event, which may
result in a weaker mesoscale PG along the gap.

96

14

E.

i
«
0.05

I

££

!

004
0.03
002

0 01

- 0.01

I
t

i

i
i
!

W

t

e
9

I
««

3•
0000

0600

1200

1800

0000

0B0C

<200

1800

OOOO 0900

1200

1800

0000

Time (LST)

Figure 4.7. (a) Wind speed during the February Event at Pam Rocks
(blue) and Squamish (red), (b) PG between Pemberton and Pam
Rocks during the February Event, (c) Potential temperature time
series during the February Event at Pam Rocks (blue), Squamish
(black) and Pemberton (dashed red), (d) Dew point time series
similar to (c). The two vertical lines represent the start and the end of
the event.

97

To understand the wind speed increase between Squamish and Pam Rocks during
both events, and to characterize the relative role of acceleration due to decreased frictional
drag versus acceleration due to hydraulic effects, a simple model of channelized wind as
balance between pressure gradient and friction is used. This model, called the “Friction
Model” in Jackson (1996) represents the wind speed that would be attained in a long channel
of constant elevation, cross sectional area, external pressure gradient and surface roughness
in which the depth o f the flowing air was constant. In the Friction Model, the wind speed u

=J

£ h , where PGF is the pressure gradient force (PGF =

with p - air density), h is

the flow depth, and C is the drag coefficient. In Jackson (1996), h/C values o f 35000 m is
used for outflows over land (Lower Fraser Valley), and 80000 m is used in Howe Sound.
Assuming that the surface roughness upwind o f Squamish is roughly similar to the surface
roughness in the Lower Fraser Valley and the PGF is 0.003 m s'2 the following results are
obtained. The steady wind over water is 15.49 m s '1 and the steady wind over land is 10.24
m s'1. Therefore the modelled wind at Pam Rocks is 33.90% higher than modelled wind at
Squamish. However, during both events (Figures 4.6a and 4.7a), the observed winds at Pam
Rocks are between 84.30% and 65.55% higher than at Squamish. . The larger observed
differences during the events are probably explained, in addition to the decreased frictional
drag force experienced at Pam Rocks, by the acceleration due to hydraulic effects between
Squamish and Pam Rocks.
The synoptic situation of both events affects the MSLP changes along the gap. Figure 4.8
shows these changes at five weather stations prior, during and after the events. The January
Event is accompanied by higher MSLP at all stations suggesting the effect of the arctic
anticyclone. The surface pressure difference is also larger between Pam Rocks and
98

Pemberton throughout the January Event compared with the February Event. The opposite
situation is seen during the February Event due to the advance of the mid-latitude cyclone
toward the channel which causes lower MSLP values at Pam Rocks than at Pemberton,
creating the PG along the channel.

1040

1035

1030

m
£ 1025
Q»

ft
Z

1020

1015

(a)

1010

0000

1200

0000

1200

0000

0000

1200

1200

0000

1200

T
05/01/10

00/01/10

07/01/10

00/01/10

00/01/10

1020

1016

i

1012

0)

1008

1010

z

1006

1004
1002

1000

0000

0600

1200

1800

0000

0600

1200

1800

0000

0800

1200

1800

0000

M /42/10

T im e (LST)

Figure 4.8. MSLP during the (a) January Event, and the (b)
February Event at Pam Rocks (blue), Squamish Airport (dashed
black), Whistler (dashed green), Whistler Mountain High-level
(pink), and Pemberton (dashed orange). The two vertical lines
represent the start and the end of the event.
4.5.3. The Outflow Events in the SAR Images
One SAR image is obtained for each outflow case. The images have different coordinates
with a spatial resolution of 1 km over water surface, but they both include Howe Sound

99

(Figure 4.9). SAR images characterize some important spatial features o f the flows. The
January Event is shown in Figure 4.9a. The image is acquired ~ 2 hours after the maximum
wind speed recorded at Pam Rocks. When zoomed in to Howe Sound, a large spatial
variability can be seen. The plume of the outflow is restricted to the channel limits until the
area where the sound widens, and then it does not occupy the whole width o f the channel.
The flow continues after it enters the Strait of Georgia. The flow starts to accelerate from
Watts point until Pam Rocks, with speeds ranging between ~ 10 - 14 m s '1 (~20 - 28 knots)
along the axis of highest speed within the channel. South o f Pam Rocks, the plume weakens
again to less than 10 m s 1 as it encounters the complex topography o f the channel. The
January Event features are probably due to the hydraulic characteristics o f the flow. The
region o f highest speeds starts after a constriction in the channel near Watts Point that would
cause acceleration and supercritical behaviour. The supercritical flow persists until it
encounters Anvil Island. However, it does not enter Thombrough Channel, as it appears
south o f Anvil Island and northeast o f Pam Rocks. South o f Pam Rocks, the flow appears
subcritical after it encounters Gambier Island. It is possible that the flow can overcome Anvil
Island, in the case of the outflow layer is higher than the island summit. This might cause a
downslope windstorm south o f Anvil Island, which appears near the northeastern coast of
Gambier Island.

100

RADARSATl SCWA_W DncamMng, 0e.jan.7010 14:33:31 UTC
WHO O etM AM Wind BarixRafXXtad at d tM tt hour UTC

nA O M SA T -iX W A jrvO m am H iig. Z J fa b Z m 14:1731 UTC
With G£344-AMWhtd B atin Bapcrtart a l doaaat hour UTC
Buoy m ndaBatxrtatl OnZr.FaOJOK) >4.(030 UTC

Figure 4.9, SAR images during the (a) January Event, and the (b) February Event. Zoomed
in images of the Howe Sound region are depicted on the left with the larger scene to the
right.

In the February Event (Figure 4.9b), the image shows less spatial variability, but is still
affected by the topography o f the sound. Higher wind speeds are shown near the southeastern
edge o f the channel before reaching Pam Rocks. It is noteworthy that the February Event
101

image is taken during the last hour of the event. The synoptic situation o f the February Event
would not cause hydraulic features as no cold air damming is formed east of the mountain
barrier such that it is less likely that the layer o f outflowing air is shallow and o f distinctly
different density than the air above it.
In the larger scene, Figure 4.9a shows a flow blowing out of the Fraser Valley toward
Juan de Fuca Strait, and then it goes into the open ocean west o f Vancouver Island with
higher wind speeds. Other northern inlets observed outflows with higher wind speed such as
in Burke Channel at the northwest comer of the image, and Johnstone Strait northeast of
Vancouver Island. However, during the February Event such outflows are not observed at
Juan de Fuca Strait or out o f the Fraser Valley. Again, this can be explained by the absence
o f an arctic anticyclone over the interior o f the continent during the February Event, which
can create outflows throughout many inlets along the BC coast (Jackson, 1996). The
horizontal features of the flows provided by the SAR images are supported by the vertical
observations at Squamish Airport which help to show the 3-D behaviour o f the flows.
4.5.4. Vertical Observations
The wind profiler data enable analysis of the 3-D atmospheric structure within the gap
about 10 km northeast o f the end of Howe Sound. The wind profiler is located hydraulically
within the subcritical region, wherein the surface wind speed is light and does not exceed 4 5 m s'1 near the surface in most cases. Figure 4.10a shows wind components presented as
vectors while the calculated along-channel outflow wind component during the January
Event is shown in Figure 4.10b. The along-channel direction is 340° at the wind profiler

102

location (from topographic maps). The along-channel outflow wind component, Vac, is
calculated as:
Vac = M cos (Wd - Aid)
where M is the reported wind speed, Wd is the reported wind direction, and Aid is the
along-channel direction (340°). Positive values o f Vac indicate wind with a component in the
outflow direction, while negative values indicate wind with a component in the inflow
direction. The height of the N-NE flow is about 2 km when the wind speed is more than 15 m
s'1 at Pam Rocks and near 6 m s 1at Squamish during the first day o f the event. Then, it thins
to less than 1 km when speeds are less than 15 m s '1 at Pam Rocks and around 3 m s _1 at
Squamish during the last day o f the event. RASS profiles are not available during the
January Event. The vertical observations during the February Event also have better quality
than those during the January Event, and the RASS profiles are available. The February
Event vertical profiles are shown in Figure 4.11.

103

1400 2000

0200

0800

1400 2000

0200

0800

1400

14002000

pOO

0800

14002000 < p »

0800

1400 2000 (jfOO 0800

04/81/10

07(01/10

2000

0200

0800

04I41/10

Time (LST)

Figure 4.10. (a) Wind components at Squamish Airport during the January
Event within the first 2 km. (b) The along-channel wind component during
the January Event. Areas in the plot that are blank indicate a lack of
reliable profiler data.

“ T—

— T“

27/02/10

26/02/10

Tim e (LST)

Figure 4.11. (a) Wind components at Squamish Airport during the February
Event within the first 2 km. (b) The along-channel wind component during
the February Event, (c) Potential temperature profile during the February
Event. Areas in the plot that are blank indicate a lack of reliable profiler
data.

A layer based between 0.7 - 1.5 km defines the top of the outflow layer in the February
Event. This layer is shown in Figure 4.1 lb as the green color where the along channel wind
component is near zero and in Figure 4.1 la as a layer o f very small vectors. The air above
this layer is not affected by the topography of the gap. Similar results are found by Jackson
and Steyn (1994a) when an outflow event is simulated by a 3D numerical model. The
potential temperature profile (Figure 4.1 lc) shows a near-surface layer of cold air. The cold
air layer approximately matches the outflow layer in Figure 4.11b suggesting that this is a
statically stable air layer. Since the wind profile is taken at one point throughout the gap, it is
not possible to identify the hydraulic features of the flow from the vertical observations .
Vertical potential temperature profiles are drawn four times during the February Event and
provided in Figure 4.12. A stable layer is detected at different elevations throughout the
event. The base o f the stable layer is determined where the potential temperature begins to
increase rapidly with height. The stable layer height ranges between 0.76 to 0.78 km during
the event (4.12).

Figure 4.12. Vertical potential
temperature profiles during the
February Event. The red line at
the start of the event (25/02/2010
- 2100 LST), the blue line 14
hours after the start of the event
(26/02/2010 - 1100 LST), and
the black line at the end of the
event (27/02/2010 - 0600 LST).

1 .2 S

1
s 0 .7 6
I
0 5
0 .2 5

7

6

O

10

11

12

13

P o te n tia l t a m p a r a tu r a (C*)

106

14

16

4.6. Conclusion
A climatological study o f the along-channel flows in the Howe Sound - Cheakamus
Valley channel was conducted over a longer period to identify the meteorological gradients
throughout the gap and the best places to detect channelized flows in both directions, and to
seek associations between channelized flows and other parameters such as pressure
gradients. This study has also examined two outflow cases in Howe Sound that occurred
during the period of intensive observations during the Vancouver 2010 Winter Olympics.
The observations included wind profiler data at Squamish Airport northeast of Howe Sound.
All stations located within the channel showed a bidirectionality in wind behaviour.
The pressure gradient between Pemberton - Pam Rocks as well as Squamish Airport- Pam
Rocks was found to have the best correlation with the outflow strength at Pam Rocks, and
the pressure gradient between Pam Rocks - Squamish Airport had the best correlation with
inflow strength at Squamish. Outflows (inflows) were accompanied by lower (higher) air
temperature and dew point.
The synoptic-scale situations o f the two outflow cases considered were different.
The January Event was caused by an arctic anticyclone that resulted in a strong acrossbarrier PG. During the January Event at Howe Sound, other outflows were also found
blowing out of many channels along the BC coast as shown in the SAR image o f the event.
The main synoptic factor causing the February Event was a mid-latitude cyclone that
advanced from the Pacific Ocean toward the southern BC coast, causing the across-barrier
PG that resulted in the outflow, which was weaker than in the January Event. Large thermal
differences were found between Pemberton and Pam Rocks during the January Event that

107

would result in a mesoscale PG enhancing the synoptic-scale situation and leading to
stronger wind speeds. However, these air temperature differences were less during the
February Event. The horizontal distribution o f the flow during the January Event showed
potential hydraulic behaviour of the flow, although the January Event showed more spatial
variability than the February Event in the SAR images. The vertical profile o f the two cases
showed an along-channel outflow layer up to 1.5 km deep during the February Event and 2
km deep during the January Event. The outflow layer had cooler potential temperatures than
the overlying layer in the February Event. Such air temperature profiles were not available
for the January Event. The outflow layer was surmounted by a thin layer which forms the
upper limits o f the gap flow. A third layer, not affected by the channeling effects of the gap,
is found above this mid-layer.

108

5. Conclusions, Further Research and Recommendations
5.1. Introduction
Wind is one of the most important meteorological parameters, especially over the sea and
in coastal and mountainous areas. This is particularly true for some locations that are
vulnerable to storms. Many o f the channels and passes along the BC coast are examples of
such locations as they frequently observe channelized winds as inflows and/or outflows. This
research is devoted to studying these flows. The findings of this research classify these flows
in separate types, which have been studied synoptically. Flows in Howe Sound have been
characterized through case studies. The synoptic-scale study illustrates some o f the
meteorological features that lead to the channelized winds, which blow through channels that
are perpendicular to the mountain barrier along the BC coast. Identifying the synoptic
features and conditions associated with these flows is one o f the main objectives o f this
study. The goal o f characterizing the detailed structure of a gap flow along the BC coast and
identifying its climatological effects was achieved through the observational study of
outflows/inflows in Howe Sound, where an enhanced observational network was installed for
the 2010 Olympic Winter Games. The following section highlights the overall findings from
this study.
5.2. Summary of Results and Conclusions
The findings of this study address its objectives that are provided in chapter 1 as well as
answer its research questions. The objective #1 “determine the synoptic-scale patterns that
are associated with outflows and inflows during different seasons and throughout several
gaps” is addressed throughout the composite analysis as follows: synoptic maps o f the

109

wintertime outflows show a down-channel MSLPG in all gaps. The synoptic conditions of
these flows at sea level are explained by the cold-air damming effect. This phenomenon
results from the advance of arctic air from the Yukon and Alaska toward central and southern
BC. The cold, stable, shallow air is blocked by the coastal chain resulting in high surface
pressure on the inland side o f the mountains. This situation leads to formation o f a PG from
the interior towards the coast that is parallel to the fjords that dissect the BC Coast
Mountains. It is found that wintertime outflows in Juan de Fuca Strait, due to the length of
the strait, are probably affected by the Coriolis force in addition to the along-channel PG.
Investigating the contribution of other potential forces that could also affect the flows
(friction, entrainment, and inertia) is not the focus o f this study and cannot be achieved with
the available observations.
The wintertime inflow composites show a low-pressure area offshore o f the BC coast at
all gaps. The composite and anomalies o f this flow type show that the isobars are parallel to
the channel orientation (rather than perpendicular as for wintertime outflows), and become
close together at the weather station locations. Since wintertime inflows accompany lows and
appear to be in geostrophic balance, they are not limited to gaps and cannot be considered as
gap winds. The composites o f the summertime flows (both outflows and inflows) are not
significantly different from the climatology. The summertime outflows are infrequent winds.
They are likely caused by summertime stormy periods in which a low approaches BC from
the SW. Summertime inflows are differentiated from the sea breeze through the synoptic
analysis.
The objective #2 “identify other weather phenomena, which do not appear in the synoptic
composites, but are potentially important fo r some o f the flow types” is addressed by the
110

frontal investigation as well as the k-means clustering as follows: the investigation of the role
of fronts on summertime inflows points out the importance o f a frontal passage in initiating
them, particularly for Qualicum and Burke Channel. In the Strait o f Juan de Fuca, an inland
MSLP low is more frequent than the presence o f a frontal system. Fronts usually are oriented
NE-SW or N-S and have an eastward movement. K-means clustering explains the variance
of the summertime inflow events better than the composites for Qualicum and Burke
Channel. A stronger MSLPG is found in the clusters than for the composite. The surface
pressure fields in the clusters explain the flows synoptically and show clearer differences
from the climatology. The results o f studying the synoptic climatology o f strong alongchannel winds along the BC coast answer the first research question of this study “what are
the synoptic patterns associated with different outflows/inflows on the BC coast?”
The objective #3 “determine the local climatology o f the along-channel flow s in one o f
the gaps - Howe Sound” is addressed by the climatological study of gap winds in Howe
Sound, and the objective #4 “investigate the different characteristics o f gap flows in Howe
Sound through studying the horizontal and vertical structure o f the flow using the case-study
method1'' is addressed by the two case studies of outflows considered in Howe Sound. The
climatological study of Howe Sound outflows and inflows shows significant effects o f the
channelized flows on the climatology of the sound. Outflows (inflows) are accompanied by
lower (higher) temperature and dew point. Winds are highly channeled within Howe SoundCheakamus Valley. Statistically significant correlations between wind speed and the pressure
gradient are found between weather stations within the gap for outflows, and between
stations within the gap and/or outside it for inflows. The detailed study o f two outflow cases
in Howe Sound, which have different synoptic conditions, show different MSLP gradients

111

along the channel as well as different wind speeds, air temperature and humidity gradients.
The comparisons between both cases lead to the conclusion that the synoptic situation affects
meteorological parameters in addition to the wind that accompany gap flows. The gap flow
layer is detected in both cases in the vertical observations. The gap-flow layer depth is up to
2 km in the outflow event that is caused by the arctic anticyclone, and up to 1.5 km in the
other case that is caused by a mean sea level cyclone that advanced toward the BC coast. The
gap flow layer is surmounted by an air layer that has a lower wind speed. The observational
case study analysis confirmed that gap winds at Howe Sound can result from different
synoptic situations, including a continental arctic anticyclone or an offshore surface low
advancing toward the southern BC coast. The results of studying along-channel winds in
Howe Sound climatologically and by the case studies answer the second research question of
this study “what are the effects o f the channelized winds on the climatology o f Howe SoundCheakamus Valley region? Moreover, what are the different characteristics o f Howe Sound
gap flo w T ’
5.3. Recommendations and Suggestions for Further Research
The current research studies channelized winds at several locations that have not been
studied before, as well as identifies some flows that need further investigation. Observational
case studies on inflows should be conducted at several locations including Howe Sound.
Inflows near Squamish are important for windsurfing during summer. Therefore, an
observational network could be installed for a summer season near Squamish and along
Howe Sound to detect inflows. The network should include vertical observations that reach
at least to the mountaintop. The ideal location for the vertical observations would be at
Squamish close to Howe Sound/Squamish River mouth. Another surface observational
112

network should be deployed further inland. Mesoscale modelling simulations would help
with the understanding o f inflows. The horizontal resolution o f a mesoscale model would
provide a better solution of the turbulent friction within the gaps. Numerical simulations
would solve a complete set of governing equations providing a better understanding o f all
forces that affect gap flows. Similar research should be conducted on inflows in other
channels such Portland Inlet and Douglas Channel on the north coast; however, Howe Sound
might represent similar channels in the southern coast such as Jervis Inlet, Toba Inlet and
Desolation Sound.
Regarding the outflows, the results of this study show an interesting horizontal
distribution of outflow winds over the water surface in Howe Sound in the two wintertime
cases. Further investigations should be conducted on the hydraulic features that have been
found in these wind cases. Shallow water modelling simulations, perhaps supplemented with
full mesoscale model simulations could be used in such an investigation. In addition,
potential observations should be extended to include autumn and spring seasons for detecting
outflow events that are not caused by arctic anticyclones.
The Qualicum Wind is a special case that is considered in this study. It is worth
mentioning that an observational network was installed during the summer o f 2012 in hopes
o f detecting Qualicum Wind events. These observations are conducted as a part of the
current study. Unfortunately, the network was unable to detect Qualicum Wind events due to
the lack o f wind events and some technical problems. Similar observations are recommended
in the future to study the Qualicum Wind. The observations should be distributed along the
mountain gap o f Home Lake/Cameron Lake, starting from Albemi Inlet and the city o f Port
Albemi further east to the north of Qualicum Beach, as this is the main path o f a Qualicum
113

Wind. Homby Island and northwest Sisters Islets, are other locations that observes Qualicum
Wind frequently. 3D mesoscale model simulations may provide a more complete
understanding of Qualicum Wind. The model simulations would explain the behaviour of
Qualicum Wind horizontally over the topographic gap and the Strait o f Georgia, and
vertically showing whether downslope windstorms accompany the Qualicum Wind.

114

Bibliography
Abbs, D. J., & Physick, W.L. (1992). Sea breeze observations and modelling: a review.
Australian Meteorological Magazine 41, 7-19.
Brewer, M. C., Mass, C. F., & Potter, B. E. (2012). The west coast thermal trough:
climatology and synoptic evolution. Monthly Weather Review 140, 3820-3843. DOI:
10.1175/MWR-D-12-00078.1.
Burlando, M. (2009). The synoptic-scale surface wind climate regimes of the Mediterranean
Sea according to the cluster analysis o f ERA-40 wind fields. Theoretical and Applied
Climatology 96, 69-83. DOI: 10.1007/s00704-008-0033-5.
Chan, K.,Snyder, B., & Neil, L. (2010, September). Using SAR winds to evaluate synoptic
and mesoscale features in forecast operations. Poster Session 9, Operationally-Driven
Satellite Research and Application Development, Annapolis, Maryland, USA.
Chenoli, S. N., Turner, J., & Abu Samah, A. (2012). A climatology of strong wind events at
McMurdo station, Antarctica. International Journal o f Climatology 33, 2667 -2681. DOI:
10.1002/joc.3617.
Colle, B. A., & Mass, C. F. (2000). High-resolution observations and numerical simulations
o f easterly gap flow through the Strait of Juan de Fuca on 9-10 December 1995. Monthly
Weather
Review,
128,
2398-2422.
DOI:
10.1175/15200493(2000)128<2398:HROANS>2.0.CO;2
de Foy, B., Clappier, A., Molina, L. T., & Molina, M. J. (2006). Distinct wind convergence
patterns in the Mexico City basin due to the interaction of the gap winds with the synoptic
flow. Atmospheric Chemistry and Physics 6, 1249-1265.
Devore, J. L. (2010). Probability and Statistics for Engineering and the Sciences [Adobe
10.1.4]. Retrieved from http://www.cengagebrain.co.uk/content/9781285314501 .pdf
Environment Canada. (2014).
http://climate.weather.gc.ca/

Historical

climate

data

Archive.

Retrieved

from:

Finnigan, T. D., Allen, S. E., & Lawrence, G. A. (1998). Physical modeling o f an outflow
event in Howe Sound, British Columbia. Journal o f Geophysical Research-Atmospheres,
103, 3937-3950. DOI: 10.1029/97JD03430
Finnigan, T. D., Vine, J. A., Jackson, P. L., Allen, S. E., Lawrence, G. A., Steyn, D. G.
(1994). Hydraulic physical modeling and observations of a severe gap wind. Monthly
Weather
Review,
122,
2677-2687.
DOI:
10.1175/15200493(1994)122<2677:HPMAOO>2.0.CO;2
Flamant, C., Drobinski, P., Nance, L., Banta, R., Darby, L., Dusek, J., ... Richard, E. (2002).
Gap flow in an Alpine valley during a shallow south fohn event: Observations, numerical
simulations and hydraulic analogue. Quarterly Journal o f the Royal Meteorological Society,
128, 1173-1210. DOI: 10.1256/003590002320373256
115

Gohm, A., & Mayr, G. J. (2004). Hydraulic aspects of Fohn winds in an Alpine Valley.
Quarterly Journal o f the Royal Meteorological Society, 130, 449-480. DOI: 10.1256/qj.03.28
Goyette, S. (2010). Synoptic conditions o f extreme windstorms over Switzerland in a
changing climate. Climate Dynamics, 36, 845-866. DOI: 10.1007/s00382-010-0820-4.
Gultepe, I., Isaac, G. A., Joe, P., Kucera, P. A., Theriault, J. M., Fisico, T. (2014).
Roundhouse (RND) Mountain Top Research Site: Measurements and Uncertainties for
Winter Alpine Weather Conditions. Pure and Applied Geophysics, 171, 59-85. DOI:
10.1007/s00024-012-0582-5
Halios, C. H., Helmis, C. G., Flocas, H. A., Nyeki, S., & Assimakopoulos, D. N. (2012). On
the variability of the surface environment response to synoptic forcing over complex terrain:
a multivariate data analysis approach. Meteorology and Atmospheric Physics 118, 107-115.
DOI: 10.1007/s00703-012-0209-5.
Harden, B. E., Renfrew, I. A., & Petersen, G. N. (2011). A Climatology o f Wintertime
Barrier Winds off Southeast Greenland. Journal o f Climate 24, 4701- 4717. DOI:
10.1175/2011JCLI4113.1.
Hughes, M., & Hall, A. (2010). Local and synoptic mechanisms causing Southern
California's Santa Ana winds. Climate Dynamics 34, 847-857. DOI: 10.1007/s00382-0090650-4.
Isaac, G. A., Joe, P. I., Mailhot, J., Bailey, M., Belair, S., Boudala, F. S., ... Wilson, L. J.
(2014). Science o f Nowcasting Olympic Weather for Vancouver 2010 (SNOW-V10): a
World Weather Research Programme Project. Pure and Applied Geophysics, 171,1-24. DOI:
10.1007/s00024-012-0579-0
Jackson, P. L. (1996). Surface winds during an intense outbreak o f arctic air in southern
British Columbia. Atmosphere-Ocean 34, 285-311. DOI: 10.1080/07055900.1996.9649566.
Jackson, P. L., Mayr, G., & Vosper, S. (2013). Chapter 3 Dynamically-driven winds. In:
Mountain Weather Research and Forecasting, Chow, De Wekker, Snyder, editors. SpringerVerlag. Submitted April 5, 2010. 99 pp. in revised form August 13, 2010, and August 15,
2011. Accepted.
Jackson, P. L., & Steyn, D. G. (1994a). Gap winds in a fjord. Part I: Observational and
numerical-simulation. Monthly Weather Review 122, 2645-2665. DOI: 10.1175/15200493(1994) 122<2645:GWIAFP>2,0.CO;2.
Jackson, P. L., & Steyn, D. G. (1994b). Gap winds in a fjord. Part II: Hydraulic Analog.
Monthly
Weather
Review
122,
2666-2676.
DOI:
10.1175/15200493(1994)122<2666:GWLAFP>2.0.CO;2.
Jain, A. K. (2010). Data clustering: 50 years beyond K-means. Pattern Recognition Letters,
31, 651-666. DOI: 10.1016/j.patrec.2009.09.011

116

Jiang, H., Farrar, T., Beardsley, R.C., Chen, R., & Chen, C. (2009). Zonal surface wind jets
across the Red Sea due to mountain gap forcing along both sides of the Red Sea.
Geophysical Research Letters 36, LI 9605. DOI: 10.1029/2009GL040008.
Jiang, N. (2011). A new objective procedure for classifying New Zealand synoptic weather
types during 1958-2008. International Journal o f Climatology 31, 863-879. DOI:
10.1002/joc.2126.
Jiang, N., Cheung, K., Luo, K., Beggs, P.J., & Zhou, W. (2012). On two different objective
procedures for classifying synoptic weather types over east Australia. International Journal
o f Climatology 32,1475-1494. DOI: 10.1002/joc.2373.
Joe, P., Doyle, C., Wallace, A., Cober, S. G., Scott, B., Isaac, G. A., ... Denis, B. (2010).
Weather Services, Science Advances, and the Vancouver 2010 Olympic and Paralympic
Winter Games. Bulletin o f the American Meteorological Society, 91, 1-3. DOI:
10.1175/2009BAMS2998.1
Joe, P., Scott, B., Doyle, C., Isaac, G., Gultepe, I., Forsyth, D., ... Boudala, F. (2014). The
Monitoring Network o f the Vancouver 2010 Olympics. Pure and Applied Geophysics, 171,
25-58. DOI: 10.1007/s00024-012-0588-z
Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., Deaven, D., Gandin, L., ... Joseph, D.
(1996). The NCEP/NCAR 40-Year Reanalysis Project. Bulletin o f the American
Meteorological
Society,
77,
437-471.
DOI:10.1175/15200477(1996)077<0437:TNYRP>2.0.CO;2.
Kassomenos, P., Vardoulakis, S., Borge, R., Lumbreras, J., Papaloukas, C., & Karakitsios, S.
(2010). Comparison o f statistical clustering techniques for the classification o f modelled
atmospheric trajectories. Theoretical and Applied Climatology 102, 1-12. DOI:
10.1007/s00704-009-0233-7.
Koletsis, I., Lagouvardos, K., Kotroni, V., & Bartzokas, A. (2009). The interaction of
northern wind flow with the complex topography o f Crete Island - Part 1: Observational
study. Natural Hazards and Earth System Science 9, 1845-1855.
Lackmann, G. (2012). Midlatitude Synoptic Meteorology: Dynamics, Analysis and
Forecasting. American Meteorological Society: Massachusetts, USA.
Lackmann, G. M., & Overland, J. E. (1989). Atmospheric structure and momentum balance
during a gap-wind event in Shelikof Strait, Alaska. Monthly Weather Review 117, 18171833. DOI: 10.1175/1520 0493(1989)117<1817:ASAMBD>2.0.CO;2.
Lange, O. S. (1998). The wind came all ways: a quest to understand the winds, waves and
weather in the Georgia Basin. Environment Canada: Vancouver, Canada.
Liu, H., Olsson, Q. O., & Volz, K. (2008). SAR Observation and Modeling o f Gap Winds in
the Prince William Sound of Alaska. Sensors, 8, 4894-4914. DOI: 10.3390/s8084894

117

Mass, C. F., & Albright, M. D. (1985). A severe windstorm in the lee o f the Cascade
Mountains o f Washington State. Monthly Weather Review, 113, 1261-1281. DOI:
10.1175/1520-0493(1985) 113<1261: ASWITL>2.0.CO;2
Mass, C. F., Businger, S., Albright, M.D., & Tucker, Z. A. (1995). A windstorm in the lee of
a gap in a coastal mountain barrier. Monthly Weather Review 123, 315-331. DOI:
10.1175/1520-0493( 1995) 123<0315: AWITLO>2,O.CO;2.
Mayr, G. J., Armi, L., Arnold, S., Banta, R. M., Darby, L. S., Durran, D. D., ... Whiteman,
C. D. (2004). Gap flow measurements during the mesoscale Alpine program. Meteorology
and Atmospheric Physics, 86, 99-119. DOI 10.1007/s00703-003-0022-2
Mercer, A. E., Shafer, C. M., Doswell, C. A., Leslie, L. M., & Richman, M. B. (2012).
Synoptic composites o f tomadic and nontomadic outbreaks. Monthly Weather Review 140,
2590-2608. DOI: 10.1175/MWR-D-12-00029.1.
Ministry o f Environment, British Columbia. (2014). BC Air Quality - Meteorological
monitoring.
Retrieved
from
http://www.bcairqualitv.ca/assessment/meteorologicalmonitoring.html
Moore, G. W. K., & Renfrew, I. A. (2005). Tip Jets and Barrier Winds: A QuikSCAT
Climatology o f High Wind Speed Events around Greenland. Journal o f Climate, 18, 37133725.
Murphy, B. D., Jackson, P. L., & Lindgren, B. S. (2004). A synoptic climatology fo r
emergence and flight o f the mountain pine beetle. Mountain Pine Beetle Initiative Reports.
Retrieved from: cirrus.unbc.ca/mpbi/docs/mpbi-annual-report-jackson-2004-05-07.pdf
Nylen, T. H., & Fountain, A. G. (2004). Climatology of katabatic winds in the McMurdo dry
valleys, southern Victoria Land, Antarctica. Journal o f Geophysical Research, 109, D03114.
DOI: 10.1029/2003JD003937.
Overland, J. E., & Walter, B. A. (1981). Gap winds in the strait o f Juan de Fuca. Monthly
Weather
Review
109,
2221-2233.
DOI:
10.1175/15200493(1981 )109<2221:GWITSO>2.0.CO;2
Pan, F. F., & Smith, R. B. (1999). Gap winds and wakes: SAR observations and numerical
simulations. Journal o f the Atmospheric Sciences. 56, 905-923. DOI: 10.1175/15200469( 1999)056<0905:G WA W SO>2.0.CO;2
Parker, S. S., Hawes, J. T., Colucci, S. J., & Hayden, B. P. (1989). Climatology o f 500 mb
cyclones and anticyclones, 1950-85. Monthly Weather Review 117, 558-570. DOI:
10.1175/1520-0493(1989)117<0558:COMCAA>2.0.CO;2.
Reed, T. R. (1931). Gap winds o f the Strait of Juan de Fuca. Monthly Weather Review 59,
373-376.
Riddle, A. C., Hartten, L. M., Carter, D. A., Johnston. P. E., Williams, C. R. (2012). A
Minimum Threshold for Wind Profiler Signal-to-Noise Ratios. Journal o f Atmospheric and
Oceanic Technology, 29, 889-895. DOI: 10.1175/JTECH-D-l 1-00173.1
118

Schar, C., & Smith, R. B. (1993). Shallow-water flow past isolated topography. Part I:
vorticity production and wake formation. Journal o f the Atmospheric Sciences, 50, 13731400. DOI: 10.1175/1520-0469( 1993)050<1373: SWFPIT>2.0.CO;2
Segal, M., & Pielke, R. A. (1985). The effect of water temperature and synoptic winds on the
development of surface flows over narrow, elongated water bodies. Journal o f Geophysical
Research-Oceans 90: 4907-4910.
Sharp, J. M. (2002). Columbia Gorge gap flow - insights from observational analysis and
ultra-high-resolution simulation. Bulletin o f the American Meteorological Society 83, 17571762. DOI: 10.1175/BAMS-83-12-1757.
Sharp, J., & Mass, C. F. (2004). Columbia Gorge gap winds: Their climatological influence
and synoptic evolution. Weather and Forecasting, 19, 970-992. DOI: 10.1175/826.1
Shimada, T., & Kawamura, H. (2011). Summertime gap winds o f the Soya Strait induced by
the developed Okhotsk high. Atmospheric Science Letters 12, 316-320. DOI:
10.1002/asl.345.
Shimada, T., Sawada, M., Sha, W., & Kawamura, H. (2010). Low-Level Easterly Winds
Blowing through the Tsugaru Strait, Japan. Part I: Case Study and Statistical Characteristics
Based on Observations. Monthly Weather Review, 138, 3806-3821. DOI:
10.1175/2010MWR3354.1
Stahl, K., Moore, R. D., & McKendry, I. G. (2006). The role o f synoptic-scale circulation in
the linkage between large-scale ocean-atmosphere indices and winter surface climate in
British Columbia, Canada. International Journal o f Climatology 26, 541-560. DOI:
10.1002/joc.l268.
Steenburgh, W.J., Schultz, D. M., & Colle, B. A. (1998). The structure and evolution of gap
outflow over the Gulf o f Tehuantepec, Mexico. Monthly Weather Review, 126, 2673-2691.
DOI: 10.1175/1520-0493( 1998) 126<2673 :TSAEOG>2,O.CO;2
Stevens, B., Duan, J., McWilliams, J C., Munnich, M., & Neelin, D. (2002). Entrainment,
Rayleigh Friction, and Boundary Layer Winds over the Tropical Pacific. Journal o f Climate,
15, 30-44. DOI: 10.1175/1520-0442(2002)015<0030:ERFABL>2.0.CO;22.
Stull, R. (2000). Meteorology fo r Scientists and Engineers, 3rd ED. Pre-publication version,
Discount textbooks: Vancouver, BC, Canada.
University of Washington. (2014). Department of Atmospheric Sciences. Retrieved from
http://www.atmos.washington.edu/cliff/Navygap5.html
Wilks, D. S. (2011). Statistical Methods in the Atmospheric Science. Academic Press:
Oxford, UK.
Yamal, B. (1993). Synoptic Climatology in Environmental Analysis. Belhaven Press:
London, UK.

119