Satellite AVHRR Observations of the Intensification of the ShelfBreak Current during an Upwelling Event off Vancouver IslandbyGordon Clifford StaplesB.Sc., University of British Columbia, 1985A THESIS SUBMITTED IN PARTIAL FULFILMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF SCIENCEinTHE FACULTY OF GRADUA 1B STUDIESDepartment of OceanographyWe accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAApril, 1993© Gordon Clifford Staples, 1993In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at The University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission forextensive copying of this thesis for scholarly purposes may be granted by the Headof my Department or by his or her representatives. It is understood that copying orpublication of this thesis for fmancial gain shall not be allowed without my writtenpermission.Department of OceanographyThe University of British Columbia2075 Wesbrook PlaceVancouver, CanadaV6T 1W5Date: April 28, 1993AbstractAVHRR satellite imagery during July 1984 was used to investigate the spatial andtemporal variability of the shelf break current during an upwelling event off VancouverIsland. The onset of the order 10-day upwelling event initially appears as a plume of coldwater at the tip of Brooks Peninsula. As the winds increase in strength, the plume of coldwater migrates equatorward and takes on a jet-like structure that remains centred justseaward along the shelf break. As the winds dramatically weaken near the end of theupwelling event, there is intense warming of the surface waters along the shelf, but a bandof cold water persists along the shelf break. Horizontal scales calculated from thermalfronts in the AVHRR imagery show that along the shelf break, the cross-shore scale isabout 20 km, which is consistent with the internal Rossby radius. Approximately 70 kmoffshore, another thermal front is present, and this front is thought to represent the offshoreextent of the shelf break current.AVHRR measurements of the SST were correlated with in situ temperaturemeasurements, and a r2 value of around 0.85 was obtained for the upper 3 m of the watercolumn. The temporal variations in the AVHRR SST structure were related to the temporalvariations of the coastal wind field, and to temporal variations in temperature measurementsobtained from coastal lightstations and subsurface moorings. The long-shore velocity wasestimated from the AVHRR imagery, and speeds that were calculated to be from 10 and 80cm/s, were in agreement with long-shore speeds of 20 to 30 cm/s measured by asubsurface current mooring, and long-shore speeds between 10 and 100 cm/s calculatedusing a time-independent barotropic model.Results indicate that wind and topography play important roles in the upwellingresponse, and under the right conditions, AVHRR observations of SST are representativeof dynamical processes within the ocean. The band of cold water along the shelf break isthought to be due to the combined processes of advection and shelf break upwelling.iiTable of ContentsAbstract^ iiList of Tables^ vList of Figures viAcknowledgments viiChapter 1Introduction^ 11.1 Coastal Processes^ 11.2 The Study Area 81.3 Thesis Objectives and Overview^ 10Chapter 2The Oceanographic Setting^ 122.1 Vancouver Island Oceanography^ 122.1.1 Bathymetry^ 122.1.2 Winds and Currents 152.2 Vancouver Island Upwelling^ 202.2.1 Overview 202.2.2 Observations^ 212.3 Summary 24Chapter 3Data Collection and Processing^ 253.1 Advanced Very High Resolution Radiometer ^ 253.1.1 Background^ 253.1.2 Sea Surface Temperature^ 273.2 AVHRR Data Processing 293.2.1 The Data Set 293.2.2 Navigation, Calibration, and SST Retrieval^ 323.3 Image Analysis^ 373.3.1 Feature Identification 373.3.2 Sea Surface Temperature Transects^ 413.4 Hydrographic, Wind, and Bakun Index Data^ 423.5 Summary^ 44Chapter 4Upwelling Theory^ 464.1 Basic Two-layer Theory^ 464.1.1 Coastal Transport 464.1.2 Two-layer Flow 484.2 Shelf Break Effects 554.2.1 Observations^ 554.2.2 Upwelling and Jets 56Chapter 5Upwelling Event: Observations and Analysis 605.1 Wind and Coastal Temperatures^ 605.2 AVHRR Observations^ 675.2.1 Stage 1: July 13-15 68iii5.2.2 Stage 2: July 16-17^ 745.2.3 Stage 3: July 18-20 785.2.4 Stage 4: July 23-24 825.2.5 Long-shore Sea Surface Temperature^ 865.3 Hydrographic Structure^ 885.4 Dynamical Interpretation 925.4.1 Cross-shore Fronts 925.4.2 Shelf-break Jet Formation 965.4.3 Surface Velocity and Shelf Break Upwelling ^ 99Chapter 6Summary and Conclusions^ 1066.1 Summary of Results^ 1066.2 Conclusion 109Chapter 7Bibliography^ 113ivList of TablesTable 3.1. Characteristics of the AVHRR/2 sensor^ 27Table 3.2. Satellite number and orbit, and time of data reception^31Table 5.1. AVHRR image features^ 68Table 5.2. Long-shore velocity estimates (cm/s) using AVHRR measurements, barotropicmodel, and measurements from the E2 current mooring^ 102List of FiguresFigure 1.0. Continental margin schematic diagram^ 2Figure 1.2. AVHRR imagery showing a) eddy and b) cold-water plume^6Figure 1.3. The Vancouver Island study area and location of place names. 9Figure 2.1. Bathymetry along Vancouver Island^ 13Figure 2.2. Bathymetric cross-sections at selected locations along Vancouver Island ^ 14Figure 2.3. Monthly mean Bakun "winds" 17Figure 2.4. Circulation along Vancouver Island a) winter regime, b) summer regime^ 18Figure 2.5. Currents off Estevan Point^ 19Figure 3.1. Emission spectra at different temperatures^ 28Figure 3.2. Typical near-surface temperature profiles 34Figure 3.3. Comparison of AVHRR SST (n6.26388) and in situ SST (Line E)^ 36Figure 3.4. Typical histogram of a thermal IR image 38Figure 3.5. Map showing the location of SST transects, CTlD stations, current meterlocations, coastal wind stations, and lighthouse stations^ 43Figure 4.1. Schematic diagram of the coastal upwelling regime showing the Ekmanlayers and the geostrophic interior^ 48Figure 4.2. The notation used to describe the constant depth, two-layer model^ 49Figure 4.3. The upwelling solution for an eastern boundary^ 54Figure 4.4. The shelf region with variable topography on both sides of the shelf break ^ 57Figure 5.1. a) Cape Scott wind b) Cape Scott Long-shore wind 61Figure 5.2. Kains Island lightstation temperature^ 62Figure 5.3. a) Estevan Point wind b) Estevan Point long-shore wind^ 64Figure 5.4. Amphitrite Point lightstation temperature 65Figure 5.5. Bakun Upwelling Index a) (51 N,131 W) b) (48 N,125 W)^ 66Figure 5.6. Stage 1 AVHRR imagery^ 69Figure 5.6. Stage 1 AVHRR imagery (continued)^ 71Figure 5.7. Stage 1 cross-shore SST transects (day 195 to 197)^ 73Figure 5.8. Stage 2 AVHRR imagery. 75Figure 5.9. Stage 2 cross-shore SST transects (day 198-199) 77Figure 5.10. Stage 3 AVHRR imagery^ 79Figure 5.11. Stage 3 cross-shore SST transects (day 200 -202)^ 81Figure 5.12. Stage 4 AVHRR imagery 83Figure 5.13. Stage 4 cross-shore SST transects 85Figure 5.14. Long-shore SST transects south of Brooks Peninsula^ 87Figure 5.15. Estevan Point a) temperature contour, b) sigma-t contour 89Figure 5.16. Estevan Point velocity and sub-surface temperature 91Figure 5.17. Cross-shore SST fronts^ 93Figure 5.18. Estevan Point long-shore flow 98Figure 5.19. Sobel edge maps. a) July 16, b) July 17^ 101viAcknowledgmentsMuch like clothing or the mask of emotion, ocean waves are the manifestation of the sea inmotion, and yet beneath the waves, there are waves, currents, and an abundance of life thatremain hidden to all but those who chip away at the surface facade, and dig deeply into itscomplexities in pursuit of understanding. During my thesis research, I have dug into theocean in an attempt to understand some of these complexities, and in the course ofunderstanding, I have developed a clarity of reason that applies to all endeavors. In thisrenaissance of clarity, I have been inspired by the dedication to understanding and theenthusiastic spirit of the people within the Department of Oceanography and beyond.I would like to thank my thesis supervisor, William Hsieh, who encouraged me to focusmy efforts when I was dealing with the vast amounts of AVHRR satellite data at the onsetof my thesis and gently nudged in the right direction when I veered off course. I wouldalso like to thank Rick Thomson at the Institute of Ocean Sciences who was a source ofinspiration, and to echo Pijush Kundu's comment of a "cheerful voice over the telephonethat was a constant reassurance that professional science can make some people happy". Inaddition, I would like to acknowledge my diving buddy, Peter Baker, who agreed with methat oceanographers should get wet on occasion, the "oceanography Dads", GuidoMarinone and Roger Pieters, who where receptive to conversations on midnight feedings,David Jones, for creative solutions to the acquisition of Vancouver real estate, Ian "Mr.Upwelling" Jardine and Scott Tinis for their stimulating discussion, and Denis Laplante,who cheerfully provided an effective fly-swatter for every computer bug that flew his way.Loving acknowledgment goes to my wife, Heather, who patiently endured the manyevenings and weekends that I spent working on my thesis, and managed to always put theimportant issues in perspective, and to my children, Mark, Courtney, and Richard, whocan sometimes teach adults to view the world in a different way. I would also like to thankmy parents, who showed me the difference between brooms and books, and were tolerantof my curiosity, in spite of their warning that under no circumstances was Ito do a valve-job on the car.Said the Polar Bear to the Snow Mouse, "I am speaking of great things to come".viiCHAPTER 1INTRODUCTION1.1 Coastal ProcessesDuring the Pleistocene Epoch, between 15,000 and 7,000 years ago, the sea levelwas depressed by some 130 m from its present level (Siebold and Berger, 1982). Largeareas of the continental shelves were dry due to the sea level drop, and rivers entered theocean along the shelf and cut back into the shelf forming the present troughs and canyonsthat are ubiquitous along most continental shelves. As the continental glaciers began toretreat during the warming period that followed the Pleistocene, the glacier melt watersreturned to the ocean and the sea level began to rise to its present day level. The nowsubmerged continental shelves cover about 5% of the earth's surface and are the mostproductive fisheries areas in the world. Ryther (1969) describes the open sea, whichcomprises 90% of the ocean, as a biological desert that produces a fraction of the world'sfish catch. Ryther continues by saying that the balance of the world's fish catch are caughtin upwelling regions and coastal waters that cover the continental shelves. In addition tothe biological productivity associated with the shelf regions, continental shelves have alarge effect on the physical oceanography of coastal waters. The shelf area which is cut bytroughs and canyons, impinged upon by coastal capes, and fed by rivers, can trap internalwaves, enhance coastal currents, and in addition to coastal upwelling, the shelf breakregion can act as a secondary source of upwelling.Figure 1.1 shows the general characteristics of the Vancouver Island continentalmargin. Extending from the coastline to the shelf break, the continental shelf averages1about 40 km wide, but has a maximum width of 65 km seaward of the Juan de Fuca Straitand is within 8 km of the coastline off Brooks Peninsula. The shelf break is characterizedby a relatively rapid change in water depth as the continental margin changes into the areacalled the continental slope.Figure 1. Continental margin schematic diagram(not to scale).Beyond the continental slope, the topography gradually changes to the continentalrise and eventually levels out and forms the floor of the deep ocean; the deep ocean floor isessentially featureless except for the occurrence of steep gradient sea mounts that can rise towithin 30 m of the ocean surface. Most of the dynamical coastal processes occur within2100 km of the coastline, roughly an area that is shoreward of the continental slope; offVancouver Island, the shelf break, shelf slope area is usually defined along the 200 misobath. For the purpose of this thesis, the shelf break and shelf slope region will bereferred to as the shelf break, and regions beyond the shelf slope will be referred to as theoffshore or deep ocean; the continental shelf area will be referred to as the shelf or coastalregion.Bowden (1981) said that the shelf break zone divided the deep ocean regime fromthe shelf waters. He continued by saying that for circulation on the shelf, the deep ocean isinvolved as a driving force with the interaction taking place across the shelf break, andfrom the deep ocean side, the shelf break region can appear as a boundary layer of thedeep-sea circulation. In contrast to the flow patterns of western boundary currents where asharp gradient between the coastal and oceanic waters is a pronounced feature, the easternboundary currents that flow offshore from Vancouver Island do not display strong frontalzones between coastal and oceanic waters. However, during an upwelling event, coldsubsurface water is upwelled along the coast and in some cases at the shelf break, andstrong thermal gradients can form between the upwelled water and the warmer surfacewater.The thermal gradients that form during an upwelling event can be readily sampledby conventional means such as using a CTD probe, but the spatial and temporal aspects ofthe sea surface temperature (SST) gradients can also be sampled by using the AdvancedVery High Resolution Radiometer (AVHRR) instrument that is on the NOAA series ofpolar orbiting satellites. The CTD data provides measurements in the water column, butunless there are many research ships sampling at a given time, a CTD survey will likelycover the study area only once during an order 10-day upwelling event.3In comparison, a satellite can sample the entire study area many times a daydepending on the particular orbit and the number of satellites that are within reception rangeof the ground station. The trade-off with the increased spatial and temporal coverage is thatthe AVHRR instrument only measures the so-called skin temperature of the ocean surface;according to Robinson (1985), the actual thickness of the skin layer varies with thewavelengths of the emitted radiation, and between 3 and 14^(the infrared range), it isless than 0.1 mm thick. However, the skin measurement is usually reasonablyrepresentative of the upper few meters of the water column, the so-called bulk sea surfacetemperature, but the agreement does however depend on such factors as the depth of theupper mixed layer, the amount of incident solar radiation, and the moisture content of theatmosphere above the sea (Stewart, 1985).During times of strong winds from the northwest , satellite imagery often revealsthe complex SST structure off Vancouver Island (some bias is introduced since clearsatellite imagery is related to high pressure and northwesterly winds). The appearance offilaments, plumes, mesoscale meanders, and eddies are common features. The July 25,1985 NOAA 9 infrared satellite image shown in Figure 1.2a is an example of an eddy thatwas situated southwest of Brooks Peninsula. The 60 km diameter eddy (E) was visible inthe satellite imagery until July 29th. and was then lost due to cloud cover. AVHRRimagery often shows eddies in the vicinity of Brooks Peninsula, and the geographiclocation of the eddies is perhaps related to the local bathymetry. Eddies off VancouverIsland have been investigated by Thomson and Gower (1985), and Thomson (1984) andthey suggest that eddy formation is due to baroclinic instabilities of a poleward flowingcoastal jet and the instabilities may have been triggered by variations in the topography.Eddy formation and mesoscale meanders related to instabilities and topography effects werealso investigated by Ikeda et al. (1984), and Ikeda et al. (1984a).4Figure 1.2b shows an example of plume formation off Vancouver Island. TheNOAA 9 satellite image taken on August 2, 1988 shows cold water that extends toapproximately 50 km offshore from northern Vancouver Island to Estevan Point; nearEstevan Point, the cold water abruptly stops and an offshore plume (P) extends nearly 100km in the offshore direction. South of Estevan Point, some clouds are visible, but the SSTis considerable warmer than further north. The abrupt change in the SST and the formationof the offshore plume is perhaps related to the longshore gradient of the wind: strong windsat the northern end of Vancouver Island coupled with weak, or decreasing winds to thesouth (R. E. Thomson, pers. comm.).The use of satellite imagery to study mesoscale processes is a tempting tool, but justas an eddy appears, or a cold water plume begins to migrate offshore, the SST structure ishidden by clouds. Since the AVHRR instrument is a passive receiver which is in contrastto an active instrument such as synthetic aperture radar, thin clouds can attenuate theradiation from the sea surface and cause an apparent reduction in the SST, and thick cloudscan result in the satellite measuring the cloud-top temperature. Depending on the frequencyof the observed dynamical process, a few days of clouds may be tolerated, but extendedperiods of thick cloud cover can result in a SST structure that is under sampled. Cloudscan therefore cause problems, and one of the initial requirements of an AVHRR data set isthat there are enough cloud-free images that adequately span a process such as upwelling.Cloud-free AVHRR satellite imagery between July 13 to July 24, 1984 wasobtained from the UBC Satellite Oceanography and Meteorological Lab (SOMeL) archive.The July sequence comprised thirteen late afternoon and early evening satellite passes, andtwo early morning passes. Due to the frequent occurrence of clouds along the west coast,it is usually difficult to obtain a reasonable long time series of cloud-free AVHRR images.5200175150125100Figure 1.2. a) July 25,1985 AVHRR imageshowing an eddy off Brooks Peninsula b) August2, 1988 AVHRR image showing a cold-waterplume. Note the clouds (coloured white) in bottomimage. The temperature scale (x104 ) is shown inthe upper right corner6In addition, the satellite imagery must be correlated with the dynamical process that isbeing observed. A few clear images between many days of cloudy imagery may show thepresence of upwelled water along a coastline, but it may not be possible to ascertain whatstage of the upwelling event that the cloud-free imagery is showing. It is thereforeadvantageous to have a time series of satellite imagery that spans the observed process.The July 1984 imagery of the west coast of Vancouver Island covered a period oftime when there was a dramatic increase in the upwelling-favourable winds. The imagerywas available before, during and after the upwelling event, so it was possible to observethe spatial and temporal variations in the SST structure as the cold, upwelled waterappeared at the surface of the ocean. In conjunction with the satellite data, coastal winddata were available from Estevan Point and Cape Scott, along the west coast of VancouverIsland, and near the end of the satellite coverage, an oceanographic cruise took place and across-shore CTD survey that included Estevan Point and Cox Point, providedhydrographic measurements within the water column (see Figure 1.3). In addition tocurrent meter measurements offshore of Estevan Point, measurements of temperature werealso available from two lighthouse sampling stations at Amphitrite Point and Kains Island.The data provided some insight into the upwelling event off Vancouver Island, and how theupwelling process relates to the appearance of cold water off Brooks Peninsula during theearly stages of upwelling, the formation of cold water along the shelf break that is probablydue in part to shelf break upwelling, the intensification of the shelf break current as thewinds increased in strength, and the cold water that remained along the shelf break during atime of intense warming over the shelf as the winds decreased in strength.71.2 The Study AreaSituated on the west coast of British Columbia, Vancouver Island is about 450 kmin length, Figure 1.3. Although reference is made to the west coast of Vancouver Island,the main area of study was from Cape Scott, at the north end of Vancouver Island, to CoxPoint in the south, a distance of 265 km. The offshore extent of the study wasapproximately 100 km (well beyond the shelf break), and this distance encompassed thespatial variability of the SST structure that was observed in the AVHRR satellite imagery.However, the width of the continental shelf varies along Vancouver Island, and at BrooksPeninsula, the continental shelf reaches its narrowest extent of about 8 km wide, with thedepth rapidly dropping to over 1000 m within about 10 km of the shelf break.Although more details will be given in the next chapter, the oceanography alongVancouver Island is highlighted by a number of characteristic features. As previouslymentioned, the continental shelf width varies along Vancouver Island and is widest in thesouthern end where the broad extent of the shelf supports an active biological community.The broad shelf region is cut by the Juan de Fuca Canyon at the entrance to Juan de FucaStrait; situated in the vicinity of the Canyon is the cyclonic Juan de Fuca Eddy (Freelandand Denman, 1982). In addition to the fresh water input from the Fraser River that makesits way to the ocean through the Juan de Fuca Strait, rivers and streams along the westcoast of Vancouver Island act as fresh water sources (LeBlond et al., 1986). In responseto seasonal changes in the wind field, the shelf break current changes direction (Ware andThomson, 1986); during the summer months, predominately northwesterly winds drive thecurrent equatorward, and during the winter months, southeasterly winds drive the shelfbreak current poleward. Inshore of the shelf break current, the Vancouver Island CoastalCurrent (VICC) flows poleward year-round (Freeland et al., 1984).8Ito^129 126 ILD ....50.............^. BrooksPeninsulaTa chu Pt...... .••OpCape ScottKains Is.Estevan Pt.Cox Pt.AmphiPt.49-trite••• •• • . .... •• ••129^128^127 126^1258Figure 1.3. The Vancouver Island study area andlocation of place names.91.3 Thesis Objectives and OverviewThe preceding comments have introduced some of the important elements in thestudy of the shelf break current response to wind forcing and serve as a guide to some ofthe thought processes that have occurred during this study. The major objectives of thisthesis will be to attempt to answer the following questions: 1) What are the spatial andtemporal scales of the upwelling event? 2) AVHRR satellite imagery reveals the complexSST pattern during an upwelling event, but is the satellite data related to dynamicalprocesses in the ocean interior or does it just represent surface features? 3) Is the cold waterthat initially appears at the tip of Brooks Peninsula and eventually migrates equatorwardduring an upwelling event due to advection by the mean currents or is there shelf breakupwelling? 4) What dynamical processes are forcing the cold water to remain centeredalong the shelf break as the water migrates equatorward from Brooks Peninsula? 5) Is theobserved upwelling event typical of the Vancouver Island response or are the observedSST patterns unique to the July 1984 event?The next chapter will look at the oceanography around Vancouver Island. Thevariability of the seasonal wind field and the ensuing seasonal variability of the currentstructure will be discussed, and as well, previous observations of upwelling events arepresented. Chapter 3 will concentrate on the satellite-data collection, processing andanalysis. The basic procedure by which AVHRR data are converted into SST is outlined,and a comparison between AVHRR SST measurements and in situ temperaturemeasurements is also presented. Chapter 3 also outlines the techniques that were used toenhance the satellite imagery, so that the spatial and temporal variations of the SST frontsare easier to identify than in the unenhanced imagery. The basic two-layer upwelling theorywill be presented in chapter 4, and the conditions when shelf break upwelling is likely to10occur are presented. In addition, observations of shelf break upwelling are summarized.Chapter 5 begins by looking at the wind field, coastal temperatures, and the BakunUpwelling Index during the upwelling event. The upwelling event is divided into fourstages, and the Chapter 5 continues by discussing the SST variations during the upwellingevent by using AVHRR imagery and cross-shore SST transects. Chapter 5 concludes byoffering a dynamical interpretation of the upwelling event based on cross-shore thermalfronts, conservation of potential vorticity, and estimates of the long-shore velocity.Chapter 6 will summarize the observations and present concluding comments.11CHAPTER 2THE OCEANOGRAPHIC SETTING2.1 Vancouver Island Oceanography2.1.1 BathymetryThe rugged west coast of Vancouver Island is cut by many long, narrow inlets.The geography of the land is reflected along the continental shelf region where numerouscanyons slice the shelf, Figure 2.1. The irregular, but broad shelf in the vicinity of BarkleySound is highlighted by the Juan de Fuca Canyon. Freeland and Denman (1982) report thepresence of a topographic upwelling centre that is driven by the interaction between thecoastal currents and the Juan de Fuca Canyon, and results from later investigations reportedby Ware and Thomson (1988) suggest that the upwelling in the canyon is linked to wind-induced upwelling along the outer shelf. The quasi-permanent eddy that is frequentlyobserved in the vicinity of the Canyon is commonly called the Juan de Fuca Eddy. Northof Barkley Sound, the shelf is more regular, but gets progressively narrower until reachingthe narrowest point just off Brooks Peninsula.AVHRR satellite imagery of the SST patterns along Vancouver Island frequentlyshows significant variation north and south of Brooks Peninsula that is due in part to thevariability in the topography along the continental shelf. Across-shore transects digitizedfrom hydrographic charts (Fig 2.2) show the dramatic differences for various locationsalong Vancouver Island. At the northern end of Vancouver Island, Kains Island has a shelf12AMPHITRITE -•PORTRENPREWSWIFTSURESANKLAPEROUSESANK49°MAPOFPLACENAMESTATCHUPTAmPHITRITEBARKLEYSOUND128°^127°W^126°^125°^124°129°VANCOUVERISLAND50°HESOUIATPENES ',111VANNOOTKAIS48°129°^ 128' 127°^ 126° 125'^124°51°width of around 15 km, and beyond the shelf the depth quickly drops to over 1500 meters(see Figure 1.3 for the location of Kains Island). Tatchu Point displays similar shelftopography to Kains Island, but with a slightly wider shelf and a more gradual decrease indepth beyond the shelf.Figure 2.1. Bathymetry along Vancouver Island.From Thomson (1990)13At Brooks Peninsula, between Kains Island and Tatchu Point, the shelf-width is narrow,and the steep topography beyond the shelf rapidly drops to over 1000 m deep and formsthe seaward head of Quoukinsh Canyon. At the southern end of Vancouver Island, CoxPoint (just northwest of Tofino) has a wide shelf region with an offshore extent of nearly50 km that gradually drops to 1500 m deep over the next 35 km.-80^-60^-40^-20^0Offshore Distance (km)Figure 2.2. Bathymetric cross-sections at selectedlocations along Vancouver Island.142.1.2 Winds and CurrentsThe currents along Vancouver Island have been studied as far back as the 1930's.In most cases, the inferred current structure was based on observations of temperature andsalinity and the use of the geostrophic method. In 1974, the first use of submerged currentmeters was used to measure coastal currents off Tofino, and between May 1979 andSeptember 1980, the Coastal Ocean Dynamics Experiment (CODE) was the first majorinvestigation of the circulation and water properties along Vancouver Island; this studyentailed numerous current moorings both on and off the shelf, and extensive hydrographicsampling (Thomson et al., 1986a, and Freeland et al., 1984). In the early 1980's,AVHRR satellite data were readily available and Emery et al. (1986) adapted techniquesthat had been used to detect ice and cloud motion and developed an objective method toestimate surface currents; their approach used a maximum cross-correlation procedurebetween a small subset of the first image and a slightly larger search area in the succeedingimage.Based on geostrophic analysis, Dodimead et al. (1963) and Favorite et al. (1976)commented on the seasonal cycle of the current structure off Vancouver Island. During thesummer, a period of weak northwest winds prevail which result in relatively weak currentsand meanders and eddies were frequently observed. In the winter, strong southeast windsproduce stronger currents than the summer conditions, and the occurrence of meanders andeddies was less pronounced. Between the summer and winter flow regimes, there werevariable currents associated with the change of the mean wind direction which result in theSpring Transition and the Fall Transition. Sea level pressure, averaged between 1946 and1988 for the months of January (winter regime) and July (summer regime), obtained from15the Comprehensive Ocean-Atmosphere Data Set archive indicate that during the winter, theAleutian Low dominates, and the coastal winds along the west coast of Vancouver Islandare predominantly from the southeast (Thomson 1981); the strongest winds are observedfrom October through March, and by April the Aleutian Low has weakened considerably.Between April and September, the North Pacific High dominates the atmospheric pressuredistribution and the coastal winds along Vancouver Island are typically from the northwest.The large scale seasonal variation in the atmospheric pressure distribution alsoappears in the seasonal structure of the winds along the coast of Vancouver Island andinfluences the dynamical processes over the shelf region. Bakun winds, (Figure 2.3) ,which are calculated from 6-hourly synoptic pressure analysis for a 3-degreelatitude/longitude grid (Mason & Bakun, 1986) are shown for the location 49° N, 126° W,approximately 10 km seaward of Amphitrite Point. The monthly mean winds clearlyshows the seasonal variation in the wind direction with winds from the southeast during thewinter months, and upwelling favourable winds from the northwest during the summermonths. Thomson (1983) observed that for periods greater than two days, the syntheticBakun winds are representative of the actual winds.The combination of a good understanding of the seasonal wind structure and anextensive field program during the CODE led Freeland et a. (1984) to resolve the seasonalcirculation pattern along Vancouver Island. During the winter (Figure 2.4a), the northwardflowing Davidson Current is observed seaward of the shelf, and the buoyancy drivenVancouver Island Coastal Current flows northward over the shelf (Hickey et al., 1991)with current speeds of 30 to 40 cm/s observed approximately 15 km from shore. Duringthe summer months (Figure 2.4b), the offshore flow reverses direction in response to thechanging winds and the southward flowing shelf-break current is observed, reaching peakspeeds of about 20 cm/s during August.16Figure 2.3. Monthly mean Bakun "winds". FromThomson et al. (1990).In contrast to the offshore flow, the Vancouver Island Coastal Current continues to flownorthward, but with currents speeds slower than in the winter. Observations of the shelfbreak current by AVHRR satellite imagery frequently shows the formation of plumes andfilaments that can extend up to 100 km offshore. These filaments, similar to those observed17in the California Current System by Mackas et al. (1991) and Ramp et al. (1991) can causea significant flux of biomass from the coastal to the offshore waters, and in addition, thefate of particulate and dissolved carbon that may be transported offshore in the filaments isone aspect of a study reported by Denman et al. (1992).Figure 2.4. Circulation along Vancouver Islanda) winter regime, b) summer regime. FromThomson (1990).18E4^E3 E2B E2^ElCurrent meter data from the Coastal Ocean Dynamics Experiment also revealed amore complete picture of the vertical current structure that was best resolved off EstevanPoint during July, 1980. As depicted in Figure 2.5, the northward flowing VancouverIsland Coastal Current is present within 20 km of the coast. In the top 300 m, centeredseaward of the shelf break, there is strong seasonal jet that reverses direction in response tochanges in the large-scale wind field. Below the seasonal jet, there is a persistentnorthward flow which is probably an extension of the California Undercurrent. In theoffshore regime, there is deep-sea circulation which fluctuates substantially, but does notshow as clear a seasonal cycle as the inshore currents.100^80^60^40^20Distance from Estevan Point in kmFigure 2.5. Currents off Estevan Point duringJuly, 1980. From Freeland et al. (1984).192.2 Vancouver Island Upwelling2.2.1 OverviewIn its simplest physical conceptual interpretation, upwelling is a process that drawscold, subsurface water into the warmer surface layer of the ocean. The extension of theconceptual model to the solution of the mathematical equations that model the upwellingprocess is an exceeding complex problem; however, the salient features of the upwellingprocess can be explained in conceptual terms. In the northern hemisphere, upwellingfavourable winds blow with the coastline to the left of the wind direction. Due to theCoriolis force, the net transport in the surface Ekman layer (the upper 20 - 50 m) is in theoffshore direction. Conservation of mass requires that the offshore flow must be balanced,and this balance is in the form of an onshore flow that upwells water along the coast(coastal upwelling) and at the edge of the continental shelf (shelf break upwelling).Due to the constraints of no flow through the coastline, the upwelled water mustmove in a vertical direction which results in the upward bending of the isopycnal surfacestoward the coast, and the resulting pressure gradient causes an increase in the equatorwardlong-shore current leading to what Charney (1955) called a coastal or surface jet. Thewidth of the jet is given by the baroclinic Rossby radius of deformation, a = cif (Gill,1982), where c is the speed of a long internal wave, and f is the Coriolis parameter, whichat mid-latitude in the northern hemisphere has a typical value of (10 4) 5-1 ; in the case of atwo-layer fluid, c2 = g`H, g' is the reduced gravity, and H is the equivalent depth given byH iH2/(H + H2). The Rossby radius increases toward the equator, and according toEmery et al. (1984), the internal Rossby radius in the vicinity of Vancouver Island istypically 15 to 20 km; in contrast, the external Rossby radius in the in the Vancouver20Island region is around 1300 km. In concert with the formation of the coastal jet, a weak,subsurface poleward-flowing undercurrent develops over the shelf. These features aretypical of an upwelling event, but spatial and temporal variability can readily occur whenother factors such as topography, stratification, the presence of capes, and variations in thewind field are taken into consideration.2.2.2 ObservationsUpwelling occurs along the continental margins in many areas of the world, notablythe coast of Africa, Peru and the west coast of Canada and the United States. In acomparative study of the upwelling regions off Oregon, northwest Africa, and Peru, Smith(1981) observes that local long-shore winds are the primary cause of the upwellingresponse, but the response is also a function of the topography and stratification. Theseasonal cycle of the winds and the similarities in the stratification group Africa and Oregontogether, but it is possible to pick out some differences. Smith continues by commentingthat in contrast to the other areas, the upwelling response off Peru shows significantdifferences. The winds off Peru are nearly always upwelling-favourable and hence do notdisplay a dominant seasonal cycle, and in addition, the year to year variations associatedwith El Nifio can affect the upwelling response. Upwelling off Vancouver Island is similarto that off Oregon, but because Vancouver Island is at the northerly limit of a substantialupwelling season, the upwelling response is not as strong as it is to the south (Freeland andDenman, 1982).Most of the upwelling studies off the coast of Vancouver Island have been doneover the last decade, due in part to the convenient access to AVHRR satellite data. Thespatial variability of the upwelling signal makes satellites an ideal platform from which21observations of SST structure during an upwelling event can be readily observed. UsingAVHRR satellite data, Ikeda and Emery (1984) observed an upwelling event off VancouverIsland during the summer of 1980. Before the intensification of the upwelling-favourablewinds, there was a cold band of water over the shelf break. As the winds increased, therewas a simultaneous decrease in the SST near the coast, and the offshore boundary of thecoastal band of cold water propagated seaward to the shelf break at a speed of about 10km/day. At the end of the upwelling event, the cold band of water remained over the shelfbreak and warmer water appeared over the shelf. The authors continue by saying that thecold band of water that remained could be an indication of shelf break upwelling or theadvection of cooler surface water from north of Brooks Peninsula. In a study by Jardine(1991), the area around Brooks Peninsula was pinpointed as a source of upwelled water inresponse to northwesterly winds, but the response was limited to an area around the southcoast of Brooks Peninsula, so it was not possible to estimate if cold water was advectedsouthward, thus accounting for the band of cold water that remained over the shelf break inthe Ikeda and Emery result.In a recent study by Fang and Hsieh (1992), Empirical Orthogonal Function (EOF)analysis of eight summers of AVHRR satellite imagery off Vancouver Island had fourdominant EOF modes. Accounting for 33% of the variance, mode 1 resembled the meanof all the images. The second mode, which represented 12% of SST variance, showed atopographic upwelling response in the form of a zero crossing line that is approximatelyaligned with the 200 m contour, thus suggesting upwelling over the shelf break. The thirdmode (9% of the variance) showed cool water extending southwestward off BrooksPeninsula, and the fourth mode (5% of the variance) showed a cool water plume extendingoff Cape Scott. The third and fourth modes could represent water that is upwelled in thevicinity of Brooks Peninsula and Cape Scott and then advected southward by the prevailingwinds.22Many of the observations of upwelling off Vancouver Island display the featuresthat are indicative of a classical upwelling response. In all cases, the importance of a long-shore coastal wind as the dominant forcing mechanism has been noted, and the appearanceof cold water along the coast, and in some instances along the shelf break, was observed.Compared to studies off the coasts of Oregon and California, investigations of upwellingoff Vancouver Island are sparse. The higher occurrence of studies off Oregon andCalifornia is perhaps related to the stronger upwelling signal and the importance ofupwelling to fisheries in these regions.In spite of less intensive studies of upwelling off Vancouver Island compared toareas to the south, the dynamical processes affecting the upwelling response off VancouverIsland are complex. Crepon et al. (1984) have shown that the intensity of the upwelling isgreater on the leeward side of a cape compared to the windward edge; the rectangular capein this study is similar to Brooks Peninsula at the northern end of Vancouver Island.Brooks Peninsula was the subject of a study by Freeland (1990) and he showed that asouthward long-shore flow (shelf break current) can support topographic Rossby leewaves, and strong offshore flow occurs on the downwind edge of the Peninsula andproduces features that look like squirts and jets which are occasionally observed inAVIIRR satellite imagery. A study by Peffley and O'Brien (1976) suggests that bottomrelief is more important than variations in coastline geometry in causing localizedupwelling. However the effect of a 20 km long headland (Brooks Peninsula) thatprotrudes into a 20 to 40 km horizontal-scale flow (shelf break current) cannot be ignored.Variations in the long-shore topography was the central issue in a study by Janowitz andPietrafesa (1982). These authors showed that the lowest-order barotropic flow was alongthe isobaths, and upwelling occurred if the slope decreases in the downstream direction sothat the isobaths diverge. The results of the Janowitz and Pietrafesa study may apply in the23region between Brooks Peninsula and the shelf region off Barkley Sound where the slopeof the continental shelf changes dramatically between these two locations.2.3 SummaryThe North Pacific Ocean sea level pressure distribution is dominated by thepresence of the Aleutian Low during the winter months and the North Pacific High duringthe summer months. The large scale pressure variations force a seasonal cycle in the windsalong the west coast of Vancouver Island. During the winter, the winds are predominantlyfrom the southeast, and the poleward flowing Davidson Current is observed along theshelf. Confined to within approximately 20 km of the coastline, the buoyancy drivenVancouver Island coastal current also flows in a northerly direction. The transition to thesummer-regime flow conditions occurs during the Spring Transition, around March orApril, and the coastal winds are driven by the North Pacific High.During the summer, the winds are predominantly from the northwest, andupwelling is a common occurrence off Vancouver Island. Longshore winds drive theupwelling, but variability in wind strength and topography can modify the upwellingresponse. The summer-regime winds also produce a change in the mean circulation patternoff Vancouver Island. The equatorward shelf break current is established in the top 300 mof the water column, and is roughly centered seaward of the shelf break. Despite thechange in the wind direction, the Vancouver Island Coastal Current continues its primarilynorthward year-round flow. Sometime in the late September, early October time frame, theFall Transition occurs, and the winds return to blowing primarily from the southeast. Inthe offshore region, there are fluctuations in the circulation, but the seasonal variabilitydoes not appear related to the coastal cycle24CHAPTER 3DATA COLLECTION AND PROCESSING3.1 Advanced Very High Resolution Radiometer3.1.1 BackgroundSince the days of using a weighted cable to measure the depth of the ocean, therehave been major advances in the development of instruments to measure various oceanproperties. Measurements of temperature have always been important in oceanography,and through the initial use of reversing thermometers to present day electronic temperaturemeasurements taken by CTD probes, the thermal structure of the ocean has become betterunderstood. However, most of the temperature measurements are discrete samples that areusually separated by tens of kilometers, and in the open ocean, the sampling separationdistance can extend to hundreds of kilometers due to the vast expanse of the oceans. Toprovide a more extensive spatial picture of the surface thermal structure of the ocean,thermal infrared (IR) scanners were developed. The first generation of IR scanners weresubject to a number of problems that included poor spatial resolution, and a high degree ofelectronic noise that introduced significant spatial and temporal errors (Legeckis, 1978).Infrared scanners were initially flown on aircraft, and were later introduced as partof the instrument package on meteorological satellites. In 1972, a Very High ResolutionRadiometer (VHRR) was placed on the NOAA series of polar orbiting meteorologicalsatellites. The VHRR scanner offered better spatial resolution and decreased electronicnoise compared to the first generation of IR scanners. With the improvements in the IR25measurements, VHRR measurements from satellites could be routinely taken, and thebroad spatial coverage that satellite radiometers provided enabled the detection of SSTfronts associated with ocean currents and upwelling. In 1978, the Advanced Very HighResolution Radiometer (AVHRR) was placed on the NOAA satellites. In contrast to theone visible and one infrared channel, and the analog data transmission scheme of theVHRR sensor, the AVHRR sensor had two visible and two infrared channels, andtransmitted 10-bit binary data in High Resolution Picture Transmission (HRPT) mode thatprovided 1024 radiance level measurements. The AVHRR/2 sensor was introduced shortlyafter the AVHRR sensor and provided a third infrared channel.Table 3.1 list some of the major characteristics of the advanced very high resolutionradiometer, specifically the AVHRR/2 sensor. The sensor has five channels that span theelectromagnetic spectrum from the visible through the infrared wavelengths. Channel 4 istypically used to measure the SST due to the balance between atmospheric transparency andthe peak of the thermal emission occurring in the 10.3 to 11.3 p.m wavelength range.Channel 3 suffers from a high degree of solar reflectance during the day, but can be readilyused for nighttime measurements of the SST. The spatial resolution of AVHRR imagery is1.1 km at the nadir point (a point on the earth directly below the satellite), but increases toaround 7 km as measurements are taken near the outside edge of the 2580 km swath width.The instrument noise is usually converted into a Noise Equivalent A T, NEAT, where ATrefers to the apparent change in temperature. The NEAT value of 0.12 K at 300 K meansthat AVHRR measurements can detect relative changes in the SST of around 0.12 K, but inpractice the NEAT value is usually higher due to the introduction of errors such ascalibration errors, and atmospheric effects (Robinson, 1985), and Kaufman and Holben(1993) comment on errors introduced by the drift of the NOAA-7 AVHRR sensor between1981 and 1990.26AVHRR/2 SensorWavebands(gm)^Channel 1^0.58-0.68Channel 2 0.725-1.10Channel 3^3.55-3.93Charmel 4 10.3-11.3Channel 5^11.5-12.5Sensitivity (IR)^NEAT 0.12 K at 300 KDigitization levels 1024Resolution^ approx. 1.1 kinSwath width 2580 kmTable 3.1. Characteristics of the AVHRR/2 sensor3.1.2 Sea Surface TemperatureThe basic principle underlying the operation of IR sensors is that the wavelengththat black bodies emit is temperature dependent. The sun emits at short wavelengths withpeak emission in the visible, and the thermal emission from the earth peaks around 101.1m,in the infrared range. The spectral characteristics of black bodies are given by Planck'sradiation law which are plotted for various temperatures as shown in Figure 3.1. Thermalemissions from the earth are in the 300 K range. The emissivity, which is the ratio ofemittance for a real surface at a given temperature to that of a black body at the sametemperature, is around 0.98 for the sea surface and thus the sea surface radiates very nearlyas a black body.27WAVELENGTH, pmFigure 3.1. Emission spectra at differenttemperatures. From Robinson (1985).Black-body radiation characteristics are utilized to measure the SST by using anAVHRR sensor. On the satellite, a black-body cavity is heated to some temperature andthis temperature is measured by four Platinum Resistance Thermometers (PRT). At the endof each AVHRR earth scan line, the AVHRR sensor views the black-body and views cold,deep space (deep space is assumed to be 4 K). Therefore, each scan line contains adigitized value of the radiance level from the earth for all five channels, plus radiance levelsfrom the on-board black-body and deep space. In addition, the temperature of the black-body is calculated by taking the average of the four PRT measurements.28Prior to the launch of the satellite, the AVHRR sensor is calibrated so that itsresponse as a function of wavelength is known. By combining the sensor responsefunction, the Planck radiation law, and the radiance level measurements from the sateffiteblack-body and deep space, a relationship between the digitized radiance levels that theAVHRR sensor measures and the temperature can be calculated (see Robinson, 1985 forfurther details). The relationship between radiance level and temperature can be applied tothe radiance levels that were obtained from viewing the earth, and the earth radiance levelscan be converted into a temperature. In summary, the AVHRR sensor scans perpendicularto the flight direction of the satellite, and for each approximately 1.1 km square area on theearth (varies with scan angle), a radiance level is measured and hence for the appropriatechannel, a temperature is calculated. The radiance levels are continuously measured fromthe sateffite, so that a radiance level map of the earth is produced that has a meridionally•extent equaling the swath width of the AVHRR sensor (2580 km), and a zonal extent that islimited by the reception range of the ground station, which for the UBC SatelliteOceanography and Meteorological Lab (SOMeL) extends from the Baja Peninsula to theBeaufort Sea.3.2 AVHRR Data Processing3.2.1 The Data SetAs previously mentioned, the initial requirement when using AVHRR satellite datais that a sufficient number of cloud-free images are available that adequately represent thespatial and temporal variability of the observed dynamical process. Along the west coast ofVancouver Island, the highest probability of obtaining cloud-free images occurs during thesummer months (June through September). In addition, the July through September timeframe coincides with the strongest upwelling signal and the peak velocity of the shelf break29current. Using the summer months as a time frame guide, approximately 400 AVHRRsatellite images between 1980 and 1991 were obtained from the SOMeL archive andsubsequently navigated and calibrated (navigation and calibration described below). Of theoriginal 400 images, 110 images over all years satisfied the minimum criteria of beingreasonably cloud-free along the west coast of Vancouver Island. The 110 images usedwere not evenly distributed over all years. Depending on the weather, some years hadconsiderable more cloud-free images and image sequences than other years. In conjunctionwith a good image sequence, there was the additional requirement of obtaining other data tosupport the satellite imagery. A sequence of imagery obtained from July 1984 provided thebest combination of satellite coverage and the availability of hydrographic and wind data.Table 3.2 lists the satellites used and the local time that the satellite was overVancouver Island during the study period in 1984. For all of the satellite passes,Vancouver Island was within a reasonable distance of the sub satellite track (a lineprojected along the earth that is directly below the satellite), and therefore, the maximum1.1 km resolution was obtained for each satellite image. The times of the satellite passeswere usually in the late afternoon or early evening, but two passes were in the morning.The SST measurements taken in the afternoon are probably comparable with each other,but care must be taken when comparing the morning passes to the other passes. Theconsistency of the afternoon and evening passes is due to the relatively small change of thenet heat flux at the sea surface during this time. However, during clear nights, there is asignificant flux of heat away from the sea surface, so the morning SST measurements mayshow an apparent decrease, and as the solar radiation from the sun increases during theday, there will be an apparent increase in the SST. The diurnal variation in the SST mustbe considered when comparing daytime and nighttime imagery, but spatial variability,where the relative SST between regions on the image does not vary significantly, can bereadily compared for nighttime and daytime imagery.30Satellite Date and TimeNOAA 6 26231 July 13 19:44 PDTNOAA 7 15779 July 14 16:51 PDTNOAA 6 26253 July 15 09:11 PDTNOAA 7 15793 July 15 16:39 PDTNOAA 7 15807 July 16 16:26 PDTNOAA 6 26274 July 16 20:06 PDTNOAA 7 15821 July 17 16:14 PDTNOAA 6 26288 July 17 19:48 PDTNOAA 7 15835 July 18 16:02 PDTNOAA 6 26302 July 18 19:24 PDTNOAA 7 15863 July 20 15:37 PDTNOAA 6 26367 July 23 09:20 PDTNOAA 7 15906 July 23 15:41 PDTNOAA 7 15920 July 24 16:28 PDTNOAA 6 26388 July 24 20:20 PDTTable 3.2. Satellite number and orbit, and time ofdata reception313.2.2 Navigation, Calibration, and SST RetrievalSatellite image navigation or geocoding, is a process by which the raw satelliteimage is transformed into a selected geographic map projection. The navigation processalso includes corrections for geometric distortions caused by satellite orbit variations,satellite altitude, and geometric distortions due the Earth shape and rotation. According toEmery et al. (1989), and Emery and Ikeda (1984) there are essentially two methods tonavigate polar-orbiting satellite data, and both methods yield similar results; in both cases,the geometric transformation scheme to rotate the image to an acceptable map projection issimilar, but the procedures to correct for geometric distortions is different. In the first case,only a limited knowledge of the satellite orbit is available and the navigation procedurerelies on the extensive use of Ground Control Points (GCP) to correct for errors. In thesecond case, highly accurate ephemeris data (orbital parameters) are used to locate thesatellite as a function of time, and ground control points are only required to correct forsatellite timing errors and altitude. For both navigation schemes, the authors report locationaccuracy in the order of 2 km. The SOMeL uses the second method of navigation, and aprocess called nudging is also used to correct for satellite timing errors. The nudgingprocedure requires the image to be shifted to whatever geographic features are present inthe image, thus obtaining location accuracy in the 2 km range. The second navigationscheme is more accurate than the first when no GCP are available; the lack of GCP caneasily occur when the satellite image does not include land.The calibration component of the images processing converts the digital radiancelevels into a measure of the temperature. As outlined above, the calibration procedure toconvert from the AVHRR sensed radiance levels to the SST is well known, but due toatmospheric effects, the radiance measured may not represent the true SST. As outlined by32Abbott and Chelton (1991), some of the atmospheric effects that can reduce thetransmission of IR radiation through the atmosphere and cause an apparent reduction in theSST include variable amounts of water vapour in the air, the presence of aerosols in thestratosphere from sources like the 1991 volcanic eruption of Mt. Pinatubo, and the largestsingle source of error is due to the presence of undetected clouds. Various methods suchas those developed by Walton et al. (1990), and McClain et al. (1982) have beenreasonably successful in using the multi-channel capacity of the AVHRR data to correct foratmospheric effects.Another source of error in the calculation of the SST is due to the variability of thethermocline structure of the ocean. As shown in Figure 3.2, the near-surface temperatureof the ocean can vary depending on the degree of wind mixing and the time of day whenthe SST is measured. Temperatures as high as 21° C have been measured in the topsurface layer in a few British Columbia fjords on calm, sunny days (Thomson, 1981). Thetemperature in the upper few meters of the water column represent the so called bulk seasurface temperature, but the AVHRR sensor only measures the top 0.1 mm, the so calledskin temperature. However, studies by Wick and Emery (1992), Schluessel et al. (1990),and Emery (1989) have shown that the skin temperature is reasonably representative of thebulk sea surface temperature, but the agreement depends on the depth of the upper mixedlayer.3.2.3 In Situ ComparisonGiven the degree of processing required and the possibility for errors, it issometimes surprising that AVHRR measurements of the SST yield the reasonably accurateresults that are routinely obtained. Studies by Tabata (1981) off Vancouver Island report3315.0 16.0CALM.SUNNY DAYNIGHTIMESUNNY WITH SOME WIND MIXING2.0Typical temperature, degCFigure 3.2. Typical near-surface temperatureprofiles. From Robinson (1985)good agreement between AVHRR measurements of the SST from satellites as compared toin situ SST measurements from ships; significant deviations between the two temperaturemeasurements are typically due to atmospheric effects, radiometer calibration errors, orerrors in shipborne measurements. At the end of the July 1984 study, an oceanographiccruise took place off Vancouver Island and shipborne measurements of the SST werecompared to those obtained from the satellite. For quality control in the comparison ofsatellite and buoy SST measurements, Walton et al. (1990) suggest that the satellitemeasurements should be within four hours of the in situ measurements. The July 1984data set contained one satellite pass that coincided with in situ SST measurements that waswithin the time guideline, and the SST from this image was correlated with the near-surfacehydrographic measurements of the temperature.34Temperature measurements were obtained from Line E (Cox Point) off VancouverIsland (see Figure 3.5); station El (over the 50 m isobath) was sampled at 7 pm PacificDaylight Time (PDT) on July 24, 1984 and station E9 (over the 1500 m isobath) wassampled at 2 am (PDT) on July 25, 1984. At 8:20 pm (PDT) on July 24, 1984, there wasan overpass of the n6 spacecraft (orbit number 26388). In future, unless otherwiseindicated, the convention will be adopted that all times will be PDT, the year will be 1984,and the satellites will be refereed to as n6 or n7 plus the orbit number; in this comparisoncase, the satellite reference name is n6.26388. The channel 4 n6.26388 image wasnavigated and calibrated, but there were no corrections for atmospheric effects applied tothe image.The latitude and longitude of each station (E1-E9) was converted into image co-ordinates, and for each pixel that corresponded with a hydrographic station, the averagetemperature of the surrounding four pixels was calculated, and the average temperature wasassigned to the central pixel value. The averaging process helped to reduce the errors dueto the conversion from geographic to image co-ordinates and errors due to the navigationprocess. The comparison between surface shipborne-measurements and the satellitemeasurements of the SST is shown in Figure 3.3a, and Figure 3.3b shows the correlationcoefficient (r) as a function of the depth. At the surface, the correlation coefficient isaround 0.9, increases slightly between 2 and 3 m of depth, and drops off significantly after3 m. One-metre averaged temperature profiles show a sharp decrease in the upper 5 m, andthe correlation increase is presumably due to the AVHRR SST, which is slightly less thanin situ surface temperatures (Figure 3.3a); the lower AVHRR-sensed measurements are inbetter agreement with CTD measurements a few metres below the surface. Although thereare only nine sample pairs, the assumption will be made that the temperature correlationagree over the entire spatial extent of the study area for a single image, and that theagreement will also hold for the entire July sequence of images.35Figure 3.3. a) Comparison of AVHRR SST(n6.26388) and in situ SST (Line E) b) Same asin (a), except AVHRR SST compared to Line Efor increasing depth.363.3 Image Analysis3.3.1 Feature IdentificationThere are numerous digital image processing techniques that help to produce clearpictures of the SST structure and aid in the identification of oceanographic features. One ofthe basic techniques is image enhancement. The process of image enhancement relies onthe distribution or histogram of pixel values for a given image. Most images are convertedfrom a 10-bit to a 8-bit format during the calibration process, which will yield, in the caseof a channel 4 IR image, a temperature range between 0 and 25.5° C (pixel values aredisplayed as integers between 0 and 255 and the decimal point is mentally inserted). In thecase of summer-time image at mid latitude, low temperature values are usually associatedwith clouds, mid-range temperatures typify the ocean, and high temperatures are associatedwith land (Figure 3.4). If the range of ocean temperatures is known, then all pixel valuesless than the minimum ocean temperature can be set to some value, say zero, and all pixelvalues greater than the ocean temperature can be set to some value, say 255. Theenhancement function is designed to set high and low pixel values (temperatures) equal tosome constant, and the pixel values that represent the ocean temperature are linearlystretched. The linear stretching increases the contrast between adjacent pixels and makesthe SST gradients easier to detect.Another technique for feature identification is an edge detector. The basic principlein edge detection is to calculate the sea surface temperature gradient in two orthogonaldirections for an image. The gradient indicates if there are changes in the pixel values,which in the case of SST, might indicate the location of a temperature front37Figure 3.4. Typical histogram of a thermal IRimage. Adapted from Robinson (1985)Simpson (1990) indicates that prior to applying an edge detector routine to animage, the image should be cloud-screened, and in addition to cloud-screening, Jensen(1986) comments that the application of a low-pass filter to the image will produce the moststriking edge enhancement. Cloud-screening was done in two different ways. The firstmethod was by visual inspection. Since the spatial variation of clouds is different than thespatial variation of the SST structure, a low temperature SST front will appear spatiallydifferent from a cloud formation. The visual inspection process works well for thickclouds, but thin clouds and fog are more difficult to detect. The second method uses themulti-channel capacity of the AVHRR data to do the cloud-screening. Liquid water dropsare strong scatters in channel 1 (visible wavelength), and cloud reflectivity can approach3870 percent (Saunders and Edwards, 1989); therefore, the presence of clouds will increasethe scatter of the incident solar radiation and hence increase the amount of solar radiationthat is reflected. A channel 1 threshold value of 75 was used to detect the presence ofclouds. When a channel 1 pixel value was greater than 75, the corresponding pixel in thechannel 4 (IR) image was set to zero. The visual inspection scheme and the multi-channelprocedure were useful for detecting the presence of clouds, although both methods requiredsome subjective interpretation of the final results.Variations in the SST usually occur over a large spatial extent, so the application ofa low-pass filter to an image will block the high spatial frequency detail. The 3x3weighted-filter, as indicated by (3.1), was developed by Wang et al. (1983); compared toan unweighted filter, the unequal size of the matrix elements helps to reduce image blurringby attenuating the loss of high frequency detail. Each of the filter weights was divided byfour so that the sum over all weights was equal to one. The application of the Wang filterresults in the low-pass images being two rows and two columns shorter than the originalimage. To solve the computer hardware display problem of the reduced image size, thepixel values at the border of the filtered image were duplicated, and the duplicated pixelvalues added to the rows and columns of the filtered image. With the addition of two rowsand two columns, the filtered image is the same size as the original image. Since thedynamical processes of interest were near the centre of the image, the artificial imageextension at the borders did not introduce spurious information.0.250.500.250.501.000.500.250.500.25(3.1)39Following cloud-screening and low-pass filtering, an edge detector was applied tothe July image sequence. There are numerous edge detectors available, but the Sobel edgedetector was used. According to Pratt (1991), the Sobel edge detector is more sensitive todiagonal edges than to horizontal and vertical edges, and the Sobel routine provides themost linear response when comparing the detected edge angle to the actual orientation of theedge. The 3x3 Sobel edge detector has the form,G(j,k) = {[GR (j,k)r + [Gc ( j,k)nu2^(3.2a)GR (j,k) = K+ 2^[(A2 + KA3 It4) (Ao KA? AO]^(3.2b)Gc (j,k)= K+^2 RA3 + Kitt + A2 ) — (A6 + KA5 + A4 )j^(3.2c)where K = 2, and the pixel-numbering convention (A0 - A7) is given by (3.3).A0^Al^A2A, G(j,k) A3As k (3.3)The motivation for the K = 2 weighting is to give equal importance to each pixel in terms ofits contribution to the spatial gradient. In most of the imagery, the SST gradients weresharp enough so that the distinction between the cold, upwelled water, and the warmershelf and offshore waters was readily apparent; however, the Sobel edge detector waseffective at delineating temperature fronts when the SST gradients were not as sharp. Theapplication of the Sobel edge detector to the SST imagery also provided an objectivemethod to measure the spatial location of a temperature front: the sharper the SST gradient,the more robust the Sobel edge detector response.403.3.2 Sea Surface Temperature TransectsTransects of the digitized data were taken perpendicular and parallel to VancouverIsland. The perpendicular (across-shore) transects were taken at five locations (see Figure3.5): Kains Island, Brooks Peninsula, Tatchu Point, Estevan Point, and Cox Point. Eachcross-shore transect was approximately 110 km in length and the geographic location wasselected to coincide with hydrographic lines that the Institute of Ocean Sciences routinelysamples. The parallel (long-shore) transects were taken to roughly line up with the 100,200, 1000, and 2000 m isobaths along Vancouver Island and were about 300 km in length.The satellite images were sampled using a bilinear interpolation algorithm developed byThomas (1987) that translates latitude and longitude into an (x,y) location, and relates the(x,y) location to pixel co-ordinates in the image matrix. The algorithm is given by,T(x,y)= dydxPi + dy(1 — dx)P2 + dx(1— dy)P3 + (1— dx)(1 — dy)P4 (3.4)T(x,y) is the temperature, P1_4 are the satellite temperatures of the four pixels nearest the(x,y) location, and dx and dy are the differences between their pixel co-ordinate in theimage matrix and the (x,y) location.Although the satellite imagery was essentially cloud-free, there were some patchesof cloud, and due to satellite data reception errors, dropout (a lost scan line) wouldsometimes occur. Both of these conditions would produce a spike in the SST transects(clouds and dropout were given a pixel value of zero). Spikes in the transect data werecorrected by interpolating the temperature across the spike by looking at the temperaturebefore and after the spike. In the along-shore direction, a maximum of five contaminatedpixels were corrected within a 50 km distance, and in the cross-shore direction, a maximumof three contaminated pixels were corrected within a 25 km distance.41The values of five and three pixels were subjectively selected, and the 50 and 25 kmdistances are the along-shore and across-shore coherence scale along Vancouver Island assuggested by Thomson (1984); however, the quality of the data was such that theinterpolation distance guideline was never exceeded.3.4 Hydrographic, Wind, and Bakun Upwelling Index DataHydrographic data were obtained from the Ocean Physics Group at the Institute ofOcean Sciences (I0S); cruise 84-13 was from July 23 to July 26, 1984 and the end of thesatellite data sequence overlapped with the beginning of the 84-13 cruise. The 84-13 cruise(Figure 3.5), was only in the southern end of Vancouver Island with sampling from theStrait of Juan de Fuca to Line G, at Estevan Point. At each station, measurements oftemperature, salinity, sigma-t, and geopotential anomaly as a function of the pressure wereobtained (Thomson et al., 1986). There were also subsurface current measurements from adepth of 30 m taken at station E2 (off Estevan Point) and in addition, continuousmeasurements of the temperature at 30 m depth for station E2, and at 5 m depth at stationT3 (off Cox Point) were also obtained.The Atmospheric Environmental Service (AES) maintains climatological stationsalong the west coast of Vancouver Island that measure wind seed and direction eitherhourly or every three hours. Wind data were obtained from June through September,1984, from two locations: Cape Scott, at the northern end of Vancouver Island, andEstevan Point, further south (see Figure 3.5). The wind data were averaged in 6-hour binsand rotated 30° from the north to produce cross-shore and long-shore components of thewind speed. In addition, daily point measurements of the sea surface temperature and thesea surface salinity were obtained from the Amphitrite Point Lightstation; similartemperature and salinity values were also obtained from the Kains Island Lightstation.42Figure 3.5. Map showing the location of: SSTtransects: Cross-shore (K=Kains Is., B=BrooksPeninsula, T=Tatchu Pt., E=Estevan Pt., C=CoxPt.) and Longshore (1=100 m isobath, 2=200 misobath, 3=1000 m isobath, 4=2000 m isobath);CTD lines: (Line G-Estevan Pt., Line E-Cox Pt.);Current meters (E2 and T3); Coastal windstations: (CSW=Cape Scott, EPW=Estevan Pt.);Lighthouse stations (KLS=Kains Is.,ALS=Amphitrite Pt.); Bakun Upwelling Index(BUI) at 51°N,131°W and 48°N,125°W.43The Bakun Upwelling Index (BUI) is based on estimating the wind stress on thesea surface at various points along a coastline. According to Bakun (1973), the stress iscalculated according to the classical relation involving the square of the wind speed (4.1);the wind speeds are estimated for 3-degree latitude-longitude grids based on 6-hourlyobjectively analyzed synoptic pressure field. The Ekman transport is calculated from thewind stress, and the component of the transport perpendicular to the coastline is taken as anindication of the amount of water upwelled through the Ekman layer to replace the waterthat is driven offshore. Positive values of the Index indicate upwelling and negative valuesindicate downwelling. Daily BUI values were obtained for two positions: one was at 48°N latitude and 125° W longitude, just south of Vancouver Island, and the other BUI valuewas obtained for 51° N latitude and 131° W longitude, northwest of Vancouver Island.3.5 SummaryFifteen AVHRR satellite images were obtained from the UBC SOMeL archive. Theimages were navigated and calibrated to produce an image that was 512x512 pixels with aresolution 1.1 km per pixel. After navigation, the images were nudged which reduced thenavigation en-or to about 2 km. The AVHRR channel 4 images were calibrated to producetemperatures between 0 and 25.5° C. Calibration errors result in a relative temperatureuncertainty (image to image comparison) of about 0.5° C, and atmospheric effects canintroduce errors in the 1.0 to 1.5° C range when comparing satellite sensed measurementsof the temperature to in situ temperature measurements (Stewart, 1985). The satellite-derived SST and the in situ temperatures were well correlated in the upper few meters ofthe water column, but the correlation decreased for depths greater than approximately 3 m.Hydrographic data were obtained that gave measurements of temperature, salinity,sigma-t, and the geopotential anomaly. The data from a subsurface current moorings werealso available, and in addition, the subsurface moorings also provided continuous44measurements of the temperature at a depth of 30 m and 5 m. Hourly and 3-hourly winddata were obtained from two climatological stations along the west coast of VancouverIsland, and two lightstations provided daily measurements of the sea surface temperatureand the sea surface salinity. Bakun Upwelling Index values were available from twolocations that were north and south of Vancouver Island.45CHAPTER 4UPWELLING THEORY4.1 Basic Two-layer Theory4.1.1 Coastal TransportWhen the wind blows over the ocean, there is a vertical flux of horizontalmomentum into the ocean surface. The mean rate of momentum transfer to the oceansurface is equal to the wind stress T,ti = CDpu2^(4.1)where p is the density of the atmosphere, u is the wind speed (usually taken at 10 m abovethe surface), and CD is the drag coefficient. According to Gill (1982), the drag coefficientincreases with wind speed, and for low wind speeds, CD has a value of about 1.1 x 10 -3 .For wind speeds between 6 m/s and 22 m/s, a commonly used version of the dragcoefficient is given by Smith (1980), and has the form shown in (4.2).CD = (0.61+ 0.0630x 10-3 (4.2)46Due to the Coriolis force, the wind stress at the ocean surface forces a surfacecurrent that is directed to the right (Northern Hemisphere) of the wind stress. The surfacecurrent rotates continuously with depth, while decaying exponentially with a scale of order10 to 20 m (Apel, 1987); the decay scale depth, which is referred to as the Ekman surfacelayer, is the region of the water column where the applied wind stress is absorbed. Theend result of the vertical transfer of momentum to the Ekman surface layer is that the net,depth-integrated transport is directed 90° to the right of the wind stress direction. When awind blows with a coastline to the left of the wind direction, the effect of the depth-integrated transport directed perpendicularly to the coastline is that upwelling occurs toreplace the water that is advected offshore in the Ekman surface layer.Huyer (1983) comments that in addition to the Ekman surface layer, the idealizedvertical structure of the water column also has an Ekman bottom layer and an interior region(Figure 4.1). The Ekman bottom layer is comparable in thickness to the surface Ekmanlayer. Between the surface and the bottom layers is the interior region, and in this region,the pressure gradient balances the Coriolis force and the flow is assumed to be geostrophic.In the open ocean, the geostrophic interior occupies most of the water column, but closer toshore the Ekman layers occupy a significant fraction of the water column. Huyer continuesby saying that during an upwelling event, most of the onshore flow is through thegeostrophic interior and the flow in the bottom layer is relatively weak.47Figure 4.1. Schematic diagram of the coastalupwelling regime showing the Ekman layers andthe geostrophic interior.4.1.2 Two-layer FlowAlthough the distributions of water properties such as density and temperature iscontinuous in the ocean, it is computationally easier to divide the ocean into layers, andeach layer is assumed to represent the variation of the vertical structure of the watercolumn. A two-layer model is the simplest approach and the minimum number of layersrequired to represent the first baroclinic, or internal mode. The baroclinic mode is definedas a flow for which the isobars and the isopycnals are not parallel, and in theory there arean infinite number of baroclinic modes. The two-layer model also represents thebarotropic, or external mode, and unlike the baroclinic mode, the barotropic mode is a flowfor which the isobars and the iospycnals are parallel. Although a two-layer model containsonly two vertical modes, LeBlond and Mysak (1977) comment that for long-periodhydrostatic motion, most of the energy is usually confined to the first two modes.48The easiest way to take into account the vertical variation of density, which is afeature of any upwelling event, is to use a two-layer model. The top layer depth is selectedto represent the density values in the mixed layer, and the bottom layer depth represents thedensity value for the deep ocean; the density in each layer is assumed to be constant. AlongVancouver Island, the mixed layer depth is typically in the upper 200 m (Thomson, 1984).Figure 4.2 shows the notation used for the two-layer model, and the developmentessentially follows Gill (1982).Figure 4.2. The notation used to describe theconstant depth, two-layer model. From Gill(1982)The upper layer density is P1, and ui and vl are the horizontal velocitycomponents. Hi is the depth of the upper layer while at rest, and z=71(x,y,t) is the surfaceelevation. Similarly, p2, u2, v2 H2, and z=-Hi+h(x,y,t) are the values for the lowerlayer, with H=H1+H2, the total depth. In the upper layer, the linear shallow waterequations with wind forcing are,49dui^an ^rx fv1(11) = g^+at(I) ax(111) Aligiv)^(4.3a)dv,^dil^ryat(,) * fu")= g"--- +°Y(m) Pilii (4.3b)and for the lower layer, the equations have the form,du2^an^, dhdt(I) fv2(H)= –8— 8 —ax(III)^dxvOV2 1 4,^an^, dh1- J"2(II)= —g aym g (INV)blItm(4.4a)(4.4b)Terms (I) are the local time rate of change of the velocity components, or theEulerian velocities. Term (II) represent the effects of the Coriolis force, where f is theCoriolis parameter. Terms (III) are the horizontal pressure gradients which by use of thehydrostatic approximation are written in terms of the free surface perturbation, -n, and g isthe acceleration due to gravity. Terms (V) are the same as terms (III) except that h is theinterface displacement between the two fluid layers, and g' is the reduced gravity, which asthe form,81=80,2 –PO/P2^(4.5)Terms (IV) represent the surface stress imposed by the wind, and are the pathways bywhich horizontal momentum is transferred into the water column.50In the upper and bottom layers, the continuity equations have the form,d(n +^h) H (au, 4.^=0atm^1 dx dy (H) (4.6a)a(h) H2(a142 + a—''Z i = 0dy (11) (4.6b)The continuity equations indicate that for terms (II), the divergence of the horizontaltransport will cause a downward displacement of the free surface and an upwarddisplacement of the interface between the two layers, terms (I).The derivation of the linear shallow water equations includes the f-plane, thehydrostatic, and the Boussinesq approximations. Additional details on the derivation areoutlined in Gill (1987), Pedlosky (1987), and LeBlond and Mysak (1977). For large-scaleocean flows, the Coriolis parameter, f=20sine, varies with latitude. However, thevariation off is only important for oceanic events of very long time scale (a few weeks), orvery long length scales (thousands of kilometers); otherwise, f can be assumed to remainessentially uniform (Kundu, 1990). When the horizontal velocities are large compared tothe vertical velocities, and the scales of horizontal motions greater than the depth, then thepressure perturbation is independent of depth, and the pressure can be calculated using thehydrostatic equation which assumes a static fluid, hence the hydrostatic approximation.The Boussinesq approximation as outlined by Pond and Pickard (1983), assumes that if thedensity variations are small, then their effect on the mass of the fluid can be neglected, buttheir effect on the weight of the fluid must be considered. In other words, the averagedensity over a region can be used in the horizontal direction, but for the vertical direction,the in situ density must be used to calculate the pressure field.51For the baroclinic mode, the free surface displacements are smaller than theinterface displacements, and the rigid lid approximation can be made which allows (4.6a) tobe written as,d(-h) + Hi ( du, ± dvi) =0atm^dx dy )(In (4.7)The velocity difference between the two layers can be obtained by subtracting (4.4) from(4.3) to obtain,X 4 , ah rat -." = 8' dx + AlliLi) =g'-(12 + I.'at -1-A^ay Pilii(4.8a)(4.8b)and the continuity equation is written as,-(1^1 ) dh dil aC—+— —+—+—=0111 H2 dt dt dtwhere,a =u,—u2 ;i; = v, — v2(4.9)(4.10)Now consider the special case when there are no variations in the along-shoredirection, ataX --> 0, a vertical wall that corresponds to the coastline at y = 0, a constant52depth, and the wind stress i x is constant and Ty = 0. Then (4.8) and (4.9) can becombined into a forced shallow water equation with a solution of the form,i; =( ;';_lip )(1— exp(1)ah = ( c; \(exp())tpg'111 j^au = ( tH(exp(ltpH,^a(4.11a)(4.11b)(4.11c)where c is the speed of a long internal wave and is given by,C =2^g'1-11 H2 (HI + H2 ) (4.12)and a = cif is the Rossby radius of deformation. Pedlosky's (1974) comments that theRossby radius is the key offshore scale, which is verified by the appearance of the Rossbyradius in the horizontal velocities and the upwelling rate in (4.11). For the case of aneastern boundary, the graphical solution of (4.11) is shown in Figure 4.3. Subsurfacewater is upwelled to replace the surface water that is advected offshore due to the Ekmantransport. In the upper layer, a coastal jet is formed that flows equatorward and intensifieswith time; the cross-shore scale of the upwelling response is given by the internal Rossbyradius of deformation. In the lower layer, a poleward countercurrent is formed that usuallyflows in the opposite direction to the wind.53In a study by Allen (1973), upwelling and coastal jets were investigated for acontinuously stratified ocean that had the same coastline geometry as the two-layer model.The results were similar to those obtained by the two-layer model, and Allen noted that thecoastal jet formed in response to an increase in the along-shore winds, and the formationtime scale was on the order of a few days. As with the two-layer case, the coastal jet wasin geostrophic balance such that the vertical velocity shear in the jet was related to theacross-shore density gradient by the thermal wind relationship.Figure 4.3. The upwelling solution for an easternboundary. From Gill (1982)544.2 Shelf Break Effects4.2.1 ObservationsIn the previous section, a two-layer model of coastal upwelling was discussed.Since the depth was constant, topographic variations in the form of a shelf break thatrepresent a more realistic coastline could not be included. The presence of the shelf breakdoes not significantly change the salient features of the upwelling response (Csanady,1982), but it does introduce the possibility of the occurrence of shelf break upwelling andthe formation of a shelf break jet. Both the shelf break upwelling and the shelf break jet aresimilar in structure, but generally weaker in magnitude than their coastal counterparts;however, these events frequently occur and they are observed in most upwelling regions.Off Vancouver Island, Ikeda and Emery (1984) observed a band of cold water thatremained roughly centered along the shelf break during an upwelling event. Along thenorthwest African shelf, Barton et al. (1977) observed cold water that appeared over theshelf and gradually migrated to the shelf break. Despite the persistence of upwellingfavourable winds, Barton commented that the cold water remained along the shelf breakuntil the winds weakened. Although the presence of cold water along the shelf break wasthought to be due to shelf break upwelling in each of the previous studies, both authorspoint out that the cold water may not be due to upwelling, but may result from water that isadvected from other regions further upstream.Further observations of shelf break upwelling were noted by Dickson et al. (1980)who used AVHRR image of the Celtic Sea to investigate a narrow ribbon of cold water thatwas aligned along the top of the continental slope; transects across the shelf break show55warmer water on either side of slope water thus suggesting that shelf break upwelling wasoccurring. Petrie (1983) studied the current response to transient wind forcing at the shelfbreak off Nova Scotia. He observed that at the onset of upwelling favorable winds,upwelling occurs at the shelf break prior to the appearance of upwelled water on the shelf,and the shelf break upwelling was confined within the Rossby radius of the break. Inaddition, Petrie commented that strong long-shore currents with large vertical shears werealso confined to the shelf break area.4.2.2 Upwelling and JetsAnalytical and numerical models have been developed to try and explain theobservations of shelf break upwelling and the formation of jets along the shelf break. As aminimum requirement, the models must incorporate a variation of topography to allow forthe change of depth that usually begins at the shelf break and extends down the continentalslope. In addition, the models should include variability of the flow that is driven by thewind stress, and stratification to incorporate the characteristic bending of the iospycnals andisotherms at the shelf break and the coastline during an upwelling event. In theobservations of shelf break upwelling mentioned previously, the upwelling at the shelfbreak was related to long-shore winds that setup the classical two-layer, constant depthupwelling circulation pattern that had water moving onshore in the interior and bottom layerto replace the offshore transport in the surface layer; however, the upwelling flow pattern ismodified in the presence of variable topography.When topographic variations in the form of a shelf break are introduced, Hill andJohnson (1974) suggest that a shear layer is formed above the shelf break, and the mainpurpose of the shear layer is to smooth out the discontinuity in the long-shore velocity thatexists on either side of the shelf break. Associated with the discontinuity in the velocity is56the production of vertical motions that lead to upwelling into the surface Ekman layer.Johnson and Nurser (1984) investigated shelf break upwelling by looking at the effect ofvarying the topography on both sides of the shelf break. They suggest that the amount ofupwelling at the shelf break depends on the slopes al, and aR, Figure 4.4.Figure 4.4. The shelf region with variabletopography on both sides of the shelf break. FromJohnson and Nurser (1984)If al, = aR (e.g. constant depth case), then there is no upwelling at the shelf break, but ifeq., is greater than aR (e.g. Hill and Johnson, 1974), there is a moderate amount ofupwelling at the shelf break. As al, becomes very much greater than aR, the shelf breakupwelling approaches a maximum, and strong currents develop along the shelf.57Using a two-layer model, Csanady (1973) investigated the forced response of asuddenly imposed long-shore wind. An equivalent depth scheme was used to smooth outthe bottom topography, and the key result of Csanady's study was the development of along-shore jet; the jet was centered along the shelf break and the jet velocity increasedlinearly with time. O'Brien and Hurlburt (1972) developed a two-layer numerical modelthat incorporated a shelf break. Their results indicate that in addition to coastal upwelling,there was secondary upwelling at the shelf break which was accompanied by a long-shorejet in the upper layer; in agreement with other studies, the upwelling and the jet wereweaker than their coastal analogs. Using potential vorticity arguments, O'Brien andHurlburt comment that the upwelling region and the jet will have the same width scale.4.3 SummaryIn the Northern Hemisphere, when the wind blows with the coastline to the left ofthe wind direction, the wind stress will drive an equatorward long-shore current, but theCoriolis force will cause the net, depth-integrated transport to be directed perpendicular tothe coastline. In the case of two-layer, constant depth model, the offshore Ekman transportin the upper layer is balanced by return flow through the lower layer. The presence of thecoastline causes a one-sided divergence situation, and the lower layer water is upwelled toreplace the water that is advected offshore. The Ekman transport sets up an offshorepressure gradient which cause the formation of a baroclinic coastal jet; both the upwellingregion and the jet have the same horizontal width scale that is given by the internal Rossbyradius of deformation.Based on theoretical studies and observations, shelf break upwelling and theformation of shelf-break jets are due to a change of the bottom topography at the shelfbreak. In the case of shelf break upwelling, vertical velocities are generated in the vicinityof the break which forces cold, subsurface water to the surface Ekman layer. The upwelled58water is confined to the shelf break region with the horizontal width scale given by theinternal Rossby radius of deformation, which is typically on order of 10 to 40 km for mostcoastal regions. In concert with the shelf break upwelling, a shelf break jet can form whichwill have peak velocities in the area that coffesponds to reasonably rapid changes in thebottom topography. As is the case for the upwelled water at the shelf break, the horizontalwidth scale of the jet is also controlled by the internal Rossby radius. The jet is assumed tobe in geostrophic balance with the cross-shore pressure gradient, which implies that verticalshear in the jet is related to the cross-shore density gradient; the jet is also confined to theoceanic surface layer. Shelf break upwelling and shelf break jet formation are a commonoccurrence, and they are observed in most upwelling regions.59CHAPTER 5UPWELLING EVENT: OBSERVATIONS AND ANALYSIS5.1 Wind and Coastal TemperaturesWind speed and direction were obtained from two climatological stations alongVancouver Island: Cape Scott, at the northern end, and Estevan Point, toward the southernend (see Figure 3.5). The Cape Scott station will be used to represent the wind conditionsat the northern end of the Vancouver Island, and the Estevan station will represent the windconditions at the southern end during the upwelling event. Wind data in the vicinity ofBrooks Peninsula would have provided a more complete picture, but it was not untilNovember 1984 that the Atmospheric Environmental Service installed an automatedclimatological station on Solander Island, at the tip of Brooks Peninsula.Figure 5.1a shows the wind at Cape Scott. The upward vertical direction is to thenorth, and the y-axis is the wind speed in m/s with the x-axis indicating the day-number for1984; day 1 corresponds to January 1, and December 31 corresponds to day 366 (1984was a leap year). The same day-number convention will be used for all time-based plotsunless otherwise indicated. Prior to the upwelling event, the winds were generally from asoutheasterly direction. Around day 195 (July 13), the winds become upwellingfavourable until nearly day 209 (July 27); however in the middle of the July 13 to July 27time frame, there is a brief period, around day 203 (July 23) when the winds weaken.Shortly after day 209, the winds return to downwelling favourable conditions.60Figure 5.1. a) Cape Scott wind (unrotated) CapeScott Long-shore wind (rotated 30° from north)61The long-shore component of the wind for Cape Scott is shown in Figure 5.1b.The upwelling favourable conditions between day 195 and day 209 are evident, and theweakening of the winds around day 203 can be clearly seen. The temperature response tothe winds at Kains Island (to the south of Cape Scott) can be seen in the Kains Islandlightstation temperature record shown in Figure 5.2. As the upwelling favourable windsincrease at Cape Scott, there appears to be a one to two day lag in the corresponding dropin the temperature at Kains Island; peak winds occur around day 199 and minimumtemperatures are near day 201. The weakening winds that follow show up in thetemperature increase, and as the winds strengthen, there is the related drop in temperature,which reaches a secondary minimum around day 208.185^190^195^200^205^210DayFigure 5.2. Kains Island lightstation temperature.The 0 symbol indicates times when satellite datawere available.62The Estevan Point winds are shown in Figure 5.3a. Upwelling favourable windsare apparent from around day 187 (July 5) through nearly day 210 (July 28), but from day187 to day 195 (July 13), the winds get progressively weaker before increasing in strengthagain after day 195. The wind structure is easier to see in the Estevan Point long-shorewind component (Figure 5.3b). Prior to the AVHRR observed upwelling event, the windsare upwelling favourable, with speeds around 10 m/s. The winds subside and reach aminimum around July 13; a period of intense northwesterly winds follows and lasts forabout eight days. The winds peak around day 200 (July 18), and dramatically reduce instrength around day 203 (July 21). Using (4.2) with the wind speed set to an average of10 m/s, a Cd values of 1.24x10 -3 is calculated, and if this Cd value is substituted into (4.1)with p = 1.293 kg/m3 , an estimate of 0.16 N/m 2 is obtained for the wind stress at EstevanPoint. A similar calculation for Cape Scott using an average of 2.5 m/s yields a wind stressestimate of 0.01 N/m 2 .The temperature time series at the Amphitrite lightstation (south of Estevan Point) isshown in Figure 5.4. In response to the upwelling favourable winds, the temperaturedrops around day 190 (July 8) and fluctuates between 11° and 12° C, but the temperaturegradually increases from day 190 through day 199 as the winds weaken. Around day 196(July 14), the winds begin to increase in strength and reach a peak of nearly 12 m/s on day200 (July 18). The temperature at Amphitrite begins to drop at day 199 and by the nextday, the temperature is around 11° C. The temperature remains at 11° C until nearly day203, but dramatically increases to over 14° C on day 206 (July 24) as the upwellingfavourable winds weaken significantly.63Figure 5.3. a) Estevan Point wind (unrotated) b)Estevan Point long-shore wind (rotated 30° fromnorth)64185^190^195^200^205^210DayFigure 5.4. Amphitrite Point lightstationtemperature. The 0 symbol indicates times whensatellite data were available.Based on the coastal wind data and the temperature measurements at the AmphitritePoint and Kains Island lightstations, the temperature decrease at the lightstations isassociated with an increase in the northwesterly winds. The wind data represents twosample points along Vancouver Island, but the wind measurements agree qualitatively withthe wind regime along the entire west coast of Vancouver Island. Figure 5.5 shows theBakun Upwelling Index for two positions: 51° N,131° W, northwest of Vancouver Island,and 48° N,125° W, south of Vancouver Island. The Bakun Index is calculated from a 3-degree latitude/longitude pressure field grid and measures the rate of upwelling ordownwelling in m3 sec- 1 (100m)-1 of coastline. The variability of the Bakun Index at the65northern site (Figure 5.5a) agrees with the variability in the long-shore component(upwelling favourable) of the wind at Cape Scott: the amount of upwelling increases beforeand after day 200 (see Figure 5.1b). To the south (Figure 5.5b), the Bakun Index has anaverage value of about 40 from around day 190 through day 197. The Bakun Indexincrease dramatically and peaks at 134 m 3 sec-1 (100 m)-1 on day 200 (July 18), and thendecreases to a value of about 25 on day 205 (July 23). The increase in the BakunUpwelling Index agrees with the strengthening of the long-shore component (upwellingfavourable) of the wind at Estevan Point (see Figure 5.3b).Figure 5.5. a) Bakun Upwelling Index (51°N,131° W) b) Bakun Upwelling Index (48° N,125°W)665.2 AVHRR ObservationsAVHRR imagery was available from July 13 to July 24 (see Table 3.2). Althoughthere was typically a late afternoon and an early evening pass each day, only thetemperature calibrated, channel 4 imagery from the late afternoon passes will be discussed.The exception to the late afternoon imagery will be for July 13, when only the earlyevening pass was available. As was shown in Figure 3.2, the near-surface temperatureprofile of the ocean varies depending on the time of day and the weather conditions;therefore, comparing imagery that was obtained at roughly the same time each day shouldyield more consistent results.The AVHRR observations of the upwelling event will be divided into four stages.The observations will be mainly south of Brooks Peninsula, along the shelf, shelf break,and offshore waters. Stage 1 is from July 13 to July 15 (day 195 to 197), and deals mainlywith the formation of cold water around Brooks Peninsula, and to a lesser extent, the coldwater formation in the vicinity of Cape Scott. Stage 2, July 16 through July 17 (day 198 to199) entails the period when the winds are increasing in strength and there appears to be anequatorward migration of cold water along the shelf break. Stage 3 occurs during the timewhen the winds are at their maximum strength, July 18 through July 20 (day 200 to 202),and there is evidence cold water over the entire shelf, and shelf break regions. Stage 4,July 23 to July 24 (day 205 to 206), is the relaxation period when the winds significantlyreduce in strength and there is intense warming over the shelf region to the south of BrooksPeninsula, but a band of cold water persists along the shelf break.675.2.1 Stage 1: July 13-15The July 13 (day 195), n6.26231 image is shown in Figure 5.6a. The SST arereasonably consistent over the shelf and offshore waters (D), but there is some evidence ofcolder water along the coastline (A). (The letters correspond to the features on the AVHRRimages, Table 5.1). There is also colder water in the vicinity of Brooks Peninsula (B), andCape Scott (C); the appearance of cold water in these two regions has been observed in theAVHRR data from other years and usually indicates the pending start of an upwellingevent. The cold water around Brooks Peninsula is perhaps due to enhanced upwelling thatis related to variations in the local topography.Letter Code^ AVHRR FeatureA^ coastal waterB Brooks PeninsulaC^ Cape ScottD offshore waterE shelf break jetF^ filamentsG La Perouse BankH Juan de Fuca St.Table 5.1. AVHRR image features. The letters areprinted on the AVHRR satellite images and relateto the corresponding feature identified in thesatellite imagery.68700175150125100Figure 5.6. Stage 1 AVHRR imagery. a) July 13(day 195) image, b) July 14 (day 196) image. Thetemperature scale (x10 -1 ) is shown in the upperright corner69The following day, July 14 (day 196), the n7.15779 image (Figure 5.6b) showsrelatively warm offshore waters (D) with temperatures in the 13 to 14° C range. Along thecoastline (A), there are isolated patches of colder water with temperatures that are typicallyaround 11 to 12° C. Water in the 11 to 12° C range is also observed around Cape Scott(C) and Brooks Peninsula (B). In the vicinity of Brooks Peninsula there are patches ofclouds that cause an apparent reduction in the sea surface temperature; however, the waterat the tip of Brooks Peninsula is still colder than the water on either side of the Peninsula.Extending equatorward from Brooks Peninsula is the indication of a jet-like structure that isroughly centered along the shelf break; the jet-like structure will be refereed to as the shelf-break jet. The jet, which will is defined to include surface water that is bounded by the 12°C isotherm, extends about 40 km to the south of Brooks Peninsula, but patches of cloudobscure the southern tip of the jet. (The 12°C isotherm was selected because it is commonto all the AVHRR images, and it can be used as an indicator of the relative change in thelong-shore SST structure during the upwelling event, but the actual core of the shelf-breakjet is in the 10 to 11° C range.)The July 15 (day 197), n7.15793 image is shown in Figure 5.6c. The southernoffshore waters (D) have increased in temperature from the previous day and are now inthe 14° to 15° C range, with temperatures exceeding 15° C in the extreme southern waters(bottom middle of Figure 5.6c). Along the coastline (A), there are isolated patches ofcolder water with a temperature range that is comparable to the previous day, but there is noindication of a uniform decrease in the sea surface temperature along the coastline. Theshelf-break jet extends approximately 50 km equatorward from Brooks Peninsula withtemperatures of 14° C on the shoreward and seaward side of the jet. Around Cape Scott(C), the sea surface temperatures are in the 12° to 13° C range and extend to the coastline.70200175150125100Figure 5.6. Stage 1 AVHRR imagery (con't). c)July 15 (day 197) image. The temperature scale(x10-1 ) is shown in the upper right corner.71The cooler water along the coastline is in contrast to the band of warm water between theshelf break and the coastline, south of Brooks Peninsula. There is also evidence of frontalinstability which can be seen in the wave-like sea surface temperature structure (F) betweenthe cooler, coastal waters and the warmer, offshore waters.Cross-shore SST transects were taken at Brooks Peninsula, Tatchu Point, EstevanPoint, and Cox Point. Each transect is about 100 km, and except for stage 1, which doesnot include the July 13 image, the transects are the average temperatures over the imagerythat entails each stage. The sharp increase in the transects that occasionally occurs at thecoastline (see Figure 5.7b) is caused by navigation errors during image processing.Navigation errors can result in the image being slightly shifted relative to the transects, andwhen this happens, the transects sample land and water near the coastline. Since the land isconsiderable warmer than the water, when the temperature of the ocean and the land areaveraged, an apparent increase in the SST occurs. Therefore, the SST minimum along thecoastline will be taken a few kilometers from the shore to avoid the artificial temperatureincrease caused by the sampling errors.The Brooks Peninsula (Figure 5.7a) transect indicates that the coldest water extendsto slightly seaward of the shelf-break, and then the SST increases in the offshore direction.At Tatchu Point (Figure 5.7b) , the SST is fairly consistent between 13° to 14° C, but thereis evidence of the shelf-break jet that can be seen in the dip of the SST just seaward of theshelf-break. Moving to the south, Estevan Point and Cox Point (Figures 5.7c & 5.7d)show little variation in the cross-shore SST; the SST are essentially constant in the offshorewaters, but the temperature decreases closer to the coastline. As previously mentioned, thetemperature decrease could be due to local upwelling, or perhaps the decrease may beevidence of the poleward flowing Vancouver Island Coastal Current.72732000 200 2000^200^1006eeaEH-100 0 0Figure 5.7. Stage 1 (day 195SST transects a) Brooks PePoint c) Estevan Point d) Coxof the isobaths (m) is indicatedeach plot.to 197) cross-shoreninsula b) TatchuPoint. The locationat the top panel of-80^-60^-40^-20Offshore Distance (km)(a)-100^-80^-60^-40^-20Offshore Distance (km)(c)2000 200 1000-100^-80^-60^-40^-20Offshore Distance (km)(d)2000 200-100^-80^-60^-40^-20^0Offshore Distance (km)(b)5.2.2 Stage 2: July 16-17The onset of stage 2 is marked by an increase in the long-shore component of thewind (upwelling favourable) at both of the coastal stations. The July 16 (day 198),n7.15807 image is shown in Figure 5.8a. Temperatures of 12° to 13° C are observedalong the coastline (A) with isolated patches of slightly cooler water. In the vicinity of LaPerouse Bank (G), a tongue of 14° to 15° C water appears to be forming, and to thesoutheast, an area of colder water exists (H) that is probably due to outflow from the Juande Fuca Strait (Griffin and LeBlond, 1990) or topographic upwelling (Freeland andDenman, 1982). The southern tip of the shelf-break jet has extended to approximately 90km equatorward of Brooks Peninsula, and the band of warmer water between the jet andthe coastline appears to be breaking down. At the tip of Brooks Peninsula (B), there areSST in the 10° to 11° C range that extend southward, but remain centered slightly seawardalong the shelf-break and form the surface core of the shelf-break jet. North of BrooksPeninsula, the SST is masked by an extensive area of cloud cover. In the offshore waters(D), the SST have not changed significantly and remain around 14° to 15° C.By July 17, the winds are nearing their peak value and the n7.15821 (day 199)image for this date is shown in Figure 5.8b. The 14° to 15°C offshore waters (D) havebeen pushed southward, and the northern extent of the 14° C water is nearly equal to theline of latitude that cuts across the northern tip of Washington State. The tongue of warmerwater is still evident over La Perouse bank (G). There appears to be cold water along mostof the coastline (A) with temperatures that have decreased from the previous day and arenow in the 11° to 12° C range, but there are also isolated patches with SST around 10° C.The isolated patches of the colder water are probably due to small-scale variations in thetopography that is causing local upwelling to occur in response to the large-scale increase74200175150125100200175150125100Figure 5.8. Stage 2 AVHRR imagery. a) July 16(day 198) image, b) July 17 (day 199) image. Thetemperature scale (x10 -1 ) is shown in the upperright corner.75in the upwelling favourable winds. R.E. Thomson (pers. comm.) has suggested that theisolated patches of cold water along the coastline may also be related to tidal mixing and theformation of buoyant plumes of cold water that extend from the coastline. The shelf breakjet is approximately 160 km equatorward of Brooks Peninsula, and there is a welldeveloped 10° to 11° C surface core of the jet that extends about 100 to 120 km along theshelf-break. In the offshore waters between Brooks Peninsula and Cape Scott, there iscontinued evidence of frontal instability (F).The cross-shore SST transects for stage 2 are shown in Figure 5.9. The SSTstructure at Brooks Peninsula (Figure 5.9a) has not changed significantly since stage 1.The coldest water is confined within approximately 10 km of the shoreline, and thetemperature gradually increases before reaching a peak of about 13° C, seaward of the2000 m contour. Evidence of the shelf-break jet can be seen in the SST transect at TatchuPoint (Figure 5.9b); the SST drops at the shelf-break and reaches a minimum just seawardof the shelf break. The cross-shore horizontal scale of the core of the shelf-break jet, asdefined by the 11° C isotherm, is about 20 km. The offshore waters are about 1° Cwarmer on average than the shelf waters, with the maximum temperature of the shelf wateroccurring about mid-shelf.The Estevan Point SST transect (Figure 5.9c) shows evidence of the shelf-break jetthat can be seen in the slight temperature drop seaward of about the 200 m contour. Aswas the case at Tatchu Point, the shelf-water temperatures are at a maximum near the mid-shelf with a temperature decrease at the coastline. The same SST structure is apparent inthe Cox Point transect (Figure 5.9d), but the temperature drop at the shelf break, indicatingthe presence of the shelf-break jet, is not as sharp as that at Estevan Point or Tatchu Point.762000 200-100^-80^-60^-40^-20^02000 200^100-100^-80^-60^-40^-20^02000^200^2000^200^100-100 -80^-60^-40^-20Offshore Distance (km)(a)0 -100^-80^-60^-40^-20Offshore Distance (km)(c)0Offshore Distance (km)(b)Offshore Distance (km)(d)Figure 5.9. Stage 2 cross-shore SST transects(day 198-199) a) Brooks Peninsula b) TatchuPoint c) Estevan Point d) Cox Point. The locationof the isobaths (m) is shown on the top of eachplot. The dashed line is the cross-shore transectfrom stage 1.77The minimum SST for Estevan Point and Cox Point occurs at the coastline; this isin contrast to the Tatchu Point transect, where the minimum SST are at the shelf break.The bimodal temperature structure roughly between the 100 and 200 m in the Cox Pointtransect is due to the tongue of warm water over La Perouse Bank; the transect intersectsthe northern tip of the tongue and causes the apparent increase in the SST (see Figure5.8b).5.2.3 Stage 3: July 18-20As was shown in Figures 5.1b, 5.3b, and 5.5, the peak winds occurred aroundJuly 18 (day 200); the peak winds at the northern end (Cape Scott) of Vancouver Islandwere at a maximum approximately one day before the peak winds near the southern end(Estevan Point). The July 18 (day 200), n7.15835 image is shown in Figure 5.10a. Thedecreased wind strength around Cape Scott (C) has resulted in an increase of the SST in theoffshore waters as compared to the previous day, but the water along the shelf hasremained essentially the same temperature. The structure of the shelf-break jet (E) does notseem as well defined as the previous images, but the southern tip of the jet is nearly 200 kmequatorward of Brooks Peninsula. The 10° to 11° C surface core of the jet has alsomigrated equatorward. There is cold water along the coastline (A), and the coastal andshelf-break jet water does not appear to be mixing with the tongue of warm water that issitting over La Perouse bank (G). In addition, the cold water at the mouth of the Strait ofJuan de Fuca (H) does not appear to be breaking down the La Perouse Bank water, but incontrast to the previous image, cold water is evident between the northern tip of the tongueand the coastline; the cold water against the coastline could be due to the Vancouver IslandCoastal Current that is advecting water northward from the Strait of Juan de Fuca outflow.78200175150125100 •Figure 5.10. Stage 3 AVHRR imagery. a) July 18(day 200) image b) July 20 (day 202) image. Thetemperature scale (x10 -1 ) is shown in the upperright corner.79The July 20 (day 202), n715863 image (Figure 5.10b) shows cold water along theentire coastline with no evidence of the thermal structure that defines the shelf-break jet.The average SST of the water along the shelf and offshore regions is in the 10° to 11° Crange. The tongue of warmer water is still present over La Perouse Bank (G), but thetemperatures have decreased and are typically 11 to 12° C. North of Brooks Peninsulathere are patches of clouds along the coastline that extend to Cape Scott.The interpretation of the July 20 image is not as obvious as the other images. Thereis a five-day gap between the July 13 to 18 image set and the July 23 to 24 image set.These images sets, with the exception of July 13, have entailed two passes per day, so thateven though there may have been a change in the absolute SST between the two-dailyimages, the relative SST would have remained essentially the same, thus permitting aspatial comparison of the daily thermal structure. In contrast, there was only one AVHRRimage available for July 20; however, multi-channel cloud detection analysis did notindicate the presence of clouds that may have caused an apparent reduction in the SSTalong Vancouver Island. Therefore, the satellite measurements obtained for July 20 wouldseem to indicate an over-all reduction in the SST as compared to the July 18 image.The cross-shore SST transects for the July 18 to July 20 time frame are shown inFigure 5.11. As was the case for stages 1 and 2, the minimum SST for the BrooksPeninsula transect (Figure 5.1 la) occur within approximately 10 km of the coastline.Gradually increasing from the 10° C minimum temperature at Brooks Peninsula, the SSTreaches an average offshore value of around 12° C. The Tatchu Point transect (Figure5.1 ib) shows that the temperature at the shelf-break has decreased slightly, but there hasbeen a substantial decrease along the coastline; the coastline temperatures, which werepreviously around 1.5° C greater than the shelf-break temperatures are now 0.5° C colder.There is also evidence of a temperature maximum at mid-shelf, between the cooler shelf8002000^200^100^ v....4...0 ....... . . . .%. ...2000 200-100^-80^-60^-40^-20^00■■•-100^-80^-60^-40^-20Offshore Distance (km)(c)Offshore Distance (km)(a)-100 -80^-60^-40^-20^0Offshore Distance (km)(d)2000 200^100Offshore Distance (km)(b)2000 200U_aeaE01---100^-80^-60^-40^-20^0Figure 5.11. Stage 3 cross-shore SST transects(day 200-202) a) Brooks Peninsula b) TatchuPoint c) Estevan Point d) Cox Point. The locationof the isobaths (m) is shown at the top of eachplot. The dashed line is the cross-shore transectfrom stage 2.81break and coastal waters. The SST increases in the offshore direction and reaches a valueof about 12° C, just seaward of the 2000 m contour.The Estevan Point SST transect is shown in Figure 5.11c. The temperature at thecoastline and at the shelf-break have each dropped about 2° C since stage 2, but there is stillevidence of a local temperature maximum at mid-shelf. The SST drops slightly at the shelf-break and then increases. The Cox Point transect displays a similar structure to the EstevanPoint and the Tatchu Point transect: cold water along the coastline, warming at mid-shelf, aslight temperature decrease at the shelf-break, and a temperature increase in the offshoredirection. The bimodal temperature structure that is roughly between the 100 m and 200 mcontour is more sharply defined than it was during stage 2.5.2.4 Stage 4: July 23-24Both the Bakun Upwelling Index and the coastal wind station data indicate that theupwelling favourable winds have significantly decreased in strength since the peak windsthat occurred around July 18 (day 200). The July 23 (day 205), n7.15906 image is shownin Figure 5.12a. Along the coastline (A), there has been intense warming of the shelfwaters with temperatures in the 13° C range. The 14° to 15° C tongue of warm water overLa Perouse Bank (G) now extends to the coastline. Since there was no satellite imageryavailable, between July 20 and 23, it is not possible to tell if the warm tongue remainedintact or mixed with the cooler, coastal water; however, the spatial location suggests that itdid remain intact. As delineated by the 12° C isotherm, the shelf-break jet (E) is welldefined and extends approximately 180 km equatorward of Brooks Peninsula, but stillremains centered along the shelf-break. On both sides of Brooks Peninsula (B), there isevidence of 13° C water, but around Cape Scott (C), colder 10° to 12° C water covers theentire shelf.82200175150125100Figure 5.12. Stage 4 AVHRR imagery. a) July 23(day 205) image b) July 24 (day 206) image. Thetemperature scale (x10-1 ) is shown in the upperright corner.83On the following day, the n7.15920 (day 206) image (Figure 5.12b) does notindicate that any dramatic changes in the SST pattern have occurred. The cooler, shelfbreak waters (E) separate the warmer offshore (D) and shelf waters (A). The equatorwardextend of the shelf-break jet is around 180 to 200 km, and the 10° to 11° C core of the jetextends approximately 120 km. Along the edge of the shelf-break jet, there are numerouscold-water filaments (F) extending into the offshore waters. There is still evidence of thewarm, 14° to 15°C water over La Perouse Bank (G), and the colder water (H) that waspresent at the mouth of the Strait of Juan de Fuca (see Figure 5.8a) has increased intemperature and is now in the 13° to 14° C range.The SST transects during stage 4 are shown in Figure 5.13. The Brooks Peninsulatransect (Figure 5.13a) is similar to the transects taken at this location during the otherstages: cold water within approximately 10 km of the coastline and a temperature increasein the offshore direction. Compared to the previous stage, the SST at the tip of BrooksPeninsula have increased about 0.5° C, but in the offshore waters, the temperature increasewas around 1.0° C. To the south of Brooks Peninsula, the Tatchu Point transect (Figure5.13b) shows the intense warming that has occurred along the shelf; the SST haveincreased to a maximum value of 14° C from a previous value of about 9.5° C during stage3. The shelf-break jet signature is evident by the temperature minimum that is seaward ofthe shelf break.The warming of the shelf waters is also apparent in the Estevan Point SST transectas shown in Figure 5.13c. The characteristic temperature drop occurs at the shelf break,followed by a gradual temperature increase in the offshore direction. The SST structure atCox Point (Figure 5.13d) also shows the warm water over the shelf, and a temperaturedecrease at the shelf break; seaward of approximately the 2000 m contour, the SST842000 200^100-100^-80^-60^-40^-20^02000^200^2000^200^100-100^-80^-60^-40^-20Offshore Distance (km)(a)02000^200-100^-80^-60^-40^-20^0Offshore Distance (km)(b)-100^-80^-60^-40^-20Offshore Distance (km)(c)Offshore Distance (krn)(d)0Figure 5.13. Stage 4 cross-shore SST transects a)Brooks Peninsula b) Tatchu Point c) EstevanPoint d) Cox Point. The location of the isobaths(m) is shown at the top of each plot. The dashedline is the cross-shore transect from stage 3.85is essentially constant. The Cox Point transect also shows a sharp temperature fluctuationaround the 100 m contour; this fluctuation is due to the warm water tongue over La PerouseBank (see Figures 5.12b).5.2.5 Long-shore Sea Surface TemperatureThe temporal variation in the longshore averaged SST during the upwelling event isshown in Figure 5.14. Transects were taken roughly along the 100 m isobath (shelfwater), 200 m isobath (shelf-break water), and the 1000 m isobath (offshore water), andextended approximately 220 km southward from Brooks Peninsula; each transect representa sample of around 800 pixels. The late afternoon satellite passes (except July 13, day195) were used to calculate the average temperature along the isobaths and the missing datapoints (days 201,203,204) were interpolated by means of a cubic spline.On day 195 (July 13), the SST along the entire coast of Vancouver Island isuniform. From day 195 to day 197 (stage 1), the shelf and offshore temperatures are thesame and increase in magnitude together. The shelf break temperature is also increasingduring the same period, but by day 197 (July 15), the shelf-break temperature is about 1° Cless than the shelf and offshore values; the temperature difference is due to the formation ofthe shelf-break jet that is gradually migrating equatorward and causing a lower averagetemperature along the shelf break. The overall-average SST reaches a maximum by day197 (July 15). Following day 197, the winds increase in strength and there is an overalldecrease in the SST (stage 2), but the coldest temperatures are still along the shelf breakand the warmest temperatures are in the shelf waters; however, the temperatures along theshelf are approaching the shelf break temperatures.86Around day 200 (July 18), the winds are at a maximum, but it is not until day 202that the minimum overall SST are observed (stage 3). By day 202, the shelf temperature isthe same as the shelf break temperature, but the temperature in the offshore waters is about1° C warmer. As the winds subside, the overall SST begin to increase (stage 4), but thereis differential warming in the shelf, shelf break, and offshore waters.6120coa)0_Ea)I—a)0 )2a)<196^198^200^202^204^206dayFigure 5.14. Long-shore SST transects south ofBrooks Peninsula. The symbols at A 0) indicatetimes when satellite data were available, and thelines (solid, dotted, broken) are the cubic splinefit to the long-shore averaged SST along the 100,200, and 1000 m isobaths.87The most dramatic temperature rise is along the shelf where increases of about 3° Care observed. Both the offshore and the shelf break temperatures increase, but not to thesame degree as the shelf temperatures. By day 205 (July 23), the shelf temperature iswarmer than the offshore and shelf break temperatures, and the offshore and shelf breaktemperatures are essentially the same.5.3 Hydrographic StructureThe Estevan Point hydrographic survey line (line G, Figure 3.5) was sampledbetween 8 am and 3 pm on July 25, one day after the last available satellite image;however, the temperature structure is probably similar to the previous few days andtherefore represents the water properties during stage 4. The Estevan Point temperatureand density structure for the top 200 m are shown in Figure 5.15. Below about 50 m, theisotherms (Figure 5.15a) and the isopycnals (Figure 5.15b) dome upward over the shelfand dip downward near the coastline and in the offshore region. The downward dip at thecoastline is presumably due to the presence of the buoyancy driven Vancouver IslandCoastal Current, and in addition, the downward dip seems to indicate that the characteristicupward tilt of the isopycnals and the isotherms associated with coastal upwelling is nolonger present. Evidence of the equatorward flowing shelf break current is apparent in thedownward tilt of the contours in the offshore direction.In the upper 50 m, roughly delineated by the 8 degree isotherm and the 25.4isopycnal, the wind influence can be seen in the water-property structure as compared tothe conditions below 50 m. Over the shelf, the temperature structure indicates wellstratified conditions, with isotherms that are essentially flat. The weak-wind conditions atthe time of the hydrographic survey account for the shelf temperature structure, andAVHRR imagery also indicated that there was intense warming along the shelf. In the880001130o..-- -----"---.----1....... -4r"^1-..-5 4"-1-\15----,Q-----":_,_____:—_._>■------ ,_ ________tact:N.CIvicinity of the shelf break, the isotherms dome upwards between the 200 m and the 1000 mcontour, and the coldest water is slightly seaward of the shelf break; AVHRR imagery alsoshows cold water in the same location (see Figure 5.12b). The isotherm structure seems toindicate the occurrence of shelf-break upwelling that was probably at a peak around thetime of the strongest winds (day 200), but has now subsided due to the weakening of thewinds.1000^200^100-60^-50^-40^-30^-20^-10Offshore Distance (km)(a)1000^200^100-60^-50^-40^-30^-20Offshore Distance (km)(b)Figure 5.15. a) Estevan Point temperaturecontours b) Estevan Point sigma-t contours-1089Six-hourly subtidal velocity records from current meters at a depth of 30 m wereobtained from station E2 (see Figure 3.5); station E2 is shoreward of the shelf break and onthe hydrographic survey line that extends from Estevan Point. The current at E2 (Figure5.16a) has a persistent southeasterly flow which is consistent with the direction of the shelfbreak current. Prior to the upwelling event, current speeds are around 20 cm/s. As thewinds increase in strength, the current increases in strength to a peak value of around 30cm/s, but continues to flow in essentially the same southeasterly direction. Following thepeak in the wind stress (around July 18), the current speed decreases to the 5 to 10 cm/srange in response to the weakening winds.The E2 mooring also recorded the six-hourly temperature at a depth of 30 m. Asshown in Figure 5.16b, temperature is around 10° C, but as the winds increase, thetemperature drops and reaches a minimum of about 8° C, near July 18 (day 200). There issome variability in the temperature as the winds weaken, but the temperature remains in the8° C range. Six-hourly temperatures were also obtained at a depth of 5 m from the T3station (see Figure 3.5). The mid-shelf, surface-layer time series at T3 (Figure 5.16c) issimilar in structure to the temperatures at E2: prior to the upwelling event, temperatures arearound 14° C and as the wind stress reaches a maximum, the temperature drops to about11° C. However, in contrast to the E2 temperature structure that did not increasesignificantly as the winds weakened, the T3 temperature increases to around 13° C as thewinds decrease in strength.The temperatures variations from the subsurface moorings agree with the AVHRRobservations and the temperature measurements from the Amphitrite Point Lightstation.The temperature decrease during the upwelling event is evident in the satellite imagery andthe lightstation measurements, but as the winds weaken, the temperature increase along theshelf is greater than along the shelf break. AVHRR imagery and lightstation temperatures90indicate intense warming along the shelf which is consistent with the T3 temperaturemeasurements, but along the shelf break, the temperature does not increase as dramatically.Satellite imagery and the temperatures at the E2 station verify the lower temperatures alongthe shelf break compared to the shelf temperatures.Figure 5.16. Estevan Point (E2) a) six-hourlysubtidal velocity vectors b) six-hourlytemperatures. Both (a) and (b) are at a depth of30 m. c) Cox Point (T3) surface layer six-hourlytemperature at a depth of -- 5 m. The baselinetemperature is 8° C and the 3° C temperature scaleis also shown. Adapted from Hickey et al. (1991)915.4 Dynamical Interpretation5.4.1 Cross-shore FrontsThe cross-shore structure of the SST south of Brooks Peninsula was examinedusing an edge detector (section 3.3); the edge detector provides an estimate of the SSTgradient and hence will indicate the location of regions where there is a sharp change in thetemperature. The output from the edge detector, as shown in Figure 5.17, indicated thatthere were four zones of strong SST gradients, or fronts: there was a shelf front (s) alongthe mid-shelf, a shelf-break front (b) that was generally along the inshore side of the shelfbreak, and an offshore front (o). There was also an indication of an offshore shelf-breakfront (d) that was on the offshore side of the shelf break and was probably the offshoreextent of the cold-core of the shelf-break jet; this front was most pronounced off TatchuPoint (Figure 5.17b). The location of the minimum SST (t) that roughly defines the centralaxis of the shelf-break jet is also shown in Figure 5.17. The fronts in the vicinity ofBrooks Peninsula are shown in Figure 5.17a. Due to the narrow continental shelf offBrooks Peninsula (see Figure 2.2), the shelf-break inshore front (b) and the minimumtemperature (t) are at the same location, essentially along the shelf break, and show littlevariation during the upwelling event.The offshore front (o) is about 40 km from the tip of Brooks Peninsula, but it doesdisplay some cross-shore variability. At the end of the upwelling event (day 205 to 206),there is a tendency for the offshore and the inshore shelf-break fronts to converge. Thecross-shore SST transects discussed previously (see Figures 5.7a, 5.9a, 5.11a, 5.13a)show that the temperature increases in the offshore direction, out to approximately 40 to 50km offshore (around the 2000 m contour), and remains fairly constant beyond that point.The offshore front probably indicates the offshore edge of the shelf break current.92Figure 5.17. Cross-shore SST fronts a) BrooksPeninsula b) Tatchu Point c) Estevan Point d) CoxPoint. s=shelf front, b=inshore shelf break front,t=minimum temperature (between the shelf breakfronts), d=offshore shelf break front, o=offshorefront. The letters (sbtdo) indicate the spatiallocation of the SST front calculated for eachsatellite image using the Sobel edge detector. Thesolid horizontal-line is the location of the shelfbreak (— 200 m isobath) at each cross-shorelocation.93Moving to the south, the fronts off Tatchu Point (Figure 5.17b) and Estevan Point(Figure 5.17c) indicate the presence of a shelf front (s) that appears to migrate in theoffshore direction. The shelf front is probably due to coastal upwelling, but betweenapproximately day 200 and day 204, only the offshore front was detected, so it is not clearhow far offshore the shelf front migrated. In a study off Vancouver Island, Ikeda andEmery (1984) quote an offshore migration rate of 10 km/day, and mention that theupwelling front they observed eventually passed beyond the shelf break. However, afterday 204 of this study, the shelf front is not present indicating that coastal upwelling hasceased.Along the shelf break at Tatchu Point, and to a lesser degree at Estevan Point, thereis a well defined inshore shelf-break front (b) and an offshore shelf-break front (d). Theminimum temperature (t) is between the fronts and remains seaward of the shelf breakduring the upwelling event. The cross-shore width of the shelf-break fronts (d - b) is about20 km, and using (4.12) with e=0.02, H1=200 m, H2=1000 m, and f=10 -4, gives aRossby radius of about 18 km which agrees with the observations; the Rossby radius of 18to 20 km is also consistent with the 15 to 20 km range for the Rossby radius as suggestedby Emery et al. (1984). As previously mentioned, when shelf break upwelling occurs, theupwelled water is confined to the shelf break region with the cross-shore width scale givenby the internal Rossby radius of deformation. Although the presence of a cross-shore scalethat is consistent with the Rossby radius does not imply that shelf break upwelling isoccurring, it does indicate that the cross-shore scale of the dynamical processes along theshelf break are characterized by the Rossby radius.Approximately 70 to 80 km offshore of Tatchu Point and Estevan Point (Figure5.17b & 5.17c), the offshore front (o) is evident. During the upwelling event, the offshorefront is reasonably stationary, but after day 204 the front appears to migrate farther94offshore. Since the cross-shore frontal structure was calculated at discrete locations alongVancouver Island, the offshore migration, particularly at Estevan Point, indicates thepresence of offshore filaments and does not indicate that the entire offshore front hasmigrated (see Figure 5.12b(F)). As was the case with Brooks Peninsula, the offshorefront is probably the offshore edge of the shelf break current; evidence of this can be seenin Figure 2.5 that shows the velocity structure off Estevan Point during July 1980 andindicates the offshore edge of the shelf break current approximately 70 km from thecoastline.The frontal structure offshore of Cox Point is shown in Figure 5.17d. The shelffront (s), perhaps indicates that coastal upwelling is occurring. Given the separationbetween the shelf and the shelf-break fronts, it is unlikely that the two fronts merged duringthe upwelling event. Along the shelf break, the minimum temperatures (t) remain seawardof the shelf-break, but after day 204, there is a tendency for the minimum temperatures andthe inshore shelf-break front (b) to migrate inshore. Approximately 70 to 80 km from thecoastline, the offshore front (o) is present.The offshore front at Brooks Peninsula was approximately 40 km from thecoastline, but at Tatchu Point, Estevan Point, and Cox Point, the offshore front wasbetween 70 and 80 km from the coastline. A study by Fang and Hsieh that used eightyears of summer AVHRR data from Vancouver Island indicated that the offshore front wastypically 70 km from the coastline, but varied depending on the upwelling response. Thevariation in the offshore front location is presumably related to the variation in the long-shore topography. The continental shelf is the narrowest off Brooks Peninsula and getsprogressively wider toward the north; to the south of Brooks Peninsula, the continentalshelf also gets wider and reaches its widest point off Barkley Sound (see Figure 2.1). Inaddition to the width of the shelf increasing, the gradient of the continental slope decreases95southward of Brooks Peninsula. As suggested by Mooers et al. (1976), the Rossby radiusof deformation can be modified by topography so that the shelf and slope widths alsobecome horizontal scales. The topographic variations that occur southward of BrooksPeninsula may account for the observed variations in the cross-shore SST frontal structure,and the tendency for the minimum temperature (t) to displaced slightly seaward of the shelfbreak when moving in the southward direction.5.4.2 Shelf-break Jet FormationThe preceding section indicated that the shelf break fronts were confined along theshelf break. Using a simple time-independent barotropic model as suggested by Thomsonand Gower (1986), the governing equation is the conservation of potential vorticity,(iil+CiII.1±47 =0dx dy H (5.1)where (u,v) are the time-averaged cross-shore (x) and the long-shore (y) velocitycomponents, Cis the vertical component of the mean relative vorticity, H is the depth, andf is the Coriolis parameter. Equation (5.1) can be expanded as,_u( 1 ( f + c1d1- + 1 ( f +H2^dx ) (0^H dx^Iii- ()H1 2 (f + al) +‘111( f ‘,1 .0ay to .111 dy (iv)^(5.2)96Using scaling arguments outlined by Freeland et al. (1984), term (i) in (5.2) ismuch greater than term (ii) and term (ii) can be neglected. If there is no variation in thelong-shore direction in H and^then term (iii) vanishes, and term (iv) is written as (H 3cos0)-1 where 9 is the angle between the coastline and a line of longitude, and (5.2) canbe written in the form,fU 1 dHV(x)= Hoficos9 H dxwhere U= uHo is the cross-shore transport in the benthic layer and the geostrophic interiorat the edge of the continental margin (depth Ho) and is equal to the cross-shore transportuE8 in the Ekman layer, where uE is the Ekman layer cross-shore velocity, and 8 is theEkman layer thickness (see Figure 4.1). Thomson and Gower further assume an Ekmanlayer balance of the form,drPoJuE = —dz (5.4)where po is the water density and ti is the long-shore component of the wind stress. If(5.4) is integrated over the surface Ekman layer depth, and with the application of (4.1) and(4.2), equation (5.3) can be written as,CpH:P) 2 1 dHV 17(x)– Hoficos0 H dx(5.3)(5.5)970 -0so^1.--E^IC.)"0a)a) oC Nco20=Cl)o) a• c?o_101' -where v(x) is the long-shore speed, CD is the drag coefficient, p a is the density of theatmosphere, p is the density of the water, and V is the long-shore component of the windvelocity. The long-shore speed v(x) using digitized topography off Estevan Point and awind speed of 7.5 m/s is shown in Figure 5.18. The results of this simple model showwhy the band of cold water observed along the shelf break will tend to remain centeredslightly seaward of the shelf break: the peak long-shore speeds are along the shelf breakwhere there is the rapid change in topography. Near the shoreline there are high speeds,but Freeland et al. (1984) comment that the Vancouver Island Coastal Current is buoyancydriven and not wind driven.shelf break-80^-60^-40^-20Offshore distance (km)Figure 5.18. Estevan Point long-shore speed from(5.5) using NO - 10-3 , H o = 2500 m, 0 = 1.5 x10-11 nr 1s -1 , 0 = 300, and V = 7.5 m/s. Thelocation of the shelf break (-- 200 m isobath) isindicated at the top of the plot.98There is also a considerable amount of shear in the cross-shore direction; the shearon the onshore side of the shelf break is greater than the shear on the offshore side of theshelf break over the same cross-shore distance. The variation in the cross-shore shear,which is related to the cross-shore variation in the topography, may account for thecontinuous presence of the onshore shelf-break front which was most pronounced offBrooks Peninsula and Tatchu Point (see Figure 5.17). Although the offshore shelf-breakfront was continuous off Tatchu Point, at Estevan Point and Cox Point, the offshore shelf-break front was not as distinct, but appeared to merge with the offshore front.5.4.3 Surface Velocity and Shelf Break UpwellingA feature tracking method was used to estimate the surface velocities usingsequential AVHRR images. The feature tracking method requires features in the satelliteimagery to remain reasonably invariant over the period that the velocities are estimated, andin the case of AVHRR imagery, the natural choice for a feature is to select a strong thermalfront. Studies by Kelly and Strub (1992), Ninths et al. (1986), and Emery et al. (1986)have used a variation of the feature tracking scheme in the form of a cross correlationtechnique to automate the velocity field extraction from sequential AVHRR images; thesurface velocity estimates they obtained were consistent with in situ current measurementsand CTD-derived geostrophic currents. A modified version of the cross correlationtechnique was used by Holland and Yan (1992) to estimate surface velocities from AVHRRimagery separated by up to 24 hours, and their AVHRR velocity estimates agreedfavourably with anchored and drifting buoy measurements.The feature tracking procedure to estimate surface velocities can be simplified if asubregion of an entire AVHRR image is selected, and the surface velocities are calculated99within that subregion. A subregion was selected that was bounded in the long-shoredirection by Brooks Peninsula and Cox Point, and in the cross-shore direction by roughlythe 100 m and 2000 m isobaths; this subregion also contains most of the spatial variation inthe SST observed during the upwelling event, and encompasses the cross-shore extent ofthe shelf break current.The Sobel edge detector was applied to the July image sequence, and the locationsof the sharpest edge detector responses, which corresponds to the strongest thermalgradients, were calculated. The spatial differences of the edges were measured betweensequential day-time satellite images (24-hour separation), and the difference converted intoa surface velocity. R. E. Thomson (pers. comm.) suggested that maximum velocities ofthe shelf break current are order 100 cm/s, so an edge decorrelation scale of 85 km wasused, which means that a given edge represents the same thermal front if the edge movedless than 85 km in 24 hours. To further simplify the velocity estimates, only the long-shore velocity was calculated; the sharpest long-shore response of the edge detectoroccurred at the equatorward edge of the shelf break jet, and the long-shore migration of thisedge presumably represents the surface expression of the intensification of the shelf breakcurrent.The results of the Sobel edge detector for July 16 (day 198) and July 17 (day 199)are shown in Figure 5.19. In the Sobel edge-maps, it was possible to identify threedominant edges, S1, S2, and S3 that roughly correspond to the 12°, 11°, and the 10°isotherms (the relationship of the edges to the isotherms was verified by checking thespatial location of the edges to the spatial location of the isotherms obtained from a contourplot of the SST). The equatorward tip of the edges was along the shelf break, and theAVHRR-derived velocities were estimated by measuring the displacement of each edge.100Figure 5.19. Sobel edge maps. a) July 16 (day198). Edge S1 (12° isotherm) and edge S2 (11°isotherm). b) July 17 (day 199). Edge Si, edgeS2 and edge S3 (10 ° isotherm). In (a), there is noevidence of edge S3 (10° isotherm), and in (b), theequatorward extend of edge S1 (12° isotherm) isdifficult to identify.101The results of the long-shore velocity estimates are shown in Table 5.2. Theimaged-derived velocities were calculated between July 14 (day 196) and July 18 (day 200)using sequential, afternoon AVHRR images (the July 15 velocity is the average from July14 to July 15, etc.); these dates correspond to stages 2 and 3, and occur during the timewhen the upwellimg favourable winds are increasing in strength. A time-independentbarotropic model was also used to estimate the long-shore velocity in the vicinity of theshelf break; using (5.5), the wind speed was varied, and speeds of 5 m/s, 7.5 m/s, 10 m/s, and 12.5 m/s, which roughly corresponds to the daily-averaged long-shore wind atEstevan Point (see Figure 5.3b) for July 15, July 16, July 17, and July 18 were used. Inaddition, the long-shore, daily-averaged current from the E2 mooring (from Figure 5.16a)is also shown in Table 5.2.Date^ avhrr^ v(x)^E2July^a^az^SI15 10 — — 10 2016 60 — — 30 3017 75 80 60 3018 — 35 30 100 30Table 5.2. Long-shore velocity estimates (cm/s)using AVHRR measurements of edges S1,S2,andS3, time-independent barotropic model, andmeasurements from the E2 current mooring. Theblank (—) entries under AVHRR indicate whenthe edges Si, S2, or S3 were not readilydiscernible. Errors in the AVHRR velocityestimates are around 15 cm/s.102The errors in the AVHRR estimates of the velocity are mainly due to imagenavigation and calculating the spatial displacement of the edge. Simpson (1990) indicatesthat in a reasonably noise-free image, the Sobel operator performs well in detecting edges,so that the edges detected are from SST gradients and not noise-generated. Emery et al.(1989) give image navigation errors on order of 2 km, and the error associated with thespatial displacement of the edge (pixel to pixel comparison) is estimated to be about 10 km.Using a conservative navigation error of 4 km, and combined with the spatial displacementerror of 10 km, yield an AVHRR velocity error-estimate of around 15 cm/s.The image-derived velocities and the velocities derived from the time-independentbarotropic model are typically larger than the results from the moored current meter, but allof the measurements show that the long-shore velocity increases as the winds increase instrength. Using (4.11c) with the previously calculated long-shore wind stress value of0.16 kg/m3 (10 m/s wind speed), p = 1025 kg/m3, Hi = 150, and y = 0, indicates that thelong-shore current should increase by about 0.1 (m/s)/day, which is consistent with thevelocity increase at the E2 current mooring, but considerable less than the increasecalculated by the other methods. Since the barotropic model lacks stratification andfriction, it is not surprising that it shows larger increases in the velocity. However, theimage-derived and barotropic-model velocity estimates, and current meter velocities are allin reasonable agreement.In the case of the image-derived velocity estimates, there is a sharp increase in thevelocity between July 15 and July 16, as measured by the displacement of Si edge. FromJuly 16 to July 17, the Si edge velocity increases slightly and is comparable to the S2 edgevelocity. Between July 17 and July 18, the image-derived velocities decrease and arecomparable to the velocity measurements obtained from the E2 current mooring. Johnsonand Nurser (1984) suggest that during the build up of coastal upwelling, there is secondary103upwelling along the shelf break, but when the coastal upwelling reaches a quasi-steadystate, the upwelling along the shelf break starts to decay. The relatively large velocitiesestimated from the AVHRR imagery (July 16 to July 17) could be evidence of shelf breakupwelling: the temperature decrease along the shelf break (as indicated by the AVHRRvelocities) is greater than the rate that the temperature decrease could be advected along theshelf break (as indicated by the E2 current mooring). However, as the coastal upwellingreaches a quasi-steady state and shelf break upwelling starts to decay, the AVHRRvelocities are similar to the velocity measurements from the E2 current mooring.Further corroboration that the image-derived velocities are indicating that shelfbreak upwelling is occurring can be seen in the longshore averaged SST structure (seeFigure 5.14). After July 15 (day 197), each of the longshore averaged SST transectsdecrease at about the same rate. Around July 17 (day 199), the rate of temperature decreasealong the shelf break (200 m isobath) begins to slow down, but the rate of temperaturedecrease along the shelf (100 m isobath) increases. Presumably, the slow down of the rateof temperature decrease along the shelf break indicates that the coastal upwelling isapproaching a quasi-steady state, and the upwelling along the shelf break is starting todecay (see Johnson and Nurser, 1984). If the coastal upwelling is approaching a quasi-steady state, the steady state probably occurs around July 20 (day 202) when the shelf andshelf break temperatures are essentially the same.In addition to the AVHRR-derived surface velocity and the longshore averaged SSTarguments that shelf break upwelling is occurring, additional evidence, that was discussedpreviously, also indicates that shelf break upwelling is occurring. When shelf breakupwelling occurs, the upwelled water is confined to the shelf break region, with thehorizontal width scale given by the internal Rossby radius of deformation; the SST fronts,that were shown in Figure 5.17, have a cross-shore scale that is consistent with the internal104Rossby radius, thus suggesting that there is shelf break upwelling. Additional evidencecan be seen in the Estevan Point cross-shore temperature contours (see Figure 5.15a).Although the Estevan Point temperature structure was calculated from data obtained at theend of the July upwelling event, the apparent signature of shelf break upwelling can beseen in the isotherms that dome upwards over the shelf break, with minimum surfacetemperatures slightly seaward of the shelf break.105CHAPTER 6SUMMARY AND CONCLUSIONS6.1 Summary of ResultsAVHRR satellite imagery of the west coast of Vancouver Island was obtained fromJuly 1984. The imagery spanned a period of time when there was a dramatic increase ofupwelling favourable winds and permitted observations of the spatial and temporalvariations of the SST structure during the upwelling event. In conjunction with the satellitedata, coastal wind data along Vancouver Island were obtained from climatological stationsat Cape Scott and Estevan Point. Lightstations at Kains Island, south of Cape Scott, andAmphitrite Point, south of Estevan Point, provided surface measurements of the oceantemperature. Near the end of the satellite coverage, an oceanographic cruise took place andhydrographic measurements within the water column were taken, and in addition,subsurface measurements of the current velocity and temperature were also available fromcurrent meter moorings. Large-scale estimates of the wind in the form of the BakunUpwelling Index were obtained from two locations that were northward and southward ofVancouver Island.The SST that the AVHRR instrument measures is based on the temperature in thetop 0.01 mm of the ocean surface, the so called skin temperature. Therefore, care must betaken to insure that appearance of temperature gradients in the SST image are related todynamical processes within the ocean, and not due to variations in the skin temperatures oran apparent SST decrease caused by cloud contamination. The imagery was cloud-106screened using subjective and multi-channel comparative techniques, and a comparisonbetween in situ measurements of the sea surface temperature and AVHRR measurementswas well correlated in the top 3 to 4 meters of the water column, but decreased significantlywith increasing depth. The correlation was consistent with the observations that the skintemperature measurement made by the AVHRR is representative of the temperature in theupper few meters of the water column, the so called bulk sea surface temperature.Coastal wind data and proxy wind data in the form of the Bakun Upwelling Indexboth indicated a dramatic increase in the upwelling favourable winds, and sea surfacetemperature data from coastal lightstations showed a decrease in the ocean temperature thatwas related to the increasing wind. Simultaneously with the increase in the strength of thewind, and the temperature decrease at the lightstations, AVHRR imagery revealed thepresence of cold water, that initially appeared in the vicinity of Cape Scott and at the tip ofBrooks Peninsula, and subsequently migrated equatorward from both locations. The coldwater was due to upwelling, and the agreement between the increase in the strength of thewind, and the appearance of cooler temperatures in the satellite imagery indicated that theAVHRR measurements of the SST were tracking the surface manifestation of the ensuingupwelling event.The AVHRR observations of the upwelling event were divided into four stages thatprimarily concentrated in the area to the south of Brooks Peninsula. Stage 1 is concernedwith the initial appearance of cold water in the vicinity of Cape Scott and Brooks Peninsula,and usually indicates the pending start of the upwelling event. Due to the topographicvariations around Brooks Peninsula, the initial appearance of the cold water is probably dueto enhanced upwelling. Stage 2 deals with the equatorward migration of the cold water asthe winds increase in strength. Satellite imagery shows that a plume of cold water extendsequatorward from Brooks Peninsula and takes on a jet-like structure (shelf-break jet); the107central axis of the shelf-break jet, which also coincides with the minimum SST, is centeredslightly seaward along the shelf-break. During stage 2, there is also evidence of isolatedpatches of cold water along the coastline which are presumably related to coastal upwellingor tidal mixing. However, most of the cold water stretches along the shelf break indicatingthat shelf break upwelling is pumping cold water into the surface Ekman layer, and the coldwater is acting as a thermal tracer that delineates the intensifying shelf break current.Stage 3 occurs when the winds are at their maximum strength, and there is evidenceof cold, upwelled water over the entire shelf, and shelf break regions. The weakeningwinds mark the onset of stage 4; during this relaxation period, satellite imagery shows thatthere is intense warming of the relatively shallow shelf-waters to the south of BrooksPeninsula. In addition, there is also warming in the offshore waters, but between the warmoffshore and shelf waters, a band of cold water persists along the shelf break that isprobably relic upwelled water that has remained along the shelf break in spite of theweakening winds. The persistence of band of cold water along the shelf break during theupwelling event was explained in terms of the conservation of potential voracity byevoking a simple barotropic model that was tied to the sharp change in the cross-shoretopography at the shelf break. Although shelf break upwelling is the likely cause of thecold water along the shelf break, it is possible that the water was advected from regionsfurther north. After a few days of relatively weak winds, the strong thermal gradientsobserved along the shelf break were no longer present in the AVHRR imagery.Analysis of the SST imagery using an edge detector indicated the presence of strongthermal fronts on the inshore and offshore side of the shelf break, and a front in theoffshore waters. The shelf break fronts were most pronounced at Tatchu Point, just southof Brooks Peninsula; to the south of Tatchu Point, only the shoreward shelf break front108was consistently observed. The cross-shore scale of the shelf-break front at Tatchu Pointwas about 20 km, which is roughly equal to the internal Rossby radius of deformation, butmoving to the south, the cross-shore scale of the shelf-break fronts increased, and theoffshore shelf-break front appeared to merge with the offshore front. The merging of thefronts to the south was probably due to the increasing width of the continental shelf and thedecreasing gradient of the shelf slope, so that the cross-shore scales were modified by thetopography. The persistence of the inshore shelf-break front may be due to the shallowshelf-waters that show greater variability due to solar heating. In addition, the inshoreshelf-break front may be related to the cross-shore shear between the equatorward flowingshelf break current and the poleward flowing Vancouver Island Coastal Current. Theoffshore front roughly coincided with the 2000 m bathymetry contour, and presumablyrepresented the offshore extent of the seasonal shelf break current; whereas, the shelf-breakfronts represent the surface signature of the cold-core of the shelf break current thatintensified during the upwelling event.Comparison of hydrographic data to AVHRR imagery showed a good qualitativeagreement. Upper layer cross-shore temperature contours at Estevan Point calculated fromCTD casts were similar to structure to the SST imagery: warmer water over the shelf andoffshore regions, and isotherms that dome upward over the shelf break with minimumtemperature just seaward of the shelf break. Similarly, continuous measurements of thetemperature taken from subsurface moorings at depths of 5 m and 30 m were alsoconsistent with the AVHRR SST structure. The longshore surface velocity was estimatedusing an AVHRR feature tracking technique and a simple time-independent barotropicmodel that was governed by the conservation of potential vorticity. The feature trackingand the barotropic model velocities agreed within an order of magnitude to velocitymeasurements recorded by a subsurface current mooring; both velocity estimates werelarger in magnitude than the current meter results, but the current meter measurements and109the velocity estimates indicated a velocity increase as the upwelling favourable windsincreased in strength. However, the AVHRR-derived velocities on July 18 (day 200) wereessentially the same as velocities measured by the E2 current mooring on the same date; thegood agreement was related to the decay of upwelling along the shelf break as the coastalupwelling approached a quasi-steady state. Overall, the in situ observations agreedfavourable with the AVHRR thermal imagery.6.2 ConclusionAt the beginning of this thesis, five major objectives were outlined, and for themost part, these objectives have been answered. The first objective was concerned with thespatial and temporal scales of the upwelling event. The cross-shore spatial scales aredominated by the presence of thermal fronts on the inshore and offshore side of the shelfbreak, and a front in the offshore waters. The shelf break front cross-shore scales aretypically 20 km, which is consistent with the internal Rossby radius of deformation. Theoffshore front cross-shore scale is around 70 km and probably represents the offshoreextent of the shelf break current. In the long-shore direction, the spatial scales presumablydepend on the strength and duration of the winds. The temporal scale of the upwellingevent was order 10 days and is probably tied to the length of time that a high-pressureatmospheric system, strong enough to cause an upwelling event, would sustain itself.The second objective dealt with the relationship between the satellite-sensed thermalstructure and dynamical processes within the ocean. Under the right conditions, when anocean-interior dynamical process manifests itself as a surface thermal feature, vis-a-vis theupwelling event, then AVHRR observations are representative of interior processes;110however, intense solar heating, combined with low winds can presumably mask interiorprocesses. In addition, the temporal and spatial scales of an upwelling event are readilyobserved using AVHRR imagery; whereas, smaller-scale dynamical processes are not asamenable to AVHRR observations due to the spatial resolution of the satellite imagery, anddepending on the period of the dynamical process, satellite observations may be limited byorbital constraints.The issue of shelf break upwelling versus advection was the subject of the thirdobjective. It was not clear which of these two processes were dominant in producing theband of cold water that persisted along the shelf break, but the cross-shore spatial scalesmeasured from the AVHRR imagery, the longshore averaged SST structure, the cross-shore temperature structure at Estevan Point, and the variability of the AVHRR-derivedsurface velocities all indicate that shelf break upwelling is probably occurring; however,since the cross-shore scales that characterize the shelf break upwelling and the shelf breakjet are essentially the same, it may be difficult to separate both processes. Other studieshave shown that the effect of a blunt headland jutting into the mean flow (BrooksPeninsula), and the effects of longshore and cross-shore variations in the topography mayalso enhance the upwelling process, and thus account for some of the observed SSTstructure. The dynamical process that forces the cold water to remain centered along theshelf break as the water migrates equatorward from Brooks Peninsula was the main issueof the fourth objective. The dynamics can be partly explained in terms of the conservationof potential vorticity that is tied to the variation of the cross-shore topography at the shelfbreak, but upwelling is a complicated event, and there is no doubt that other dynamicalprocesses are playing a role.The uniqueness of the upwelling event was addressed by the fifth objective, and itappears that the July, 1984 upwelling event is indicative of the spatial and temporal111variation of the SST that can occur during the summer months; qualitative investigations ofAVHRR imagery from other summers confirms the similarity to the 1984 observations, butinterannual variability frequently occurs that is typified by the formation of eddies, and inthe offshore migration of plumes of cold, upwelled water. The interannual variabilitypresumably has a significant effect on the transport of nutrients into the La Perouse Bankfishing grounds. As was the case during July, 1984, the equatorward extent of theupwelled water was significant (see Figure 5.12a), and there was probably a considerabletransport of nutrients in the long-shore direction; however, at other times, offshore plumesform, and the nutrients that are transported to the surface during the upwelling process areadvected offshore (see Figure 1.2b) and do not reach the region in the vicinity of LaPerouse Bank. One of the unanswered questions from this study dealt with the issue ofshelf-break upwelling versus long-shore advection, and the resolution of this question mayyield some estimate as to whether or not nutrients are upwelled at specific locations alongVancouver Island and transported by long-shore advection, or if nutrients are upwelledcontinuously along the shelf break.The results of this study may also act as a starting point to develop a comprehensivepicture of the interannual variability of the shelf break current. A reasonably long timeseries of AVHRR data are difficult to obtain, but a time series of coastal wind data andBakun Upwelling Index data are readily available. By the combining the temporal andspatial scales obtained from the July, 1984 AVHRR data set with wind data from otheryears, it may be possible to interpolate the SST structure when satellite imagery is notavailable.112CHAPTER 7BIBLIOGRAPHYAbbot, M.R., and D. B. Chelton, Advances in passive remote sensing of the ocean, inU.S. National Report 1987-90, International Union of Geodesy and Geophysics,American Geophysical Union, 571-587, 1991.Allen, J.S., Upwelling and coastal jets in a stratified ocean, J. of Phys. Oceanography, 3,245-257, 1973.Allen, J.S., Models of wind-driven currents on continental shelves, Annu. Rev. FluidMech., 12, 389-433, 1980.Apel, J.R., Principles of Ocean Physics, Academic Press, London, 634 pp., 1987.Bakun, A., Daily and weekly upwelling indices, west coast of North America, 1946-71,NOAA Tech. Rep. 16, NMFS-SSRF-671, 96 pp., U.S. Dept. of Commerce,1973.Bang N.D. and W.R.H. Andrews, Direct current measurements of a shelf-edge frontal jetin the southern Benguela system, J. Mar. Res., 32, 407-419, 1974.Barton, E.D., A. Huyer, and R.L. Smith, Temporal variation observed in the hydrographicregime near Cape Coveiro in the northwest African upwelling region, February toApril 1974, Deep Sea Res., 24, 7-23, 1977.Bowden, K.H., Summing-up, Philos. Trans. R. Soc. London, Ser A, 302, 683-689,1981.Charney, J.G., The generation of ocean currents by wind, J. Mar. Res., 14, 477-498,1955.^ 113Crepon, M., C. Richez, and M. Chartier, Effects of coastline geometry on upwelling, J.Phys. Oceanography, 14, 1365-1382, 1984.Csanady, G.T., Circulation in the Coastal Ocean, Reidel, Holland, 278 pp., 1982.Dickson, R.R., P.A. Gurbutt, and V.N. Pillai, Satellite evidence of enhanced upwellingalong the European continental slope, J. Phys. Oceanography, 10, 823-819, 1980.Denman, K., R. Forbes, A. Gargett, D. Mackas, R. Thomson, and S. Calvert, Processescontrolling vertical and horizontal exchanges of carbon along the Canadian Pacificcontinental margin, JGOFS News, in Canadian Met. Ocgy. Society AnnualReview 1992, 26 pp., 1992.Dodimead, A.J., F. Favorite, and T. Hirano, Salmon of the North Pacific Ocean-Part II:Review of the oceanography of the subarctic Pacific region, Int. North Pac. Fish.Comm., Bull. 13, 195 pp., 1963.Emery, W.J., M. Ikeda, A comparison of geometric correction methods for AVHRRimagery, Canadian J. of Remote Sensing, 10, 46-56, 1984.Emery, W.J., W.G. Lee, and L.Magaard, Geographic and seasonal distributions of Brunt-Vaisala frequency and Rossby radii in the North Pacific and North Atlantic, J.Phys. Oceanography, 14,294-317, 1984.Emery, W.J., A. Thomas, M. Collins, W. Crawford, and D. Mackas, An objectivemethod for computing advective surface velocities from sequential infrared satelliteimages, J. Geophysical Res., 91, 12,865-12,878, 1986.Emery, W.J., J. Brown, and Z. Nowak, AVHRR image navigation: Summary and review,Photogrammetric Eng. Remote Sensing, 55, 1175-1183, 1989.Emery, W.J., P. Schluessel, Global differences between skin and bulk sea surfacetemperatures, Eos, 70(14), 211-213, 1989.114Favorite, F., A.J. Dodimead, and K. Nasu, Oceanography of the subarctic Pacific region,1960-72, Int. North Pacific Salmon Comm., Bull 33, 187 pp., 1976.Fang, W., W.W. Hsieh, Summer upwelling off Vancouver Island from satellite SST data,submitted to J. Geophysical Res. (Oceans), July 1992.Freeland, H.J., K.L. Denman, A topographically controlled upwelling centre off southernVancouver Island, J. Mar. Res., 40, 1069-1093, 1982.Freeland, H.J., W.R. Crawford, and R.E. Thomson, Currents along the Pacific coast ofCanada, Atmosphere -Ocean, 22, 151-172, 1984.Freeland, H.J., The flow of a coastal current past a blunt headland, Atmosphere -Ocean,28, 288-302, 1990.Gill, A.E., Atmosphere-Ocean Dynamics, Academic Press, London, 662 pp., 1982.Griffin, D.A., P.H. LeBlond, Estuary/ocean exchange controlled by spring-neap tidalmixing, Estuarine, Coastal, and Shelf Science, 30, 275-297, 1990.Hickey, B.M., R.E. Thomson, H. Yin, and P.H. LeBlond, Velocity and temperaturefluctuations in a buoyancy-driven current off Vancouver island, J. GeophysicalRes., 96, 10,507-10,538, 1991.Hill, R.B., J.A. Johnson, A theory of upwelling over the shelf break, J. Phys.Oceanography, 4, 12-24, 1974.Holland, J.A., X.H. Yon, Ocean thermal feature recognition, discrimination, and trackingusing infrared satellite imagery, IEEE Trans. Geosci. Rem. Sensing, 30, 1046-1053, 1992.115Huyer, A., Coastal upwelling in the California current system, Prog. Oceanography, 12,259-284, 1983.Ikeda, M., L.A. Mysak, and W.J. Emery, Seasonal variability in meanders of theCalifornia current system off Vancouver Island, J. Geophysical Res., 89, 3487-3505, 1984.Ikeda, M., L.A. Mysak, and W.J. Emery, Observation and modeling of satellite-sensedmeanders and eddies off Vancouver Island, J. Phys. Oceanography, 14, 3-21,1984a.Ikeda, M., W.J. Emery, A continental shelf upwelling event off Vancouver Island asrevealed by satellite infrared imagery, J. Mar. Res., 42, 303-317, 1984.Janowitz, G.S., L.J. Pietrafesa, The effects of alongshore variation in bottom topographyon a boundary current (topographically induced upwelling), Cont. Shelf Res., 1,123-141, 1982.Jardine, I., Upwelling off Vancouver Island, M.Sc. Thesis, University of BritishColumbia, 87 pp., 1991.Johnson, A.J., A.J. Nurser, A model of secondary upwelling over the shelf break. II,Geophys. Astrophys. Fluid Dynamics, 28, 161-170„1984.Jensen, J.R., Introductory Digital Image Processing-A Remote Sensing Perspective,Prentice Hall, New Jersey, 385 pp., 1986.Kaufman, Y.J., B.N. Holben, Calibration of the AVHRR visible and near-IR bands byatmospheric scattering, ocean glint, and desert reflection, Inter. J. of Rem.Sensing, 14(1), 21-52, 1993.Kelley, K.A., P.T. Scrub, Comparison of velocity estimates from advanced very highresolution radiometer in the coastal transition zone, J. Geophysical Res., 97, 9653-9668, 1992.116Kundu, P.K., Fluid Mechanics, Academic Press, San Diego, 638 pp., 1990.Legeckis. R., A survey of worldwide sea surface temperature fronts detected byenvironmental satellites, J. Geophysical Res., 83, 4501-4522, 1978.LeBlond, P.H., L.A. Mysak, Waves In the Ocean, Elsevier, Amsterdam, 602 pp., 1977.LeBlond, P.H., B.M. Hickey, and R.E. Thomson, Runoff driven coastal flow off BritishColumbia, In S. Skrerslet (ed.) The role of freshwater outflow in coastal marineecosystems, NATO ASI Series, G7, 309-317, 1986.Mackas, D.L., L.Washburn, and S.L. Smith, Zooplankton community patterns associatedwith a California current cold filament, J. Geophysical Res., 96, 14,781-14,797,1991.Mason, J.E., A. Bakun, Upwelling index update, U.S. west coast, 33N-48N latitude,NOAA-TM-NMFS-SWFC-67, U.S. Comm. Dept., 21 pp., 1986.Mooers, C.N.K., C.A. Collins, and R.L. Smith, The dynamic structure of the frontal zonein the coastal upwelling region off Oregon, J. Phys. Oceanography, 6, 3-21, 1976.Ninnis, R.M, W.J. Emery, and M.J. Collins, Automated extraction of pack ice motionfrom advanced very high resolution radiometer imagery, J. Geophysical Res., 91,C9, 10,725-10,734, 1986.O'Brien, J.J., H.E. Hurlburt, A numerical model of coastal upwelling, J. Phys.Oceanography, 2, 14-26, 1972.Pedlosky, J., On coastal jets and upwelling in a bounded basin, J. Phys. Oceanography, 4,3-18, 1974.117Pedlosky, J., Geophysical Fluid Dynamics, Springer-Verlag, New York, 710 pp., 1987.Petrie, B.D., Current response at the shelf break to transient wind forcing, J. GeophysicalRes., 88, 9567-9578, 1983.Peffley, M.B., J.J. O'Brien, A three-dimensional simulation of coastal upwelling offOregon, J. Phys. Oceanography, 6, 164-179, 1976.Pond, S., G. Pickard, Introductory Dynamical Oceanography, Pergamon Press, Oxford,319 pp., 1983.Pratt, W.K., Digital Image Processing, 2 ed., John Wiley & Sons, New York, 578 pp.,1991.Ramp, S.R., P. Jessen, K. Brink, P. Niiler, F. Daggett, and J. Best, The physicalstructure of cold filaments near Point Arena, California, during June 1987, J.Geophysical Res., 96, 14,859-14,883, 1991.Robinson, I.S., Satellite Oceanography: An introduction for oceanographers and remote-sensing scientists, Ellis Horwood, England, 455 pp., 1985.Ryther, J.H., Photosynthesis and fish production in the sea, Science. 166, 72-76, 1969.Saunders, R.W., and D.P. Edwards, Atmospheric transmittances for the AVHRRchannels, Applied Optics, 28, 19, 4154-4160, 1989.Schlussel, P., W. Emery, H. Grassi, and T. Mammen, On the bulk-skin difference and itsimpact on satellite remote sensing of sea surface temperature, J. Geophysical Res.,95, 13,341-13,356, 1990.Siebold, E., W.H. Berger, The Sea Floor: An Introduction to Marine Geology, Springer-Verlag, New York, 1982.118Simpson, J.J., On the accurate detection and enhancement of oceanic features observed insatellite data, Remote Sensing Environment, 33, 17-33, 1990.Smith, S.D., Wind stress and heat flux over the ocean in gale force winds, J. Phys.Oceanography, 10, 709-726, 1980.Smith, R.L., A comparison of the structure and variability of the flow fields in three coastalupwelling regions: Oregon, Northwest Africa, and Peru, in Coastal Upwelling,Coastal Estuarine Science Series , vol. 1, edited by F.A. Richards, 107-118, AGU,Washington, D.C., 1981.Stewart, R.H., Methods of Satellite Oceanography, University of California Press,Berkeley, 360 pp., 1985.Tabata, S., On the accuracy of satellite-observed sea surface temperatures, inOceanography From Space, edited by J.F.R. Gower, 145-157, PlenumPublishing, New York, 1981.Thomas, A.C., Relationships between near-surface plankton distributions, hydrography,and satellite measured sea surface thermal patterns, Ph.D. Thesis, University ofBritish Columbia, 147 pp., 1987.Thomson, R.E., Oceanography of the British Columbia Coast, Can. Spec. Publ. Fish.Aquat. Sci., 291 pp., 1981.Thomson, R.E., A comparison between computed and measured oceanic winds near theBritish Columbia coast, J. Geophysical Res., 88, 2675-2683, 1983.Thomson, R.E., A cyclonic eddy over the continental margin of Vancouver Island, J.Geophysical Res., 88, 2675-2683, 1984.Thomson. R.E., J.F.R. Gower, A wind-induced mesoscale eddy over the VancouverIsland continental slope, J. Geophysical Res., 90, 8981-8993, 1986.119Thomson, R.E., B.M. Hickey, and P.H. LeBlond, Water property observations from theVancouver Island Coastal Current experiment: June, July, and October, 1984, Can.Data Rep. Hydrogr. Ocean Sci., 46, 505 pp., 1986.Thomson, R.E., W.R. Crawford, H.J. Freeland, and W.S. Hugget, Low-pass filteredcurrent meter records for the west coast of Vancouver Island: Coastal OceanDynamics Experiment, 1979-81. Can. Data Rep. Hydrogr. Ocean Sci. 40: 102 pp.,1986a.Thomson, R.E., B.M. Hickey, and P.H. LeBlond, The Vancouver Island coastal current:Fisheries barrier and conduit, in Effects of Ocean Variability on Recruitment an• anEvaluation of Parameters Used in Stock Assessment Models, edited by R.J.Beamish and G.A. McFarlane, Can. Spec. Publ. Fish. Aquat. Sci., 108, 265-296,1990.Walton, C.C., E.P. McClain, and J.F. Sapper, Recent changes in satellite-based multi-channel sea surface temperature algorithms, MTS '90, Marine Technology SocietyMeeting, Washington, D.C., 7 pp. September 1990Wang, D.C., A.H. Vagnucci, and C.C. Lin, Digital Image Enhancement: A Survey,Computer Vision, Graphics, and Image Processing, 24, 363-381, 1983.Ware, D., R.E. Thomson, La Perouse project: First annual report, 1985, Dep. Fish.Oceans Canada, 25 pp., 1986.Ware, D., R.E. Thomson, La Perouse project: Third annual report, 1987, Dep. Fish.Oceans Canada, 64 pp., 1988.Wick, G., W.J. Emery, A comprehensive comparison between satellite-measured skin andmultichannel sea surface temperature, J. Geophysical Res., 97, 5569-5595, 1992.120