JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 88, NO. C10, PAGES 5960-5972, JULY 20, 1983 Near-Inertial Motions Off the Oregon Coast IAIN ANDERSON United States Coast Guard, Atlantic Area, GovernorsIsland, New York 10004 ADRIANA HUYER AND ROBERT L. SMITH Schoolof Oceanography, OregonState University,Corvallis,Oregon97331 Near-inertial motions were observedat all current metersin an array of five mooringsspanningthe continentalmarginoff centralOregonduringOctober1977to January1978.All mooringswerebetween 10 and 130 km from shore,in water depthsbetween100 m and 2500 m. Largestnear-inertialamplitudes (> 30 cm/s)wereobserved at the uppermost currentmetersof the offshoremoorings,althoughthesewere belowthe surfacemixedlayer. The near-inertialenergygenerallydecreased with increasingdepth,and therewaslessnear-inertialenergyover the continentalshelfthan at similardepthsoffshore.The energy levelsobservedover the shelfwereabout the sameas observedthereduringsummer1973,and the energy levelsobservedoffshorewere comparablewith thoseobservedin the open North Atlantic. Horizontal coherence scales werelarge,exceeding 115km duringthefirsthalfof theobservational periodandabout60 km duringthe secondhalf;estimates of thehorizontalwavelength (50 km duringthefirsthalf and 20 km duringthe secondhal0 suggest that thecoherence scaleis of theorderof threewavelengths. Althoughwe did not have current data in the surfacemixed layer nor wind measurements over each mooring, the availablemeteorological data (synopticpressurechartsand hourlywind and pressureat Newport)suggest that mostof the near-inertialenergywasforcedby the local wind. INTRODUCTION Pure inertial motion on the rotating earth is circular,rotating clockwisein the northernhemisphere with a periodof 2n/f, wheref is the local Coriolis parameter.Such pure inertial scalesof about 60 km in the North Atlantic [Pollard, 1980; Fu, 1981]. In this paper,we investigatenear-inertialmotionsin current observationsmade during late fall and winter at five sites spanningthe continentalmargin off central Oregon. We demotions are seldom observed in the ocean. There is almost always someother force acting on the fluid that cannot be scribe the near-inertial motions in terms of their amplitude, completelyneglected.However,motionsthat are very nearly coherence,and phaseby usingboth spectraland band-passing inertial in character(called 'near-inertialmotions')with very techniques,and attempt to accountfor their generationby the nearly circularrotation in the proper senseand with a fre- local wind. Where appropriate,the resultsare comparedwith quencynearf have been observedat all depthsin the oceans thoseof previousstudies. and large lakes of the world [Webster, 1968]. The most common characteristics of near-inertial motions are well docu- mented [Kroll, 1975; Kundu, 1976; Fu, 1981; Thomsonand Huggett,1981].They are highlyintermittentand generallylast only a few oscillationswith a maximumvelocityof 20-30 cm/s; the observedfrequency,o•, is generallyslightlygreaterthanf; vertical coherencescalesare of the order of only a few tens of meters; and horizontal coherencescalesare of the order of a few tens of kilometers. Previousstudiesof inertial currentshave usuallybeen conductedeitherin the open ocean[e.g.,Pollard, 1980;Fu, 1981] or over the continentalshelf[e.g.,Johnson,1976; Thomsonand Huggett, 1981]. The few availablestudiesof near-inertialmotions over the Oregon shelf are based exclusivelyon data collectedin summer.Maximum amplitudesobservedover the Oregon shelfwere about 15 cm/s during the summerof 1973 [Johnson,1976; Kundu, 1976], considerablylower than the open-oceansurface-layer valuesof about 50 cm/sat site D in the North Atlantic [Pollard, 1980] and 35 cm/s in the Mediterranean[Gonella,1971]. Horizontal coherencescalesoverthe Oregonshelfwerebetween10 and 35 km duringsummer1972 [Kindle, 1974] and about 15 km duringsummer1973[Johnson, 1976]; theseare smallin comparisonwith horizontalcoherence Copyright1983by the AmericanGeophysicalUnion. Paper number3C0480. 0148-0227/83/003C-0480505.00 5960 OBSERVATIONS During 1977 and 1978, Oregon State Universityand the Universityof Washingtonconducteda joint fieldprogram,the SlopeUndercurrentStudy(SUS),to studythecirculationalong the continentalslopeoff centralOregon.As part of thisstudy, an array of currentmeterswasmooredacrossthe continental marginat 45ø20'NbetweenOctober 1977 and January1978 (Figure1).Occasional hydrographic sections weremadealong the mooredarray, and coastalwindsweremeasuredcontinuouslyat Newport, Oregon. The SUS array consisted of fivemoorings:'Skunk,'overthe 2600m isobathjust seawardof thefoot of thecontinentalslope with current metersat lea, 600, and 1850 m; 'Leopard,' over the 1lea m isobathat midslopewith currentmetersat 70 and 820 m; 'Ocelot,'overthe 600 m isobathon the upperslopewith current meters at 141, 243, and 371 m; 'Puma,' at the 300 m isobath near the shelfbreak with current meters at 64, 164 and 261m; and'Elephant,'overthe 110m isobathat mid-shelfwith current metersat 23, 49, and 89 m. Ocelot and Elephant were maintainedby the Universityof Washington,and Skunk,Leopard,and Pumaweremaintainedby OregonStateUniversity. All mooringsweredeployedin October,with intendedrecovery in January,but the originalPuma mooringwas lost and replacedin November,andOcelotwasrecovered prematurely in early December.Althoughthere was only a very brief (390hour) periodwith data from all five moorings,thereare two A•DEgS08 ET AL.: NEAR=INERTIAL MOTI08 OFF OgF,OO8 .5961 46 ø N Elephant Leopard ßSkunk ß 126ø W ß Oce lot 125• 124 • Fig. 1. Locationof themoorings of theSlopeUndercurrent Study.Theshaded areasindicatethelocationof priorstudies of inertial currents. periods of about 55 days each with data from four separate moorings (Figure 2). These two periods are limited by the durationof Ocelotand Puma,respectively (Figure2). All of the mooringsof the SUS array were sub-surface,and Aanderaa current meters were used throughout. All instruments recorded mean speed and instantaneousdirection at inervalsof 40 min. Time seriesof speedwere plotted for error detection,and spuriousvalueswerereplacedby linear interpolation. After removal of obviouserrors,the speedand direction were used to determine the eastward (u) and northward (v) componentsof the current.Recordgapsof 120 hoursbeginning at 0900 October 10 in Ocelot371 m and 136hoursbeginningat 0200 December N 2 = -g Op/l•Oz(wherep is density,g is theacceleration due to gravity,and z is verticallyupward),werecomputedfrom the hydrographicdata by usinga centeredfinite differenceof 20 m. These profiles (e.g., Figure 3) show that most of the current metersin the array werebelowthe surfacemixedlayer,during both the first (Ocelot) and second(Puma) periods.The mixed layer wasfairly shallow(lessthan 25 m) duringthe first period. I" FIRST PERIOD ' ,I I 1/85o I I I Leopora 'l ' , I Ii i Ocelot , I I •l' I /oo,,, I•øø I 70m I 8ZO I, ' i' I I ß I ! Puma I I Elephant •' [I I I0 • II '1 OCT SECOND PERIOD • '1 Skunk, 9 in Puma 261 m were were filled with zeroes. The data were filtered with a half power point of 2.5 hours to suppress high frequencysignalsand decimatedto hourly values to yieldthe processed data. Vertical profilesof the squareof the Brunt-V•iis•il•ifrequency, ' i' I• ' I [ 49 iß 20 NOV I0 20 I. •6/ II DEC .' I . I0 20 JAN I0 20 Fig. 2. Summary ofthetimespanofthecurrentmeterdata.Thetwotimeperiods usedforspectral analysis areindicated. 5962 ANDERSONET AL.' NEAR-INERTIAL MOTION OFF OREGON I25øL 126 ø W I I I 0 E I• I I I 124 ø I]• : /o -- 5OO - -• _ - ct 1977 13_ - iooo _ 40 cph 2 i _ I -- 500 _ _ _ i 15o I IOO S* 126 o w N•EFROM SHORE (km) 50 _ msoL I ! ' ' •000 0 P E I It ' II ß •,•o I)lø -1- Dec I/'. ß 1977 - ,ooo (*26Jan1978) - . -- 15OO _ _ •CE FROM SHORE (km) i 15o I I 50 ioo _ _ I 2000 0 Fig. 3. Profiles oftheBrunt-V/iisiilii frequency, N 2,neareachmooring. The only currentmeter possiblyin the mixed layer was Elephant23 m; the top currentmeterof eachothermooringwasin the pycnocline(Figure3). During the secondperiod,the mixed layer reachedalmost to the bottom over the shelf,placing Elephant23 m and 49 m in the mixedlayerand Elephant89 m in the pycnocline;again,the top currentmeter of each other mooringwasin the pycnocline. Time seriesof the eastwardcomponentof the currentat all current meters(Figure 4) show frequentepisodeswith motion that hasa periodof about two-thirdsof a day (i.e.,roughlythe localinertialperiodof 17 hours).Someof theseepisodesappear to beginimpulsively(e.g.,at Leopard 70 m on December13), but othersshow a gradual increasein amplitude (e.g.,at Leopard 70 beginningabout October25). The amplitudeof these oscillationsappears to decreasewith depth at all moorings (Figure4). Althoughonly the uppercurrentmetersat Elephant are in the surfacemixed layer, all of the currentrecordsfrom the SUS array contain a substantialamount of near-inertial motion. SPECTRAL ANALYSIS The vector time series from each current meter were used to calculate rotary spectra,which separate the clockwise-and counterclockwise-rotatingfrequency components [Mooers, 19731. Pure inertial motion at this latitude would have a clockwise-rotating peak at 0.0592cph and no counterclockwise energy.Sincethe commonrecordlengthfor all five moorings was too short for meaningfulspectralanalysis,we used two time periodsof 1280 hours,which both had simultaneousdata s Ioo S 600 S 1850 L 70 L 82.0 0 141 0 24:5 0 371 •tober 1977 November December Jenuery 1978 Fig. 4a P 64 P 164 P 261 •':[ l•J•,.•,,•,,n,i•Al••, •,,,,.,,,,.,,n,,•., ,A,.. ,,.,An,,,,, ,.,•,,.,. .... ,,,, ,.,,,,_ •,_ •,,•,,,.,._ ,,.., ,,, ,••,..,,,,,•,,._ :,....•, :,,.,. ,.,...,,,.,.. _•,o[I •r!y"uv"?"•I'p,"'•'•Qlrl•'Vl'lF"'•"nrpV•'llp,:•'•Ir,'¾" ',•",""•,"v•'•'•",,T•,,•"'?l'• '••' •':•..,,AI,.,,,,, _, •,I,,,,.,,,,,,.,,,A.,,&,•I.LIII,II•b I•,tlJlil•bll dlld]A. ,,[11,iAnlllbn,,,J ,,,J ,.t hi,a, ,I,l.,q ILb A,.,,. _.,, ,•,q. ,.•,,..., ,..,,, ,,.,,t.,,,,•,,,,,,11 .,.,,,_ .u ,t,,,,.,, •,q I0 20 October 1977 •0 I I0 20 I November I0 20 December 30 I I0 January1978 Fig. 4b Fig. 4. Theeastward (u)components of thecurrents at thecurrentmetersat (a)Skunk,Leopard,andOcelot,and(b)Puma and Elephant. 20 5964 ANDERSONET AL.' NEAR-INERTIAL MOT•O• OFF OREGO• Leopard70 Elephant23 Elephant49 clockwise F•rst Period •' Io nz i 01 uJ I00 o• I0 i • Second Period I0 i 0 Iø 0 08 i 0 0 04 0 08 I 0 I o 04 I I I o 08 cph Fig. 5. Rotaryspectra ofselected current meterrecords forboththefirst(Oct.13toDec.5)andsecond (Nov.20toJan.12) timeperiods. Thelocalinertialfrequency f, andthesemidiurnal frequency M 2areindicated. from four of the moorings(Figure 2). The first time period,with data from Skunk, Leopard, Ocelot, and Elephant, began at 0400 UT October 13 and ended at 1200 UT December 5; this period includedan episodeof strongnear-inertialmotion at the end of October.The secondperiod startedat 0400 UT November 20, ended 1200 UT January 12, and included data from Skunk, Leopard, Puma, and Elephant. We allowed the two periodsto overlap by 15 days to useas much of the available data as possible.Spectrawereobtainedfor eachtime periodby ensemble-averagingfive periodogramscalculated from 256hour recordsegments. The resultingspectrahave 10 degreesof freedomand a spectralbandwidthof 0.0039cph,about 6% off. The rotaryautospectra for the firsttimeperiod(e.g.,Figure 5) all showa prominentclockwise-rotating peaknearthe local inertialfrequencyand very little counterclockwise energyin that frequencyband.In all cases,the near-inertialenergyis equalto or greaterthan the semi-diurnalenergy.The peak frequency of thenear-inertialmotionvariesslightlyamongthe differentcurrentmeters;in mostcases, it is somewhatgreater thanf. Similarfrequencyshiftshavebeenobservedelsewhere in the ocean[e.g.,Fu, 1981; Gonella,1971; Kundu,1976]. The peak energydecreases with increasingdepth at eachlocation exceptoverthe continentalshelf(Table1, Figure6). The ratio betweenthe majorand minoraxescomputedfromthe rotary TABLE 1. Characteristics of theClockwiseNear-InertialPeakin theRotarySpectrumfor EachCurrentMeter DuringBothTime Periods SecondPeriod (Nov. 20 to Jan. 12) First Period (Oct. 13 to Dec. 5) Peak Peak Peak Peak Current Energy, Frequency, Width Major Axis/ Energy, Frequency, Width Meter cTn 2/S2 cph cph MinorAxis cTn 2/S2 cph cph 26.9 5.1 1.9 0.0625 0.0625 0.0625 0.0098 0.0085 0.0066 1.01 1.01 1.02 19.9 11.3 2.7 0.0586 0.0586 0.0625 0.0106 0.0097 0.0101 1.01 L 70 m L 820 m 147.0 3.4 0.0586 0.0625 0.0060 0.0062 1.01 1.02 56.1 3.1 0.0625 0.0586 0.0054 0.0054 1.00 1.01 O 141 m O 243 m 40.1 18.6 0.0586 0.0625 0.0084 0.0082 1.03 1.02 O 371 m 16.5 0.0586 0.0097 1.03 S 100 m S 600 m S 1850 m P 64 m P 164 m P 261 m -• E 23 m 25.8 0.0664 0.0122 1.10 E 49 m E 89 m 9.9 13.2 0.0625 0.0625 0.0107 0.0069 1.18 1.03 -• -• Major Axis/ Minor Axis 1.02 1.02 • • 13.7 0.0625 0.0100 2.3 0.0508 0.0108 8.8 2.2 0.0625 0.0625 0.0074 0.0097 1.06 1.51 1.03 1.20 4.4 3.4 1.4 5.7 0.0625 0.0508 0.0547 0.0586 0.0099 0.0118 0.0156 0.0069 1.46 7.02 1.41 1.22 ANDERSONET AL.: NEAR-INERTIALMOT•O•4OFF OREC, O•4 SPECTRAL ENERGY at 0.0625cph (cm2/sec z) I0 I0 I00 E // I000 // ioo //// /// // / • ,/I/ , I I SPECTRAL ENERGY ot 0.062.5cph (cm2/sec •) I.O • ioo IO IOO IOOO , , //, '//'• .//.•E ///'// ,// /,/•L /// / p / w / // /S •// // SECOND PERIOD (20Nov-12 Jan) / / I0,000 / / / / , ,4 • , Fig. 6. Log-logplot of the clockwisespectralenergyversusdepth for both time periods.Along the solidline energywould vary inversely with depth.The separationbetweenthe dashedlinesindicatesthe 95% confidenceintervalfor eachspectralestimate. spectra[Mooers,1973] showsthat the near-inertialmotion is very nearly circular offshore,with slight ellipticity over the continentalshelf(Table 1). The bandwidthof the near-inertial peak, definedas the differencebetweenfrequencieswhere the energyfalls to one-half of the peak value [Fu, 1981], is about 0.008 cph (twice the spectralbandwidth)offshore,and somewhat greater over the continental shelf (Table 1). The bandwidth of the peak is inverselyproportionalto the persistence of near-inertial motions [Munk and Phillips, 1968]; the persistence seems to be about 120 hours offshore and 100 hours over the shelf. 5965 period; elsewherethe mean flow was much weaker.During the first period,meancurrentswereweak everywhere. During both time periods,the near-inertialenergydecreases monotonicallywith depth at each of the offshoremoorings (Table 1). A log-logplot of the energyversusdepth at the most commonpeak frequency(Figure 6) showsthat the pointsfrom Skunk, Leopard, and Ocelot lie roughly along a straight line correspondingto an inverserelationshipbetweenenergyand depth.During both time periodsthe currentmetersat Elephant (i.e., over the shelf) have much lessnear-inertial energy than offshore current meters at the same depth (Figure 6). The current meters at Puma, at the edge of the continental shelf, have lessnear-inertialenergythan thosein the deep water but more than thoseover the shelf(Figure 6). Data from previous studiesshow a similar variation of near-inertial energy with depth. The energylevelsobservedin the open North Atlantic [Pollard, 1980; Fu, 1981] seemto be comparableto those at Skunk, Leopard, and Ocelot (i.e., thesedata seemroughly to obey the sameinverserelationshipbetweenenergyand depth). The near-inertial energy observedover the Oregon shelf in summer 1973 [Kundu, 1976] are similar to thoseat Elephant, and energy levels from the vicinity of the 200-m isobath in Queen Charlotte Sound [Thomson and Huggett, 1981] are comparableto thoseat Puma. Thus, it may be generallytrue that there is lessnear-inertial energy over the continental shelf than at similardepthsin the openocean. Coherenceand phasespectrawere computedfor all current meter pairs and both time periods.With 10 degreesof freedom, coherencesquaredvaluesof 0.53,0.58,and 0.68 are significantly different from zero at the 95, 97, and 99% levels,respectively [Thompson,1979]. There was significantcoherenceat the 95% level at a clockwise-rotatingfrequency between 0.0508 and 0.0664cph between23 out of 55 currentmeterpairsduring the first period and between 19 pairs during the secondperiod. During both periods, there was generally higher coherence betweenhorizontally or diagonally separatedcurrent meters than betweenverticallyseparatedcurrentmeters(Figure 7). The horizontal coherence scale exceeded the extent of the array (115 km) during the first period and was at leasthalf the extent of the array during the secondperiod (Figure 8). The large horizontal coherencescale of the first period exceeds thoseobservedin the open oceanat comparabledepths [Fu, 1981] and is comparableto the coherencescalesobservedin the surfacemixedlayer at siteD in the North Atlantic [Pollard, 1980] and in the relativelyshallowwatersof Queen Charlotte Soundoff BritishColumbia [ThomsonandHuggett, 1981]. Phasespectrawerealsocomputedfor both time periods,and confidencelimits for phasewere determinedby the procedure describedby Koopmans[1974, p. 284-285]. This method of computingconfidence intervalsis conservative, sinceit assumes During the secondperiod,therewereprominentnear-inertial peaks at all current meters of the three offshore moorings (Skunk, Leopard,and Puma),similar to thoseobservedduring the first time period,exceptthat the energywasgenerallylower and the bandwidth was generallygreater (Table 1). Over the continentalshelf,the inertial energywas much smallerduring that the time seriesdata have a Gaussian distribution; if one insteadthat the data consistsof a coherentsignalplus the secondtime periodthan duringthe first(Table 1, Figure5). assumes The spectrumof the top current meter at Elephant (Figure 5) incoherentnoise,the phaseestimateswould be exact [Schott hastwo near-inertialpeaks,centeredat 0.0625cph (very nearf) and Diiing, 1976].The phasedifferencesbetweencurrentmeter and 0.0508 cph (about 14% belowf), respectively.A similar pairscan be usedto determinethe directionof phasepropagasplit was observedat Puma 64, near the shelfedge(Table 1). tion and the wavelengthof the motion, if the separationbeElephant 49 m has a relatively broad, flat peak centeredat tween current meters(horizontally or vertically) is small com0.0547 cph, about 7% belowf All other currentmetershave a pared with the wavelengthof the near-inertial motions. For singlepeakfrequencythat is very nearlyequaltofor somewhat example,Kundu [1976] useddata from 11 current meterssepgreaterthanf (Table 1).The downwardfrequencyshiftsmay be aratedverticallyby 4-20 m to estimateupwardphasepropagaa result of Doppler shiftingof the near-inertialmotion in the tionof 0.1-0.4cms- • andverticalwavelengths of 55-225m. In are diagonal, presenceof a mean flow [Thomsonand Huggett, 1981]' There our case,mostof the currentmeterseparations was strongmean alongshoreflow, about 20-30 cm/s, at Ele- making it difficult to estimateeither the vertical or horizontal phant 23 m, Elephant 49 m, and Puma 64 during the second propagationspeeds.To add to the difficulty,most of the sepa- 5966 ANDERSONET AL.' NEAR-INERTIAL MOTION OFF OREGON S 126 ø W 1•5oL 0 i ioo 200 300 5OO I000 _ ß FIRST . PERIOD -- _ (13 Oct-5 Dec) _ -- 1500 _ _ - ß DISTANCE FROM SHORE (krn) i 150 I ioo 2000 50 ,•5oL 126' W _ I 0 P E 124*o ioo 200 300 5OO IOOO ß.. SECONDPERIOD ß ß (20 Nov-12 Jan) ß ß ß 1500 ß . 150 DISTANCE I00 FROM SHORE (km) I 50 I 0 200O Fig. 7. Distribution ofcurrent meterpairscoherent in theclockwise rotatinginertialfrequency bandof0.0508-0.0664 cph. Pairsconnected bysolidlineshavea squared coherence ofat least0.58,the97%significance level. rations(both verticaland horizontal)arc largerthan the expectedverticaland horizontalwavelengths. Nevertheless, wc used the phase spectrato estimatethe horizontal east-west (cross-shelf) wavelength by makingthefollowingassumptions: dcnccinterval of 36-56 km. Similarly,the phasedifferenceof 380ø(20ø + 360ø)betweenPuma and Elephant(19.7km apart) yieldsa wavelengthof 19 km (the 80% confidenceinterval is 18-20 km). By addingintegral numbersof 360ø to the phase (1)phasepropagation wasfromeastto west(i.e.,energypropa- differencesbetweenthe other pairs, wc were able to obtain gatedonshore); (2) the verticalseparation (maximumof 71 m) consistentestimatesof the cross-shelf wavelengthfor eachtime between Elephant89m, Puma64 m, Ocelot141m, Leopard70 period(Table 2). The averagewavelengthobtainedwas 50 km m, and Skunk100m wasinsignificant whencomparedwith the for the first periodand 20 km for the secondperiod.The 50-km horizontalseparation; (3)duringthefirstperiod,thehorizontal wavelengthfor the first period is consistentwith the large wavelength waslargerthanthesmallest horizontalseparation; coherencescaleobservedthen: Only three wavelengthsspan and (4) duringthe secondperiod,the horizontalwavelength the entire array. During the secondperiod,both the estimated was sligtitlysmallerthan the smallestseparation.With these wavelength and the coherence scale arc shorter. In both assumptions, the phasedifference of 115ø betweenLeopard70 periods,there is significantcoherencebetweencurrent meters m and Ocelot 141 m (13.9 km apart)duringthe first period separatedby two horizontalwavelengths or less. yieldsan cast-westwavelengthof 44 km with an 80% confi: Wc also consideredestimatingthe verticalwavelength,but ANDERSON ET AL..' NEAR-INERTIAL MOTION OFF OREGON 1.0 than 0.048 cph and greater than 0.077 cph, tapering linearly within the edges(0.002 cph wide) of the filter. The band passed by the filter lies entirely betweenthe semidiurnaland the diurnal tidal frequencies.For computationalefficiency,we filtered only 2048 hours of the long records,1408 hours of the Ocelot 0.8 F•rst Period '-' 0.6•).• 0 "' 0 records, and 1280 hours of the Puma records. In each case, there were at least 32 Fourier componentswithin the band o (n 5967 0.4 passedby the filter. The possibilityof peak smearingdue to the sharp filtering was testedby usinga synthetic1024-hour serieswith 5, 10, or o 15 cyclesof motion at 0.0625 cph, with the remainder of the seriesset to zero. Like Kundu [1976], we found that peak 0 25 50 75 I00 125 km smearingis lesspronouncedwhen the number of cycleswas Ax greater:When the input seriescontainedmore than 10 cycles, Fig. 8. The squared coherenceat 0.0625 cph between pairs of the amplitudeof the band-passed output did not vary greatly current meters with roughly horizontal separation: Skunk 100 m, from the input series.The phaseof the near-inertialmotion was I•opard 70 m, Ocelot 141 m, Puma 64 m, and Elephant 89 m. The 90%, 95%, and 97% significancelevelsare 0.44, 0.53, and 0.58, respec- not alteredby the band-passfilter, regardlessof the number of tively. cycles. The time seriesof the eastward component of the bandpassedcurrent at all current meters(Figure 9) showsthere is there was significantcoherencebetweenonly a few vertically somenear-inertialmotion throughoutmost of eachrecord.The separatedpairs.The spacingof the currentmeterswastoo large energyseemsto occur predominantlyin burstsor 'events'of to resolve vertical wavelengthsof the magnitude (50-500 m) generallyobservedin the ocean.Although we could have ob- severaldays duration. Someof theseeventsappear to begin tained estimates by assuming particular values of integral impulsivelyand decaygradually,but othersincreasegradually wavelengthsbetween current meters, the choice would have and persistfor more than a week. The maximum value (39 been completelyarbitrary, and, hence,the resulting estimates cm/s) of the band-passeddata occurredat Leopard 70 at the end of October.At eachcurrentmeter,the band-passed northwould be meaningless. ward currentis verysimilarto the eastwardcomponent,except that it is 90øout of phase(e.g.,Figure 10). BAND-PASSED DATA The time seriesof the band-passeddata (Figure 9) demonTo examine the time variability of the near-inertial motion, strateexplicitlysomeof the conclusions drawn from the specwe followedKundu's[1976] exampleand usedthe techniqueof tral analysis:The strongestnear-inertial currentsoccur at the band-passingrather than complex demodulation.(While the offshoremoorings,and at eachlocationthe amplitudegenerlatter has the advantage of using both componentsof the ally decreases with depth.During the firstperiod,October13 to original vector time series,the former has the advantage of December5, the near-inertialmotion is very persistentat the showingthe currentvariability more clearly.In a caselike ours, offshoremooringsand lesspersistentover the continentalshelf. where the motion in the near-inertialfrequencyband is essen- At mostcurrentmeters,the largestamplitudesoccurduringthe tially circularand clockwise,the two techniquesgiveequivalent first period rather than the second.During the secondtime results.)The current records were band passedby using a period,November20 to January12, there is very little nearnarrow filter centeredat 0.0625 cph, the most common fre- inertial energyover the continentalshelf. quency of the near-inertial peak in the autospectra.The proDuring each period of high near-inertialenergy,the phase cedurewas to determinethe Fourier componentsof each cur- differencebetweenthe band-passedcurrent vectorsat Skunk rent record by using a Sande-Tukey fast Fourier transform, 100m and Leopard70 m, and betweenLeopard70, and Ocelot pass100% of the energyin the frequencyband between0.050 141m, remainedfairly constant(Figure 11).During the strong and 0.075 cph, and suppressall of the energyat frequenciesless near-inertial event of October 27 to November 13, Skunk 100 z rr02 od A i TABLE 2. Horizontal East-West(Cross-Shell)WavelengthsCalculatedFrom PhaseDifferencesat 0.0625cph, for Current Meter Pairs With Nearly Horizontal Separations. SecondPeriod (Nov. 20 to Jan. 12) First Period (Oct. 13 to Dec. 5) Current Separation Meter Pairs Distance, km S 100-L 70 S 100-O 141 52.3 66.4 Coherence Squared Phase Difference N 0.64 0.57 331ø q- 21ø 78ø q- 25 ø 0 1 S 100-P 64 95.6 S 100-E 89 115.3 0.73 91ø+__17ø 2 13.9 0.57 115ø _+25 ø 0 L 70-0 141 Wavelength, km m J Coherence Squared 57 55 0.30 -- 0.10 50 0.03 Phase Difference N Wavelength, km 163ø q- 49 ø 2 21 21 ø q- 31ø 2 21 20 ø q- 25ø 1 19 44 L 70-P 64 43.1 m L 70-E 89 62.8 0.63 128ø q- 22ø 1 46 0.48 0.16 O 141-E 89 P 64-E 89 48.9 19.7 0.68 -- 25 ø + 20 ø 1 46 -- 0.57 The 80% confidence intervalis shownfor phase'N is the assumed numberof integralwavelengths betweeneachpair. 5968 ANDERSON ET AL.' NEAR-INERTIAL MOTION OFF OREGON •0•cm/sec S -Ii. ., I00 S 600 ........ AAAAAAAA .... A..,.AAA ...... A .... AA,,A,,_,,AAAAAA,,,,,,AA ..... A,• ............ •v........... ,•AAAAAAAAAAAAAA,,,.,• o•.... ,AAA .vv ............ -::-'• -,,..wvvvv•,.-v,,vvv-,,,,-v-v-.-vvvvv--vvvvvvv,,.v., ..... v.... v....... ......... ,,-vvwv•vvvul/VVV,,VVV-VVVVVvv ................... S 1850 O• ....'............. ::...... AAAAA ...... ,A#,,_.,,,A.• ...................................... AAA. AAAAAAAA ..... A_AAAAAAAA ........... •M:•_•v•vA• ' ............... vvvvvvv'-:-'•%%-:'•%%':--: --v v--v v--vvv ....... ,,v,,---,,............ ' -v,,•vvvvvvvvvvvvv,,.,,-vvvvvvvvvv L 7O L 820 0 141 o 245 0 571 -•o;VVVVVVVVVVVVV"Vo"VV"'v"' ...... ',vvvvvvVVV" ....... '................................................ o•..:#::-: •--....-..AA A A,•,,,,AA AAAAAAAA.AAAA,,.AAAAAAAAAAAA .,•,,A,,,,AAAAAAAAAAAAA,,,..AAAAAAA vvvv•'"vvvvllVvl/gg"vgvv"vvvvvvvvl/vv•'""'vvvvgVVVV. vv v""'WVV1/VVVVVVVV': .................................................. I0 20 30 I I0 October 1977 20 I November I0 20 December 30 I I0 January 1978 Fig. 9a P 64 ............................ .... -............... P 164 0[...................................... AA AAAAA,,AAA. AA .............. AAA,, ^AAAAAAAAAAA ,•A.,A AA ......... ......... AAAAA .... ..... vvvvvvv...,.vv ..............wv.vvvgl/vvvgg v.... vvv---v ....AA vv........ 'vv................... P 261 0•.............................................. E 23 E 49 q½-:'•';:--:-:::#• •-::•::;: ........ vvvv .......v,................. vVg v...... ..... ^AA ............................. AAAA ....... .,rAy .,,.,AAA...AAAhAAAAAAAI •.,•..,,.AAhA.AA,,AAA,,,,.,,..,,AAAA,,•,AAA ......A,,^.. .......AAA,__AAA ....,.v,,, ....,v, .................... ,...vvv.vvv,ViV,V g E 89 ,o •o October 1977 •o, ,o •o , November K• •0 December •0, ;0 January 19• Fig. 9b Fig. 9. Theband-passed timeseriesof theeastward component of thecurrentat thecurrentmetersat (a) Skunk,Leopard, and Ocelot,and (b) Puma and Elephant. ANDERSON ET AL.' NEAR-INERTIAL MOTIONOFF OREOON 50T 5969 Leopard 70m BandPassed I • •l•J•, ]',,•,•. Fig. 10. A 10 day (Oct. 27 to Nov. 6) segmentof the band-passed eastward(u) and northward(p)componentsof current at Leopard70 m. The approximatelyequalamplitudeof the componentsand the 90ø lag of u behindp indicateclockwise circular motion. m lagsbehindLeopard70 m by about 330ø,and Leopard 70 m lagsbehind Ocelot 141 m by about 120ø;thesenumbersagree very well with the phasespectrafor the first period (Table 2). Thus, the horizontal wavelengthestimateof 50 km seemsto be valid throughout the 20-day duration of strong near-inertial motion beginningabout October 27. Several current meters showed high near-inertial energy during December 13-28 (Figure 9). At Leopard 70 m, there seemto be two separatenear-inertialeventsthat begin about December 13 and December 24; the near-inertial energy at 0 Skunk 100 m seemsto be marc nearly continuous.The phase differencebetween Skunk 100 m and Leopard 70 m varies considerablyduring this period: It is about 180ø on December 14-16, about 75ø on December 18-22, and about 270ø on December24-28; only the first of theseagreeswell with phase spectrafor the secondperiod (Table 2). Thus, the near-inertial phasedifferencebetweena pair of current metersseemsto bc well defined and fairly constant during a particular nearinertial event but varies considerablyfrom one event to another. To the extent that differencesin phase reflect different 2ø t 141 -2 $60 240 L 70-0 141 120 : : : : : ................................................ L 7O -40 $60 -• 240 S I00-L 70 *'• 120 I0 20 October •) I I0 20 November •) I I0 20 December January 2o S I00 Fig. 11. Thehourlydifference in direction between theLeopard 70m andtheSkunk100m or Ocelot141band-passed current vectors. I0 5970 ANDERSON ET AL.' NEAR-INERTIAL MOTIONOFF OREGON i2 dynes/cm W•nd Stress Magnitude Nov Dec Jan Elephant 23m .......... ...... JAAAAAfiAAAfiAA .... nAA.AAAhAIA.AAA, ...... ........ nAAA.. ........................ ,lb,.. Leopard 70rn Fig. 12. Themagnitude of theNewport windstress, andthemagnitude andcastward component of thePollardand Millardmodel, andcastward components oftheband-passed current atElephant 23m(inthemixed layer), andLeopard 70 m (about30m belowthemixedlayer). wavelengths, thesevariationsimply that the wavelengthof near-inertialmotionsvariesconsiderably betweendifferenten- mooringsin late OctoberandearlyNovember.Sincethemodel ergeticepisodes. is very sensitive to small-scalevariations in the wind field WIND FORCING Thegeneration ofnear-inertial motionin themixedlayerby energetic near-inertial motion observed at the three offshore [Pollard,1980],we mightexpectpoor agreement betweenobservedcurrentsand modelresultswhenever thereare strong atmospheric frontsin theregion. Frontalpassages wereverycommonoff Oregonduringlate October and early November (Figure13):betweenOctober23 successfully by PollardandMillard [1970].Althoughmostof with low pressure cenour currentobservations werewellbelowthemixedlayerand andNovember15, 17frontsassociated the nearestwind observations weremadeon the Newport tersof varyingsizeandintensitypassedeastwardovertheSUS southjetty 85 km southof the array,we usedthe Pollardand array. Many of thesefrontal passageswere recordedin the Millard modelto obtaina qualitativeestimateof the currents Newport pressuredata and causedclockwiserotation of the that might havebeengeneratedby the observedwinds.The windat Newport(Figure13).Sincetheextentof theSUSarray was115km,thearrivaltimeof eachfrontwouldnot generally modelequationsare be the sametime at all moorings.Whereverthe rotationof the u,--fv = •:"/(pZ)- cu (1) near-inertialmotioncausedby eachfrontalpassagewas in v, + fu = •:'/(pZ)- cv (2) phasewith the pre-existingnear-inertialmotion,therewouldbe constructive interference [Pollard,1970] whichwouldincrease wheref is thelocalinertialfrequency, u andv areeastwardand the near-inertialenergy.To look for phaseshiftsin the obnorthward components of thecurrent,c is thedamping coef- servednear-inertialmotion,wecalculated the phasedifference ficient,p isthewaterdensity,x0:",•:Y) is thewindstress, andZ is betweenthe band-passed currentvectorsand a referencevector the mixedlayerdepth.The windstress, x = Cap,,JVJV,was rotatingclockwise at 0.0625cph:vectorswith no phaseshifts computed fromtheNewporthourlyvector,V, thedensityof will have a constantphasedifferenceif they have the same air,p• = 1.25x 10-3gm/cm 3,anda dragcoefficient, Ca= 1.5 frequency as the reference vectoror a linearlyincreasing and local wind forcingat siteD in the North Atlanticwasmodeled x 10-3.Weassumed a constant mixed layerdepthof25m.We decreasing phasedifference if theirfrequency is slightlydifferuseda simplefinitedifference formwith a 4-s timestepto ent.Timeseriesof thisphasedifference for Leopard70 m and computehourly valuesof the modelcurrent. Elephant 23 m (Figure13)showthatno shiftsin thephaseof ThePollardandMillardmodelshowsa strongnear-inertial thenear-inertial motionoccurred at Leopard70m duringlate response to thewindstress peaksof December 14andJanuary October/early November,whileseveral phaseshiftsoccurred at 4 (Figure12),and a weakerresponse to someof the weaker Elephant 23duringthesameperiod.Thissuggests thatmostof storms between November 7 and December 10. Near-inertial the wind-generated near-inertial motionat Leopardin late motionwasobserved morefrequently at Leopard70m thanin October/early Novemberwasnearlyin phase,causingthe thePollardandMillardmodelresults (Figure12).Near-inertial strong persistent near-inertial currents there. On the other motion occurredfrequentlyat Elephant23 m from mid- hand,someof thenear-inertial motiongenerated at Elephant Octoberto early December, but not immediately after the may havebeenout of phasewith pre-existing near-inertial strongstormsof December 14andJanuary4, althoughbothof thesestormsseemedto generatenear-inertialmotionsat Leo- pard70 m. Themodelfailedto account for thepersistent and motions,causingdestructiveinterferenceon about October 27 and again on November 4 and 6. Thus, the observations stronglysuggestthat the strongand persistentnear-inertial ANDERSON ET AL.' NEAR-INI•TIAL MOTIO8 OFF OREC,O• 5971 I 45' N /000rob •' 40 ø 1020 I"W I12øNøo•v •...!.I 29 Oct 125 ø 130øW 125" i:•o"w 125, i:•o•w 1:•5o IODec / I oooo / 130øW / •".' I / 125' 10401 I010 NEWPORT 980 Barometric Pressure mb Oct. Nov. NI •WPORT Dec. Low-Poss Jon. Wind Direction t........ /L/ 360 ø' lD ::::::::::::::::::::::::: LEOPARD : ........................... 70rn Phose Difference 40 ELEPHANT 23m 360 ø 18 i........................................ ::............................ ELEPHANT23m PhoseDifference Dec, Jori Fig. !3. Examplesof the synopticatmospheric pressur e chartsfrom the period of the near-inertial'event'of late October/earlyNovemberandtimeseriesof atmospheric pressure and the directionof the low-passed (< 0.025cph)windat Newport,withtheeastward components andthephasedifferences of theband-passed currentvectors at Leopard70m and Elephant 23m relativeto a reference vectorrotatingclockwise at 0.0625cph. 5972 ANDERSON ET AL.' NEAR-INERTIAL MOTION OFF OREGON motionsobservedat the offshoremooringsin late October and early November were generatedby the local wind. Without directwind observations over the moorings,this tentativeconclusion cannot be verified. CONCLUSIONS Near-inertial motions were observed at all current meters in an array of five moorings which spanned the continental margin off Oregon betweenOctober 1977 and January 1978. The currentmeterswerebetween23 and 1850m deep,in water depths between 100 and 2500 m, at locations between 10 and 130 km from shore.Although almost all of the currentmeters werebelowthe mixedlayer,largeamplitude(~ 20 cm/s)nearinertial motionswereobservedon severaloccasions. The largest amplitude(> 30 cm/s)near-inertialmotionswere observed at the shallowestcurrentmetersof the offshoremoorings;these currentmeterswerein the pycnoclineat depthsbetween70 and 141m. Althoughtherewerecurrentmetersat shallowerdepths (23 and 49 m) within the surfacemixed layer over the continentalshelf,therewasmuchlessnear-inertialenergythere.The amount of near-inertial energy observedover the shelf was similar to that observedthere in the summerof 1972, and the amount of energyin the deep water was comparableto that observed in the North Atlantic. Most of the near-inertialmotion had a frequencyslightly greater than the local inertial frequency,but someof it had a frequencylessthan the local inertial frequency.The downward frequencyshiftoccurredin the presenceof a strongnorthward current. The near-inertialmotion at somecurrent meterswas very persistent(in one caseit persistedfor more than 15 days).In this case,the amplitudeseemedto increasegraduallybeforeit decayed.In other cases,the amplitude increasedimpulsively with gradualdecay,asusuallyobservedelsewhere. Spectralanalysisshowedthat the horizontal coherencescale exceededthe extent of the array (115 km) in October and Novemberand was about half the extent of ,thearray (i.e.,60 km) in Decemberand early January.Thesecoherencescalesare much larger than thosepreviouslyobservedover the Oregon shelfin summer.Estimatesof the horizontalwavelengthof the motion changedfrom 50 km during the first half of the records to about20 km duringthe second.Theseestimatessuggest that thereis significantcoherence betweencurrentmetersseparated by two wavelengthsor less. The near-inertialphasedifferencebetweena pair of current meters varied considerably between separate near-inertial events,even though the stratification remained much the same throughoutthe period of observation.This suggests that the wavelengthis the resultof the particularcircumstances generating the motion, rather than of the oceanicenvironment. Because of the lack of current measurements Pollard and Millard modelof local forcing,usingthe Newport wind observations(85 km south of the array) as the forcing function, indicates that some of the near-inertial motion ob- servedbelow the mixed layer (e.g., at Leopard 70 m) was generatedimpulsivelyby strongpeaksin the wind stress(e.g., December14 and January4). The strongand persistentperiod of near-inertialmotion in late October and early November may have beenthe resultof a sequenceof fronts that happened to pass over the region in the proper phase for constructive interference.Thus, our data are consistentwith the hypothesis that the local wind can force strong near-inertial motions at depthswell belowthe mixedlayer. Acknowledgments.We wish to thank Barbara Hickey of the Universityof Washingtonfor providingthe data at Ocelot and Elephant. We also thank Rick Romea and Jim Richman for their comments on earlier drafts and JosephBottero and William Gilbert for assistance in the analysis.This study was completedwhile the first author was studyingfor a degreeat Oregon State University.Financialsupport was providedby the United StatesCoast Guard and by the National ScienceFoundation throughgrantsOCE-7925019 and OCE-8026131. REFERENCES Fu, L., Observationsand modelsof inertial wavesin the deep ocean, Rev.Geophys.SpacePhys.,19, 141-170, 1981. Gonella,J., A local studyof inertial oscillationsin the upperlayersof the ocean,DeepSea Res.,18, 775-788, 1971. Johnson,W., Cyclesondemeasurementsin the upwelling region off Oregon, Tech. Rep. 76-1, RosentialSchoolof Mar. and Atmos.Sci., Univ. of Miami Florida, 1976. Kindle, J. C., The horizontal coherence of inertial oscillations in a coastalregion,Geophys. Res.Lett., 1, 127-130, 1974. Koopmans, L. H., The SpectralAnalysisof Time Series,Academic, New York, 1974. Kroll, J., The propagationof wind-generatedinertial oscillationsfrom the surfaceinto the deepocean,J. Mar. Res.,33, 15-51, 1975. Kundu, P. K., An analysisof inertial oscillationsobservednear the Oregoncoast,J. Phys.Oceanogr., 6, 879-898, 1976. Mooers,C. N. K., A techniquefor the crossspectrumanalysisof pairs of complex-valuedtime series,with emphasison propertiesof polarizedcomponentsand rotationalinvariants,DeepSeaRes.,20, 11291141, 1973. Munk, W., and N. Phillips, Coherenceand band structureof inertial motionin the sea,Rev.Geophys., 6, 447-471, 1968. Pollard, R. T., On the generationby winds of inertial wavesin the ocean,DeepSeaRes.,17,795-812,1970. Pollard,R. T., Propertiesof near-surface inertialoscillations, J. Phys. Oceanogr.,10, 385-398, 1980. Pollard, R. T., and R. C. Millard, Comparisonbetweenobservedand simulatedinertialoscillations, DeepSeaRes.,17, 815-821, 1970. Schott, F., and W. Diiing, Continental shelf waves in the Florida Straits,J. Phys.Oceanogr., 6, 451-460, 1976. Thompson,R. O. R. Y., Coherencesignificance levels,J. Atmos.Sci.,36, 2020-2021, 1979. Thomson,R. E., and W. E. Huggett,Wind driveninertial oscillations oflarge spatialcoherence,Atmos.Ocean,19, 291-306, 1981. Webster,F., Observationsof inertial-periodmotionsin the deepsea, Rev. Geophys.,6, 473-490, 1968. in the mixed layer and the absenceof direct wind measurementsover each mooring, we could not rigorouslytest the hypothesisof local generation of the near-inertial motions. However, the simple (ReceivedJuly 6, 1982; revisedJanuary31, 1983; acceptedMarch 21, 1983.)