1458 MONTHLY WEATHER REVIEW VOLUM- 110Response of Stratified Flow in the Lee of the Olympic Mountains~BERNARD A. WALTER, JR., AND JAMES E. OVERLANDPacific Marine Environmental Laborator~/NOA,4, Seattle, W,4 98105(Manuscript received 2 September 1981, in final form 24 June 1982) ABSTRACT The behavior of stratified air flowing around an isolated mountain is dependent on an internal Froudenumber (F), which indicates the relative importance of upstream velodty and vertical stratification. Threecases of the flow in the lee of the Olympic Mountains in the State of Washington are studied where themeasured F was in the range 1.0-1.4 but apparently dominated by stable stratification. This study combinedmeasurements of spatial variation of low-level winds and other parameters from a NOAA P-3 researchaircraft with a dense network of surface stations including eight meteorological buoys and six upper-airstations. Results from these cases show the presence of an area of light winds in the lee of the OlympicMounta/ns. The characteristics of the flow are shown to be similar to laboratory results for low Froudenumber flow around an isolated obstacle where the flow is confined to quasi-horizontal planes. These casesare contrasted with a situation which led to the formation ofa mesoscale low-pressure area and h/gh surfacewinds in the lee of the mountains. The latter case was the Hood Canal Bridge storm on 13 February 1979where local winds in the lee of the Olympic Mountains were in excess of 50 ms-'. The flow at the surfacewas produced by down-pressure-gradient acceleration in the confined channels of Puget Sound toward theorographically produced low-pressure center. The measured internal Froude number in this situation was4.6, and the pressure fields are shown to agree with the linear hydrostatic model developed by Smith (1980)for F > 1. It is suggested that the Froude number calculated from routine, upper-air sounding data is anindex that forecasters can use to determine the potential for severe wind conditions over the inland watersof Puget Sound,1. Introduction The problem of defining the flow field of a stratifiedfluid past ideal obstacles such as cylinders, cones, infinitely long ridges and hemispheres has occupiedefforts in theoretical and experimental fluid dynamicsfor many years. The extension of this work to geophysical scales, where the effects of complex obstaclessuch as mountains and elevated islands on atmospheric flow are considered, has implications for thesolution of a number of important meteorologicalproblems. Between the extremes of continental-scalemountain ranges influencing the general circulationof the atmosphere and small-scale hills on the orderof hundreds of meters modifying the dispersion ofpollutants, lie isolated orographic features on the order of 50-100 km in horizontal extent and 1-2 kmin height. These latter obstacles have dominant influence on local wind fields, precipitation patterns,and other parameters affecting local climate (Chopra,1973; Mass, 1981). A specific situation occurs in northwestern Washington State where, during southerly to southwesterlywind conditions, the Olympic Mountains (Figs. la Contribution No. 527 from the NOAA/ERL Pacific MarineEnvironmental Laboratory.and l b) act as an obstacle to the flow and producea mesoscale wake in the lee of the mountains overthe eastern Strait of Juan de Fuca. The sinking of theHood Canal Bridge on 13 February 1979 during surface winds in excess of 50 m s-I dramatically demonstrated the impact of such a phenomenon. Reed(1980) analyzed the meteorological conditions pre.sent at that time and found that the gradient of sealevel pressure at the bridge, although strong becauseof the synoptic pressure gradient, was greafiy enhanced by the formation of a mesoscale low-pressurearea in the lee of the Olympics. In the restricted channels of Puget Sound the airflow at the surface accelerates toward low pressure producing the observedhigh wind speeds (Overland and Walter, 1981). During the winter (December-February) synopticweather patterns cause S-SW flow conditions overthe Puget Sound region 75% of the time, accordingto a classification given by Maunder (1968), and thusthe opportunity for lee effects. Reed (1980) mentioned that "marine weather forecasters in westernWashington have long suspected that a mesoscalevortex tends.to form in the lee of the Olympics duringperiods of strong SW flow." It is thus important todefine the conditions that lead to the formation of anintense lee-side low as opposed to those that do not.From 10 February-lO March 1980 a mesoscaleOCTOBER 1982 BERNARD A. WALTER, JR., AND JAMES E. OVERLAND 1459i %~ :~i~~- BRITISH COLUMBIA .?125~W 124- 123~ 122~F~G. la. Location map of western Washington State and southwest British Columbia.wind study was carried out in the Puget Sound regionof the State of Washington using a NOAA WP-3Dresearch aircraft to measure low-level winds and turbulent fluxes. During the experiment, data were obtained for, three cases (18 and 28 February and 9March) during S-SW wind flow conditions where lowwind speeds were observed in the lee of the OlympicMountains over the Strait of Juan de Fuca. For allthree cases the boundary layer was nearly saturatedand slightly stable to about 650 mb. Above this levelthe stability increased. In studies of stratified flow Over obstacles the valueof the internal Froude number F is used to indicatewhether the flow will, in general, go around or overthe obstacle. F is an indicator of the relative importance of upstream velocity and vertical stratificationand is defined as F = V/hN, where V is upstream flowvelocity, h the vertical length scale taken as the heightof the mountain and N the Brunt-V'fiis/il~/frequency.Durran and Klemp (1981) indicate that the BruntV~/is//1/i frequency Ne applicable to a saturated regioncan be written N] = g 'Ym a(lnOe) , 3~a dzwhere g is gravity, -~ the most adiabatic lapse rate,'~a the dry adiabatic lapse rate, 0e equivalent potentialtemperature, and z the vertical coordinate. As will beseen later the value of F for the Hood Canal Bridgecase was significantly larger than values calculated forthe Puget Sound Wind Study cases due to both lowerstability (giving a smaller Ne) and higher upstreamwinds. The data obtained from these cases will becompared to theoretical and experimental results forboth stable and nearly neutral conditions with theemphasis on contrasting the Hood Canal case withother winter storms.2. Previous studies on stratified flow over isolated mountains Lilly (1973), using the Boussinesq approximation,derived a set of nondimensional momentum, continuity and conservation of buoyancy equations forsteady-state, incompressible inviscid flow in a formwhich illustrated the importance of various processesfor stratified flow interacting with topography. Thehorizontal coordinates x and y were scaled by L, acharacteristic horizontal mountain dimension; thevertical coordinate z by h, the mountain height; thehorizontal velocities u and v by V, a characteristicvelocity of the oncoming flow; the vertical velocityw by Vh/L; the perturbation pressure p -/5(z), by thehorizontal kinetic energy Oo V2; and the perturbationdensity p - b(z), by ~ooV2/gh, where po is a referencedensity. The resulting equations are [I/2(gt2-t" 112) q-/7] -- V rq-~O0 + W~ = O, (1)0-300 meters300-1500 metersover 1,500 metersFIG. lb. Topographic map of the region.1460 ,MONTHLY WEATHER REVIEW VOLUME 110 ..... Numerical Results ~.~,~ Experimental Results .:~,~q~'""~.-'J'-~ ~.~~""'""~'~'~X~(cm)~~ ...". ;.r' ~ BG. 2. Comparison of experimental and numerically computed (Drazin's solution) streak-lines in the hodzont~ plane at mid-level of a Gaussian-sha~d obstacle for a Froude number of 0.154 (from Riley et al.,1976). 1 )_o [,/2(u2 + v2) + p] + u + + w o,(2)oy (3) Ou Ov Ow ~xx+~y+~z=O, (4) F2 .~+v~y W~zz -w=0, (5)where ~ = (Ov/Ox - Ou/Oy), the relative vorticity. The scaling introduces three dimensionless parameters: the mountain aspect ratio h/L, the internalFroude number F = V/hN, and the R0ssby numberRo = V/f L, where f is the Coriolis parameter. Asmall aspect ratio indicates that the hydrostatic approximation is valid. Very stable conditions (F ~ 1)imply that the flow is essentially horizontal (w ~ 0),and a large Rossby number implies that the Coriolisforce can be neglected. For flow over isolated obstacles, the situations involving very stable flow (small internal Froude number) and slightly stable flow (large internal Froudenumber) have received the most consideration because of simplifying assumptions that can be made.The theory for very stable flow (F ~ 1) was developedby Drazin (1961) and has been studied in laboratoryexperiments by Brighton (1978) and Riley et al.(1976). Drazin (1961) treated the flow of a nonrotating, incompressible fluid around a three-dimensional obstacle where the upstream flow was paralleland horizontal and velocity and density were functions of height alone. He expressed the vertical displacement of the streamlines from their upstreamposition and the velocity of the flow in terms of aseries expansion in F~. The primary flow occurs inquasi-horizontal planes where the flow in each planeis given by a two-dimensional potential solutionabout an obstacle defined by the contour of the terrainat that level. Small vertical displacements ~ (where~ ~ h) of the streamlines passing around the side ofthe obstacle are a result of the decrease in the potentialenergy necessary to balance the increase in kineticenergy of the flow around the cylinder. Drazin's(1961) solutions for the secondary flow are not validin a region of thickness hF near the top of the obstacle. Laboratory studies by Riley et al. (1976) confirmedthe validity of Drazin's (1961) flow equations for verylow Froude numbers except for the region of thewake, where the flow separates. The results (Fig. 2)show that the numerical calculations from Drazin'sequations tend to underpredict streak-line displacement. Riley et al. (1976) attribute this to two effects:1) the displacement effect of the boundary layer whichis not taken into account in the inviscid numericalmodel, and 2) the displacement effect of the separatedwake. Brighton (1978) extended Drazin's.theory toinclude rotation and presented the results of laboratory studies for very stable flow (F = 0.03-0.25).Brighton (1978) indicated that rotation plays only aminor role because it is important only in secondarysolutions; therefore, results from laboratory studiesmay reasonably represent the behavior of the atmosphere. Brighton's tank experiments showed the presence of a "cowhorn eddy" in the lee near the top ofthe obstacle with a separated wake, region below theeddy where the flow had only very small vertical displacements [Fig. 3a(i)]. There was a strong velocityshear between the level of the eddy and the stagnantregion below it. Hunt and Snyder (1980) investigated the behaviorof flow over an isolated obstacle in a laboratory studythat considered a wide range of stability conditions(F = 0.1-1.7, ~). Their focus was on studying theOCTOBER 1982 BERNARD A. WALTER, JR., AND JAMES E.' OVERLAND 1461characteristics of flow separation for different Froudenumbers and also on defining criteria for determiningwhether a plume emitted upstream at a certain heightwill go around the sides of a hill or over the top [Figs.3a(ii) and 3a(iii)]. Their results confirmed the applicability of Drazin's (1961) theory for F ~< 0.4. Outsidethis range of F the location of the point of separationmoved down the lee slope and a hydraulic jumpmoved progressively downstream with increasingFroude number. For values of 1.0 and 1.7 separationoccurred near the top of the hill with recirculatingregions present over the lee slope. Care must be taken,however, in comparing separation features observedfrom the laboratory study, where the obstacle had anaspect ratio of 0.2, with those of geophysical scalessuch as the Olympic Mountains, where the aspectratio is 0.04. Another consideration as stated by Huntand Snyder (1980) is that the Reynolds numbers foratmospheric flows may be several orders of magnitude larger than in the tank studies. For the case where F > 1, Smith (1980) developeda hydrostatic linear model for stratified flow over anisolated mountain. A critical assumption was settingthe vertical velocity at the mountain equal to theupstream velocity times the slope of the mountain,which was a result of boundary condition linearization. This provided an explicit boundary conditionfor vertically integrating the continuity equation (4).Smith's results for a bell-shaped mountain showed anasymmetric perturbation surface pressure field (Fig.3b) with high pressure over the windward slope andlow pressure on the lee side. This pressure distributioncaused the surface streamlines to be deflected aroundthe sides of the mountain and to maintain a permanent outward deflection downstream of the mountain. This horizontal divergence in the lee of themountain was partially responsible for the sinking ofwarm air from aloft and a decrease in the verticaldistances between downwind low-level isentropic surfaces. Upper-level tlow was composed of verticallypropagating mountain waves with maximum amplitude over the mountain and with considerable energytravelling downstream along parabolically shapedwakes. On geophysical scales, aircraft measurements havebeen made by Marwitz et al. (1969) of the flowaround Elk Mountain, a relatively isolated mountainin Wyoming with an elevation of 3300 m MSL and'a surrounding plain at 2100 m MSL. The internalFroude number for this case was 1.5. The aircraftmeasurements (Fig.~3c) showed a depression ofstreamlines and the possibility of a hydraulic jumpin the lee of the mountain. Aircraft measurementsmade by J. Connell2 around Elk Mountain and wind 2 Connell, J. R., personal communication, Dept. of AtmosphericSciences, Battelle Pacific Northwest Laboratories, Richland, WA99352.(a)(b)4000-~ Elk Mt, (C) 18 March 1969 I J FIG. 3a. (i) Three-dimensional sketch of the most prominentfeatures of strongly stratified flow past an isolated hill. In regionI, the fluid approaches the obstacle in the nearly horizontal potential flow described by inviscid flow theory. The flow separates fromthe sides of the obstacle and in the wake (region IV) vortices (V)may be shed. Near the summit (region III) the fluid passes overthe top. Downstream lee waves are formed and also sometimes acowhorn eddy (CE). In region II, well above the top of the obstacle,the fluid is little affected by its presence (from Brighton, 1978).(ii) Plumes from upwind stacks at various elevations, for F = 0.2.How is constrained to move principally in horizontal planes. Aslight hydraulic jump occurs just downstream from the top of thehill (Hunt and Snyder, 1980). (iii) As in 3a(ii) but for F = 0.4. Theflow is able to move farther in the vertical direction and the regionwhere the flow goes over the top is broader. Flow going over thetop separates about half-way down the lee slope (Hunt and Snyder,1980). FIG. 3b. Solid lines show the topographic contours of a bellshaped mountain. Dashed lines show the isobars of perturbationsurface pressure p' for hydrostatic flow from Smith's (1980) linearmodel. The pressure coefficient is defined as % -= P'/Po VNh. FIG. 3c. Streamline distributions over Elk Mountain, Wyoming,estimated from aircraft observations. Height is in meters. (Marwitzet aL, 1969.)1462 MONTHLY WEATHER REVIEW VOLUME II0tunnel studies by Kitabayashi et al. (1971) alsoshowed the presence of hydraulic jumps downwindof the peak. The Coriolis force plays a role in the dynamics ofthe stratified flow for obstacles with horizontal scaleslarger than hundreds of kilometers. Results of Merkine and Kalnay-Rivas (1976) showed that when thehorizontal momentum in the horizontal momentumequation was approximated by the geostrophic momentum, a topographically bound anticyclonic vortex was present over an obstacle which increased thevelocity field on the left side of the mountain in theNorthern Hemisphere, facing downstream, and decreased it on the right side. The results from Drazin's (1961) small Froudenumber theory and Smith's (1980) large Froude number theory provide a framework for discussing thePuget Sound Wind Study cases and the Hood CanalBridge case.3. Case Studiesa. Data sources Fig. 4 shows the locations of surface data sourcesin the Puget Sound Wind Study. Eight meteorologicalbuoys were deployed to provide direct measurementsof overwater winds and air temperature. Groundbased data were collected from United States andCanadian Coast Guard, Weather Service stations andFAA facilities; measurements included wind, temperature, dewpoint and pressure. All U.S. surfacepressure stationswere compared with a standard pressure sensor and found to be within +0.3 mb of thestandard. Additional upper-air sounding stationswere established to supplement the two routinelyavailable observation sites. Soundings were takendaily at 0000 GMT, and at 1800, 2100 and 0000GMT on days with aircraft flights. A NOAA WP-3Daircraft was based in Seattle and flew six missions toobtain measurements of various parameters at aheight of 50 m above the water surface. Aircraft dataincluded wind, temperature, dewpoint temperatureand vertical wind speed at flight level, sea level pressure, sea surface temperature, and vertical profiles oftemperature, moisture and wind from omega dropwindsondes. A typical tlight pattern is shown in Fig.5; deviations from this pattern occurred on any specific flight because of limitations of visibility or restricted areas. The total time to fly these tracks wason the order of 3 h. Flight-level winds were computedby combining inertial navigation system (INS) outputs with measurements of air speed, angle of attackand angle of side slip. After correcting the data fororientation of the aircraft with respect to the winddirection .and drift of the INS, the absolute accuracyof each wind component i~ better than +1.0 m s-~(Merceret and Davis, 1981).126- 125- 124- 123- 122-49--49-48-47-46ONLEGEN~DCANADIANU.S. COAST GUARDNATIONAL WEPMEt~ BUOYFAAUPPER AIR-48-El I~1 .47o-46o 2G-W 12Go 124o 123- 1~2oFIG. 4. Surface and upper air data sources and locations.b. Parameters characterizing the fiow regimes Values of the Brunt-V~iisfilfi frequency Ne in Table1 for cases from the Puget Sound wind study werecalculated using sounding data where it was felt thatthe data quality was good and that land influenceson the profiles were a minimum (i.e., the special station at Ocean Shores (Fig. 11), or overwater dropsondes from the P-3). The value of dOe/dz was computed as an average for the 950-700 mb layer: 1 dOe I [0e (700 rob) - 0e (950 mb)l Oe dz Oe Azwhere be was the average equivalent potential temperature for the layer and Az the 950-700 mb layerthickness. A table of values of %, as a function ofpressure and temperature was given by Hess (1959)and is used in calculating the values of Ne. Minimumand maximum values of 3% were computed for the950-700 mb layer, and the mean of these values wasused in computing Ne2. Likewise a mean layer veloc,ity was determined from the sounding, and this valuewas used to compute the Froude number F = l?/hNefor each case. The value of h used for computing Fwas 1800 m, the mean height of the high peaks in theOlympic Mountains. For the 18 February case the P-3 dropwindsondeat 48-31'N, 124-57'W was used to computeThere was an inversion present at about 600 mb. The0000 GMT 19 February Quillayute sounding was /OCTOBER 1982 BERNARD A. WALTER, JR., AND JAMES E. OVERLAND 1463FIG. 5. Typical flight tracks for the NOAA P-3. Total flight distance was on the order of 1000 km.very similar to the dropsonde for the 950-700 mblayer. Mean winds for the layer were 12.5-15 m s-~from 205-225-. For the 28 February case, the P-3dropsonde at 48-32'N, 124-40'W was used.to compute Ne. The Ocean Shores sounding was very similarto the dropsonde for the 950-700 mb layer. Bothsoundings showed an inversion at 650 mb. Meanwinds were 15 m s-~ from ~210-. For the 9 Marchcase, the P-3 dropsonde at 48-2 I'N, 124 - 16'W wasused to compute Ne. Soundings at Quillayute andGray Field were both similar to the dropsonde profilewith all soundings showing an inversion at 730-750mb. Mean winds were 15 m s-l from 220- below theinversion and 17.5-20 m s-~ from 275--290- abovethe inversion. Fifteen meters per second was used inthe Froude number calculation. Data for two othercases are also shown in Table 1, 12 January 1968 and13 February 1979, the latter being the day of theHood Canal Bridge storm. For 12 January 1968, thesounding for 0000 GMT 13 January from Quillayutewas used. Mean winds were about 30 m s-I. For theHood Canal Bridge storm the sounding from Salem,Oregon (1200 GMT 13 February) was used in computing F. Mean winds here were 25 m s-l. Also shown in Table 1 are values of the Rossbynumber Ro for the different cases. The horizontallength scale L is taken to be 40 km, the approximatehalf-width of the mountains, and jr = 1. x 10-4 S-1.The Rossby numbers computed were of order 3.137.5 so that the effects of the Coriolis force were minimal and thus will be ignored.c. 18 February 1980 case At 0000 GMT 19 February 1980 a surface lowpressure system was centered just offshore north ofVancouver Island (Fig. 6a) giving southerly low-levelwinds over the Puget Sound region. At 500 mb atrough was oriented NW-SE over Washington andVancouver Island with winds at this level from theWSW (Fig. 6b). Fig. 7 shows the mesoscale sea-level pressure fieldat 2100 GMT 18 February, which is based upon sealevel pressures measured by land-based stations and1 min average values measured by the P-3. In orderto compare all of the pressures at a common time theaircraft-derived pressures were adjusted to 2100 GMTby correcting for the time change of the synoptic pressure field. In this case pressures were increasing at 0.4mb h-~ at the start of the flight period and 0.8 mbh-I at the end. From station intercomparisons weconsider the accuracy of the land stations to be betterthan 0.3 mb and the accuracy of the aircraft pressuresto be on the order of I mb. Fig. 7 also shows thetlight-level (50 m) winds measured,by the'P-3 aircraftand the mean surface wind reports at stations indicated by open circles during the period 2045-2315GMT 18 February. The blocking effect of the Olym TABLE 1. Parameters characterizing the flow regimes. Shown arevalues of N~, the Brunt V/iis~l~i frequency; V, the upstream windvelocity; F, the internal Froude number; and Ro, the Rossby number. Ne v v = v Ro = v Day (s-l) (ms-~) hNe fL18 Feb 80 0.007 12.5 0.992 3.1328 Feb 80 0.006 15 1.39 3.759 Mar 80 0.009 17.5 1.08 4.3812 Jan 68 0.012 30 1.39 7.5013 Feb 79 0.003 25 4.63 6.25146450-,140o 135o MONTHLY WEATHER130- 125- 120- 516 50~,REVIEW 140-135-130-VOLUME 110125- 120o55--55-45-. 45-.-50--50-40-` 40-.~45o~45o130--40-125- 120- 115-I~G. 6. Analyses of sea-level pressure (a) and 500 laoo ~5o 120o (b)tab height (b) for 0000 GMT 19 February 1980.115-40-pics is evident in the aircraft wind fields in the Straitof Juan de Fuca from 123-124-W, where the windswere light (generally <2 m s-l) and variable. Windswere southerly at 10-15 m s-l from northern PugetSound through the Strait of Georgia. The mountainson Vancouver Island cause the flow crossing thenorthwest corner of the Olympic Peninsula to be deflected eastward and form a narrow jet along thenorth shore of the central part of the Strait of Juande Fuca. Inspection of the mesoscale sea-level pressure pattern shows no indication of an area of fig' nificant low pressure in the light wind region in thelee of the Olympics although there is a slight tendencyfor troughing along the Strait of Juan de Fuca witha slightly higher pressure gradient over the entranceto Puget Sound and for the formation of a slight ridgeof high pressure just inside the entrance to the Strait.The higher winds over the entrance to Puget Soundand the eastern Strait of Juan de Fuca show the response of the flow field to the larger pressure gradient. Fig. 8 shows plots of the vertical profiles of tem perature, dewpoint temperature and velocity at sev eral stations around the Puget Sound area and from two dropsondes released from the P-3 aircraft. The dropsondes and Gray Field and Quillayute soundings all show the winds to be from the SSW at all levels. Open-ocean speeds were of order 15 m s-l. The sounding at Gray Field shows a low-level inversionat 950 mb whereas the other soundings show that theatmosphere, up to at least 600 mb, was slightly stablystratified with no marked inversions present. Allsoundings Show the air to be nearly saturated exceptfor a somewhat drier layer at Whidbey Island from750-900 mb.d. 28 February 1980 case At 0000 GMT 29 February 1980 a surface lowpressure system was located at 52-N, 132-W offtheBritish Columbia coast (Fig. 9a) giving S-SW surfacewinds over the Puget Sound region. This situation isquite similar to that present on 18 February 1980. At500 mb (Fig. 9b) a low was situated above the surfacelow with a trough oriented NW-SE lying just off theBritish Columbia coast. Winds at this level over thePuget Sound area were WSW. The mesoscale pressure field from the aircraft dataand ground-based stations is shown in Fig. 10. Thesedata have been adjusted to the common time of 1800GMT 28 February by correcting the aircraft data forthe time rate of change of the synoptic pressure fieldwhich was increasing at the rate of 0.85 mb h-~. Alsoshown are the 50 m level winds measured by theP-3 during the period 1745-2120 GMT 28 ~:ebrnaryand the mean surface winds observed during thistime. As in the 18 February case the blocking effectOCTOBER 1982 BERNARD A. WALTER, JR., AND JAMES E. OVERLAND 1465126-W 125- 124- 125- 122o49ON48*47*~48o ~6' ~5' ~2n. ,~o 122. lqG. 7. Local sea-level pressure field computed from land-basedand aircraft data. Non-simultaneous airerat~ measurements wereinterpolated to 2100 GMT 18 February 1980 based upon pressuretendencies from land records. Right level wind measurements areshown as small wind arrows. Bold arrows with circles indicate surface station or buoy reports. Wind convention is in knots.of the Olympics was noted on the winds in the Straitof Juan de Fuca from 123-124-W. In this case theregion of very light winds (42 m s-') was confinedto the area 122-50'-123-20'W on the flight leg closestto the south shore. A thin fog bank was present overthis area and visual observations from the aircraftindicated that the water surface was "glassy" smooth.Again, inspection of the mesoscale pressure analysisshows no indication of an area of significant low pressure in the light wind region in the lee of the Olympics, but some troughing over the eastern Strait ofJuan de Fuca. The gradient over the eastern Strait isweaker than in the 18 February case and likewise thewinds are lighter. The soundings taken at 0000 GMT 29 Februaryat several stations in the Puget Sound region and bythree dropsondes released by the P-3 (Fig. 11) showedthat winds at sonde 1 in the open ocean were fromthe SSW and were quite constant at 15 m s-' fromthe surface to 650 mb. Winds from the soundingsover land were light near the surface (3-7 m s-') andmerged with the SW flow of 20 m s-I at 500 mb. Allprofiles again show slightly stable conditions from thesurface to above 700 mb.e.. 9 March 1980 case The sea-level pressure analysis for 0000 GMT 10March 1980 (Fig. 12a) shows that there was a surfaceridge present over the State of Washington with lowlevel winds from the S-SW. At 500 mb (Fig. 12b)there was a broad ridge just offthe Washington coastwith winds at this level from the WNW. Fig. 13 shows the mesoscale, sea-level pressure fieldat 2100 GMT 9 March. Again the aircraft data werecorrected for the time rate of change of the synoptic $~)NDE SONDE 2-20 0 -20 0 SYDNEY~ 800n~ I O0 ~ -20 0~ AIR TEMPERATURE............ DEW POINT TEMPERATURE QUI LLAYUTE WHIDBEY I$.-20 -20 O TEMPERATURE (*C) G. RAY FIELD SAND PT. '%'%%.. 2 ~ I I i i -20 0 -20 0 TEMPERATURE (*C) 126'W 49'N ~ 00 GMT *so,d, 2 $ Y dne !ills'~ 19 FEB 80 48- Whidbeyl,, Oui[l~ S~nd Poinf/ 47- Gr~y Field~ ~ ~ ~ 126* 124*~G. 8. Atmosphedc mundin~ for ~00 GMT 19 Febm~ 1980.124' 122' ~~ 49* 48* 47* 122'146650-, MONTHLY WEATHER 'REVIEW140-' 135o 130o 125- 120- 140o 135- 130- 50-~VOLUME 110~25' 120-55-~55-45-- 50-45-.50-40-.40-.-45- 130-(a) 40- 120- II 5- 130- 125- 120- (b)FIG. 9. Analyses of sea-level pressure'(a) and 500 mb height (b) for 0000 GMT 29 February 1'980.115o-40-pressure field, which was decreasing at a rate of about0.6 mb h-~ during the measurement period. Alsoshown are the 50 m level winds measured by theP-3 during the period 2000-2200 GMT 9 March andthe mean surface winds measured during this time.As in the other two cases, there was no indication ofan area of significant low pressure over the Strait ofJm/n de Fuca, but slight troughing with a weak pressure gradient occurred over the eastern Strait of Juande Fuca. Winds in the western part of the Straitchanged from southwest near the entrance to westnear 124-W. In the eastern Strait the winds were fromthe east to about 123-40'W where the flow convergedwith the air from the outer Strait. Fig. 14 shows the temperature and wind profilesfrom both the dropsondes and the land-based radiosondes. The presence of an inversion at about 700750 mb (800 mb at Gray Field) is indicated by a dropin the dewpoint temperature. The layer from the surface to that level was slightly stable. The wind profilesat sonde 1 and Quillayute showed the wind near theinversion to be from the west at 17.5-20 m s-~ andbelow that from the west-southwest at 12.5-15 mf. 12 January 1968 case Another case of SW flow around the OlympicMountains is presented here for comparison with theabove cases. The sea-level pressure analysis for 0000GMT 13 January 1968 (Fig. 15) showed an elongatedlow lying offshore of British Columbia with an occluded front oriented north-south lying severalhundred kilometers of[ the west coast. The 0000GMT 13 January sounding at Quillayute (Fig. 16)showed the atmosphere was uniformly stable to a126-W 125* 124- 123* 122*49ON Jd dddd48-47~-48- -47-,~o ,~5o ,~4o ,~,- ,zz FIG. '10. Local wind field on 28 February and sea level pressureanalysis for 18 GMT 28 February 1980. Surface observations aiedenoted by bold wind barbs.OCTOBER 1982 BERNARD A. WALTER, JR., AND JAMES E. OVERLAND 1467 SONDE I SYONEY SONDE .B GRAY FIELDt~ 800~ IO00 ' -2~ ' ~) ~ -20 0 -20 0 ~0 ~ TEMPERATURE (*C) SONDE 2 QUI LLAYUTE WHIOOEY OCEAN SHORES C5~ ..,... - '-.....%..~ 800 ' '"..~ IO00 , ~ -20 0 - -20 0 -20 0 TEMPERATURE (-C) 126-W~ AIR TEMPERATURE............ DEW POINT TEMPERATURE49-NO0 GMT 48-29 FEB 8047- -~onde ~;'od.";:~OuitlayuleWhldbeGray Field-Ocean $ hores-~4126' 124.0 122- 49- 48- - 47~124- 122-FIG. 11. Atmospheric soundings for 0000 GMT 29 February 1980.50*08140- 135- 130- 125- i20- 140- 155- 130- 125- 120- o/5:3450 4.55--55-45- 45-,-50-~50*40-. 40-~-45-- 45--40-~40- 130-(a)125- 120- 115- 130- 125- 120- 115- (b)FIG. 12. Analyses of sea-level pressure (a) and 500 mb height (b) for 0000 GMT 10 March 1980.1468t26-W 125o )24-48- 126' 125~ 124' 123' 122o FIG.' 13. Local wind field on 9 March and sea-level pressureanalysis for 2100 GMT 9' March 1980. Surface observations aredenoted by bold wind barbs.height far above the mountains. The atmosphere wasmore stable here (higher Ne) than in the three previouscases but the upper level winds were much stronger,i.e., 30 m s-~. The mesoscale surface pressure fieldat 2100 GMT on 12 January (Fig. 17) showed sig- nificant troughing along the Strait of Juan de Fucainduced by the flow over the Olympic Mountains. In- this case the orientation of the isobars due to thesynoptic-scale field is more north-south than in theMONTHLy WEATHER REVIEW VOLUME 110 ~z~o ~2- above cases. The maximum pressure gradient is lo cated between the area east of Port Angeles, WA and Sydney, B.C., and this is where the maximum winds were observed. Victoria, B.C. recorded steady winds of 22.5 m s-s by 2100 GMT with gusts to 30 m s-t. g. Hood Canal Bridge storm The surface pressure analysis in Fig. 18 (Reed, 1980) showed that at 1200 GMT 13 February 1979 a 976 mb low-pressure system was located just offthe coast of Vancouver Island, B. C., with a cold front over central Washington. Surface winds over the Puget Sound region were from the south at 12-15 m s-l. The 1200 GMT SOunding from Salem, Oregon (Fig. 19), showed that the atmosphere was weakly stable throughout a large depth; the sounding from Quillayute was not available for this time. The weak stratification gave a value of the Brunt-V//is/iEi fre quency much smaller than any of the other cases (Ne = 0.003). This combined with the strong winds throughout the depth of the lower atmosphere, how ever, gave a value of the Froude number much larger than in any of the other cases (F = 4.63). The rue s-scale surface pressure field at 1200 GMT from Reed (1980) is shown in Fig. 20 and is quite different from the other mesoscale pressure analyses. A strong, rue s-scale, closed low-pressure area was located in the lee of the Olympics near the.Hood Canal Bridge. The response of the low-level momentum balance to this orographically induced low pressure was the accel eration of the wind to 50 m s-~ in the vicinity of the bridge. SONDE I J 8=2eoo~ '~ ~ ! SONDE 2 QUILLAYUTE~ ,ooo~ -~0 0 -~0 0 -~0 0 TEMPERATURE (~C) -20 0 SYDNEy~O00L TEMPERATURE (-C) ~ AIR TEMPERATURE .' .......... DEW ROINT TEMPERATURE O0 GMT I0 MAR 80 /26-W49-N124~ 126- 124oFIG. 14, Atmospheric soundings for 0000 GMT 10 March 1980.122o122~ 49~ 48* 47~OCTOBER 1982 BERNARD A. WALTER, JR., AND JAMES E. OVERLAND 1469FIG. 15. Sea-level pressure analysis for 0000 GMT 13 January 1968.4. Discussion In all five cases described above, the atmospherewas slightly stable and continuously stratified over alarge depth with Brunt-V/ii~lii frequencies varyingfrom 0.003 to 0.012 s-I. In four of the cases (18 and28 February, 9 March 1980 and 12 January 1968)the internal Froude numbers were 1.0-1.4 and theflow around the Olympic Mountains was such thata weak trough in the pressure field was present over400500'60070080090(I00012-i2E i3-OOZ 'FIG. 16. Soundings at Quiilayute, Washington for 1200 GMT12 January and 0000 GMT 13 January 1968. Solid lines are temperature and dashed lines are dewpoint.the eastern part of the Strait of Juan de Fuca. In thelee of the mountains winds were light and disorganized, indicating the presence of a wake region there.In the fifth case, 13 February 1979, the internalFroude number was 4.6 due to a combination of bothlow stability and high wind speed. In this case, themesoscale pressure analysis done by Reed (1980) indicated the presence of a significant lee-side, closedlow-pressure area. As the Froude number increases from values lessthan some critical value (strong stability) to valuesgreater than that critical value (weak stability) theflow around the mountain changes from being quasihorizontal in nature to having strong downslope motion in the lee of the obstacle. This was shown in thelaboratory studies of Hunt and Snyder (1980) in Figs.3a(ii) and (iii). This same behavior is shown qualitatively in the cases considered here. Cross-sectionplots of constant potential temperature 0 surfacesalong the east side and in the lee of the olympics forthe three Puget Sound Wind Study cases (Figs. 21 a2 lc) show that the flow is indeed quasi-horizontal inagreement with results for stable flow conditions. Inaddition, the depression in the 0 surfaces over theeastern Strait of Juan de Fuca near Whidbey Islandis consistent with the surface pressure analysis showing slight troughing over this area. The cross-sectionplots were derived from soundings taken at GrayField, Sand Point, Whidbey Island, Salem, Oregon,Sydney, B.C., and by P-3 dropwindsondes. Wind profiles were also plotted where available along with asurface wind recorded about 1600 GMT on Hurricane Ridge. The magnitudes of the isotherm displace50~45~JAN 12, 1968 08'2100 GMT Meters~"~:~'~ 900 ~ Above~0-300 I~fi-120'50*45*FIG. 17. Local pressure field for 2100 GMT 12 January 1968. Isobars are drawn at I mb intervals.1470 MONTHLY WEATHER REVIEW VOLUME I1012o' ~ 56.28 FIG. 18. Sea-level pressure analysis for 1200 GMT 13 February1979. Storm track is given by arrows. Dots give position of centerat indicated times (Reed, 1980).ments showed that the flow was approaching the transition region between small and large Froude numbertheory. A plot of the surface pressure field for large F fromSmith (1980) was shown in Fig. 3b. Combining thisperturbation pressure field with a synoptic scalenorth-south pressure field for the Hood Canal stormgives the result shown in Fig. 22 (Smith, 1981). Thecharacteristics of this field have many similarities tothat analyzed by Reed (1980). As shown by Overland and Walter (1981) the surface flow in the confined channels of the Puget Soundregion is primarily down pressure gradient and canbe closely represented by the balanceAU2 Ap-- - , (7)2 pwhere Au2 is the increase in the square of the velocity,Ap the difference in sea level pressure, and p air density. Reed (1980) indicates that in the vicinity of theHood. Canal Bridge on 13 February 1979 the pressuregradient was a maximum of 6 mb in 15 km or 0.4mb km-'. Table 2 compares the observed winds atthe time of the bridge collapse (from Table 1, Reed,1980) with those calculated using (7) and a value ofthe pressure gradient of 0.4 mb km-I. Calculationsgive values close to those observed at the time of thebridge collapse. Thus, in the Hood Canal storm theperturbation pressure field from Smith (1980), combined with the synoptic field, is able to create an extreme pressure gradient in the lee of the Olympics.The flow at the surface in the restricted channels ofPuget Sound responds by a rapid acceleration towardlow pressure to give extreme wind speeds in a localized area.EbJ500 ix\ ,~</,,' ~X , ~ , ' ~_5400 i .X-" ', \ 48-- 600 .0 7OC _ _ 7NN 3000 ~ :: g x :: ~ ~4oo ~ 800 8O0 ~ ' 1200 900 6O0 IO00 ~ ~ ~ ~ - 40 -30 -20 - I0 O IO ~ TEMPERATURE ~ ~G. 19. Sounding for S~em, Oregon, 1200 GMT 13 Febma~ 1979. Solid lines:tem~mture and dew ~int; dmhed lines: d~ adiabat, Fa and moist adiabat Pro;dotted line: mixing ratio (g kg-~) (R~, 1980).OCTOBER 1982 BERNARD A. WALTER, JR., AND JAMES E. OVERLAND 1471 FIG. 20. Regional surface map for 1200 GMT 13 February 1979. Isobars are drawn at 1 mbintervals (dashed at higher elevations). Winds are in knots. Maximum gusts (G) at observationtime and in preceding hour are plotted below stations (Reed, 1980). We hypothesize that when the internal Froudenumber has a value.of 0.5 to 1.5, as calculated fromone of our upper-air stations, the flow response in thelee of the Olympics will be relatively benign withslight troughing in the eastern Strait of Juan de Fucaand a region of light winds directly in the lee of themountains. On the other hand, when the Froudenumber is much larger (i.e., 4-5), a deep low-pressureregion can be induced in the lee of the mountains,and the formation of high winds caused by stronggradients in the sea-level pressure field is possible. TheFroude number can be calculated from the routineatmospheric sounding at Quillayute; this provides anindex that forecasters can use to determine if a particular offshore storm may lead to severe surfacewinds in the Puget Sound basin.~ 500 3o5 3o8 3o2~ 600 ~oo ......... 3oo - -g~ 300 700 296~7 .... 295 ~ ~296~'''' 295 ........... 294 - - ~ ~ ........290- ~ ~ pOO 5~ 292/ 286---~ -- ~ OLYMPl~ 288~ 45" 47- 48- ~9~ LAT FiG. 21a. No~h/south cross ~ction over the Puget Sound re,on east of theOlympic moun~ns sho~ng the ~havior of surfaces of consent potenti~ tem~rature at 0000 GMT 19 Febma~ 1980.1472 MONTHLY WEATHER REVIEW VOLUME IIObC500'6OC)'700'800'900'IOO0.FIG. 2lb. As in Fig. 21a except at 0000 GMT 29 February 1980. ------~312~ ..... .~ ~ - ~ - ..... 298 ~ ~ ~,0o0i ~,~.~ ,', ~o ~.~ ~, // ~.~.45~ 47~ ~oSA~M GRAY SONDE WHI~Y SYDNEY 0R FIELD 2 I EIG, 21C, AS in Fig. 21a except at ~00 GMT 10 March 1980.49g LAT5. Summary We have investigated the mesoscale surface windregimes in the lee of the Olympic Mountains for five'cases where the flow was from the S-SW. In all casesthe atmosphere was 'continuously stratified throughout a deep layer. In four of the cases the wind fieldswere generally light and variable in the lee of themountains, and surface analyses showed an area ofrelatively flat pressure over the same region. The internal Froude numbers for these cases were in therange 1.0-1.4 and the response of the flow agreesquite well with both theoretical and laboratory resultsfor low-Froude-number flow. In the other case the surface winds were very strongin the lee of the mountains, and surface analysesshowed the presence of a major mesoscale low-pressure area. The internal Froude number for this caseO 50 IOOkm-4-3-I 0 I 3 4rnbFIG. 22. Sum of the perturbation pressure field shown in Fig. 3b and a hypothetical synoptic pressure field (from Smith, 1981).OCTOBER 1982 BERNARD A. WALTER, JR., AND JAMES E. OVERLAND 1473 TABLE 2. Comparison of winds reported by Reed (1980) in HoodCanal at 1400-1500 GMT 13 February 1979 and those calculatedfrom Eq. (7). The initial velocity used for the calculation was 22.5ms-1. Observed Calculated Location (m S-~) (m s-I) Bridge 40 40.5 9.25 km SSW 30 33.0 18.5 km SSW 22.5 22.5** Assumed.was 4.6 because of strong upper-level winds (~25m s-~) and lower stability, and the response of thepressure field is in agreement with the results ofSmith's (1980) linear hydrostatic model. The flow atthe surface is produced by a down-pressure-gradientacceleration in the confined channels of Puget Sound.It is hypothesized that the internal Froude numbercan be used as an index to indicate the potential forextreme winds in the lee of the Olympic Mountains. Acknowledgments. This work is a contribution tothe Marine Services Project at the Pacific MarineEnvironmental Laboratory. We wish to thank: themembers of the NWS Seattle Ocean Services Unit fortheir aid in providing forecasts which were crucial inplanning aircraft flights; Don Faulkner, Steve Niklevaand Ron McLaren of the Canadian Atmospheric Environment Service in Vancouver, B.C., for takingatmospheric soundings at Sydney, B.C., and providing us with encouragement as well as surface observations from Canadian Marine Stations; Cdr. EarlKerr and Mr. Mallory for arranging to have soundings taken at the Whidbey Island Naval Air Stationand Ft. Lewis, Washington, respectively; and especially to the crew of the NOAA P-3 for their assistancein the collection and processing of the aircraft data. We wish to thank Drs. Cliff Mass, Richard Reed,Ron Smith and James Connell for helpful commentsand suggestions. We also thank the Journal of FluidMechanics (Cambridge University Press) for permission to use Figs. 3a(ii) and (iii); Tellus for permission to use Fig. 3b and Quarterly Journal of theRoyal Meteorological Society for permission to useFig. 3a(i). We acknowledge the assistance of Professor Richard Reed, Department of Atmospheric Science,University of Washington, in providing a copy of anunpublished manuscript by V. F. Morris titled "Astudy of mesoscale mountain barrier effects in western Washington and Vancouver Island" from whichthe information contained in Figs. 15, 16 and 17 wasexcerpted. REFERENCESBrighton, P. W. M., 1978: Strongly stratified flow past three-di mensional obstacles. Quart. J. Roy. Meteor. Soc., 104, 289 307.Chopra, K. P., 1973: Atmospheric and ocean flow problems in troduced by islands. Advances in Geophysics, Vol. 16, Aca demic Press, 297-421.Drazin, P. G., 1961: On the steady flow of a fluid of variable density past an obstacle. Tellus, 13, 239-251.Durran, D. R., and J. B. Klemp, 1981: The effects of moisture on trapped mountain lee waves, Preprints 2nd Conf Mountain Meteorology, Steamboat Springs, CO, Amer. Meteor. Soc., 106-113.Hess, S. L., 1959: Introduction to Theoretical Meteorology. Holt, Rinehart and Winston, 362 pp.Hunt, J. C. R., and W. H. Snyder, 1980: Experiments on stably and neutrally stratified flow over a model three-dimensional hill. J. Fluid Mech., 96, 671-704.Kitabayashi, K., M. M. Orgill and J. E. Cermak, 1971: Laboratory simulation of airflow and atmospheric transport-dispersion over Elk Mountain, Wyoming. Fluid Dyn. Diff. Lab. Rep. No. CER 70-71 KK~MMO-JEC-65, Colorado State University, 90 PP.Lilly, D. K., 1973: Calculation of stably stratified flow around com plex terrain. Res. Note No. 40, Flow Research, Inc., Kent, WA 98031, 7 pp.Marwitz, J. D., D. L. Veal, A. H. Auer and J. R. Middleton, 1969: Prediction and verification of the airflow over a three-dimen sional mountain. Natural Resour. Res. Inst. Tech. Rep. No: 60, University of Wyoming, 26 pp.Mass, C., 1981: Topographically forced convergence in western Washington State. Mon. Wea. Rev., 109, 1335-1347.Maunder, W. J., 1968: Synoptic weather patterns in the Pacific Northwest. Northwest $ci., 42, 80-88.Merceret, F. J., and H. W. Davis, 1981: The determination of navigational and meteorological variables measured by NOAA/ RFC WP3D aircraft. NOAA Tech. Memo., ERL RFC-7 NOAA/ERL, Research 'Facilities Center, Miami, 21 pp.Merkine, Lee-Or, and E. Kalnay-Rivas, 1976: Rotating stratified flow over finite isolated topography. J. Atmos. Sci., 33, 908 922.Overland, J. E., and B. A. Walter, 1981: Gap winds in the Strait of Juan de Fuca. Mon. Wea. Rev., 109, 2221-2233.Reed, R. J., 1980: Destructive winds caused by an orographically induced mesoscale cyclone. Bull. Amer. Meteor. Soc., 61, 1346-1355.Riley, J. J., H. T. Liu and E. W. Geller, 1976: A numerical and experimental study of stably stratified flow around complex terrain. EPA Environ. Monitoring Ser. Rep. No. EPA-600/4 76-021, National Environmental Research Center, Environ mental Protection Agency, Research Triangle Park, NC, 27709, 30 pp.Smith, R. B., 1980: Linear theory of stratified hydrostatic flow past an isolated mountain. Tellus, 32, 348-364.--, 1981: An alternative explanation for the destruction of the Hood Canal Bridge. Bull. Amer. Meteor. Soc., 62, 1319-1320.
Abstract
The behavior of stratified air flowing around an isolated mountain is dependent on an internal Froude number (F), which indicates the relative importance of upstream velocity and vertical stratification. Three cases of the flow in the lee of the Olympic Mountains in the State of Washington are studied where the measured F was in the range 1.0–1.4 but apparently dominated by stable stratification. This study combined measurements of spatial variation of low-level winds and other parameters from a NOAA P-3 research aircraft with a dense network of surface stations including eight meteorological buoys and six upper-air stations. Results from these cases show the presence of an area of light winds in the lee of the Olympic Mountains. The characteristics of the flow are shown to be similar to laboratory results for low Froude number flow around an isolated obstacle where the flow is confined to quasi-horizontal planes. These cases are contrasted with a situation which led to the formation of a mesoscale low-pressure area and high surface winds in the lee of the mountains. The latter case was the Hood Canal Bridge storm on 13 February 1979 where local winds in the lee of the Olympic Mountains were in excess of 50 m s−1. The flow at the surface was produced by down-pressure-gradient acceleration in the confined channels of Puget Sound toward the orographically produced low-pressure center. The measured internal Froude number in this situation was 4.6, and the pressure fields are shown to agree with the linear hydrostatic model developed by Smith (1980) for F > 1. It is suggested that the Froude number calculated from routine, upper-air sounding data is an index that forecasters can use to determine the potential for severe wind conditions over the inland waters of Puget Sound.