OCTOBER 1982 ECKLUND, GAGE, BALSLEY, STRAUCH AND GREEN 1451Vertical Wind Variability Observed by VHF Radar in the Lee of the Colorado RockiesW. L. ECKLUND, K. S. GAGE, B. B. BALSLEY, R. G. STRAUCHl AND J. L. GREENAeronomy Laboratory, National Oceanic and Atmospheric Administration, Boulder, CO 80303(Manuscript received I March 1981, in final form 7 June 1982)ABSTRACT During March 1981 the Sunset and Platteville VHF clear-air radars located in Colorado to' the east of thecontinental divide observed vertical winds continuously over a three-week period. The vertical winds at theselocations contain fluctuations with periods from a few minutes to several hours and with magnitudes rangingup to a few meters per second. The Sunset radar, which is located in the foothills, observed systematicallylarger vertical velocities than the vertical velocities observed by the Platteville radar, which is located on theplains, some 60 km to the east. Although periods of enhanced vertical wind activity were observed to occurat the same time at both sites, attempts to correlate vertical wind structures o.ver the two sites in detail weregenerally not successful. The magnitude of vertical velocity fluctuations seen by both radars show large day-to-day variations with"active" periods alternating with "quiet" periods. An examination of upper level maps reveals that theoccurrence of active and quiet periods are linked to the large-scale wind field. During the March experimentthe magnitude of the vertical velocity variance was well correlated with the 500 mb zonal (west) wind.1. Introduction Vertical velocity plays an important role in thedynamics of atmospheric circulation systems on allscales. However, until very recently there has beenno direct means to continuously measure verticalvelocities. As a result our present knowledge of therole of vertical velocity in weather systems has beenderived indirectly from kinematic analysis of largescale synoptic wind fields. These analyses provide insight into the broad-scale features of vertical velocityon the synoptic scale but give us no informationabout the variability of vertical velocity at any particular location. The local vertical velocities can beexpected to contain contributions from convection,gravity waves (including stationary lee waves) andturbulence, in addition to the much smaller synopticscale vertical wind. Doppler radars sensitive enough to obtain usefulechoes from clear-air refractive structure have provided new information on winds, waves, turbulenceand atmospheric stability over the past several years(Gage and Balsley, 1978; R~ttger et al., 1978; Greenet al., 1979; Balsley and Gage, 1980; James, 1980).As part of this .effort, several groups have reportedlimited observations of vertical winds in the troposphere and lower stratosphere (Woodman and Guilleft, 1974; Balsley et al., 1977; RiSttger et al., 1978;Fukao et al., 1978; Green et al., 1978; Peterson andBalsley, 1979; Gage et aL, 1981). These vertical windWith the Wave Propagation Laboratory.observations have often shown oscillations with periods from about 5-30 min with amplitudes rangingfrom 0.1-1 m s-I, which are apparently associatedwith propagating internal gravity waves. In additionto these oscillations, fairly steady vertical winds of upto I m s-~ have occasionally been observed to persistfor several hours. These steady vertical winds aremost likely due to non-propagating gravity wavessuch as lee waves. More recently, Ecklund et al.(1981 a) presented over 30 days of nearly continuousobservations of vertical winds up to 20 km obtainedusing the MST radar at Poker Flat, Alaska. Generally,fluctuations were found to be controlled by propagating planetary-scale waves that modulate the largescale wind field. The strongest vertical winds werealmost always associated with the enhanced wind andwind shear found in conjunction with intense baroclinc zones. In this paper we present the results from a verticalwind experiment conducted in Colorado duringMarch 1981. In the experiment two VHF clear-airradars separated by 63 km were operated continuously over a three-week period in order to obtainvertical winds up to an altitude of ~20 km with temporal resolution of a few minutes. There were twomajor motivations for conducting this experiment.First, since one site was located in the mountainfoothills while the other site was located on the plains,we wanted to determine what effects local topographyhad on vertical wind fields. Second, we wanted todetermine whether observed vertical wind structurecould be correlated over the 63 km horizontal dis1452 MONTHLY WEATHER REVIEW VOLUME 1104000 - Q' '% ~ e) '~ ~ .L~ . ~ o ~ r , g-~'"- Sunset ~- Platteville a -- ~ / ' ttt ~-~t- ~ ~ I - ~ ~ ~ t ~km ] ~. ~ _~ ~' ~-~ ~ _ ~300C - '~gIOOC zAC AFIG. 1. Map of the experimental area.tance between the two -sites. Although attempts tocorrelate vertical wind structures over the two sitesin detail were generally not successful, "active" periods of enhanced vertical velocity fluctuations wereobserved to occur at the same time at both sites. These"active" periods were found to correlate well withenhanced 500 mb zonal (west) wind.2. The radar experiment The relative locations of the two radar systems usedin this experiment are indicated on the map in Fig.1. Note that the site alignment is approximately eastwest, i.e., in line with the prevailing wind pattern. Anoutline of the profile of the ground surface height onthe east-west line between the two sites appears at thebottom of Fig. 1. The location of both sites relativeto the continental divide and the beginning of theplains region is evident. Sunset is located in the foothills close to the divide, while Platteville is situatedon the plains, but only 40 km from the edge of thefoothills. The line-of-site distance between the tworadars is ~63 km. The Platteville (50 MHz) and Sunset (40 MHz)radars are pulsed, VHF Doppler systems operatingat peak transmitted powers of 15 and 50 kW, respectively, with antennas comprised of phased arraysof dipole elements. A photograph of the Plattevillearray is shown in Fig. 2 to demonstrate the simplicityof this type of antenna. Detailed descriptions of theSur/set and Platteville radars can be found in Greenet al. (1979) and Ecklund et al. (1979). For the present experiment both systems operatedalmost exclusively with vertically directed beams, theexception being that Sunset performed a three-position scan every 12 h to measure the horizontal windfield as well. This procedure did not cause an appreciable deterioration of the vertical data and affordeda much more local picture of the horizontal windprofile than that which could be obtained from thetwice-daily balloon soundings at Denver. Data from both systems were processed in similarways to afford the best comparison. The time resolution in the present data set has been standardizedby appropriate computer averaging. The Sunset radarused a vertical resolution of 1 km and the Plattevilleradar used a vertical resolution of 2.3 km. To. facilitatecomparison, both radars were sampled at altitude inOCTOBER 1982 ECKLUND, GAGE,BALSLEY,STRAUCHAND GREEN1453FIG. 2. Photograph of the Platteville antenna array.tervals of 1.2 km. Thus the Sunset radar data wereslightly undersampled and the Platteville data wereoversampled by a factor of about 2.3. The observed variability of the vertical wind A composite picture of the vertical wind field overboth Sunset and Platteville for the complete experimental period is shown in Fig. 3. Both data sets consist of 15 rain averaged data and are plotted on thesame time axis with the same vertical velocity scaleto facilitate comparison. The vertical range of the database extends from below 5 km to above 20 km (MSL).The upper limit is controlled primarily by the radarsystem sensitivity; more sensitive VHF radars (i.e.,bigger antenna areas or higher average transmitterpower) would extend this upper limit [see Green etal. (1979) and Balsley and Gage (1980) for details].Obvious blank spots in the data arise from systemoutages; less obvious data dropouts occur occasionally on the Sunset data when the wind magnitudeexceeds the nominal 2.0 m s-~ limit used in thisfigure. There are a number of obvious similarities in thetwo data sets as well as a number of clear dissimilarities. These are treated in the following paragraphs. Perhaps the most obvious similarity in the Sunsetand Platteville results is the coincident pattern of active periods and quiet periods in the vertical windsover both locations. Here we define active periods asthose times with enhanced vertical wind activity centered very roughly, for example, at 16, 19-20, 22-23and 26-27 March. Similarly, quiet periods are typified by times with minimal vertical wind activity centered, for example, on 17-18, 21, 24-25 and 28March. As shown in the next section, the overall activity in the vertical wind appears to correlate wellwith the magnitude of the westerly component of thetropospheric wind field. This suggests that the verticalwind structure arises at least in part from orographically generated gravity waves, since the sharp profileof the continental divide lies to the west of both sites(cf. Fig. 1). Moreover, if orography is the Controllingfactor, the activity would be expected to be considerably stronger at Sunset than that at Platteville, asobserved. Since most of the vertical wind structuresshown in Fig. 3 extend over several kilometers, thedifferent vertical resolution at Sunset (1 km) andPlatteville (2.3 km) cannot explain the stronger activity observed at Sunset. The point to be made hereis that the activity occurs essentially simultaneouslyat both sites. Attempts to correlate concurrent details in the vertical wind pattern at Sunset and Platteville meet withonly marginal success. Possible evidence of a coherentfeature over the two sites occurs in the forenoon of1454MONTHLY WEATHER REVIEWPlatteville Radar Vertical WindsVOLUME 1 10202 km19.0 kin.16.6 km15,4 kin.14.2 kin.13.011.8 kin.10.6 km 9.4 km 8.2 ?.0 kin. 5.8 km 4.6 km I0 I i I I- II 12~ 13 14T2 m/sJ-Upward 20.Skm~193km1 ~ : e!~ -"~ I~"~ ~I1. I't J'~ ....... ~ ~ ~.l "-,., -, ' "~' ~ .... '-- -'l'' '" .".'m '"_ '~"'~1 ~"" ~. '"'~1 ~ '~g''~'~'''~'~ 14Jkm1 ~ ~ ~ ~ ~ ~T q~ w'~ ..... tl. ~ .~__ ~.~,,. ~ :,.~.~,s .... ~,~ ~ .~ ~ ~ ~r -, ~-- '~v~--~"~~ __ , -,,,~ ~'~"1 '~~' ~t :~"~" - ' ~ ~ ~ ~ ~ -~,?~.~--- .~f,, ~,l~ 'o.*'-1 ~ ~~ -'~I ~~' -'~' - j' TL~.~ .... '- m"n~ '~'~ '"'"1 2:~T~ ::,~ *-'~'~:"~?' ''-~ ~ s.5 km1 ..... ~ ?~... ~ [~'~~ "~1 ~'~~ 4= ~ ~*~"~-"~-~'~'~'~ ~- ~ , -- q '~ 6.1kml~~,*. ",,. ' '~J '~' ""! ....... !,: 213 1415 16 ff 18 19 20 21 ~ 2~ 24 25 ~ March 198127 28 292 m/sUpwardFIG. 3. Fifteen-minute-averaged vertical velocities over Platteville and Sunset, Colorado, for the complete experimental period.19 March, where the downward flow pattern between7 and 14.2 km at Platteville corresponds to the (stronger) downward flow over Sunset below 15.7 km. Thevertical flow reverses at higher heights (16.6 km atPlatteville; 16.9 km at Sunset). However, evidence forcomparable structures over the two sites is generallynot convincing. This may be due to the fact that thegravity waves (including lee waves) typically have spatial scales smaller than the 63 km spacing betweenthe radars. Comparison may also be complicated bythe fact that the two stations do not necessarily liealong the direction of the mean wind. A more detailed plot of one active' day (23 March1981) over Sunset and Platteville appears in Fig. 4.The time resolution for this plot is 5 min. Here thevertical scales have been adjusted to reproduce windvelocities up to 5 m s-L (Note that the "periodic"structures at 0400 and 1700 MST on the Sunset dataare due to vertical data dropouts' when the antennawas scanned to obtain oblique wind information.) Some general correlation between the gross verticalmotion at the two sites can be seen in this record. Forexample, the period of enhanced flow between 1300and 1700 above about 8 km appears on both records;the sense of the general vertical flow, however, is reversed (Sunset flow is downward below the 13.3 kmlevel while Platteville is upward below the 16.6 kmlevel; the reversed situation is seen above these levels).This reversed picture is also seen at the ~ 15-16 kmlevel on both systems between 0000 and 0200. A possible explanation for the reversed vertical winds at thetwo sites during this period is that each radar wasobserving a different phase of a large-amplitudemountain lee wave. The horizontal wavelength of theOCTOBER 1982 ECKLUND, GAGE, BALSLEY, STRAUCH AND GREEN 1455proposed lee wave cannot be determined because theradars are too far apart to eliminate spatial ambiguities. Future experiments should use three or morestations (if possible) with closer spacing along theeast-west line.4. Day-to-day variability of vertical wind and the large-scale wind field As discussed above, the most striking feature of thevertical wind structure seen by the Platteville andSunset radars is the large overall day-to-day variationin the magnitude of the vertical velocity fluctuationswith "active" periods alternating with "quiet" periods. In order to gain some perspective on therelationship between day-to-day variations in themagnitude of vertical velocity fluctuations and synoptic-scale weather features we have examined theupper-level synoptic maps. Selected synoptic mapsfor the 500 mb level are reproduced in Fig. 5. Thesemaps are keyed to the synoptic times A-H indicatedon the graph qf vertical winds in Fig. 3. Map A showsthe large-scale wind field at 1700 MST 12 March1981. At this time the winds over the western UnitedStates were very weak. This synoptic situation persisted for several days and coincided with the lack ofactivity in the vertical winds at the beginning of theobserving period 10-14 March 1981. The first activeperiod occurred on 15-16 March, and at this timemap B (0500 MST 16 March) shows that the largescale gradient in the height field has increased considerably and a ridge has built over the mountainstates. Map C (1700 MST 17 March 1981) shows amuch more complex synoptic pattern. At this timelow pressure has developed to the east of Coloradoand the zonal wind over eastern Colorado is veryweak. This corresponds to a quiet period in the vertical wind activity. The next-active period is shownon the synoptic map D for 0500 MST 19 March. Atthis time the gradient in the height field is quitestrong. This situation is also revealed on the map bythe strong winds observed at Denver. The remainingtwo cycles of active and quiet periods shown in Fig.3 also show strong zonal flow during active periodswith an upper level low typically located to the eastwith weak zonal flow over Colorado during quietperiods. In the preceeding section we have shown an association between periods of enhanced/suppressedactivity in vertical winds with strong/weak zonal flow.To further examine this relationship throughout theobserving period, we have compared the variance ofthe vertical wind observed by the Platteville radar(the radar vertical wind variance was obtained over4 h periods bracketing the rawinsonde synoptic observing tim.es) with the 500 mb zonal wind component observed by the Denver rawinsonde. This comparison is shown in Fig. 6. For convenience, we havePlatteville Radar Vertical Winds20.2 krn'19.0 km -17.8 kin'16.6 km15.4 km.14.2 kmI~,0 km'11.8 km10.6 krn.8.4 kin.8.2 krn'7.0 km.5.8 km4.6 km~-5 m/s Upward Sunset Radar Vertical Winds ~0.5 km~ '.,-,'" --- .- I * I,a,",'~l '~' ' ~-~ ' '/"' , ...... .... '6.8 km J~l~W.Jl~`dd, ,lqjjj~l~w,~*-~' ",qjI ~ .....~": 15.7 km ~ 'lll~J"Im'~m~lml, ,'r"w~ .... I ,.,'..~,.. ,. .1~ ,4.s kin- ----..-t .,,,_.~,~mVi..w...~m. - ~""~1_ I~.~krn;---'~.---' .-.~~ .... 'T---"-,--'.-'~-- ~ ,z, ~.:-------..i,.,~d.~-L_.--~ ,,...~._.....__.3:10.9 km - '~-~.-'~' "" ,~JJllJm...~..~ I ,'"""'3-'~'~ ~.~J illlJ~_~__ i. Ilk[.---- 8., ,m- ,; ..... _ .t I._ ,m q.~m-~ ,W~. ,,~' Ir.X . . I ' . I. ," .'- 0 (~ST) 6 12 18 March 198lRo. 4. Five-minute-averaged vc~ical velocities over Plattcvillcand Sun~t for the amivc pcd~ 23 ~arch 1~81.~5 m/s Upwarddivid6d the radar data into three height ranges. Eachof the height ranges is an average of two radar heights.A comparison between the vertical wind variance atthe three separate height ranges and the 500 mb zonalwind shows a very good correlation. The best correlation appears to occur near the tropopause level.Since the vertical wind variance at Platteville andSunset appears to be highly correlated (cf. Fig. 3) itis reasonable to expect that the result shown in Fig.6 for Platteville would also have been obtained byusing Sunset data. The Platteville data were used inthis example only because they were available for alonger period of time.5. Concluding remarks In this paper we have presented the vertical windsobserved by the Platteville and Sunset radars duringa three-week observing period in March 1981. Ouranalysis shows that during this period the overall dayto-day variability of vertical wind activity is affectedprimarily by synoptic-scale weather patterns in general and by the magnitude of the zonal wind in par1456 MONTHLY .WEATHER REVIEW VOLUME 1101700 MST 12 Mlr 1981 ~) 1700 MST 17 MIr 1981ill - ,~) 0500 MST 16 Mar 1981~..__~ 0600 MST 21 MIr 18811700 MST 23 MIr 19810400 MST ESMM' lee1~) 0500 M8T 26 Mar 1981FIG. 5. 500 mb synoptic maps for the observing period of Fig. '3. The circled letters (A)-(H) at the top of each map correspond to the times denoted by the circled letters in Fig. 3.OCTOBER1982 ECKLUND, GA~E, BALSLEY, STRAUCH AND GREEN 1457 0.18 0.12 0.08 0,0~ 0~~-~ -- 500 rnb Zonal Wind ~ Average Variance at 15:4 and 17,8 km -- 500 mb Zonal W, ind 5 ~E Average Variance at IO.6 and 13.O krnv o.2~ o.~ol /~ 25 ~ o.~6~,2 .~ 0.08 ~0'~ Q04 5 ~.........._ 8 ~ 5OOmb Zonol Wi~ ~ ~ 0.~ ~ Average Varion~ al ~8 ond 8.2 km 024 ~ ~ 02C ' ~5 o.o8 / Iio ~04 5 0 - ~ - 0 ~RCH i0 12 I~ 16 t8 ~ 22 24 26 Z8 N~CH 50 Dey FIG. 6. A comparison of ve~ica] velocity variance at three heists with the 500 mb zonal wind speed.ticula~. Although not discussed in this report, at timesduring the observation period vertical winds appearedto be also affected by convective activity as evidencedby the presence of cumulus clouds (Ecklund et at.,1981b). The day-to-day variability of vertical wind activityobserved at Platteville can be compared to observations of vertical wind activity at Poker Flat, Alaska,reported by Ecklund et al. (1981a). Although themagnitude of the vertical wind activity in both Alaskaand Colorado alternates between quiet and activeperiods in response to the large-scale wind field, thereare important differences in the way this variationtakes place. At Poker Flat the variation in verticalwind activity is correlated generally with wind speedand wind shear and not directly with wind direction.At Platteville, on the other hand, we find a very goodcorrelation between the magnitude of vertical windactivity and the magnitude of the zonal wind (i.e.,there appears to be a direction dependence). The difference in behavior at the two locations may be ascribed to differences in orography, since at Poker Flatrough terrain is fairly uniformly distributed in alldirections while in Colorado there is a pronouncedeast-west asymmetry in roughness. In Colorado windfrom the west flows over the continental dividewhereas wind from other directions comes primarilyover the plains. It remains to be determined to whatextent vertical wind activity varies over regions ofsmooth terrain. The fairly steady vertical winds ranging up to several meters per second and lasting for several hoursobserved in this experiment are probably due tomountain lee waves. However, because of the difficulty in correlating concurrent details of vertical windpatterns at the two radars, it has not been possibleto estimate the relative importance of stationary gravity waves (lee waves) and propagating gravity wavesin generating the observed vertical wind variability.Multiple radars with horizontal spacing considerablyless than the 63 km spacing used in this experimentshould allow this question to be resolved in futureexperiments. Acknowledgments. The Sunset radar is operated byNOAA's Aeronomy Laboratory. The Platteville radaris currently a joint operation between the AeronomyLaboratory and the Wave Propagation Laboratory.REFERENCESBalsley, B. B., N. Cianos, D. T. Farley and M. J. Baron, 1977: Winds derived from radar measurements in the Arctic tro posphere and stratosphere. J.Appl. Meteor., 16, 1235-1239.--, and K. S. Gage, 1980: The MST radar technique: Potential for middle atmospheric studies. Pure AppL Geophys., 118, 452-493.Ecklund, W. L., D. A. Caner and B. B. Balsley, 1979: Continuous measurements of upper atmospheric winds and turbulence using a VHF Doppler radar: Preliminary results. J. Atrnos. Terr. Phys., 41, 983-994.--, K. S. Gage and A. C. Riddle, 1981a: Gravity wave activity in vertical winds observed by the Poker Flat MST radar. Geo phys. Res. Lett., 8, 285-288. , -- and B. B. Balsley, 1981b: A comparison of vertical wind variability observed with the Platteville VHF radar and local weather conditions. Preprints 20th Conf. Radar Meteo rology, Boston, Amer. Meteor. Soc., 1.-6.Fukao, S., S. Kato, S. Yokoi, R. M. Harper, R. F. Woodman and W. E. Gordon, 1978: One full-day radar measurement of lower stratosphere winds over Jicamarca. J. Atmos. Terr. Phys., 40, 1331-1337.Gage, K. S., and B. B. Balsley, 1978: Doppler radar probing of the clear atmosphere. Bull. Arner. Meteor. Soc., 58, 1074-1093.---, D. A. Caner and W. L. Ecklund, 1981: The effect of gravity waves on specular echoes observed by the Poker Flat radar. Geophys. Res. Lett., 8, 599-602.Green, J. L., K. S. Gage and T. E. VanZandt, 1978: Three-di mensional wind observations of a jet stream using a' VHF Doppler radar. Preprints 18th Conf. Radar Meteorology, At lanta, Amer. Meteor. Soc., 88-93. , -- and---, 1979: Atmospheric measurement by VHF pulsed Doppler radar. IEEE Trans. Geosci. Electron., GE-14, 262-280.James, P. K., 1980: A review of radar observations of the tropo sphere in clear air conditions. Radio Sci., 15, 151-175.Peterson, V. L., and B. B. Balsley, 1979: Clear air Doppler radar measurements of the vertical component of wind velocity in the troposphere and stratosphere. Geophys. Res. Lett., 6, 933 936.Rtttger, J., J. Klostermeyer, P. Czechowsky, R. R'dster and G. Schmidt, 1978: Remote sensing of the atmosphere by VHF radar experiments. Naturwissenschafien, 65, 285-296.Woodman, R. F., and A. Guilltn, 1974: Radar observations of winds and turbulence in the stratosphere and mesosphere. J. Atmos. Sci., 31, 493-505.