A VHF boundary layer radar' First results

Transcription

1 Rado Scence, Volume 33, Number 4, Pages , July-August 1998 A VHF boundary layer radar' Frst results R. A. Vncent, S. Du!!away, A. MacKnnon, I. M: Red, and F. Znk Department of Physcs, Unversty of Adelade, Adelade, South Australa, Australa P. T. May Bureau of Meteorology Research Centre, Melbourne, Vctora, Australa B. H. Johnson Atmospherc Radar Systems, Adelade, South Australa, Australa Abstract. The development of a novel VHF radar desgned to measure wnds and temperatures n the planetary boundary layer s descrbed. The radar operates at 54.1 MHz and s compact and easly transportable. The antenna system conssts of 12 Yags grouped nto three subarrays arranged n the form of an equlateral trangle. Transmsson takes place on the whole array, and recepton takes place on the three subarrays, wth wnds measured by the spaced antenna technque over a heght range between 300 and 3000 m. Results from feld trals conducted n southern Australa n a varety of meteorologcal condtons are presented. Comparsons wth hgh-resoluton radosondes launched from the radar ste show excellent agreement, wth rms dfferences between radosonde and radar wnd components beng about 1.5 m s -1. Observatons carred out n ran show that echoes from precptaton are clearly dstngushable from clear-ar echoes. Unlke UHF radars, ths means that vertcal ar veloctes can be measured durng precptaton, and the evoluton of drop-sze dstrbutons can be studed down to low alttudes. It s shown that temperatures derved from a rado acoustc soundng system are measured up to heghts near 2 km, dependng on background wnd condtons. 1. Introducton Stratosphere-troposphere (ST) radars or wnd proflers are powerful tools for atmospherc research and operatonal meteorology [e.g., Gage, 1990]. Whle the ST radar feld s relatvely mature, development has proceeded n step wth advances n technology. In the last few years, the advent of cheap, powerful computers, together wth nexpensve computer memory for storng raw data and relable sold-state transmtters, have opened up new possbltes for ground-based radar studes of the atmosphere. Developments n conceptual technques nclude the use of nterferometry to mprove estmates of pontng Copyrght 1998 by the Amercan Geophyscal Unon. Paper number 98RS / 98 / 98RS drectons and correct such mportant parameters as vertcal ar velocty. The ultmate am s to derve as many atmospherc parameters as possble n real tme to ad research and forecastng. The last decade has seen the development and large-scale deployment of boundary layer (BL) proflers. These low-power systems have been desgned for the express purpose of observng many of the mportant meteorologcal phenomena that occur n the lowest part of the troposphere. Typcally, frequences near I GHz are used for UHF boundary layer radars. Examples are the wdely used radar system developed at NOAA's Aeronomy Laboratory whch uses a frequency of 915 MHz [e.g., Ecklund et al., 1988, 1990; Carter et al., 1995] and the L-band (1.357 GHz) system developed by the Kyoto Unversty group [e.g., Hashguch et al., 1995]. Systems such as these are beng used to nvestgate boundary layer wnd and temperature felds and, snce they are senstve to hydrometeors, the precptaton feld. 845

2 846 VINCENT ET AL.: A VHF BOUNDARY LAYER RADAR Radars operatng n the lower VHF band (- 50 MHz) have prmarly been used to measure wnds to as hgh an alttude as possble, whch requres the use of physcally large antennas and hgh transmtter powers. However, effects such as long recovery tmes n transmt-receve systems and reflectons or rngng n antennas and transmsson cables mean that these larger systems generally cannot make measurements n the lowest 1-2 km of the atmosphere. However, poneerng work by the French group n Toulon show that t s possble to use small antennas and make useful BL measurements at VHF [e.g., Crochet et al., 1994; Fllot et al., 1997]. Operatng at UHF offers many advantages. These nclude (1) the use of antennas that are physcally small but relatvely large compared wth the radar wavelength, so that narrow beams can be generated; (2) low external nose; and (3) the avalablty of wde bandwdths, whch means good heght resoluton. On the other hand, when precptaton or heavy cloud s present, echoes from hydrometeors domnate echoes from the clear ar, whch makes measurements of vertcal velocty dffcult. Ths s a partcular lmtaton when rado acoustc sound- lmtaton wth RASS at UHF s that the acoustc frequences ( KHz) suffer rapd attenuaton, whch lmts the upper heght to whch temperatures can be obtaned [May et al., 1988]. The advantages n operatng at VHF nclude the followng: 1. Precptaton echoes are comparable n strength to, or weaker than, the clear-ar echoes, and the precptaton and clear-ar echoes are usually well separated n the Doppler spectrum. Hence a sngle VHF radar can smultaneously study the dynamcs and cloud mcrophyscs of precptatng clouds [e.g., Rajopadhyaya et al., 1993]. 2. Measurement of the vertcal gradent of the vertcal velocty gves the dvergence of the wnd feld, especally n storm condtons when the vertcal motons and gradents are large. 3. VHF B L radars are unlkely to suffer the problems wth contamnaton by brd and bat echoes that UHF BL radars often encounter [Wlczak et al., 1995; May, 1995]. 4. A VHF/RASS boundary layer system should also have a better heght coverage than a smlar UHF system because the lower-frequency sound waves used suffer much less attenuaton. Thus a 50-MHz BL system has many advantages for the study of shallow weather systems, such as summertme cold fronts across southern Australa [Wlson and Stern, 1985]. Other applcatons nclude sea-breeze fronts whch trgger convecton. Fnally, there s much that we do not understand about scat- terng processes from the atmosphere. Measurements should be made at a range of frequences n the VHF and UHF bands to nvestgate these processes as well as explot the research and operatonal potental of all systems. The Atmospherc Physcs group at the Unversty of Adelade has been developng radars at MF/HF and VHF for studes of both the upper and lower atmospheres. A major desgn goal has been to develop systems that have as many features n common as possble. Here we descrbe a VHF (54.1 MHz) radar that has been developed snce 1994 for boundary layer studes. The am s to study wnds and temperatures n the lowest 2-3 km of the atmosphere wth good heght and tme resoluton and to extend the scope of boundary layer studes to nclude precptaton and vertcal ar motons n the presence ng systems (RASS) are used to measure atmospherc temperatures at UHF wth vertcal pontng beams, of precptaton. The overall system desgn s brefly snce the sound velocty should be corrected for the descrbed n secton 2, and some of the frst wnd and vertcal wnd [e.g., Angevne et al., 1994a]. Another RASS measurements are presented and dscussed n secton System Desgn A number of factors were taken nto account n the development of the Adelade VHF boundary layer (BL) radar. The prme requrements were that the system should be flexble n operaton, relatvely easy to transport, and power effcent so that t could be used at remote stes. A lower heght lmt of about 300 m was specfed, and t was requred to measure wnds regularly to heghts of 2-3 km n order to overlap wth observatons made wth a ST VHF radar located at the Buckland Park feld ste [Vncent et al., 1987]. To acheve these ams, the system shares a number of features common to other radar systems recently developed by our group n Adelade. In order to obtan hgh mean power t was decded to use a compact, modular, sold-state transmtter that s capable of a hgh duty cycle (10%). The modular nature of the transmtter means that transmt-receve anten- nas can be confgured n dfferent ways to allow dffer-

3 VINCENT ET AL.' A VHF BOUNDARY LAYER RADAR 847 ent wnd measurng technques (e.g., Doppler, spaced antenna, and nterferometrc) to be nvestgated and used as approprate. A recever and data acquston system (RDAS) was developed wth wde dynamc The arrangement s a compromse between makng the subarrays as small as possble to mnmze rngng and large enough to ensure an adequate poweraperture product. Three subarrays, each consstng range and,, ffc en ' + of memory (R. of,,, of four three-element Yags, - the temporary storage of raw data. The data are then downloaded to the host computer for on-lne n an equlateral trangle. The whole array s used for transmsson and each group s used for recepanalyss/permanent storage whle the next sequence ton. An equlateral trangle arrangement reduces of data s beng acqured. For operatonal flexblty, as many operatng parameters (e.g., recever gan, heght range, heght ncrement, etc.) as possble are placed under computer control. Ths enables a wde range of experments to be confgured n software and operated sequentally. System control and data analyses are undertaken wth a PC computer that s easly networked. It s therefore both smple and nexpensve to add more computng power as requred. Data are analyzed wth a propretary data analyss and dsplay package and stored on dsk for further off-lne processng. Raw data can also be stored for later analyss n other ways. the chances of systematc bases n SA wnd measurements, as can occur f other confguratons, such as a rght-angle trangle, are used. The optmum spacng of antennas n SA mode s when the mean cross correlaton at zero lag s about 0.5 [Brggs, 1984]. Prelmnary trals showed that the basc spacng should be at least 1.5A (8.32 m), and ths was chosen as the ntal spacng. A reevaluaton of the spacng took place after the radar had operated n a varety of meteorologcal condtons, as dscussed n secton 3.1. The confguraton adopted ensures that gratng lobes on transmsson are not sgnfcant. Fgure 2 shows plots of the one-way polar dagrams for the recevng and transmttng arrays. The transmttng 2.1. Antenna Confguraton The low-heght performance of 50-MHz ST systems s lmted by the use of large antenna arrays and problems assocated wth rngng after the transmtter pulse, whch hampers recever recovery. Thus a VHF BL radar requres small antennas, whch mples wde beam wdths, so that ground clutter and array has a half-power half wdth of 10 ø, correspondng to an effectve area of about 250 m, whle the respectve values for the recevng antennas are 18 ø and about 80 m. However, as May [1990] has shown, the overall effectve area of the system depends on the scatterng mechansm. For the stuaton of sotropc volume-scatter the effectve area s gven by nterference to and from other operators may be a 2ATA problem. Fast transmt-receve swtches must be used and antenna rngng mnmzed to ensure recep- Aeff - (AT + Ae) (1) ton at low heghts. It was decded to utlze a spaced where AT and As are the respectve areas for the antenna (SA) confguraton so that only vertcally transmttng and recevng antennas. In ths case, pontng beams are used. Ths arrangement s more Aeff 120 m. One problem wth such relatvely compact than Doppler systems, and t elmnates the heght-smearng problem that occurs when oblquely pontng wde beams are used for Doppler soundng wde beams s the possble effects of ground clutter, but by sutably spacng and orentatng the Yag antennas t s possble to partly null out clutter and [Fllot et al., 1997] as well as potental problems assonterference that arrve at low elevatons. cated wth ansotropc backscatter whch complcate wde-beam Doppler measurements [RSttger, 1984]. Another potental advantage of the SA confguraton 2.2. Operatng Parameters Table I summarzes the range of possble operatng s that the data can be used for nterferometrc stud- parameters. In the evaluaton of the system a 1-/ s es as well as conventonal spaced antenna wnd measurements. Interferometrc technques can then be used to study the scatterng rregulartes and have the potental to correct for possble contamnaton of the vertcal velocty measurements caused by tlted layers [Vncent and RSttger, 1980]. Fgure I shows the antenna confguraton adopted for ntal measurements wth the VHF BL system. pulse length was used to acheve the desred 150-m range resoluton, and a 2-MHz recever bandwdth was selected to ensure that fast recever recovery occurs and the desred lower heght lmt of about 300 m s attaned. A partcular feature of the system s the hgh degree of computer control of the system confguraton, such as heght coverage, range resoluton, and so on. If necessary, a number of dfferent ex-

4 $45 VINCENT ET AL' A VHF BOUNDARY LAYER RADAR VHF BOUNDARY LAYER WITH RAss LAYOUT YAGI C YAGIB YAGIA 8.32m [,, (m) RASS SPEAKER (Approx. locaton) Fgure 1. Plan vew of VHF boundary layer radar antennas. perments can be programmed to take place n any desred sequence. Ths s partcularly useful whle, as at present, the optmum operatng parameters for SA wnd and RASS measurements were beng determned. Intal tests suggest that n the spaced antenna mode the operatng parameters shown n Table 2 gve satsfactory performance. These values gve a lower detectable wnd speed of 01 m s - and an upper lmt of about 80 m s -. Wth these settngs the mean power s about 20 W, gvng a power-aperture product of at least 2500 W m2o The complex ampltudes receved at the three antennas are analyzed n the standard method usng the full correlaton analyss (FCA) dscussed by Brggs [1984]. The qualty of the data was assessed usng the standard crtera descrbed by Brggs [1984], although only data wth sgnal-to-nose ratos (SNR) greater than 0 db (after coherent ntegraton) were accepted, as modelng studes show that the accuracy of SA wnd determnatons decreases rapdly for smaller sgnal-to-nose ratos [Holdsworth, 1995]. 3. Intal Results Intal tests of the BL system were carred out at the Unversty of Adelade's Buckland Park feld staton (35øS, 138øE), where t s located mmedately adjacento a VHF ST profler [Vncent et al., 1987]. The ste s very fat and upwnd of any sgnfcant topography. However, the ste s rather crowded wth buldngs, antenna masts, etc., and there s the potental for sgnfcant ground clutter, whch can make the detecton of the clear-ar echoes dffcult. Fortu- nately, clutter has not proved to be a major problem, except at the lowest range gates of m, and s easly dstngushed from atmospherc echoes by ts characterstc slow fadng. A number of algorthms are beng tested to reduce the effects of ground clutter and ntermttent clutter due to arcraft, ncludng the use of waveletransforms [Jordan et al., 1997]. In order to assess the performance of the radar, a seres of ntercomparsons were made durng the perod July 31 to September 26, 1997, wth smultaneous measurements made wth hgh-resoluton ra-

5 VINCENT ET AL' A VHF BOUNDARY LAYER RADAR (a) Table 1. Operatng Parameters of the VHF Boundary Layer Radar Parameter Value Zenth Angle Transmtter power Transmt/receve swtch Antenna bandwdth Recever bandwdth* Pulse length* Maxmum duty cycle Sample heght* Pulse repetton frequency* Coherent ntegraton* Dgtzers Sgnal averagers Memory 3 x 350 W peak envelope power passve,.. l-ps recovery tme -4 MHz 2 MHz to 250 khz ps 10% 100-m mnmum, 25-m ncrements up to 50 khz up to 4096 ponts 12-bt 24-bt I Mbyte per recever -! (b) *Software selectable. -2o -$0 I / / \ / \ / / 31 and August 1, campagn II took place between September 12 and 15, and the fnal campagn occurred on September 26, Results from all three campagns were used to evaluate the performance of the radar, such as accuracy of wnd measurements. To llustrate the capabltes of the system, we focus on some of the results from the frst two campagns, whch were conducted under very dfferent weather condtons. -40 Table 2. Typcal Spaced Antenna Parameters Used n Campagns - 0 I o Zenth Angle Fgure 2. Antenna polar dagrams. The sold lnes ndcate the E plane, and the dashed lnes ndcate the H plane. (a) Recevng antenna. (b) Transmttng antenna. dosondes (Vasala RS80-15) launched from Buckland Park. Three campagns were carred out n a varety of meteorologcal condtons, ncludng sgnfcant precptaton. Campagn I occurred between July Parameter Parameter Pulse repetton frequency* Number of coherent ntegratons* Number of data ponts per sample* Heght coverage* Record length* Range resoluton* Maxmum lag* Sample rate *Software selectable. Value Value 20,480 Hz 1,024 1, m 52.4 s 150 m 4s 0.05 s

6 ß 850 VINCENT ET AL.: A VHF BOUNDARY LAYER RADAR 3.1. Radar and Radosonde Wnd Comparsons A prmary goal of the campagns was to assess the accuracy of the radar wnd measurements by comparng them wth radosondes equpped wth medum- to hgh-resoluton wnd-fndng systems. For campagn I a Global Postonng System (GPS) trackng system was used, whle for the campagns startng on September 12 and 26, the balloons were tracked us- ng VLF Omega navgaton sgnals. For each flght the radar wnds at each 150-m range gate were averaged over a perod of 20 mn, centered on the tme at whch the balloon had reached an alttude of I km. At least three radar observatons were requred at a gven level to produce a mean value. No consensus averagng [Strauch et al., 1984] was used n producng the mean wnd veloctes. In all, there were 29 flghts and 144 samples. Fgure 3 shows scatterplots of the zonal (u) and merdonal wnd (v) components. It s apparent that the observatons are hghly correlated (r- 0.95), wth the ponts dstrbuted reasonably evenly around the lne of unt slope. Only n the case of the v component are there any sgnfcant outlers, wth sx ponts lyng about 5 m s -1 away from the lne. These ponts came from two consecutve soundngs n whch there appeared to be sgnfcant curvature n the wnd feld. Table 3 summarzes the results of the comparsons. There are two lnes for each wnd component, wth the second lne correspondng to the case where the two soundngs mentoned prevously were dscounted. Whether these sound- ngs are ncluded or not, the comparsons are very good, wth rms dfferences between the radar and radosonde wnds beng about 1.5 m s -1. These results compare very favorably wth the results from smlar studes. Many earler comparsons were between profler and radosonde release stes separated by many tens of klometers, and so some of the dfferences could be ascrbed to spatal varablty. For example, Vncent et al. [1987] compared 50-MHz ST profler wnds measured at Buckland Park wth radosondes released from Adelade lo -20 2O (a)., Radosonde (ms") -20 (b) +q-, -lo o lo 20 Radosonde (ms") Fgure 3. Scatterplot of wnd veloctes measured by radar and radosondes. (a) zonal component and (b) merdonal component. Lnes of unt slope are shown. Arport, stuated some 36 km to the south. They found rms speed dfferences n the 2-6 km heght range of 3-4 m s-1. The most extensve ntercomparson usng collocated soundngs s that of Weber and edtng ther data to remove outlers, they found rms Wuertz [1990], who compared 915-MHz profler and component dfferences of about 2.5 m s -1. radosonde observatons conducted at Stapleton Ar- There are fewer comparsons of boundary layer port n Denver, Colorado, over a 2-year perod. After measurements. Hashguh et al. [1995] report rms -

7 VINCENT ET AL' A VHF BOUNDARY LAYER RADAR 851 Table 3. Statstcs of Radosonde/Profler Comparsons for Zonal (u) and Merdonal (v) Wnd Components Correlaton Mean Dfference, ms -1 Intercept, ms -1 Slope RMS, ms (0.96) (-0.10) (0.74) (0.83) (1.4) (0.98) (-0.13) (0.26) (0.94) (1.3) Profler observatons are averaged over 20 mn from the tme of launch of the radosondes, and comparsons were made at each radar range bn whenever there were three or more radar observatons n that bn. There were a total of 29 profles for each nstrument. The ntercept and slope refer to the weghted least squares ft straght lne to the scatterplots shown n Fgure 3, whle the rms values ndcate root-mean-square dfference between radosonde and radar data. The values n parentheses refer to comparsons wth two flghts removed (see text). component dfferences of better than 3 m s -1 when they compared L-band boundary layer profler wnds wth wnds derved from radosondes released from the radar ste. An extensve comparson usng Trop- cal Ocean-Global Atmosphere/Coupled Ocean-Atmosphere Response Experment (TOGA-COARE) data taken wth 915-MHz proflers and radosondes reveals rms dfferences of less than I m s -1 [Rddle et al., 1996]. Angevne and MacPherson [1995] compared 915-MHz profler wnd measurements wth smultaneous arcraft observatons and found rms dfferences of less than I m s -1. The dfferences are ste dependent. For example, May [1995] found dffer- ences of ~2-3 m s -1 n a hghly convectve boundary layer durng the day n central Australa. Comparsons were even worse at nght because of scatterng from bats. The excellent agreement between the wnd veloctes measured here by the two technques probably reflects the low spatal varablty nherent n ths comparson. As noted, Buckland Park s stuated on the fat Adelade Plans upwnd of any sgnfcant topography, so spatal varablty was mnmzed as the balloons drfted away from the ste. It should be noted, however, that ths comparson gnores the errors nherent n the radosonde soundngs. One fac- tor that s often gnored n these comparsons s the smoothng requred to reduce the effects of the pendulum moton of the sonde package. Ths smoothng acts as a low-pass flter that reduces the ampltude of short-vertcal-wavelength wnd varatons. The length of the flter n the GPS-sonde measurements was 60 s and was 240 s n the Omega-sonde wnd determnatons. Gven the ~4 m s -1 ascent rate of the balloons, the effectve heght resoluton of the radosondes s several hundred meters. Ths reduces the small-scale wnd varatons relatve to those ob- served by the radar. Some of the Omega-sonde data were reprocessed wth a shorter flter length of 150 s to see how sgnfcant ths effect was. The rms devaton from the wnd speeds processed wth the normal 240-s flter was about m s -1. As expected, the devatons were largest where the wnd curvature was greatest. There are a number of factors that nfluence the ac- curacy of radar wnd measurements usng the spaced antenna technque. As noted above, the optmum antenna spacng s where the spatal correlaton s about 0.5. Ground-pattern statstcs obtaned from the full correlaton analyss (FCA) were used to test how well ths was satsfed wth the present arrangement, as well as to provde other useful nformaton on the scatterng rregulartes. In the FCA, the ground dffracton pattern s modeled as a seres of concentrc ellpses of constant correlaton, p. The sze and orentaton of the characterstc ellpse for whch p are convenent measures of the average pattern scale and orentaton. Pattern-scale statstcs obtaned from spaced antenna measurements n the three campagns are summarzed n Table 4. The most probable value of 8.5 m s dentcal to the an- tenna spacng of 8.3 m (1.5,k), whch shows that the spacng was close to optmal. Table 4 also shows another useful pattern parameter, the axal rato, whch

8 852 VINCENT ET AL.' A VHF BOUNDARY LAYER RADAR Table 4. Characterstcs of Ground Dffracton Pattern lar dagrams as a functon of s - snt, respectvely, then Most Probable Mean Pattern scale, m Axal rato s a measure of the elongaton of the characterstc ellpse. The most probable and mean values are both less than 1.5, whch n practce means that the average pattern s not sgnfcantly dfferent from crcular [Wrght and Ptteway, 1978]. Profles of the pattern scale wth heght, such as those gven n Fgure 4 for the frst two campagns, suggest that there s a small ncrease n pattern scale wth ncreasng heght. As the pattern scale and the angular spectrum or wdth of the backscattered radaton are a Fourer transform par [e.g., Brggs, 1992], ths ndcates that there s a narrowng of the angular spectrum wth heght and hence a change n the aspect senstvty of the scatterng rregulartes. In order to make a quanttatve estmate of the wdth of the angular spectrum t s necessary to make some assumptons about the angular dependence of the antenna polar dagrams and of the angular spectra. The almost crcular nature of the correlaton functons suggests that energy s backscattered almost unformly from around the zenth and s just a functon of the zenth angle T. Assumng that the antenna polar dagrams and angular spectrum have a Gaussan dependence and that the rregulartes are confned to narrow layers, then the relatonshp between the spatal correlaton functon and angular spectrum has the smple form gven by I I I 1 = + + (6) In the present case, T " = 12ø and TR = 21.8 ø, so, from the measured values of pattern scale, P0.5, t s straghtforward to estmate Tz, the angular wdth of the atmospherc echoes. The scale at the top of Fgure 4 gves the converson of the pattern scale to angular wdth. An nfnte value corresponds to sotropc scatter. The rather narrow angular wdths observed n the lowest range gates are probably caused by ncomplete removal of slow fadng clutter. However, the decrease n Tz wth heght s most lkely real. There are at least two reasons for the effect. Frst, the backscatterng rregulartes may be more aspect senstve n the more stable free atmosphere above the well-mxed boundary layer. Second, t s possbly a selecton effect, as rregulartes whch scatter energy preferentally from overhead are more lkely to be observed by the vertcally pontng radar at longer ranges, where the SNR are smallest oo Angular Wdth (deg). p(r) - exp(-ws or2), (2) where So - snto and r s the antenna separaton n radar wavelengths [Brggs, 1992]. If a(s) (x exp(-s2/s ), (3) ,! ' x, GT(S) (x exp(--s /s ), (4) lo Pattern o Scale (m) G.a(s) (x exp(-s2/s ), (5) descrbe the angular spectrum s of the rregulartes and GT and G s are the transmtter and recever po- Fgure 4. Vertcal profles of mean pattern scale for campagn I (sold lne) and campagn II (dashed lne). The scale at the top gves the angular wdth of the backscattered echoes.

9 VINCENT ET AL.: A VHF BOUNDARY LAYER RADAR 853 3OOO! /! ß OO O I, /,, I,,,, I... I,, Theta (K),,, I,,, I,, q (g kg') Fgure 5. Vertcal profles of (left) potental temperature and (rght) specfc humdty derved from radosonde soundngs at 1206 LT (sold lne) and 0206 LT (dashed lne) Campagn I Durng ths perod a strong hgh-pressure regon (~1035 hpa) was centered over the ste. The BL observatons were nterleaved wth observatons wth another radar also undergong tests at Buckland Park, so the effectve sample perod was 2 mn. Eleven radosondes were launched at ntervals of approxmately 3 hours. Day and nght profles of potental temperature and specfc humdty q are shown n Fgure 5. There s a marked subsdence nverson clearly evdent n all profles. The daytme boundary layer s reasonably well mxed but contans a localzed stable layer at a heght of about 800 m and a correspondng maxmum n specfc humdty. The nghttme profle shows a resdual mxed layer overlyng a typcal ground-based radaton nverson. Fgure 6 shows the tme-heght cross secton of the sgnal-to-nose ratos (SNR). Values were smoothed over 30 mn to reduce some of the short-term varablty. There s a clear relatonshp between the strong subsdence nverson and the layer of enhanced reflectvty, analogous to the results of Angevne et al. [1994b] wth a 915-MHz profler. Ths example s partcularly clear compared wth typcal 915-MHz estmates of boundary layer depth, but t s not evdent f t s the result of the dffer- ence n radar frequency or of the ntense subsdence nverson. The nghttme profles show the growth of a nocturnal nverson at alttudes below 400 m and a fossl mxed layer (resdual layer) that s very evdent n q at heghts below 1200 m. Durng the nght, there s no heatng from below to produce turbulent fluctuatons of temperature and mosture (and hence scatterng rregulartes) n the fossl mxed layer, a condton exacerbated by the lght wnds SNR (db) 31 July- 1 August LT Fgure 6. Tme-heght cross secton of 30-mn average sgnal-to-nose ratos observed n campagn I O

10 ß _! VINCENT ET AL.: A VHF BOUNDARY LAYER RADAR Meteorologcal parameters recorded durng the radosonde ascents can be used to gan some nsght nto radar performance. The receved power Pn s related to the range R and radar reflectvty r/by I ' ' ' < (a) LT Fgure 8. Half-hour average horzontal wnds measured between July 31 and August 1, A half barb represents 2.5 m s -, and a full barb represents 5ms , I, I...-I r--'r 7' :..-'! SNR (db) (b) SNR (db) Fgure 7. Vertcal profles of reftactvty from radosondes (sold lne) and radar (dashed lne) measurements for (a) 1200 LT July 31, 1997, and (b) 0206 LT August 1, The reflectvty s related to the turbulent refractve ndex C, whch, n turn, can be expressed n terms of the outer scale of turbulence and the gradent of rado refractve ndex, M, where M -77'610-6 p Oz (8) [ q T ( 1 20lnO/(Oz) 101nq/(Oz))] and z s heght [VanZandt et al., 1978]. Vertcal profles of M 2 were constructed from profles of temperature T, pressure p (measured n hectopascals), potental temperature q, and specfc humdty. The observatons were made wth 2-s tme resoluton, equvalent to approxmately 8-10 m heght resoluton. Fgure 7 compares SNR values n decbels observed by the radar and estmated from the radosonde soundngs durng both day and nght condtons. The range-corrected reflectvty profles constructed from the sonde values have been smoothed to gve a heght resoluton comparable to the radar and have been ncreased by an arbtrary factor of 175 db to brng them nto approxmate concdence wth the radar

11 VINCENT ET AL.' A VHF BOUNDARY LAYER RADAR SNR (db) September 1997,.o ß - ; :.; {- -:.':./.".:.....,..... : Fgure 9. S e O0 I O0 1 O0 I LT as for Fgure 6, bu for campagn II. values. It s evdent that the profles agree rather well n shape and ndcate that the tme delays n the radar system were calbrated reasonably correctly. Both the sonde and radar profles are characterzed by strong peaks at heghts near 1500 m and very weak refectvtes at nght at heghts between 500 and 1200 m. The correspondng horzontal wnd feld s shown n Fgure 8. The wnds were very lght, rarely exceedng 5 m s -1. No wnds were recorded n the lowest range gate at 300 m due to slower recovery n one of the recevers. However, wnds were frequently measured at the second lowest range gate at 450 m, and observatons extended to heghts near or above 2 km. Notceable gaps n the wnd measurements occurred durng perods centered on 0000 and LT on August 1. In the latter nterval, no wnds were measured at heghts near 0.5 km, despte the fac that the SNR was near 10 db. The cause was the presence of stronger than average slow fadng ground clutter supermposed on the atmospherc echoes. Ths emphaszes the need for mproved clutter suppresson. The gap centered on mdnght was caused by SNR values that fell as low as-20 db n the heght range between 600 and 1500 m Campagn II The second case study centered on the passage of a cutoff low across the radar ste. A tme-heght cross secton of SNR s shown n Fgure 9, and Fgure 10 shows the wnd feld, where ts vortex nature s 2O clearly evdent. The SNR plot shows lttle evdence of the layerng so evdent n Fgure 6, and ths s' also evdent n profles of potental temperature and humdty (Fgure 11), whch show a much smoother varaton wth heght than s evdent n the profles n Fgure 5. The hgh SNR means that there were often suffcent useful echoes to produce wnds at most tmes at heghts between 450 m and 3000 m. Durng ths campagn, qute heavy ran fell on occason. The ran gave the opportunty to test the capablty of the profler to dstngush ran echoes from the sgnals scattered from the clear ar despte the expected large amount of spectral broadenng due to the wde beam wdth. That ran echoes can be detected s llustrated n Fgure 12, whch shows stacked plots of Doppler spectra normalzed to the strongest sgnal n each spectrum. The spectra, whch were constructed by averagng nne spectra obtaned n a 16-mn nterval, show strong precptaton echoes wth downward veloctes n the range 5-10 m s -1 n the heght range between 500 and 2000 m. At ths tme the precptaton system was qute shallow. Satellte measurements of cloud top temperatures over Adelade at ths tme were only about -10øC, wth deeper cloud about 300 km to the east. The mportant pont s that the clear-ar echo can be used to measure the vertcal ar velocty as a functon of heght. The ablty to dstngush the clear-ar and precptaton echoes s also an mpor- 3.0 " (' / Fgure 10. pagn II. O0 12 O0 12 O0 12 LT Same as for Fgure 8, but for cam-

12 856 VINCENT ET AL.' A VHF BOUNDARY LAYER RADAR 3000! ' ' '......,],,,,/ 25OO 2r o 1000 \ \ \ \ \ /! 285 /, Theta (K) q (g kl ') Fgure 11. Velocty profles of (left) potental temperature and (rght) specfc humdty derved from soundngs on September 13 at 1057 LT (sold lne) and 2348 LT (dashed lne). o.oo tant factor n the deconvoluton procedures requred to deduce drop-sze dstrbutons [Rajopadhyaya et al., 1993]. All these factors llustrate the capablty of the VHF profler for observng mesoscale detal wthn weather systems, ncludng vertcal wnd measurements n ran. mprove the sgnal-to-nose ratos, whch means that 3.4. RASS Measurements the tme resoluton of the temperature retrevals s 4- In common wth other profler-based RASS systems, the Adelade BL RASS uses a FM CW acoustc exctaton to ensure that a range of temperatures 5 mn. The spectral parameters for the acoustc echo are computed by the moment method, and the vertcal velocty derved from the clear-ar echo s used can be measured [May et al., 1990]. Pseudorandom, to correct the sound speed before t s converted to sawtooth, and trangular sweep types are software se- vrtual temperature [e.g., Angevn et al., 1994a]. lectable, together wth approprate bandwdths and dwell tmes to ensure a good match at the Bragg frequency [May et al., 1990; Angevn et al., 1994a]. The use of fast dgtzers, and lmted amounts of coherent ntegraton gve an effectve sample rate of 320 Hz and ensure a Nyqust frequency sgnfcantly hgher than the acoustc echo. The power spectra computed from the 1024 pont samples at each heght encompass both the acoustc-echo peak near 120 Hz and the clear-ar echo near 0 Hz. A number of dfferent modes of operaton of the RASS have been evaluated. The acoustc source, whch s free runnng to ensure a fat spectrum over the range of operaton [Angevne et al., 1994a], s fed to a 1-kW audo amplfer whch drves aver- tcally pontng stadum horn. Table 5 summarzes the mode of operaton currently n use. Data are recorded for 6.4 s and then transferred to the com- puter; allowng for data transfer and power spectral analyss, the effectve sample tme s about 10 s. Usually, about spectra are ncoherently summed to Comparsons wth RASS measurements and vrtual temperatures computed from the temperature and humdty profles derved from collocated radosonde soundngs are shown n Fgure 13. RASS measurements were made over a perod of mn near the tme of the release of the sondes, and the values shown are the result of averagng four to fve data sets so that the bars gve an dea of the degree of varablty n the retreved temperatures. It s apparent that the agreement s excellent. The plots also llustrate that temperatures can be measured from heghts as low as 500 m to as hgh as at least 2500 m. However, they also llustrate some of the dff- cultes that can be encountered wth the present expermental setup. Bauer and Peters [1993] and May

13 ,, VINCENT ET AL.' A VHF BOUNDARY LAYER RADAR 857 o 13.. "o E o z Doppler Spectra, I,, I,,,,,,,, I, Velocty (m/s) Relatve Power I I I I I I I I 11 ; 15 \' Illllflllllllllllllllllllllltllllllllll db E Fgure 12. Stacked Doppler spectra obtaned between 0400 and 0416 LT on September 14, At each heght the frst moment of the clear-ar spectral peak, ndcated by a plus sgn, gves the vertcal velocty. The dashed lne shows the poston of the peak of the precptaton echo. The vertcal profle of relatve echo power of the clear ar peak s shown n the rght panel ? ular n nature so that t can be operated to use a range of wnd-measurng technques. A notable feature s that very useful wnd and temperature measurements were obtaned wth qute small antennas Ol-... bl-klll llll loll : : llill -' 1 2G 2 pt on. k_., antennas a compromse between wantng to make them as small as possble to mnmze rngng and thus acheve the desred lower heght lmt and makng them large enough to obtan suffcent gan to attan the desred upper heght lmt and to reduce clutter and nterference. Intal measurements made wth the new bound- ary layer VHF radar show that ts performance meets the desgn specfcatons, wth wnd and temperature measurements beng made wth good accuracy on a regular bass to over 2 km n alttude. The ablty to dstngush ran echoes from the clear-ar echo has also been demonstrated. Ths s an mportant attrbute, whch wll allow the study of precptaton drop-sze dstrbutons to be extended downward nto the boundary layer. Overall, the performance compares well wth UHF proflers operatng near I GHz at a smlar stage of ther development [Ecklund et el., 1990; Heshguch et el., 1995]. The next stage s to mprove the heght coverage. Wnd and temperature measurements are made reg- Table 5. Typcal Rado Acoustc Soundng System Operatng Parameters Parameter Value et el. [1996] have shown that the RASS sgnal s focused onto a dffracton-lmted spot approxmately the sze of the transmttng antenna. The small antennas used here mean that the acoustc spot s also small and s easly advected out of the radar beam by the background horzontal wnds. It was necessary to carefully place the speaker upwnd of the transmttng array to ensure the good heght coveragevdent n Fgure 13a. Future developments nclude multple acoustc sources. 4. Conclusons Here we have dscussed the development of a VHF (50 MHz) boundary layer radar and llustrated ts capabltes wth some of the frst results obtaned wth the system. The system s compact and rood- Pulse repetton frequency* Number of coherent ntegratons* Number of data ponts per sample* Heght coverage* Record length Range resoluton* Nyqust frequency Acoustc frequency range* Acoustc power Acoustc exctaton* Acoustc steps* Frequency step* Dwell tme (per frequency step)* *Software selectable. 20,480 Hz 64 2, m 6.4s 150 m 160 Hz Hz FM W CW saw- tooth sweep 2OO 0.04 Hz 100 ms

14 . = = = 858 VINCENT ET AL.' A VHF BOUNDARY LAYER RADAR 3.0 (a) Temperature ('C), 0.0 I,,,, I,,,, I... I,,, I, Temperature ('C) Fgure 13. (a) Profle of vrtual temperature derved from a radosonde soundng at 1530 LT on October 18, Rado acoustc soundng system measurements made from 1500 to 1515 LT are ndcated by damonds. (b) Same as for (a), but for 1750 LT on October 19, 1997 wth RASS measurements made from 1732 to 1805 LT. ularly to heghts as low as 450 m and sometmes as low as 300 m. Recent testng shows that t s possble to get to lower heghts by usng pulse lengths of 0.75 ps and explotng the capacty of the system to make observatons wth a 100-m range resoluton. Smlarly, the upper heght lmt s mproved by ncreasng the mean power, ether by usng pulse codng or usng longer pulses to gve a 300-m resoluton. Usng 300-m pulses, t has been possble to acheve observatons as hgh as 5000 m. It should also be possble to use somewhat larger antennas for both transmsson and recepton, whch wll mprove the power-aperture product and hence over-all system senstvty. Tests of the system wth these mprovements are now n progress, as are tests usng nterferometrc technques to better defne the domnant drectons of scatter and correct the vertcal veloctes. Acknowledgments. The development of the radar was made possble by the support of grant A from the Australan Research Councl. The help and provson of radosonde equpment by J. Stckland, Australan Bureau of Meteorology, and M. J. Reeder, Monash Unversty, are gratefully acknowledged. ' References Angevne, W. M., and J. I. MacPherson, Comparson of wnd profler and arcraft wnd measurements at Chebogue Pont, Nova Scota, J. Atmos. Oceanc Technol., 12, , Angevne, W. M., W. L. Ecklund, D. A. Carter, K. S. Gage, and K. P. Moran, Improved rado acoustc soundng technques, J. Atmos. Oceanc Technol., 11, 42-49, 1994a. Angevne, W. M., A. B. Whte, and S. Ko Avery, Boundary layer depth and entranment zone characterzaton wth a boundary layer profler, Boundary Layer Meteorol., 68, , 1994b. Bauer, M., and G. Peters, On the alttude coverage of temperature proflng by RASS, n 26th Interna- tonal Conference on Radar Meteorology, pp , Am. Meteorol. Soc., Boston, Mass., Brggs, B. H., The analyss of spaced sensor records by correlaton technques, n Handbook for MAP, vol. 13, edted by R. A. Vncent, pp , SCOSTEP, Secr. Unv. of Ill., Urbana, Brggs, B. H., Radar measurements of aspect senstvty of atmospherc scatterers usng spacedantenna correlaton technques, J. Atmos. Sol. Terr. Phys., 5, , 1992.

15 VINCENT ET AL.: A VHF BOUNDARY LAYER RADAR 859 Carter, D. A., K. S. Gage, W. L. Ecklund, W. M. Angevne, P. E. Johnston, A. C. Rddle, J. Wlson, and C. R. Wllams, Developments n UHF lower tropospherc wnd proflng at NOAA's Aeronomy Laboratory, Rado $c., 30, , Crochet, M., C. Bourder, and E. Spano, VHF ST radar for low alttude nvestgatons, n Proceedngs oj the 6th Workshop on Techncal and Scentfc Aspects oj MST Radar, edted by B. Edwards, pp , SCOSTEP, Secr. Unv. of Ill., Urbana, Ecklund, W. L., D. A. Carter, and B. B. Balsley, A UHF wnd profler for the boundary layer: Bref descrpton and ntal results, J. Atmos. Oceanc Technol., 5, , Ecklund, W. L., D. A. Carter, B. B. Balsley, P. E. Currer, J. L. Green, B. L. Weber, and K. S. Gage, Feld tests of a lower tropospherc wnd profler, Rado $c., 25, , Fllol, J.-M., P. Broche, and M. Crochet, Retreval of stratfed atmospherc reflectvty and wnd velocty usng nverse methods: Applcaton to a VHF ST mn-radar, J. Atmos. Sol. Terr. Phys., 59, , Gage, K. S., Radar observatons of the free atmosphere: Structure and dynamcs, Radar n Meteorology, edted by D. Atlas, pp , Am. Meteorol. Soc., Boston, Mass., Hashguch, H., S. Fukao, T. Tsuda, M.D. Yamanaka, D. L. Tobng, T. Srbmat, S. W. Harjono, and H. Wryosumarto, Observatons of the planetary boundary layer over equatoral Indonesa wth an L-band clear-ar Doppler radar: Intal results, Rado Sc., 30, , Holdsworth, D. A., Sgnal analyss wth applcatons to atmospherc radars, Ph.D. thess, 371 pp., Unv. of Adelade, Adelade, South Aust., Australa, Jordan, J. R., R. J. Latats, and D. A. Carter, Removng ground and ntermttent clutter contamnaton from wnd profler sgnals usng wavelet transforms, J. Atmos. Oceanc Technol., 1, , May, P. T., Spaced antenna versus Doppler radars: A comparson of technques revsted, Rado $c., 25, , May, P. T., The Australan nocturnal jet and durnal varatons of boundary-layer wnds over Mt. Isa n north-eastern Australa, Q. J. R. Meteorok $oc., 121, , May, P. T., R. G. Strauch, and K. P. Moran, The alttude coverage of temperature measurements usng RASS wth wnd profler radars, Geophys. Res. Left., 15, , May, P. T., R. G. Strauch, K. P. Moran, and W. L. Ecklund, Temperature soundng by RASS wth wnd profler radars: A prelmnary study, IEEE Trans. Geosc. Remote $ens., 28, 19-28, May, P. T., T. Adach, T. Tsuda, and R. J. Latats, The spatal structure of RASS echoes, J. A tmos. Oceanc Technol., 13, , Rajopadhyaya, D., P. T. May, and R. A. Vncent, A general approach to the retreval of ran dropsze dstrbutons from VHF wnd profler Doppler spectra: Modellng results, J. Atmos. Technol., 10, , Oceanc Rddle, A. C., W. M. Angevne, W. L. Ecklund, E. R. Mller, D. B. Parsons, D. A. Carter, and K. S. Gage, In stu and remotely sensed horzontal wnds and temperature ntercomparsons obtaned usng ntegrated soundng systems durng TOGA COARE, Betr. Phys. Atmos., 69, 49-61, RSttger, J., The MST radar technque, n Handbook for MAP, vol. 13, edted by R. A. Vncent, pp , SCOSTEP, Secr., Unv. of Ill., Urbana, Strauch, R. G., D. A. Merrtt, K. P. Moran, K. B. Earnshaw, and D. Van de Kamp, The Colorado wnd proflng network, J. Atmos. Oceanc Technol., 1, 37-49, VanZandt, T. E., J. L. Green, K. S. Gage, and W. L. Clark, Vertcal profles of refractvty turbulence structure constant: Comparson of observatons by the Sunset radar wth a new theoretcal model, Rado $c., 13, , Vncent, R. A., and J. RSttger, Spaced antenna VHF radar observatons of tropospherc veloctes and rregulartes, Rado $c., 15, , Vncent, Ro A., P. To May, W. K. Hockng, Wø G. Elford, B. H. Candy, and B. H. Brggs, Frst results wth the Adelade VHF radar: Spaced antenna studes of tropospherc wnds, J. Atmos. Sol. Terr. Phys., 9, , Weber, B. L., and D. A. Wuertz, Comparson of raw- nsonde and wnd profler radar measurements, J. Atmoso Oceanc Technol., 7, , Wlczak, J. M., et al., Contamnaton of wnd profler data by mgratng brds: Characterstcs of corrupted data and potental solutons, J. Atmos. Oceanc Technol., 12, , 1995.

16 860 VINCENT ET AL.: A VHF BOUNDARY LAYER RADAR Wlson, K. J., and H. Stern, The Australan summer- of Adelade, Adelade, SA 5005, Australa. (e-mal: tme cool change, 1, Synoptc and sub-synoptc as- pects, Mon. Weather Rev., 113, , B. H. Johnson, Atmospherc Radar Systems, 1/12 Wrght, J. W., and M. L. V. Ptteway, Computer sm- Sterlng St., Thebarton, SA 5031, Australa. (e-mal: ulatons of onospherc radar drft measurements b com.au) and ther analyss by correlaton methods, Rado P. T. May, Bureau of Meteorology Research Cen- $c., 13, , tre, PO Box 1289K, Melbourne, VIC 3001, Australa. (e-mal: p.may@bom.gov.au) S. Dullaway, A. MacKnnon, I. M. Red, R. A. Vncent, and F. Znk, Department of Physcs, Unversty (Receved December 17, 1997; revsed March 5, 1998; accepted March 10, 1998.)