1. INTRODUCTION the Wall Cloud Is a Lowering of Cloud Base Associated

Home , Cloud base, Inflow (meteorology), Rear flank downdraft, Tornadogenesis, Wall cloud

4A.2 OBSERVATIONS OF WALL CLOUD FORMATION IN SUPERCELL THUNDERSTORMS DURING VORTEX2

Nolan T. Atkins, Eva M. Glidden, and Timothy M. Nicholson Lyndon State College, Lyndonville, Vermont 05851

1. INTRODUCTION formation is based on idealized modeling results of Rotunno and Klemp (1985) and visual observations. The wall cloud is a lowering of cloud base In their idealized simulation of a supercell, Rotunno associated with the updraft of a thunderstorm. The and Klemp (1985) showed that the air entering the focus of this study concerns wall clouds formed storms updraft region where the wall cloud formed within supercell thunderstorms (Bluestein 1983; was being ingested from behind the gust front. This Davies-Jones 1986; Bluestein 1993). Early obser- rain-cooled air had descended from mid levels with- vational studies suggested that the supercell wall in the storm and subsequently saturated below the cloud is the visual indicator of a strong updraft core primary cloud base as it ascended with the low-level and may exhibit cyclonic rotation. (Moller et al. 1978; updraft that was dynamically forced by an upward Bluestein 1984). Recent studies have also revealed directed pressure gradient. Observations by storm the existence of anticyclonic wall clouds (Atkins et intercept teams have since confirmed that cloud al. 2012). National Weather Service storm spot- tags associated with cooler air behind the gust front ter training documents indicate that persistent wall often rise into the updraft. Often this occurs near or clouds that develop strong rotation and exhibit rapid in association with the distinct “tail cloud” that pro- upward vertical motion are often regarded as pre- trudes from near the base of the wall cloud toward cursors to tornadogenesis (http://www.nws.noaa. the cooler air. This mechanism is described in Na- gov/om/brochures/adv_spotters.pdf). While many tional Weather Service storm spotter training doc- studies of supercell thunderstorms have photo-documents (http://www.nws.noaa.gov/training/wxspot. umented the wall cloud, amazingly little is known php). about their formation. To date, no observational study has been Our current understanding of wall cloud published confirming the Rotunno and Klemp (1985) 5 June, 2009 Photo Time: 2157:45 UTC (a)

5°

Cloud Base ≈ 1100 m ARL 3° Wall Cloud RFD 900 m parcel Mesonet estimated wall cloud base: 530 m ±50 m FF 600 m parcels 1° 300 m

272° 274° 276° 277.4° 278° 280° 282° 284° 286° 288° 290° 292° FIG 1. Photograph of the wall cloud taken at 2157:45 UTC on 5 June 2009. Photogrammetric estimates of cloud base height and wall cloud vertical extent are shown in black. Mobile mesonet estimated wall cloud base height is shown in purple. Wall cloud boundary is shown in red.

______Corresponding author address: Nolan T. Atkins, Department of Atmospheric Sciences, Lyndon State College, Lyndonville, VT 05851 E-mail: [email protected]

1 5 June, 2009 2156 - 2158 UTC z = 600 m -1 Reflectivity (dBZ) W (ms ) -1 horizontal winds p’ (mb)

1 280° fr om CAMA location of minimum OW 277.4°

275° -6

approximate -3 wall cloud location -1

0 mb

4 km (a) 30 ms -1 (b)

FIG 2.0 Dual-Doppler 5 10 15 20 data 25 dBZfrom 352156:00 40 45 – 2158:00 50 55 UTC at -20 600 -15m ARL -10 is shown.-5 0 ms (a)-1 Radar 10 reflectivity 15 20 from DOW 7 (color) and dual-Doppler winds (ms-1; black vectors). Cyan lines are the left, center, and right azimuths of the estimated wall cloud location based on photogrammetry. The thick red line is the approximate wall cloud location. The location of minimum Okubo-Weiss number is filled purple. (b) Vertical velocity (ms-1; color) and perturbation pressure (black contours) are shown. Wall cloud location and azimuths are same as in (a).

mechanism for wall cloud formation. It is not known lect observations in and around supercell thunder- what fraction of air creating observed wall clouds storms. To increase the number of storm intercepts, originates in the storm mid levels or from other the experiment was mobile covering the South- source regions such as the inflow. Given that some ern, Central, and Northern Plains states during the wall clouds may be collocated with the low-level me- spring seasons of 2009 and 2010. More information socyclone and therefore contain significant low-level on VORTEX2 can be found in the review article by rotation, it is possible that some or all of the wall Wurman et al. (2012). cloud lowering is created by the adiabatic cooling Analyses of two supercells observed during associated with the pressure deficit at the circula- VORTEX2 are presented herein. The first formed tion center. While this mechanism may not explain on 5 June 2009 over Goshen CO Wyoming (here- all of the lowering for wall clouds associated with after referred to as the Goshen County supercell). weak rotation, it may explain a significant portion of This supercell produced a well-defined wall cloud the lowering associated with strongly rotating wall and attendant EF-2 tornado. The second supercell clouds. No study has systematically examined this was observed on 11-12 June 2009 west-northwest wall cloud formation mechanism. of La Junta CO (hereafter referred to as the La Junta supercell). While this supercell was tornado warned 2. VORTEX2 during the time of VORTEX2 data collection, it did not spawn a tornado despite producing a well-de- The requisite visual, kinematic, and thermo- fined wall cloud. dynamic data to examine the aforementioned wall cloud formation mechanisms was collected during 3. DATA ANALYSIS METHODS the Verification of the Origins of Rotation in Torna- does Experiment II (VORTEX2). VORTEX2 was The wind field in and around the wall clouds a large multi-agency field project designed to col- was generated from dual-Doppler synthesis of radial

2 velocities collected by the mobile X-band Doppler cup anemometer. All data, including GPS derived On Wheels (DOW) 6 and 7 radars operated by the latitude and longitude, were collected at one-second Center for Severe Weather Research (CSWR; Wur- intervals. Data collected within the window of five man 2001). After the data were edited and navigat- minutes before and after the dual-Doppler analysis ed, it was objectively interpolated to a Cartesian grid time were used in the analyses. It was assumed using a two-pass Barnes scheme. The horizontal that storm evolution was not significant during this and vertical grid spacing was set to 100 m over a ten-minute time period. All mobile mesonet data 15 km x 15 km horizontal domain. Vertical veloc- were space-time adjusted to account for storm mo- ities were calculated by upward integration of the tion. mass continuity equation. Dual-Doppler volumes The visual extent and evolution of the wall were collected every two minutes. Examples of du- clouds was determined by photogrammetry analysis al-Doppler wind syntheses from VORTEX2 data and of still images captured by two photo teams. Pho- further details of the dual-Doppler technique can be togrammetry is the process of superimposing azi- found in Markowski et al. (2012a, b), Atkins et al. muth and elevation angle grids on photographs by (2012), and Kosiba et al. (2013). computing or determining the precise azimuths of The thermodynamic properties of low-level landmarks in the horizon relative to the photogra- parcels entering the wall cloud, was based on mo- pher. Once the landmark locations are known, the bile mesonet (Waugh and Fredrickson 2010) obser- effective focal length and tilt angle of the photograph vations. The mobile mesonets carried roof-mounted can be computed. Spherical trigonometry is then instrumentation that included temperature, relative used to create the azimuth-elevation grid. Once the humidity, and pressure sensors along with a three- photo has been gridded, it can then be combined

5 June, 2009 Photo Time: 2200:24 UTC CAMA (a)

Mesonet estimated 3° wall cloud base: 530 m ±50 m 560 m Wall Cloud

400 m

1°

274° 276° 278° 280° 281.4° 282° 284° 286° 288°

5 June, 2009 Photo Time: 2200:33 UTC CAMB (b)

3°

Wall Cloud

1°

336° 338° 340° 342° 344° 346°

FIG 3. Photographs of the wall cloud taken at (a) 2200:24 UTC from the CAMA and (b) 2200:33 UTC from the CAMB locations, respectively on 5 June 2009. Photogrammetric and mobile mesonet estimates of wall cloud base height are shown in black and purple, respectively. Wall cloud boundary is shown in red.

3 5 June, 2009 2200 - 2202 UTC z = 600 m Reflectivity (dBZ) W (ms-1) horizontal winds p’ (mb)

-1 285° from CAMA location of minimum OW 281.4°

276° -3

approximate wall cloud location 344° from CAMB 344° from CAMB -1 -1 0 mb

1 337°

337° 1 4 km (a) 30 ms -1 (b)

0 5 10 15 20 25 dBZ 35 40 45 50 55 -20 -15 -10 -5 0 ms-1 10 15 20 FIG 4. Same as Fig. 2 except for the time period 2200:00 – 2202:00 UTC. with the dual-Doppler wind field and radar reflectivity Consistent with Fig. 1, the wall cloud was embed- observations. More details on photogrammetry can ded in precipitation that was wrapping around and be found in Wakimoto et al. (2011) and Atkins et al. embedded in the low-level mesocyclone (Fig. 1a). (2012). Video (not shown) revealed that the wall cloud contained significant rotation. The strongest rotation 4. GOSHEN COUNTY TORNADIC WALL CLOUD in the dual-Doppler data was located by computing and locating the minimum Okubo-Weiss number 4.1 Visual and Dual-Doppler Observations (Okubo 1970; Weiss 1991; Markowski et al. 2011). It was located at an azimuth of 277.4 degrees from The Goshen County wall cloud observed at the DOW7/photo location (Fig. 2) near the wall cloud 2157:45 UTC is shown in Fig. 1. The view in Fig. 1 center. The wall cloud lowering was located at the is from the DOW7 location on highway 85 looking same azimuth (Fig. 1). to the west (see Fig. 1 in Wakimoto et al. (2011) for It has long been accepted and shown by the radar and photo team locations relative to the Rotunno and Klemp (1985) that the wall cloud is em- hook echo). The wall cloud was approximately 17 bedded within low-level updraft. The updraft is nec- km from the radar/photo location. A radar-detected essary to dynamically force the negatively buoyant tornado formed at about 2152 UTC but was not yet evaporatively-cooled air to wall cloud base. The wall visible. The wall cloud lowering was centered on cloud position relative to the vertical velocity field in azimuths 276-278° and extended nearly 800 m be- Fig 2b suggests that the wall cloud is on the gradient low the primary cloud base of 1100 m above radar of vertical velocity and contains some downdraft air level (ARL; relative to DOW 7). The primary cloud on the southern and eastern flanks. This downdraft base was determined photgrammetrically and with a air is likely associated with the rear-flank downdraft nearby inflow sounding (not shown). The wall cloud (RFD) that is wrapping around the southern side of lowering was asymmetric as it gradually rose to the the wall cloud. It has been hypothesized that the north. Rain curtains were observed to the north and RFD may play a role in tornadogenesis (e.g., Lem- south of the wall cloud suggesting that it was em- on and Doswell 1979). The relationship between bedded within precipitation. The location of the wall the RFD, tornado, low-level mesocyclone, and wall cloud relative to the hook echo is shown in Fig. 2. cloud, is poorly understood.

4 05 June 2009 Backward Trajectories 2156:00 UTC - 2148:02 UTC 4000

6000 a) 50 b)

3500

5000 40 3000 Forward Flank 4000 30 2500

3000 20 Inflow 2000 y (meters) 2000 z (meters) 10 1500

1000 0 Rear Flank 1000 0 Downdraft

−10 500

−1000

−20 0 −3000 −2000 −1000 0 1000 2000 3000 4000 −8 −7 −6 −5 −4 −3 −2 −1 0 x (meters) time (minutes)

4000 c) d)

3500

3000

−8 min 2500

−6 min −6 5000 2000 z (meters) 4000 −4 −4 4000 −4 1500 −2 2000 −6 3000

z (meters) −2 0 0 1000 0 −8 2000 0 −2 y (meters) −3000 −2000 1000 500 −1000 0 0 1000 0 x (meters) 2000 −8 −7 −6 −5 −4 −3 −2 −1 0 3000 time (minutes) FIG 5. Radar reflectivity from DOW 7 (color; 500 m ARL) and backward trajectories for parcels entering the wall cloud. The radar reflectivity data is from the 2156:00 – 2158:00 dual-Doppler analysis. Backward trajectories were calculated from 2156:00 to 2148:02 UTC. (a) Plan view of radar reflectivity and trajectories. Inflow, forward flank, and rear-flank downdraft parcels are colored pink, green, and purple, respectively. The approximate wall cloud location is shown in black. (b) Height versus time plot of the trajectories shown in (a). (c) Three-dimensional perspective of the radar reflectivity field shown in (a) along with representative parcel trajectories from the inflow, forward flank, and rear-flank downdraft locations. Black dashed lines represent the ground-relative location of the respective trajectories. Black time labels are minutes before the initial time of 2156:00 UTC. (d) Height versus time diagram of the three representative trajectories shown in (c).

Similar results were observed a couple min- in Fig. 3 and superimpose the position on the radar utes later with photos taken at 2200:24 and 2200:33 data collected from 2200 - 2202 UTC. As shown in UTC at the DOW7 and DOW6 locations, respective- Fig. 4 and consistent with the results in Fig. 2, the ly. Much of the wall cloud base was found at about wall cloud was located in a region of precipitation 560 m ARL (Fig. 3a). An additional 160 m of lower- associated with the hook echo and appeared to be ing was observed just north of the wall cloud center. located within the low-level mesocyclone. Notice It was possible to triangulate the wall cloud location that the strongest rotation was not located at the

5 05 June 2009 Backward Trajectories 2200:00 UTC - 2150:02 UTC 3500 a) b) 7000 50

6000 3000 40 5000 2500 4000 30 Forward Flank 3000 2000 20 2000 z (meters) y (meters) Inflo 1500 w 10 1000

0 1000 Rear Flank 0 Downdraft −1000

−10 500 −2000

−3000 −20 0 −3000 −2000 −1000 0 1000 2000 3000 4000 5000 6000 7000 −10 −9 −8 −7 −6 −5 −4 −3 −2 −1 0 x (meters) time (minutes)

4000 c) d) 3500

3000

2500 −8 −10 min −8 −6 −4 2000

6000 z (meters) 4000 −2−2 −4 −6 −2 −6 1500 2000 0 −4 4000 0 0 −8

z (meters) 1000 0 2000 −10 −2000 0 y (meters) 500 0 2000 x (meters) 4000 −2000 0 −10 −9 −8 −7 −6 −5 −4 −3 −2 −1 0 6000 time (minutes) FIG 6. Same as Fig. 5 except that the trajectories were run from 2200:00 – 2150:02 UTC and the radar reflectivity was generated from a dual-Doppler analysis valid from 2200:00 – 2202:00 UTC.

center of the wall cloud. Rather, it was displaced is necessary to identify regions of the storm from to the north and was collocated with the local wall which parcels entering wall cloud base are coming cloud lowering shown in Fig. 4. This was the loca- from. This was partially accomplished by calculat- tion of the developing tornado. The tornado funnel ing backward parcel trajectories for hundreds of par- cloud made visual contact with the ground at about cels entering the wall cloud base. Trajectories were 2206 UTC. computed by interpolating the dual-Doppler data As with the wall cloud in Fig.1, the wall cloud every 20 seconds in time. Spatial interpolation was at 2200 UTC was partially embedded in downdraft. trilinear. Results of this analysis are shown in Fig. 5 (Fig. 4b). It appeared to be the rear-flank downdraft for the wall cloud observed at 2157:45 in Fig. 1. The that wrapped around the southern and eastern side parcel trajectories were computed backward in time of the wall cloud. for ten minutes and were released within the wall cloud location shown in Fig. 2 at a height of 550 m 4.2 Wall Cloud Parcel Source Regions ARL. The trajectories in Fig. 5a,c illustrate that parcels entering the wall cloud come from three source In order to test the wall cloud formation hy- regions. They are the inflow, rear-flank downdraft, pothesis put forth by Rotunno and Klemp (1985), it and the forward flank with the majority coming from

6 the forward flank region. The inflow parcels originat- 5 June 2009 Radar: 2156 - 2158 UTC Mesonet: 2152 - 2202 UTC ed at low levels, wrapping around the northern and -1 -1 Mixing Ratio (g kg ) 13.81 .8 30 ms θ 303.330030 .3 western flanks before ascending to wall cloud base v (K) dd winds - 300m ARL 12.612262..66 13.6 302.330277 304.8 (Fig. 5b,d). dBZ - 500m ARL 1010.7.7 13.9131 .9 301.303011.44 304.303 44.1 It was surprising to observe a number of 11.411.44 302.30202.66 11.4111 .4 parcels reaching cloud base that had descended 302.303 22.4 in downdraft. A few parcels descended within the rear-flank downdraft while others originated in the forward-flank region at altitudes greater than 1000 11.0 m ARL and then descended to wall cloud base (Fig. 304.3 305.303055..88 5b). The fact that parcels were descending to wall 1111.9..99 11.7 cloud base is consistent with the results shown in 305.3305.4 306.2 11.41111.4 Figs. 2b. and 4b. Namely, the wall cloud was par- 306.0606.6.2 11.81111.8 306.303 6.5 11111.51.. tially embedded in downdraft. It is not known how 306.303060 .4 306.306 8 12.1 10 km these parcels are descending. One might expect 304.4 that they are descending moist adiabatically. The microphysical characteristics of the supercell rear- −2200 −1100 0 10 20 dBZ 30 40 50 flank downdraft, however, are poorly understood. FIG 7. Radar reflectivity from DOW 7 (color; 500 The remaining forward flank parcels originated at m ARL) and dual-Doppler winds (black vectors; 300 low-levels (Fig. 5b,d) and ascended to wall cloud m ARL) from 2156:00 – 2158:00 UTC. Mobile me- base in a manner consistent with the idealized mod- sonet winds (black barbs; ½ flag = 5 ms-1, full flag = eling results of Rotunno and Klemp (1985). 10 ms-1), mixing ratio (g kg-1), and virtual potential Parcel trajectories entering the wall cloud at temperature (K) are shown from 2152:00 – 2202:00 2200 UTC (Fig. 4) were computed to examine the UTC. The semi transparent region represents the generality of the results shown in Fig. 5. These tra- location of forward-flank parcel trajectories shown in jectories are shown in Fig. 6 and are consistent with Fig. 5. the results in Fig. 5. Additional analyses of a supercell wall cloud that exhibited weak rotation on 26 May 2010 also revealed the same three wall cloud source regions (not shown). these low-level forward flank parcels would saturate at approximately 530 m ARL. This altitude is nearly 4.3 Wall Cloud Formation identical to the observed wall cloud base of 560 m ARL at 2200:24 and 2200:33 UTC (Fig. 3) and is The question of wall cloud formation is now within the range of 300-900 m ARL for the sloping addressed. The requisite data exists to test the hy- wall cloud at 2157:45 UTC. pothesis put forth by Rotunno and Klemp (1985) Apparent in Fig. 3a is the localized lowering that evaporatively cooled air in the forward flank just north of the wall cloud center. This lowering is region saturates at a lower altitude as it is ingest- evident in Fig. 3a between azimuths of 280 and 282 ed in the updraft. Indeed, Figs. 5 and 6 showed degrees. It extended about 150-160 meters below that low-level air in the forward flank region entered the primary wall cloud base. It is hypothesized that wall cloud base. The thermodynamic properties of this lowering was associated with the pressure defi- this air can be approximated with the mobile me- cit associated with the rotation in the wall cloud. A sonet data (Fig. 7) despite the sparse coverage in pressure retrieval using the dual-Doppler wind field and around the hook echo due to the limited road was performed using techniques established by network over southeastern Wyoming. The rear and Gal-Chen (1978) and Hane and Ray (1985). The forward flank regions appear to be about 3-4 K vir- perturbation pressure fields are shown in Figs. 2b tually cooler and about 1 g kg-1 drier than the inflow and 4b for the 2156-2158 UTC and 2200-2202 UTC as partially sampled by the mobile mesonet (Fig. volume scans, respectively. In Fig. 2b, the lowest 7) and a nearby mobile sounding (not shown). The pressure perturbation was approximately -8 mb potential temperature and mixing ratio deficits in the and was centered on an azimuth of 277.4 degrees. forward flank region where the parcels entering wall The lowest extent of the wall cloud was observed cloud base originate (gray shaded region in Fig. 7) at 2257:45 UTC at this same azimuth (Fig. 1). The were approximately 4.0 K and 1.5 g kg-1 relative to pressure perturbation in Fig. 4b was about -10 mb, the inflow, respectively. Assuming adiabatic ascent, centered on 281.4 degrees. The localized lowering

7 12 June, 2009 Photo Time: 0028:01 UTC (a)

17° Cloud Base ≈ 1300 m ARL

15°

13° Mesonet estimated 11° wall cloud base: Wall Cloud 9° 680 m ± 110 m 7° 595 m 580 m 5° 520 m 430 m 3° ± 90 m 1° 344° 348° 352° 356° 0° 4° 8° 12° 16° 20° 24° 28° 32°

FIG 8. Photograph of the wall cloud taken at 0028:01 UTC on 12 June 2009. Photogrammetric estimates of cloud base height and wall cloud vertical extent are shown. Mobile mesonet estimated cloud base height is shown in purple. Photogrammetric estimated cloud base in shown in black. Wall cloud boundary is shown in red.

in Fig. 4b was located between azimuths of 280-282 cloud was not collocated with the low-level meso- degrees. Thus, the lowest portion of the wall cloud cyclone. The dual-Doppler analysis and WSR-88D base is collocated with the lowest pressure within observations from near Pueblo CO suggested that the wall cloud. Assuming that the air near wall cloud the mesocyclone was approximately 3 km to the base is close to or saturated, a -10 mb pressure northwest of the wall cloud (Fig. 9a). The mesocy- drop would result in about 140 m of lowering using clone was relatively weak, consistent with the lack the US Standard Atmosphere. Thus, the pressure of a well-defined hook echo at the time shown in deficit within the wall cloud can explain nearly all of Figs. 8 and 9. Video observations confirmed that local lowering observed in Fig. 3. It is possible that the wall cloud was not rotating at 0028 UTC. These the pressure deficit is larger than what is shown in observations suggest that the La Junta supercell Figs. 2 and 4 since circulation on the tornado scale had produced strong outflow that had displaced the is not resolved in the dual-Doppler wind field. updraft from the mesocyclone. There were no large pressure perturbations evident in the wall cloud (Fig. 5. LA JUNTA NON TORNADIC WALL CLOUD 9b). Only weak vertical motion was observed near wall cloud base (Fig. 9b). Stronger updraft at high- 5.1 Visual and Dual-Doppler Observations er levels was observed within the wall cloud region, approaching magnitudes of 20 ms-1 at 2 km ARL A well-defined wall cloud at 0028:01 UTC (not shown). was observed with a supercell sampled by the VOR- TEX2 teams on 12 June 2009 west-northwest of La 5.2 Wall Cloud Parcel Source Regions Junta, CO (Fig. 8). The wall cloud extended 700- 750 m below the primary cloud base. Wall cloud Parcel trajectories for the La Junta supercell base varied from approximately 520 to 600 m ARL wall cloud were calculated in the same manner as from northeast to southwest (Fig. 8). There are no- for the Goshen County supercell. The results are table differences between the wall clouds observed shown in Fig. 10. It is clear that most of the air en- on 5 and 11-12 June 2009. First, triangulation of tering wall cloud base is coming from the forward photographs placed the La Junta wall cloud to the flank precipitation region of the storm (Fig. 10a, c). south of the storms precipitation region. This is in The trajectories suggest that the storms outflow was contrast to the 5 June 2009 wall cloud that was em- surging out ahead of the precipitation region. While bedded in precipitation wrapping around the low-lev- some trajectories exhibited large vertical displace- el mesocyclone. Also notice that the La Junta wall ments at they moved toward the wall cloud (Fig.

8 12 June, 2009 0028 - 0030 UTC z = 1100 m -1 Reflectivity (dBZ) W (ms ) 1 horizontal winds p’ (mb) 1

mesocyclone 1 mb at 2 km ARL

348° from CAMA -1

62° from CAMB 0 0 0 mb

approximate wall cloud location

-2 0

71°

31°

4 km (a) 30 ms -1 (b)

0 5 10 15 20 25 dBZ 35 40 45 50 55 -20 -15 -10 -5 0 ms-1 10 15 20

FIG 9. Dual-Doppler data from 0028:00 – 0030:00 UTC at 1100 m ARL is shown. (a) Radar reflectivity from DOW 7 (color) and dual-Doppler winds (ms-1; black vectors). Cyan lines are the left and right azimuths of the estimated wall cloud location based on photogrammetry. The thick red line is the approximate wall cloud location. (b) Vertical velocity (ms-1; color) and perturbation pressure (black contours) are shown. Wall cloud location and azimuths are same as in (a).

10b, d), the majority of the parcels ascended into southwestern portion of the wall cloud. Computing wall cloud base and from the precipitation region. It mean potential temperature and mixing ratio values is also clear from Fig. 10 that the there was only one in the colder and warmer forward flank regions yield- source region of parcels entering the La Junta wall ed cloud base heights of approximately 430 and cloud. Unlike the Goshen County supercell, there 680 m, respectively. These cloud base heights are appeared to be few or no parcels originating in the consistent with the phtogrammetrically-determined inflow or rear-flank downdraft regions of the La Jun- wall cloud base heights (Fig. 8). The observations ta supercell. shown in Figs. 8-11 suggest that the La Junta wall cloud formed as low-level parcels from the forward 5.3 Wall Cloud Formation flank region ascended into the low-level updraft.

Mobile mesonet data was again examined 6. DISCUSSION AND SUMMARY to approximate the thermodynamic properties of low-level parcels entering the wall cloud from the The formation of wall clouds observed with- forward flank region. This data is summarized in in three supercells sampled by VORTEX2 has been Fig. 11. Notice that the air is cooler in the north- examined. Results for two of the cases were pre- eastern portion of the forward flank (light green sented herein. The first supercell occurred in Gosh- observations) region and relatively warmer to the en County Wyoming on 5 June 2009 and produced southwest (blue observations). The cooler parcels an EF2 tornado. The Goshen County wall cloud ascended into the northeastern portion of the wall was embedded in precipitation wrapping around the cloud (Fig. 10a) where cloud base was lower (Fig. low-level mesocyclone. The wall cloud appeared 8). The warmer parcels ascended into the higher, to be centered on the low-level mesocyclone and

9 12 June 2009 Backward Trajectories 00:2800 UTC - 00:1800 UTC 2200

a) 50 b) 2000 6000 dBZ

40 1800

4000 cooler 1600 30

2000 1400 20 warmer 1200

0 z (meters)

y (meters) 10 1000

800 −2000 0

n 600 −4000 −10 400 approximate wall cloud locatio −20 200 −2000 −1000 0 1000 2000 3000 4000 5000 6000 7000 −10 −9 −8 −7 −6 −5 −4 −3 −2 −1 0 x (meters) Time (minutes)

1800 c) d) 1600

1400 2000

1500 −8 min 1200 −6 1000 −10 −6 1000 −8 0 z (meters) z (meters) 500 −4−2 −4 −10 800 0 0 5000 −2 4000 −8 −6 600 3000 7000 2000 0 6000 1000 −2 −4 5000 0 4000 400 −1000 3000 −2000 −3000 2000 y (meters) 1000 x (meters) 200 −4000 −10 −9 −8 −7 −6 −5 −4 −3 −2 −1 0 Time (minutes) FIG 10. Radar reflectivity from DOW 7 (color; 1100 m ARL) and backward trajectories for parcels entering the wall cloud. The radar reflectivity data is from the 0028:00 – 0030:00 dual-Doppler analysis. Backward trajectories were calculated from 0028:00 to 0018:00 UTC. (a) Plan view of radar reflectivity and trajectories. The approximate wall cloud location is shown in black. (b) Height versus time plot of the trajectories shown in (a). (c) Three-dimensional perspective of the radar reflectivity field shown in (a) along with representative parcel trajectories. Black dashed lines represent the ground-relative location of the respective trajectories. Black time labels are minutes before the initial time of 0028:00 UTC. (d) Height versus time diagram of the three representative trajectories shown in (c).

therefore, contained strong rotation. A developing The fact that some parcels reaching wall tornado was observed in the wall cloud. Air entering cloud base are descending in downdraft is perhaps this wall cloud came from three source regions; the not that surprising. Previous studies have shown inflow, forward flank, and rear-flank downdraft. Most that the low-level mesocyclone forms as horizontal parcels originated in the forward flank region. The vorticity is tilted at the updraft/downdraft interface. majority of the parcels from the inflow and forward Furthermore, it has also been suggested that down- flank regions originated at low levels. Some parcels draft is necessary to abruptly tilt vortex lines verti- in the forward flank originated at higher levels and cally near the ground to produce a tornado. So, it is descended into the wall cloud as did the rear-flank possible that tornadic wall clouds collocated with the downdraft parcels. These wall cloud source regions low-level mesocyclone contain air that is descend- were also observed with a wall cloud on 26 May ing in downdraft. It is not known how this air is de- 2010 that exhibited weak rotation (not shown). scending. Perhaps high resolution modeling may

10 12 June 2009 Radar: 0028 - 0030 UTC Mesonet: 0024 - 0034 UTC Parcels entering wall cloud base appeared Mixing Ratio (g kg-1) to come largely from the surging outflow. Many of θ (K) 11.5 11.8 v 11.1 301.301.6 the parcels originated at low-levels. dd winds - 300m ARL 304.8 301.9 dBZ - 500m ARL cooler The La Junta wall cloud appeared to form 12.62..66 12112.1.1 304.044.7 303.303033.77 10.9100.9 9.8 from the lifting of evaporatively cooled forward flank 302.30242.44 300.3000.4 parcels. Wall cloud base increased from northeast 10.1 to southwest. This was a result of the forward flank 302.2 11.2 11.1 301.5 300.9 downdraft being colder to the northeast. Parcel tra- 9.9 7 302.30302288 jectories from this portion of the storm entered the 10.4 warmer northeastern portion of the wall cloud saturating at a 304.4 lower altitude that the warmer parcels that ascended 11.0 1010.50..5 303.9 into the southwestern portion of the wall cloud. 303.303033.9 11.4 11.4 303.30303.40303..4 10110.8.8 There are important similarities and differ- 303.303 33.2 10.9110099 ences between the conclusions of Rotunno and 11.4 303.3 303.9 2 km Klemp (1985) and this study concerning wall clouds. 30 ms-1 Both studies have shown that wall clouds form as

−20 −10 0 10 20 dBZ 30 40 50 evaporatively-cooled air from the forward flank region ascends into the low-level dynamically forced FIG 11. Radar reflectivity from DOW 7 (color; 500 updraft. This study has also shown that additional m ARL) and dual-Doppler winds (black vectors; 300 lowering may be attributed to the pressure deficit m ARL) from 0028:00 – 0030:00 UTC on 12 June associated with strongly rotating wall clouds. Fur- 2009. Mobile mesonet winds (black barbs; ½ flag thermore, it was shown herein that some parcels = 5 ms-1, full flag = 10 ms-1), mixing ratio (g kg-1), reaching wall cloud base may be descending from and virtual potential temperature (K) are shown from the forward flank region and rear-flank downdraft. 0024:00 – 0034:00 UTC. The source region for par- Our analyses suggested that this will only be true cels entering the wall cloud is shaded gray. for wall clouds that are collocated with the low-level mesocyclone. Future work should examine more cases to assess the generality of these results. A modeling shed more light on the details of the forward and study would help to understand the important forcing rear-flank downdrafts to address this question. mechanisms for parcels descending from the for- The observations suggested that the Gosh- ward flank region and rear-flank downdraft. Finally, en County wall cloud formed as low-level evapora- work is underway to examine the three-dimensional tively-cooled air in the forward flank region ascend- structure of wall clouds and their relationship with ed into the dynamically forced low-level updraft. the storm updraft, precipitation region, and low-level Using mobile mesonet data to approximate the ther- mesocyclone. modynamic properties of the low-level forward flank parcels, saturation occurred at an altitude that was Acknowledgments: This research was supported by consistent with observed cloud base. This mecha- the National Science Foundation under grants AGS- nism is consistent with that proposed by Rotunno 0757714 and AGS-1242339. and Klemp (1985). It was also shown that additional lowering may be attributed to the adiabatic cooling REFERENCES associated with the pressure deficit at the center of the rotating wall cloud. Atkins, N. T., A. McGee, R. Ducharme, R. M. A second wall cloud associated with a su- Wakimoto, and J. Wurman, 2012: The La- percell west-northwest of La Junta CO on 11-12 Grange tornado during VORTEX2. Part II: Pho- June 2009 was examined. The La Junta wall cloud togrammetric analysis of the tornado combined was well defined and extended approximately 700- with dual-Doppler radar data. Mon. Wea. Rev., 800 m below the primary cloud base. The wall cloud 140, 2939 – 2958. formed on the southern periphery of the forward Bluestein, H. B., 1983: Surface meteorological ob- flank precipitation region. The outflow appeared to servations in severe thunderstorms. Part II: be surging away from the precipitation region. As Field experiments with TOTO. J. Climate Appl. a result, the wall cloud was not collocated with the Meteor., 22, 919–930. mesocyclone. ——,1984: Photographs of the Canyon, Texas

11 storm on 26 May 1978. Mon. Wea. Rev., 112, The LaGrange tornado during VORTEX2. Part 2521–2523. I: photogrammetric analysis of the tornado ——, 1993: Synoptic-Dynamic Meteorology in Mid- combined with single-Doppler radar data. Mon. latitudes. Vol. 2, Observations and Theory of Wea. Rev., 139, 2233–2258. Weather Systems, Oxford Press, 594 pp. Waugh, S., and S. E. Fredrickson, 2010: An im- Davies-Jones, R.P., 1986: Tornado dynamics. Vol. proved aspirated temperature system for mobile 2, Thunderstorms: A Social and Technological meteorological observations, especially in se- Documentary. 2d ed. E. Kessler, Ed., University vere weather. Preprints, 25th Conf. on Severe of Oklahoma Press, 197–236. Local Storms, Denver, CO, Amer. Meteor. Soc., Gal-Chen, T., 1978: A Method for the Initialization of P5.2. [Available online at https://ams.confex. the Anelastic Equations: Implications for Match- com/ams/25SLS/webprogram/Paper176205. ing Models with Observations. Mon. Wea. Rev., html.] 106, 587–606. Weiss, J., 1991: The dynamics of enstrophy transfer Hane, C. E., P. S. Ray, 1985: Pressure and Buoy- in two-dimensional hydrodynamics. Physica D, ancy Fields Derived from Doppler Radar Data 48, 273–294. in a Tornadic Thunderstorm. J. Atmos. Sci., 42, Wurman, J., 2001: The DOW mobile multiple-Dop- 18–35. pler network. Preprints, 30th Int. Conf. on Ra- Kosiba, K., J. Wurman, Y. Richardson, P. Markows- dar Meteorology, Munich, Germany, Amer. Me- ki, P. Robinson, and J. Marquis, 2013: Genesis teor. Soc., P3.3. of the Goshen County, Wyoming, Tornado on 5 ____, D. Dowell, Y. Richardson, P. Markowski, June 2009 during VORTEX2. Mon. Wea. Rev., E. Rasmussen, D. Burgess, L. Wicker, H. B. 141, 1157–1181. Bluestein, 2012: The Second Verification of the Lemon, L. R., and C. A. Doswell, 1979: Severe Thun- Origins of Rotation in Tornadoes Experiment: derstorm Evolution and Mesocyclone Structure VORTEX2. Bull. Amer. Meteor. Soc., 93, 1147– as Related to Tornadogenesis. Mon. Wea. Rev., 1170. 107, 1184–1197. Markowski, P., and Coauthors, 2012: The Pretor- nadic Phase of the Goshen County, Wyoming, Supercell of 5 June 2009 Intercepted by VOR- TEX2. Part I: Evolution of Kinematic and Sur- face Thermodynamic Fields. Mon. Wea. Rev., 140, 2887–2915. ____, and Coauthors, 2012: The Pretornadic Phase of the Goshen County, Wyoming, Supercell of 5 June 2009 Intercepted by VORTEX2. Part II: In- tensification of Low-Level Rotation. Mon. Wea. Rev., 140, 2916–2938. ____, M. Majcen, Y. Richardson, J. Marquis, and J. Wurman, 2011: Characteristics of the wind field in a trio of nontornadic low-level meso- cyclones observed by the Doppler on Wheels radars. Electronic J. Severe Storms Meteor., 6 (3), 1–48. Moller, A. R., 1978: The improved NWS storm spotters’ training program at Ft. Worth, Tex. Bull. Amer. Meteor. Soc., 59, 1574 – 1582. Okubo, K., 1970: Horizontal dispersion of floatable particles in the vicinity of velocity singularities such as convergences. Deep-Sea Res., 17, 445-454. Rotunno, R., and J. B. Klemp, 1985: On the rotation and propagation of simulated supercell thunderstorms. J. Atmos. Sci., 42, 271–292. Wakimoto, R.M., N. T. Atkins, and J. Wurman, 2011: