sign in sign up
<<
Home , Supercell

VOLUME 142 MONTHLY REVIEW SEPTEMBER 2014

PICTURE OF THE MONTH

A Multiple- in Southeastern

ERNANI DE LIMA NASCIMENTO Grupo de Modelagem Atmosférica de Santa Maria, Departamento de F´sica,ı Universidade Federal de Santa Maria, Santa Maria, Rio Grande do Sul, Brazil

GERHARD HELD AND ANA MARIA GOMES Instituto de Pesquisas Meteorológicas, Universidade Estadual Paulista ‘‘Júlio de Mesquita Filho,’’ Bauru, Sao Paulo, Brazil

(Manuscript received 9 October 2013, in final form 18 April 2014)

ABSTRACT

During the late afternoon hours of 24 May 2005 a outbreak occurred in the state of São Paulo, southeastern Brazil. Severe were observed ahead of a surface cold front, including a (Southern Hemisphere) cyclonic left-moving that produced a multiple-vortex tornado in the outskirts of the town of Indaiatuba, Brazil (23.18S, 47.28W). A documentation of the multivortex structure of the tornado and of the -base features is performed using still images from a video that recorded the event. Characteristics of the tornadic and the synoptic-scale environment in which it developed are examined using Doppler radar data, geostationary satellite imagery, surface and upper-air observations, and data from the National Centers for Environmental Prediction’s Forecast System Reanalysis. The cloud base of the thunder- displayed morphological features associated with midlatitude tornadic , including a low-level and a ‘‘clear slot’’; however, the rear-flank downdraft did not obscure the view of the tornado from the western flank of the storm. The tornadic storm developed in a moist prefrontal environment with a low-level jet. Limited mesoscale observations hampered the quantitative analysis of the local thermodynamic forcing, but the available data suggest that the supercell developed under moderate conditional instability. Strong speed and directional vertical shear were observed, while the local boundary layer displayed very high relative hu- midity and low surface-based lifting condensation level.

1. Introduction , northeastern , and southern Brazil as one region where atmospheric conditions favorable Severe weather events associated with deep moist for severe thunderstorms and tornadoes are found. convection such as flash floods, damaging , large Satellite-based thunderstorm climatologies also demon- , and tornadoes are reported every year in Brazil, strate that subtropical South America hosts some of the especially south of 208S (e.g., Silva Dias 2000, 2011). In strongest thunderstorms in the world (Zipser et al. 2006; fact, the midlatitudes and subtropics of South America, Cecil and Blankenship 2012). east of the Andes mountain range, have been recognized The propensity to severe convection in that part of the as prone for severe convective for quite some world is partially explained by the frequent establishment time (Fujita 1973; Schwarzkopf 1982; Velasco and of a northerly low-level jet (LLJ), east of the Andes, Fritsch 1987). More recently, Brooks et al. (2003) and particularly during the warm (Marengo et al. 2002; Brooks (2006) identified the area enclosing , Vera et al. 2006). This circulation has a twofold impact on midlatitude South America; transporting moisture from the Amazon basin to higher latitudes (e.g., Berbery and Corresponding author address: Ernani de Lima Nascimento, Barros 2002), and increasing the curvature and length of Grupo de Modelagem Atmosférica de Santa Maria, Departamento de Física, Universidade Federal de Santa Maria, Avenida Roraima low-level hodographs (Doswell 1991; Nascimento 2005). 1000, CEP 97050-331, Santa Maria, RS, Brazil. In addition, it is not uncommon for the LLJ to become E-mail: [email protected] dynamically associated with geopotential height falls

DOI: 10.1175/MWR-D-13-00319.1

Ó 2014 American Meteorological Society 3017 Unauthenticated | Downloaded 09/28/21 08:21 PM UTC 3018 MONTHLY WEATHER REVIEW VOLUME 142 induced by migratory troughs over the La Plata basin of São Paulo State University. This S-band radar, hereafter (Salio et al. 2002; Seluchi et al. 2003); these same systems referred to through its acronym BRU, is located 624 m can also promote steep midlevel lapse rates by large-scale above sea level at 22.368S, 49.038W in the town of Bauru, accent. All of these aspects can bring together the at- 200 km northwest of Indaiatuba. In its operational config- mospheric ingredients necessary to the development of uration in 2005 it completed a full set of plan position in- severe thunderstorms (Doswell and Bosart 2001). dicators (PPIs) at 11 elevations every 7.5 min when in The regular occurrence of convective ingredients high- volume scan mode. The beamwidth is 28, and the range lights the need for better documentation of severe con- resolution is 1 km. Postprocessing of the clutter-filtered vective events in South America (Nascimento and Doswell radar data was conducted using the storm tracking algo- 2006). One dramatic example of a significant severe rithm Thunderstorm Identification, Tracking, Analysis and weather episode in Brazil is described in this article. In the Nowcasting (TITAN; Dixon and Wiener 1993). late afternoon hours of 24 May 2005 [mid- to late- Data from the National Centers for Environmental in the Southern Hemisphere (SH)], a left-moving cyclonic Prediction’s Climate Forecast System Reanalysis (NCEP supercell thunderstorm developed in the countryside of CFSR; Saha et al. 2010) were used in the synoptic anal- the state of São Paulo (SP), southeastern Brazil. This storm ysis. NCEP CFSR data have 0.58 horizontal grid spacing produced a multiple-vortex tornado in the outskirts of the and 37 vertical levels from the surface to 1 hPa and town of Indaiatuba located at 23.18S, 47.28W(Fig. 1), which were obtained online (http://nomads.ncdc.noaa.gov/data. was documented by a video surveillance camera during its php#CFSR-data). Figures of NCEP CFSR fields and mature stage. skew T–logp diagrams were generated using the Center A brief analysis of the tornado structure and cloud-base for Ocean–Land– Studies (COLA) Grid morphology is performed based on selected still images Analysis and Display System (GrADS). extracted from the videographic documentation; other characteristics of the parent storm are discussed through 3. Discussion radar data analysis. In addition, the synoptic-scale con- a. General overview and visual characteristics of the ditions that prevailed around the time of the tornadic tornadic episode event are investigated from an ingredients-based per- spective (Doswell et al. 1996; Moller 2001). The 24 May 2005 Indaiatuba tornado was spawned at approximately 1725 LST (2025 UTC) by a supercell thunderstorm that developed ahead of a surface cold 2. Meteorological data front. During its life cycle the tornado followed a path Data used in this study include hourly aviation routine approximately 15 km in length, from west-northwest to weather reports (METARs) from six airports located in east-southeast, and with maximum width reaching 200 m, SP: Presidente Prudente (SBDN; 22.128S, 51.388W, ele- as found from an in situ assessment. This was conducted vation: 435 m), São José do Rio Preto (SBSR; 20.808S, 48 h after the event by a damage survey team from the 49.408W, 543 m), Bauru (SBBU; 22.328S, 49.078W, Natural Disasters Study Group of the Federal University 590 m), Ribeirão Preto (SBRP; 21.128S, 47.778W, 549 m), of Santa Catarina (GEDN-UFSC, in Portuguese; Isabela São Paulo–Campo de Marte (SBMT; 23.528S, 46.638W, P. V. Marcelino 2005, personal communication). Figure 2a 722 m), and Campinas (SBKP; 23.018S, 47.138W, 661 m), depicts the tornado damage path according to the the last one being the METAR station closest to GEDN-UFSC survey team, who rated the tornado in- Indaiatuba (Fig. 1b). Soundings from the SBMT upper-air tensity as F3 in the Fujita scale.1 site (Fig. 1b) performed operationally at 0000 UTC [2100 local standard time (LST)] and 1200 UTC (0900 LST) are also examined. METARs were obtained from the Mete- orological Network of Brazil’s Military Air Force (http:// 1 It is worth mentioning that prior to reaching Indaiatuba, the www.redemet.aer.mil.br), and sounding data from the parent storm also produced damage in the southern sections of the town of Capivari located 32 km northwest from Indaiatuba, with Wyoming Weather Web of the University of Wyoming’s one casualty due to a collapsing wall being reported. Locals at- Department of (http://weather.uwyo. tributed the damage inflicted to Capivari to a tornado. By the time edu/upperair/sounding.html). the parent storm moved over Capivari it already displayed a cy- Also included in the analysis are enhanced infrared im- clonic circulation, as indicated by IPMet’s BRU radar imagery (not agery from the Geostationary Operational Environmental shown). However, in the lack of further ground truth evidence regarding the nature and damage path of the phenomenon that hit Satellite-12 (GOES-12), and composite reflectivity and Capivari, we limit our attention to the Indaiatuba tornado episode Doppler velocities from the operated by the as documented by the video camera and damage surveyed by the Meteorological Research Institute (IPMet, in Portuguese) GEDN-UFSC research team.

Unauthenticated | Downloaded 09/28/21 08:21 PM UTC SEPTEMBER 2014 P I C T U R E O F T H E M O N T H 3019

FIG. 1. (a) Geographical position of São Paulo state in the Brazilian map (bottom-right corner inset), and a detailed map of São Paulo state (larger panel); thick (thin) gray contours are topographic curves at 500 (100) m intervals; elevation of Indaiatuba: approximately 620 m MSL; (b) close-up view of the region enclosed by a dotted box in (a), locating Indaiatuba city (solid black circle) with respect to the city of São Paulo; the crisscrosses (3) indicate the position of the SBKP METAR site and SBMT upper-air site.

Unauthenticated | Downloaded 09/28/21 08:21 PM UTC 3020 MONTHLY WEATHER REVIEW VOLUME 142

rotation in the SH). In Fig. 3a the video camera faces the east (E)-northeast (NE) (cf. the relative positions of WT, WH, and highway SP-075 with Fig. 2b), and as the video operator followed the tornado motion, the camera grad- ually turned to the east (Fig. 3d). The Indaiatuba tornado produced subvortices embedded in the main tornadic circulation (Figs. 3b,c), which characterizes the first multiple-vortex tornado ever documented on video in Brazil. The bright light in Fig. 3c is one of the power flashes produced by the tornado as it downed power lines. Surprisingly, no casualties were reported with this signif- icant tornado despite the lack of a sys- tem in Brazil. Figure 3d shows a wide-angle view of the tornado, with the ‘‘clear slot’’ being also discernible. While the low-level structure of this storm displayed morphologi- cal characteristics typical of midlatitude tornadic su- percells (Moller 2001), it is interesting to note that the rear-flank downdraft (RFD) wrapping around the tor- nadic circulation (Fig. 4) did not obscure the tornado as seen from the vantage point of the camera, which is from the west (Fig. 3d). This indicates an absence of pre-

FIG. 2. (a) Path of the tornado according to a damage survey cipitation in that sector of the RFD despite being typi- conducted 48 h after the event by GEDN-UFSC team; tornado mo- cally associated with the portion of the with tion is from the top-left corner to bottom-right corner of the panel; the higher reflectivity as illustrated by the conceptual model white circle indicates the position of the video camera (CAM) that depicted in Fig. 4. captured the tornado video; background map from Google maps (http://maps.google.com). (b) Close-up view of the area enclosed by b. Radar data analysis the rectangle in (a) showing the relative position of CAM to the warehouse (WH) and white tower (WT) that appear in the tornado The first echo of the storm that would eventually evolve video and to Indaiatuba’s industrial district hit by the tornado [dashed into Indaiatuba’s tornadic supercell was detected ap- line enclosing a white shading in both (a) and (b)]; background sat- proximately 30 km southeast of BRU radar at 1737 UTC. ellite image obtained from Google Earth (http://www.google.com/ earth) with image date of 1 Apr 2013. TITAN’s identification algorithm, with a reflectivity threshold of 40 dBZ over 16 km3, detected this cell in the subsequent volume scan at 1744 UTC and tracked it until Just after crossing a highway (SP-075), the Indaiatuba 2059 UTC (i.e., including the tornadic stage) when it tornado was captured in video by the operations control reached the limit of the quantitative 240-km volume scan center of a concessionary company responsible for high- range. Figure 5 shows TITAN’s full tracking sequence of way traffic surveillance and management (Rodovias das the 40-dBZ composite reflectivity area at 7.5-min in- Colinas S.A.). Figure 2b is a recent high-resolution satellite tervals depicting the storm movement that was at an av- 2 image illustrating the area surrounding the documented erage speed of 64 km h 1 toward the southeast in the first video footage, including a warehouse (WH) and a white half of the trajectory and then to the east-southeast after tower (WT) as reference points that appear between the a slight deviation to the left. In this later stage the storm camera and the tornado. moved approximately 108 to the left of the prevailing 0–6- Figures 3a–d depict a selected time sequence of still km mass-weighted mean wind vector (u6 5 12.3, v6 5 2 frames from the tornado video (the full video can be 29.9 m s 1) as computed from the NCEP CFSR data found online at www.youtube.com/watch?v58rcBZldO valid at 1800 UTC for a 28328 grid-sized box centered ZBM, uploaded by the authors). The camera was located over Indaiatuba. The cell was discrete during its entire life on the northwestern sector of the parent thunderstorm as cycle. schematically indicated in Fig. 4, and the storm motion After a rapid development in both intensity and size 2 was to the east-southeast at approximately 64 km h 1 during its early stages, the storm maintained a rather 2 (17.8 m s 1), as determined by IPMet’s BRU radar. steady behavior with maximum reflectivity of 55 dBZ Figure 3a shows the tornado and its parent low-level during the first 2 h of its life cycle. However, from 1944 to mesocyclone displaying a clockwise rotation (i.e., cyclonic 1952 UTC, at approximately 35 km west of Indaiatuba

Unauthenticated | Downloaded 09/28/21 08:21 PM UTC SEPTEMBER 2014 P I C T U R E O F T H E M O N T H 3021

FIG. 3. (a)–(d) Still frames in chronological order of the 24 May 2005 Indaiatuba tornado video (full video found at www.youtube.com/watch?v58rcBZldOZBM). The wide arrow in (a) indicates the clockwise rotation of the tornado and of the parent low-level mesocyclone; in this frame the estimated distance from the camera to the tornado is approximately 1300 m; also indicated are highway SP-075 and two reference points: a white tower (WT) and a warehouse (WH), also shown in Fig. 2b. (b),(c) Black arrows in the close-up views indicate subvortices embedded in the main tornadic circulation. (d) The ‘‘clear slot’’ is highlighted by a black arrow. In (a) the camera faces E-NE and in (d) it is facing E, such that the left (right) edge of each image is upstream (downstream) of the tornado. Local standard time is shown on the top of each still frame. (Courtesy of Rodovias das Colinas S.A.)

(see Fig. 5), the storm experienced a growth in the position indicators (CAPPIs) of ground-relative radial 40-dBZ area and an intensification in reflectivity, velocities (right column) from volume scans completed reaching 57.5 dBZ. It is not clear what caused the storm at 2022, 2029, and 2037 UTC (i.e., just before and around intensification in that stage, but interaction with local the time of the tornado video). For convenience, the mesoscale boundaries is one possibility (Markowski tornado damage path and video camera position are also et al. 1998a; Atkins et al. 1999). The discrete storm indicated in the images. Figures 6a and 6c show the high trailed behind previous convective activity in the form of reflectivity core of the storm passing south of Indaiatuba multicells (not shown) that may have produced outflow with maximum values reaching 55 dBZ,whicharenot boundaries not detected by the available low-resolution particularly high values for supercells (e.g., Kumjian and surface data and satellite imagery. In addition, the Tietê Ryzhkov 2008); at 2029 UTC (Fig. 6c) the relative posi- River valley, along which the storm moved during most tion of the video camera with respect to the reflectivity of its life cycle (Fig. 5), may also have been a source for field is in general agreement with the conceptual model local circulations (Acevedo et al. 2007) that could have sketched in Fig. 4. It should be noticed that at this stage influenced the storm evolution. Interestingly, the storm the cell was 200 km away from BRU and thus, the low- maintained or even increased its strength while follow- levels of the storm were not effectively sampled; in addi- ing the Tietê River valley, a similar behavior also de- tion, at such a distance from the radar, the spatial resolution scribed in Held et al. (2010) for other supercell storms of the 28-wide radar beam is low. These combined aspects that occurred in the state of SP. may account for the lack of a clear-cut hook echo in the The storm remained strong and displayed a cyclonic images; a closer look in Fig. 6c, however, shows a hint of circulation in the radial velocity field (not shown) as it aweakechoregionjusttotheeastofthelabelCAM.At caused damage and one casualty in the southern sub- 2037 UTC (Fig. 6e) the storm showed signs of weakening; urbs of Capivari (Fig. 5). Figure 6 shows composite a trend that would continue until the last volume scan reflectivity (left column) and constant altitude plan tracked by TITAN.

Unauthenticated | Downloaded 09/28/21 08:21 PM UTC 3022 MONTHLY WEATHER REVIEW VOLUME 142

FIG. 5. Movement of the tornadic cell, as detected by BRU radar and processed by the TITAN tracking algorithm with a 40-dBZ threshold for reflectivity and 16-km3 threshold for volume. The alternating color shades enclose areas of composite reflectivity $40 dBZ at 7.5-min intervals. The three areas outlined in black south of Indaiatuba highlight the volume scans around the tornado time that are shown in Fig. 6. Relevant times are indicated: 1744 and 2059 UTC are the first and last volume scans tracked by TI- FIG. 4. Idealized sketch adapted for a left-moving cell in the TAN; 1944–1952 UTC is the time window in which the cell expe- Southern Hemisphere providing the approximate position of the rienced strengthening. Also shown are the 200-km BRU range ring, video camera (CAM) with respect to the tornadic thunderstorm the Tietê River, and some of the nearby towns. and its main structures [rear-flank downdraft (RFD); front-flank downdraft (FFD); updraft (U); tornado (T)]; the vector in the top- right corner indicates the actual storm motion for the Indaiatuba at 2022 UTC for the lower elevation PPIs. Although cell, while streamlines depict the conceptual model for the storm- BRU radar is not a WSR-88D, this result is meaningful relative near-surface flow. [Sketch adapted from Markowski and given that the actual cyclonic circulation, poorly resolved Richardson (2010) and based on the original model by Lemon and 8 Doswell (1979).] by BRU’s 2 beamwidth, was likely stronger than the one indicated in Fig. 6b. However, the assessment of Stumpf et al.’s criteria regarding the depth and longevity of the The large volume of the radar pulse 200 km away from cyclonic circulation for a rigorous characterization of BRU and thus, the scarcity of scatterers filling it, pre- was not possible because of numerous clude a precise analysis of the radial velocity fields. The missing data in the higher elevation PPIs at 2022 UTC and best depiction for the midlevel circulation within the in PPIs from the precedent 2015 UTC scan and sub- storm is provided by CAPPIs of the ground-relative sequent 2029 and 2037 UTC scans. radial velocities interpolated at 6.5 km (for the 2022 and Regarding the longevity of the circulation, it should be 2037 UTC volume scans; Figs. 6b and 6f, respectively) stated that radial velocity patterns similar to that shown and 8.0 km for the 2029 UTC volume scan (Fig. 6d). At in Fig. 7 were found in PPIs from volume scans preceding 2022 UTC the radial velocities display a broad area of 2015 UTC (not shown). In addition, the in situ confir- (SH) cyclonic circulation to the southwest of Indaiatuba mation of a strong tornado and of its parent low-level (Fig. 6b). Inbound and outbound velocities above mesocyclone (Fig. 3) points to the existence of a meso- 2 11 m s 1 are found and TITAN analysis indicates shear aloft around the time of volume scans shown in 2 2 of 4.5 3 10 3 s 1 with this structure. Similar quantitative Fig. 6 (Trapp et al. 2005). Hence, the rather diffuse cy- results are obtained when individual PPIs are evaluated, clonic circulation pattern shown in Figs. 6b,d and 7 most especially for the lower elevation scans such as the one likely results from a midlevel mesocyclone sampled by illustrated in Fig. 7 for the 0.88 elevation scan (approx- a low-resolution radar beam (Wood and Brown 1997). In imately 6 km above ground level) at 2022 UTC. At line with a discussion conducted by Brotzge and Donner 2029 UTC (Fig. 6d) the general structure of the cyclonic (2013) for the , the difficulties found in this circulation is still discernible south of Indaiatuba, while work regarding the quantitative analysis of the radial at 2037 UTC (Fig. 6f) paucity of data hampers a mean- velocity fields have important operational implications to ingful inference about the velocity field. the nowcasting of tornadic storms in regions where radar Nearly all of the objective criteria described in Stumpf data coverage is poor. et al. (1998) for the detection of velocity couplets using Reflectivity cross sections along the A–B segments in- Weather Surveillance Radar-1988 Doppler (WSR-88D) dicated in Fig. 6 are shown in Fig. 8. An interesting feature systems, for a storm approximately 200 km away seen at 2022 UTC (Fig. 8a) is the downshear leaning of from the radar, are satisfied for the Indaiatuba storm the high reflectivity core. At 2029 UTC (Fig. 8b)such

Unauthenticated | Downloaded 09/28/21 08:21 PM UTC SEPTEMBER 2014 P I C T U R E O F T H E M O N T H 3023

FIG. 6. (a),(c),(e) Composite reflectivity and (b),(d),(f) CAPPIs of radial velocities from BRU radar around the time of the Indaiatuba tornado. The embedded polygon in light blue in the images outlines the 40-dBZ threshold. Positive (negative) values of radial velocities 2 are outbound (inbound) velocities in m s 1. The tornado path is marked with cross signs, and CAM indicates the position of the video camera. Line segments A–B represent base lines for the vertical cross sections shown in Fig. 8 and are oriented along the tornado damage track. All indicated times refer to the end of the respective volume scan. a structure is no longer seen along the same A–B segment 40-dBZ reflectivity threshold shown in Figs. 8b and 8d is and the depth of the high reflectivity core is lower since rather modest, the same being found for cross sections this cross section is now placed closer to the edge of the through the central portion of the storm (not shown). It is 40-dBZ polygon, which has reduced in size (Fig. 6c). our experience that severe storms observed in central SP Compared to supercells documented in other parts of the during austral autumn are relatively low topped (Held world (e.g., Kumjian and Ryzhkov 2008) the height of the et al. 2010).

Unauthenticated | Downloaded 09/28/21 08:21 PM UTC 3024 MONTHLY WEATHER REVIEW VOLUME 142

FIG.7.AsinFig. 6b, but showing the respective PPI in polar coordinates for the 0.88 elevation scan. c. The synoptic-scale environment

1) SATELLITE IMAGERY To describe the synoptic-scale environment in which the tornadic thunderstorm evolved, it is useful to start with the enhanced thermal infrared images from GOES- 12 satellite shown in Fig. 9, valid a few hours around the tornado event (no sectorized geostationary satellite image was found for the time of the tornado). At 1730 UTC (Fig. 9a) deep convection was already under way ahead of an advancing surface cold front posi- tioned in between Santa Catarina and Paraná states, south of SP, as determined by subjective analysis. A gradient FIG. 8. (a)–(c) Vertical cross sections of reflectivity from BRU in brightness temperature is discernible across SP, with radar along the line segments A–B indicated in Fig. 6 for the three selected times. Reflectivity color scheme is as in Fig. 6. The hori- southern and eastern sections displaying cloudy conditions, zontal length scales are not constant. and even fully developed thunderstorms, while western and northern areas of the state displayed fewer and patches of clear skies. Combining this information with the just around the time of the thunderstorm initiation and radar analysis described above, it appears that storm ini- a couple of hours before the tornado) for the entire La tiation occurred at the transition region between partially Plata basin (Fig. 10a, only) and for south-southeastern cloudy skies to the west and north of SP and cloudy skies to Brazil (Figs. 10b–e and 11a–d). Figure 10a shows the east and south of the state. 500-hPa geopotential heights and areas with negative The conditions a few hours after the demise of the omega velocity, and indicates that SP was located just tornado (Fig. 9b) showed widespread convective activity downstream of a broad migratory SH trough. As ex- over the totality of southern and eastern SP, with new pected, a large area of upward motion was induced east discrete cells developing to the north-northwest. By this of the moving trough, where 500-hPa heights were fall- time, the city of São Paulo was under very heavy rainfall, ing. At higher levels (not shown), the westerly leading to a significant flash flood event (Vasconcelos displayed a strong cyclonic curvature just east of the and Cavalcanti 2010). Andes around 258S, while over the southern Brazilian coast, around 308S, a change to anticyclonic curvature 2) MAP ANALYSIS AND ATMOSPHERIC PROFILES was evident on the jet, as a response to an upper-level Figures 10 and 11 show atmospheric fields extracted positioned just south of the equator over from NCEP CFSR valid at 1800 UTC 24 May 2005 (i.e., Brazil. No important feature of the jet stream structure

Unauthenticated | Downloaded 09/28/21 08:21 PM UTC SEPTEMBER 2014 P I C T U R E O F T H E M O N T H 3025

FIG. 9. Enhanced thermal infrared imagery from GOES-12 satellite over São Paulo state (outlined by a thick black contour) and surrounding areas, valid at (a) 1730 UTC 24 May 2005 and (b) 0000 UTC 25 May 2005. Areas with brightness temperatures below 2308C are enclosed by black dotted lines (below 2708C are indicated in dark gray with closed white circles). The approximate position of the surface cold front and of the surface baroclinic trough is also illustrated. The closed black circle in both panels is the location of Indaiatuba, while the open circle and black arrow in (a) depict the storm initiation point and the direction of storm motion, respectively. The white horizontal line in (a) is a data artifact. (Data source: DSA-CPTEC- INPE, NOAA.). was positioned over SP at that time, except for some Seluchi et al. 2003). The combination of the displaced indication of a weakly diffluent flow. NAL and the developing produced Surface variables over southeastern Brazil showed two a frontogenetical configuration as confirmed by the posi- low pressure centers in the mean sea level pressure field tive values of the frontogenetic function (computed fol- (Fig. 10b): one to the west and another one to the south of lowing Satyamurty and Mattos 1989) within the elongated SP. The latter was a developing extratropical cyclone as- dot-shaded areas with northwest–southeast orientation sociated with the migratory upper-level trough; the for- shown in Fig. 10b. mer characterized a baroclinic trough the genesis of which Figure 10c depicts air temperature at 2 m, which dis- occurred several hours before just east of the Andes played both a synoptic-scale gradient across the south of mountain range before drifting eastward also in associa- Brazil and extreme northern Paraguay and an important tion with the upper-level forcing. This was a manifestation smaller-scale temperature gradient across SP with lower of the so-called northwestern Argentinean low (NAL; (higher) temperatures to the east (north and west) of the

Unauthenticated | Downloaded 09/28/21 08:21 PM UTC 3026 MONTHLY WEATHER REVIEW VOLUME 142

Unauthenticated | Downloaded 09/28/21 08:21 PM UTC SEPTEMBER 2014 P I C T U R E O F T H E M O N T H 3027 state. This pattern is consistent with the prevailing cloud and western SP, NCEP CFSR underestimated both cover (Fig. 9a). The city of Indaiatuba was located temperature and dewpoint by as much as 58C. In eastern within the cloudy sector with lower temperatures, while SP the differences are less but still with potential impli- the tornadic thunderstorm initiation occurred farther to cations regarding the assessment of the prevailing ther- the northwest where less contributed to modynamic conditions just preceding the tornado. warmer surface temperatures. NCEP CFSR fields show Moisture availability in a deeper layer is inspected by SP under northwesterly surface winds (barbs in Fig. 10c) examining the NCEP CFSR average dewpoint temper- except for the eastern section that includes Indaiatuba, ature in the lowest 30 hPa and the 1000–800 hPa verti- where surface winds are indicated from the northeast. cally integrated moisture convergence (MCONV; Fig. Albeit weak, these northeasterlies may have played 10d). Most of SP, including Indaiatuba and the genesis some role in producing a mesoscale environment con- area of the tornadic storm, was within a northwest– ducive to the tornadic event. southeast-oriented sector with average dewpoints above To assess how well the NCEP CFSR is representing 168C, characterizing a moist environment. In fact, given the local surface conditions we examine Fig. 12a, which that NCEP CFSR underestimated surface dewpoints, it indicates plots of the METAR hourly reports conducted is likely that the 30-hPa-averaged dewpoints shown in from 1600 to 2000 UTC (i.e., just prior to the tornadic Fig. 10d are also underestimated. The MCONV field in event) at the Viracopos International Airport in the Fig. 10d highlights the linearly oriented convergent nearby city of Campinas (SPKP), approximately 20 km patterns to the south and west of SP associated with the northeast of the tornado touchdown location (see Fig. ongoing frontogenesis; in contrast, over central SP the 1b). Surface winds were continuously reported from the MCONV field was more diffuse, indicating less linearly northeast, with cloudy skies and relatively low temper- organized forcing for convective initiation, which may atures for midafternoon hours. have been a factor supporting a more discrete mode of A broader view of METAR reports at 1800 UTC is convection. This further contributes to the evidence that provided in Fig. 12b, with temperature and dewpoint the Indaiatuba tornadic storm developed as a part of temperature being compared to the respective NCEP a prefrontal convective activity. CFSR values. A temperature gradient was evident tow- Given that the east-central portion of SP was under ard the north and west portions of the state. Dewpoint a thick layer of cloud around 1800 UTC, the impact of temperatures were more variable, but with some in- cloud shading upon the magnitude of the conditional in- dication of moister conditions to the west. With the stability for a surface air parcel should also be clear. The exception of station SBRP, METAR sites from the north shaded field in Fig. 10e is the surface-based CAPE valid 2 and west reported winds from the northwest or north- at that time. Values above 1000 J kg 1 are indicated to the northwest (SBBU), while to the east SBKP and SBMT north and south of Indaiatuba, while a narrow stretch of reported northeasterly winds. Thus, qualitatively speak- low CAPE is discernible in east-central SP. Midlevel ing, there is a good agreement between NCEP CFSR lapse rates, also depicted in Fig. 10e, display no high surface fields at 1800 UTC (Fig. 10c) and the actual local values over SP, becoming significant only over extreme observations. A more quantitative examination, however, southern Brazil. indicates important differences in terms of temperature To perform an evaluation of the NCEP CFSR’s con- and moisture. For most sites, especially those in central ditional instability fields valid at 1800 UTC 24 May 2005

FIG. 10. Meteorological fields from NCEP CFSR valid at 1800 UTC 24 May 2005 over (a) the La Plata basin and over (b)–(e) south-southeastern Brazil. (a) 500-hPa: geopotential heights (thick solid 2 2 contours, in dam), upward vertical motion (gray shading, in Pa s 1), and winds (barbs, in m s 1); (b) mean sea level pressure (thick solid contours, in hPa), and the surface frontogenetic function, only 2 2 where is equal to or greater than 4 3 10 10 K(ms) 1 (densely dotted shading); (c) winds at 10 m 2 (barbs, in m s 1) and air temperature at 2 m (gray shading, in 8C); (d) mean dewpoint temperature in the first 30 hPa, only where is equal to or greater than 168C (thick solid contours, in 8C) and moisture divergence (only where is negative) integrated over the 1000–800-hPa layer (gray shading, 2 2 2 2 in 10 3 ms 1 kg kg 1); and (e) surface-based CAPE (gray shading, in J kg 1) and lapse rates for 2 the 700–500-hPa layer, only where is equal to or greater than 6.58Ckm 1 [thick solid contours, in 2 (213)8Ckm 1]. In all figures the state of São Paulo is outlined by a thick dotted contour and the location of Indaiatuba is depicted by a solid black circle. The box in (a) shows the close-up area for (b)–(e).

Unauthenticated | Downloaded 09/28/21 08:21 PM UTC 3028 MONTHLY WEATHER REVIEW VOLUME 142

FIG. 11. As in Figs. 10b–e, but for (a) 850-hPa: geopotential heights (thick solid contours at every 1 dam, and labeled at every 6 dam), winds 2 2 (barbs, in m s 1), and wind speed (gray shading, in m s 1); (b) 0–6-km (bulk) vertical , only where is equal to or greater than 2 2 2 20 m s 1 (thick solid contours, in m s 1), and 0–1-km (bulk) vertical wind shear (gray shading, in m s 1); (c) 0–3-km storm-relative helicity 2 2 (gray shading, in m2 s 2), with the white dashed lines enclosing regions where the 0–1-km storm-relative helicity is less than 2200 m2 s 2,and 2 with barbs indicating the estimated storm motion in m s 1; and (d) dewpoint depression at 2 m, only where is equal to or less than 38C(thick solid contours, for 18 and 38C) and height of the lifting condensation level (for a surface air parcel) (gray shading, in m).

Unauthenticated | Downloaded 09/28/21 08:21 PM UTC SEPTEMBER 2014 P I C T U R E O F T H E M O N T H 3029

FIG. 12. Surface observations in the state of São Paulo on 24 May 2005. (a) Hourly METARs from SBKP site (see Fig. 1b for location relative to Indaiatuba) a few hours preceding the Indaiatuba tornado; (b) map of METAR reports at 1800 UTC; in parentheses are the 2-m temperature and dewpoint temperature extracted from CFSR’s grid points nearest to each corresponding METAR site. The dashed line indicates the trajectory of the tornadic storm 2 within the indicated times. In both (a) and (b) barbs are in m s 1 and temperatures in 8C. we first examine the 1200 UTC 24 May 2005 and The reanalysis-derived sounding valid at 1200 UTC 0000 UTC 25 May 2005 soundings of São Paulo city (solid lines in Fig. 13a) shows a profile that is saturated or (SBMT; Figs. 13a,b) located approximately 75 km nearly saturated for most of the troposphere. While the southeast of Indaiatuba (see Fig. 1b for location) and temperature profile shows a good agreement with the compare them with the respective profiles from NCEP observations above 800 hPa, the presence of cloudiness in CFSR. Overall, the actual soundings (dashed lines) dis- the NCEP CFSR ‘‘sounding,’’ not present in the real play a moist tropical-like troposphere devoid of deep sounding, may explain the underestimated temperature at layers of strong lapse rates. Nevertheless, an increase in the surface. At 0000 UTC 25 May (Fig. 13b), one partic- the 975–700-hPa is discernible for the 0000 UTC ularly relevant feature is the weaker lower-tropospheric 2 25 May profile, which displays some CAPE [309 J kg 1 lapse rate in the NCEP CFSR profile as compared to 2 (363 J kg 1) for the surface (most unstable) air parcel] observations. The important distinctions described above along a rather shallow layer. It can be argued that none highlight some difficulties in relying solely in reanalysis of these represents a true proximity sounding (Potvin data to assess the conditional instability observed around et al. 2010) of the Indaiatuba tornadic event, and that the 1800 UTC, when no operational sounding was available. 0000 UTC 25 May 2005 is, at least partially, contami- The finding that NCEP CFSR underestimated surface nated by ongoing convection over São Paulo (Fig. 9b). temperatures and dewpoints at 1800 UTC (Fig. 12b) also However, some relevant assessment of the NCEP CFSR raises questions about the interpretation of the surface- profiles can be carried out. based CAPE values (Fig. 10e).

Unauthenticated | Downloaded 09/28/21 08:21 PM UTC 3030 MONTHLY WEATHER REVIEW VOLUME 142

FIG. 13. Skew T–logp diagrams depicting the actual soundings conducted at SBMT site (thick dashed lines for both T and Td) and the CFSR vertical profiles for the nearest grid point (thick solid lines for both T and Td) valid at (a) 1200 UTC 24 May 2005 and (b) 0000 UTC 25 May 2005.

However, as illustrated for other reanalyses (e.g., of the surface air parcel when observations are used. For Allen and Karoly 2014; Gensini et al. 2014), surface the eastern section of the state (SBKP and SBMT) where modified profiles from the NCEP CFSR can provide lower temperatures were observed, the distinctions are valuable information where observations are limited. To less but still present and always suggesting a CAPE un- illustrate that we analyze NCEP CFSR profiles valid at derestimation (Figs. 14e,f). 1800 UTC 24 May 2005 for the six METAR sites de- Given the points above, the highly detailed CAPE picted in Fig. 12b and compare the adiabatic ascent of features shown in Fig. 10e may not be realistic. Based on a surface air parcel using NCEP CFSR data (labeled 1) the mismatch with the surface observations, it is rea- and the METAR reports (labeled 2). The results are sonable to state that at least moderate instability was in indicated in Fig. 14. Since the real profiles are unknown place in the sector where NCEP CFSR shows low CAPE it is not possible to perform a complete quantitative values in between SBBU and SBKP (see Fig. 12b). It evaluation of NCEP CFSR, but in all six locations the should also be mentioned that tornadic environments in METAR-based surface air parcel is more unstable than some parts of the world and times of the year do not the NCEP CFSR counterpart. This is particularly true display extreme values of instability, but rather strong for SBDN site (Fig. 14a) in far west SP (Fig. 12b) where low-level wind shear (Hanstrum et al. 2002). In this the 58C difference in both temperature and dewpoint context, we shall now turn our attention to the analysis was found. Perhaps more importantly is Fig. 14c for the of the kinematic fields. SBBU site in the central part of SP within the stretch of Figure 11a illustrates 850-hPa geopotential height and low CAPE values shown in Fig. 10e. This site is close to wind fields. The developing extratropical cyclone over the storm initiation point around 1744 UTC (Fig. 12b). the southern coast of Brazil is evident, with strong Figure 14c displays a potentially significant destabilization northwesterly and northerly flow over the warm sector

Unauthenticated | Downloaded 09/28/21 08:21 PM UTC SEPTEMBER 2014 P I C T U R E O F T H E M O N T H 3031 to the north and east of the cyclone. This wind pattern The wind profile for the SBMT upper air site is de- was entirely within the region of high moisture content picted in Fig. 15 by environmental hodographs from the (Fig. 10d), highlighting the relevant role played by the surface to the mid- to the upper troposphere. First, Fig. lower-tropospheric flow in transporting moisture pole- 15a superimposes two observed hodographs; the thin ward. The 850-hPa winds on the western part of the (thick) line refers to the 1200 UTC 24 May (0000 UTC domain were particularly strong reaching more than 25 May) sounding. As expected, both show a coun- 2 24 m s 1, with a LLJ structure (Vera et al. 2006). The terclockwise turning of the wind with height in the LLJ over southern Brazil was a response to height falls lower troposphere, indicative of warm advection at (as strong as 240 gpm just off the Brazilian coast from low levels (SH). Some striking features when com- 1200 to 1800 UTC) induced by the approaching migra- paring the two hodographs are the remarkable tory system seen in Fig. 10a. Comparatively speaking, strengthening of the northwesterly flow around over SP the 850-hPa northwesterly flow was not as 2900 m AGL leading to a much longer and more strong at this time (Fig. 11a), despite some indication of curved hodograph, and the hodograph lengthening in a local wind speed maximum over central SP. the first 950 m for the later sounding. Computing SRH The magnitude of the bulk vertical wind shear is using such profile and the observed storm motion 2 shown in Fig. 11b for both 0–6-km (contours) and 0– leads to high magnitudes of both SRH3 (2397 m2 s 2) 2 1-km (shaded areas) layers. It is interesting that, while and SRH1 (2347 m2 s 2) for SRH1. The pattern most of southern Brazil displayed strong deep-layer shown in Fig. 15a confirms that a strong increase in 2 bulk shear (DLS), above 20 m s 1, over SP the DLS LLS was observed over eastern SP from morning to field displayed a horizontal gradient, becoming stronger evening hours on 24 May 2005. on the eastern half of the state. A similar pattern is Figures 15b and 15d compare the observed hodo- discernible in the low-level bulk shear field (LLS), graphs with the NCEP CFSR counterparts, while Fig. 2 reaching values above 16 m s 1 only over east central SP, 15c depicts the 1800 UTC NCEP CFSR hodograph including the Indaiatuba region. Under a moderately alone. The lower-tropospheric portion of the CFSR strong 850-hPa northwesterly flow, the higher elevation NCEP 1200 UTC hodograph (Fig. 15b) agreed with the of the eastern half of the state may have had an in- observations in terms of the maximum wind speed fluence in the increase in LLS from western SP to east- within the 0–1-km layer, although this maximum oc- central SP. curred 200 m lower than in the actual hodograph. Sur- Given the combination of strong DLS and LLS asso- face winds were stronger in the reanalysis, and above ciated with a northwesterly flow aloft (see also Fig. 10a) 2000-m deviations from the observations became more and the northeasterly surface winds in that same area evident. The 0000 UTC 25 May hodograph (Fig. 15d) (Figs. 10c and 12b), strong speed and directional vertical showed that NCEP CFSR replicated the significant en- shear appeared to be in place over east-central SP. To hancement of the LLS as well as the curvature of the analyze this, 0–3- and 0–1-km storm-relative helicity hodograph associated with the northwesterly LLJ. Despite (SRH3 and SRH1, respectively) were considered, as differences in the surface winds and above 4000 m, the well as the estimated storm motion (ESM; Fig. 11c). It principal aspects associated with the lower-tropospheric should be kept in mind that, for the SH, negative values wind profiles, namely, shape and length of the hodograph, of storm-relative helicity (SRH) are the ones relevant were relatively well represented by NCEP CFSR. for evaluating severe storm potential associated with An evident trend in increasing speed and directional left-moving cells. The ESM field was determined using LLS is discernible in the NCEP CFSR wind profiles (Figs. the Davies and Johns (1993) algorithm (adapted to the 15b–d), depicting an evolution that is in accordance with SH), which could be regarded as inappropriate for the establishment of a synoptic-scale LLJ (Doswell 1991). South American supercell storms. Nevertheless, a com- This represents one relevant aspect because, while the parison of the ESM over east-central SP (u 5 12.8; y 5 role played by the LLJ in conditioning environments 2 24.3 m s 1) with the radar-determined storm motion conducive to severe thunderstorms is widely documented 2 discussed earlier (u 5 15.9; y 528.1 m s 1), shows, for (e.g., Johns 1993; Doswell and Bosart perhaps surprisingly, a fairly good agreement between 2001), in Brazil this documentation is still poor (Silva Dias them. In fact, if one employs the observed storm motion 2000), particularly when it comes to tornadic supercells. for the computation of SRH the change will be a less The qualitative agreement between NCEP CFSR than 10% (magnitude) decrease in SRH3, from 2321 to wind profiles and observations at SBMT at 1200 UTC 24 2 2293 m2 s 2, and a 6% (magnitude) increase in SRH1, May and 0000 UTC 25 May, especially at low levels, 2 from 2202 to 2214 m2 s 2, for the grid point closest to provides confidence that the substantially negative Indaiatuba. values of SRH3 and SRH1 shown over Indaiatuba in

Unauthenticated | Downloaded 09/28/21 08:21 PM UTC 3032 MONTHLY WEATHER REVIEW VOLUME 142

FIG. 14. Skew T–logp diagrams displaying vertical profiles of T and Td (thick solid lines) valid at 1800 UTC 24 May 2005 extracted from NCEP CFSR grid points nearest to each of the six METAR stations indicated in Fig. 12b.Ineach panel the thick dotted lines depict adiabatic ascents for two surface air parcels: using T and Td at 2 m from CFSR (air parcel 1), and T and Td from the respective METAR report (air parcel 2). METAR station identifiers are indicated on the top of each panel.

Unauthenticated | Downloaded 09/28/21 08:21 PM UTC SEPTEMBER 2014 P I C T U R E O F T H E M O N T H 3033

FIG. 14. (Continued)

Fig. 11c are realistic. This, combined with high values of 4. Summary and final remarks DLS and LLS, indicates a kinematic profile consistent This study described a rare significant tornado over with a significant tornadic environment (Rasmussen and southeastern Brazil. The video documentation of the Blanchard 1998; Rasmussen 2003; Thompson et al. 2003; episode showed that the base-level of the parent storm Markowski et al. 2003). displayed morphological characteristics commonly as- Studies have shown that the tornadic mode in super- sociated with midlatitude tornadic supercells. The cell environments tends to be favored in regions with multiple-vortex structure of the tornado is also evident strong LLS and low lifting condensation level (LCL) (or, in the images, which adds scientific value to the video equivalently, boundary layers with high relative hu- given the scarce visual documentation of significant tor- midity; Markowski et al. 2002; Thompson et al. 2003; nadoes in South America. An interesting feature present Craven and Brooks 2004; among others). In that sense, it in the video is the absence of within the is interesting to note that the METAR observations at RFD wrapping around the tornadic circulation on the SBKP (Fig. 12a) did report near-saturated conditions western flank (i.e., rear) of the low-level mesocyclone. continuously during several hours prior to the tornado Given the absence of finescale observations sampling the event. Similarly, Fig. 11d shows that LCL heights were environment near and within the storm (e.g., Kosiba et al. lowest over central SP and associated with a very narrow 2013), the cause(s) for the precipitation-free RFD and its dewpoint depression of less than 18C. (their) possible relation to the genesis and strength of Regarding the synoptic-scale conditions that favored the Indaiatuba tornado are unclear. It is worth mention- the development of a significant tornado, our analysis ing, though, that the study by Wakimoto and Cai (2000), shows that intense speed and directional wind shear comparing observed storm-scale structures between a combined with low LCLs over the Indaiatuba region nontornadic supercell and a tornado-producing counter- have contributed to the conditioning of the pretornadic part in the United States, found ‘‘more extensive pre- environment. cipitation echoes behind the rear-flank gust front’’ in the

Unauthenticated | Downloaded 09/28/21 08:21 PM UTC 3034 MONTHLY WEATHER REVIEW VOLUME 142

FIG. 15. Environmental hodographs from surface (SFC) to mid–upper troposphere for SBMT: observations at (a) 1200 UTC 24 May 2005 (thin line) and 0000 UTC 25 May 2005 (thick line); NCEP CFSR hodographs valid at (b) 1200 UTC 24 May 2005, (c) 1800 UTC 24 May 2005, and (d) 0000 UTC 25 May 2005 are represented by thin lines with the corresponding observations superimposed to them with thick lines in (b) and (d). All indicated heights are with respect to the surface (or model surface in the case of NCEP CFSR). The star in (a) depicts the observed storm motion. nontornadic storm. Future studies on the Indaiatuba the 28 width of the radar beam. Nevertheless, some in- tornadic cell should include a realistically configured dication of a weak echo region on the northwestern flank high-resolution numerical simulation of the parent storm of the storm was found in the reflectivity field. As the and surrounding environment, which may provide im- storm approached Indaiatuba, its midlevel cyclonic cir- portant insight regarding the mechanisms that led to culation was also detected, although a typical mesocy- . clone detection algorithm (Stumpf et al. 1998) would Weather radar analysis confirmed that the parent most likely not have flagged the structure as a mesocy- storm was a left-moving cell. Reflectivity values were clone, given the poorly resolved radial velocity field. not particularly high and the storm was rather low top- As for operational implications (Brotzge and Donner ped as compared to ‘‘classic’’ supercells documented in 2013), forecasters working in tornado-prone areas other parts of the world (e.g., Kumjian and Ryzhkov 2008). Interestingly, the study by Held et al. (2010) re- around the world, some of them not densely covered by ports further examples of supercells with relatively low radars, must be reminded that features such as meso- tops over SP during the cold (austral autumn and hook echoes may be difficult to detect and ). Despite the strong tornado confirmed on when tornadic storms are distant from the radar and/or the ground, no evident hook echo was observed; its de- when the radar beamwidth is greater than 18. Thus, tection may have been hampered by a combination of knowing the configuration of the specific radar system is the long distance between the storm and the radar and important.

Unauthenticated | Downloaded 09/28/21 08:21 PM UTC SEPTEMBER 2014 P I C T U R E O F T H E M O N T H 3035

Assessment of the synoptic-scale atmospheric condi- (e.g., Rasmussen et al. 2000), as well as the mesoscale tions prevailing around the time of the event found that and storm-scale variability of weather parameters within the tornadic thunderstorm developed under moderately the inflow layer of the storm (e.g., Markowski et al. strong synoptic forcing, in a moist prefrontal environment 1998b; Parker 2014) should be assessed in future studies and under a developing synoptic-scale LLJ. Evaluation of in order to fine-tune the characterization of the pro- the thermodynamic forcing based on the combination of cesses that favored the tornado formation. Of particular NCEP CFSR data and observations suggests that the interest is the fact that convective activity was observed severe thunderstorm initiation occurred in warm and ahead of the tornadic cell, which may have produced moist conditions in central SP under moderate condi- surface boundaries and modified the air mass effectively tional instability. This quantitative analysis, however, was ingested by the severe storm; in addition, the storm limited by the lack of observations on the mesoscale. followed a path that coincided quite well with a river Farther downstream, over central-eastern SP, where valley, which is a topographic feature also capable of the tornado was observed, surface temperatures were locally modifying an air mass (Acevedo et al. 2007). lower but the combination of a northwesterly LLJ and Our results stress the value of the ingredients-based northeasterly surface winds promoted strong speed and approach and of a judicious choice of weather parame- directional vertical wind shear, especially at low levels. ters to characterize the South American severe weather In addition, the local boundary layer displayed very high environment (Nascimento 2005). Moreover, it is im- relative humidity leading to low surface-based LCL portant to point out a few distinctions in the synoptic heights. These findings agree with previous studies that patterns conducive to severe storms and tornadoes be- showed that substantial lower-tropospheric vertical tween South and North America. The Indaiatuba tor- wind shear and near-saturated boundary layer condi- nado occurred in a prefrontal environment far northwest tions are among the best atmospheric discriminators of from the low center of an extratropical cyclone at the tornadic environments once a severe storm is formed surface (i.e., very far from the surface warm front; Fig. (Craven and Brooks 2004). It is also interesting to 10b). This synoptic pattern, also found in other studies mention Parker (2014) who examined several soundings (Nascimento and Foss 2010), is somewhat distinct from that sampled the near-storm conditions of tornadic and the classic conceptual model that describes a weather nontornadic supercells at different stages of their life pattern favorable for severe thunderstorms in North cycles during the Verification of the Origins of Rotation America under strong synoptic forcing (Fig. 5 of Johns in Tornadoes Experiment 2. For the later stages of 1993), where the most favorable region usually is much the supercells’ development Parker (2014) described closer to the center of the surface cyclone and to the intensification of the low-level wind shear and the warm front. Additional investigations focusing on SRH concomitant with the drop (increase) in surface South American supercells are encouraged, ranging temperature (relative humidity) as the boundary layer from the preconvective environment to their internal underwent the evening transition toward nocturnal sta- structure. ble conditions. This trend in turn led to enhanced values of parameters indicative of severe weather potential. Acknowledgments. This study was concluded while It is tempting to trace a parallel between this time- the first author was in a postdoctoral sabbatical year at dependent transition in the near-storm atmospheric the Center for the Forecast of Severe Weather by Re- conditions with the very similar (but mostly space de- mote Sensing and Numerical Modeling, University of pendent) environmental transition that the Indaiatuba L’Aquila, Italy, funded by Brazil’s Centro Nacional de storm experienced when moving from central SP to the Desenvolvimento Científico e Tecnológico (National eastern portions of the state, described above. At this Center for Scientific and Technological Development; stage, however, the parallel traced with the work by CNPq) under Grant 245576/2012-6. The authors are Parker (2014) is just speculative. grateful to the three anonymous reviewers for their In summarizing the synoptic-scale analysis, it appears constructive comments that led to the improvement of that the magnitude of the kinematic parameters played the manuscript. The authors also wish to thank the fol- a more prominent role than the instability parameters in lowing: the Operations Center from Rodovia das Col- creating conditions conducive to this tornadic supercell, inas S. A. for authorizing the publication of the still similar to what is found for other regions in the world frames from the tornado video, National Center for (Hanstrum et al. 2002; Grams et al. 2012). However, this Atmospheric Research for making available the latest investigation does not provide a complete picture of the version of TITAN and assisting with its implementation near-storm atmospheric conditions. Possible storm in- at IPMet, Mrs. Isabela Pena Viana de Oliveira Marcelino teraction with low-level boundaries or convergence lines for sharing detailed information of the tornado damage

Unauthenticated | Downloaded 09/28/21 08:21 PM UTC 3036 MONTHLY WEATHER REVIEW VOLUME 142 assessment, and Mr. Eduardo de Paula from Instituto Fujita, T. T., 1973: Tornadoes around the world. Weatherwise, 26, Tecnológico SIMEPAR for preparing some of the map 56–62, doi:10.1080/00431672.1973.9931633. figures. Gensini, V. A., T. L. Mote, and H. E. Brooks, 2014: Severe thun- derstorm reanalysis environments and collocated radiosonde observations. J. Appl. Meteor. Climatol., 53, 742–751, doi:10.1175/ REFERENCES JAMC-D-13-0263.1. Grams, J. S., R. L. Thompson, D. V. Snively, J. A. Prentice, G. M. Acevedo, O. C., O. L. L. Moraes, R. Silva, V. Anabor, D. P. Hodges, and L. J. Reames, 2012: A climatology and compar- Bittencourt, H. R. Zimmermann, R. O. Magnago, and G. A. ison of parameters for significant tornado events in the Degrazia, 2007: Surface-to-atmosphere exchange in a river United States. Wea. Forecasting, 27, 106–123, doi:10.1175/ valley environment. J. Appl. Meteor. Climatol., 46, 1169– WAF-D-11-00008.1. 1181, doi:10.1175/JAM2517.1. Hanstrum, B. N., G. A. Mills, A. Watson, J. P. Monteverdi, and Allen, J. T., and D. J. Karoly, 2014: A climatology of Australian C. A. Doswell III, 2002: The cool-season tornadoes of Cal- severe thunderstorm environments 1979-2011: Inter-annual ifornia and southern . Wea. Forecasting, 17, 705–722, variability and ENSO influence. Int. J. Climatol., 34, 81–97, doi:10.1175/1520-0434(2002)017,0705:TCSTOC.2.0.CO;2. doi:10.1002/joc.3667. Held, G., A. M. Gomes, and K. P. Naccarato, 2010: The structure Atkins, N. T., M. L. Weisman, and L. J. Wicker, 1999: The influence and behaviour of supercell storms in the state of São Paulo, of preexisting boundaries on supercell evolution. Mon. Wea. Brazil. Sixth European Conf. on Radar in and Rev., 127, 2910–2927, doi:10.1175/1520-0493(1999)127,2910: Hydrology, Sibiu, Romania, ERAD. [Available online TIOPBO.2.0.CO;2. at http://www.erad2010.org/pdf/oral/wednesday/nowcasting/ Berbery, E. H., and V. R. Barros, 2002: The hydrologic cycle of the 05_ERAD2010_0376.pdf.] La Plata basin in South America. J. Hydrometeor., 3, 630–645, Johns, R. H., 1993: Meteorological conditions associated with doi:10.1175/1525-7541(2002)003,0630:THCOTL.2.0.CO;2. development in convective storms. Wea. Fore- Brooks, H. E., 2006: A global view of severe thunderstorms: Esti- casting, 8, 294–299, doi:10.1175/1520-0434(1993)008,0294: mating the current distribution and possible future changes. MCAWBE.2.0.CO;2. Severe Local Storms Special Symp., Atlanta, GA, Amer. Kosiba,K.,J.Wurman,Y.Richardson, P. M. Markowski, P. Robinson, Meteor. Soc., J4.2. [Available online at http://ams.confex.com/ and J. Marquis, 2013: Genesis of the Goshen County, ams/pdfpapers/102202.pdf.] Wyoming, tornado on 5 June 2009 during VORTEX2. Mon. ——, J. W. Lee, and J. P. Craven, 2003: The spatial distribution Wea. Rev., 141, 1157–1181, doi:10.1175/MWR-D-12-00056.1. of severe thunderstorm and tornado environments from Kumjian, M. R., and A. V. Ryzhkov, 2008: Polarimetric signatures global reanalysis data. Atmos. Res., 67–68, 73–94, doi:10.1016/ in supercell thunderstorms. J. Appl. Meteor. Climatol., 47, S0169-8095(03)00045-0. 1940–1961, doi:10.1175/2007JAMC1874.1. Brotzge, J., and W. Donner, 2013: The tornado warning process: Lemon, L. R., and C. A. Doswell III, 1979: Severe thunderstorm A review of current research, challenges, and oppor- evolution and mesocyclone structure as related to torna- tunities. Bull. Amer. Meteor. Soc., 94, 1715–1733, doi:10.1175/ dogenesis. Mon. Wea. Rev., 107, 1184–1197, doi:10.1175/ BAMS-D-12-00147.1. 1520-0493(1979)107,1184:STEAMS.2.0.CO;2. Cecil, D. J., and C. B. Blankenship, 2012: Toward a global clima- Marengo, J., M. Douglas, and P. L. Silva Dias, 2002: The South tology of severe hailstorms as estimated by satellite passive American low level jet east of the Andes during the LBA- microwave imagers. J. Climate, 25, 687–703, doi:10.1175/ TRMM and WET-AMC campaign of January-April 1999. JCLI-D-11-00130.1. J. Geophys. Res., 107, 8079, doi:10.1029/2001JD001188. Craven, J. P., and H. E. Brooks, 2004: Baseline climatology of Markowski, P. M., and Y. Richardson, 2010: Mesoscale Meteorol- sounding derived parameters associated with deep moist ogy in Midlatitudes. J. Wiley and Sons, 407 pp. convection. Natl. Wea. Dig., 28, 13–24. ——, J. M. Straka, and E. N. Rasmussen, 1998a: The occurrence of Davies, J. M., and R. H. Johns, 1993: Some wind and instability tornadoes in supercells interacting with boundaries during parameters associated with strong and violent tornadoes. The VORTEX-95. Wea. Forecasting, 13, 852–859, doi:10.1175/ Tornado: Its Structure, Dynamics, Prediction, and Hazards, 1520-0434(1998)013,0852:TOOTIS.2.0.CO;2. C. Church, C. Doswell, and R. Davies-Jones, Eds., Amer. ——, ——, ——, and D. O. Blanchard, 1998b: Variability of Geophys. Union, 573–582. storm-relative helicity during VORTEX. Mon. Wea. Rev., Dixon, M., and G. Wiener, 1993: TITAN: Thunderstorm Identifi- 126, 2942–2958, doi:10.1175/1520-0493(1998)126,2942: cation, Tracking, Analysis, and Nowcasting—A radar-based OOLLBG.2.0.CO;2. methodology. J. Atmos. Oceanic Technol., 10, 785–797, ——, ——, and ——, 2002: Direct surface thermodynamic obser- doi:10.1175/1520-0426(1993)010,0785:TTITAA.2.0.CO;2. vations within the rear-flank downdrafts of nontornadic Doswell, C. A., III, 1991: A review for forecasters on the applica- and tornadic supercells. Mon. Wea. Rev., 130, 1692–1721, tion of hodographs to forecasting severe thunderstorms. Natl. doi:10.1175/1520-0493(2002)130,1692:DSTOWT.2.0.CO;2. Wea. Dig., 16, 2–16. ——, C. Hannon, J. Frame, E. Lancaster, A. Pietrycha, R. Edwards, ——, and L. F. Bosart, 2001: Extratropical synoptic-scale processes and R. L. Thompson, 2003: Characteristics of vertical and severe convection. Severe Convective Storms, C. A. wind profiles near supercells obtained from the Rapid Up- Doswell III, Ed., Amer. Meteor. Soc., 27–69. date Cycle. Wea. Forecasting, 18, 1262–1272, doi:10.1175/ ——, H. E. Brooks, and R. A. Maddox, 1996: Flash flood fore- 1520-0434(2003)018,1262:COVWPN.2.0.CO;2. casting: An ingredients-based methodology. Wea. Fore- Moller, A. R., 2001: Severe local storms forecasting. Severe Con- casting, 11, 560–580, doi:10.1175/1520-0434(1996)011,0560: vective Storms, C. A. Doswell III, Ed., Amer. Meteor. Soc., FFFAIB.2.0.CO;2. 433–480.

Unauthenticated | Downloaded 09/28/21 08:21 PM UTC SEPTEMBER 2014 P I C T U R E O F T H E M O N T H 3037

Nascimento, E. L., 2005: Severe storms forecasting utilizing con- Schwarzkopf, M. L., 1982: Severe storms and tornadoes in Argen- vective parameters and mesoscale models: An operational tina. Preprints, 12th Conf. on Severe Local Storms, San Antonio, strategy adoptable in Brazil (in Portuguese)? Braz. Meteor. TX, Amer. Meteor. Soc., 59–62. Mag., 20, 121–140. [Available online at http://www.scielo.br/ Seluchi, M., A. C. Saulo, M. Nicolini, and P. Satyamurty, 2003: pdf/rbmet/v25n4/09.pdf.] The northwestern Argentinean low: A study of two typical ——, and C. A. Doswell III, 2006: The need for an improved doc- events. Mon. Wea. Rev., 131, 2361–2378, doi:10.1175/ , . umentation of severe thunderstorms and tornadoes in South 1520-0493(2003)131 2361:TNALAS 2.0.CO;2. America. Symp. on the Challenges of Severe Convective Storms, Silva Dias, M. A. F., 2000: Storms in Brazil. Storms, R. Pielke Jr. and R. Pielke Sr., Eds., Vol. II, Routedge Press, 207–219. Atlanta, GA, Amer. Meteor. Soc., P1.18. [Available online ——, 2011: An increase in the number of tornado reports at https://ams.confex.com/ams/Annual2006/techprogram/ in Brazil. Wea. Climate Soc., 3, 209–217, doi:10.1175/ paper_102247.htm.] 2011WCAS1095.1. ——, and M. Foss, 2010: A 12-yr climatology of severe weather Stumpf, G. J., A. Witt, E. W. Mitchell, P. L. Spencer, J. T. Johnson, parameters and associated synoptic patterns for subtropical M. D. Eilts, K. W. Thomas, and D. W. Burgess, 1998: The South America. 25th Conf. on Severe Local Storms, Denver, National Severe Storms Laboratory mesocyclone detection CO, Amer. Meteor. Soc., 12A.5. [Available online at https://ams. algorithm for the WSR-88D. Wea. Forecasting, 13, 304–326, confex.com/ams/25SLS/techprogram/paper_175790.htm.] doi:10.1175/1520-0434(1998)013,0304:TNSSLM.2.0.CO;2. Parker, M. D., 2014: Composite VORTEX2 supercell environ- Thompson, R. L., R. Edwards, J. A. Hart, K. L. Elmore, and ments from near-storm soundings. Mon. Wea. Rev., 142, 508– P. Markowski, 2003: Close proximity soundings within supercell 529, doi:10.1175/MWR-D-13-00167.1. environments obtained from the Rapid Update Cycle. Wea. Potvin, C. K., K. L. Elmore, and S. J. Weiss, 2010: Assessing the Forecasting, 18, 1243–1261, doi:10.1175/1520-0434(2003)018,1243: impacts of proximity sounding criteria on the climatology of CPSWSE.2.0.CO;2. significant tornado environments. Wea. Forecasting, 25, 921– Trapp, R. J., G. J. Stumpf, and K. L. Manross, 2005: A reassessment 930, doi:10.1175/2010WAF2222368.1. of the percentage of tornadic mesocyclones. Wea. Forecasting, Rasmussen, E., 2003: Refined supercell and tornado forecast 20, 680–687, doi:10.1175/WAF864.1. parameters. Wea. Forecasting, 18, 530–535, doi:10.1175/ Vasconcelos, F. C., and I. F. A. Cavalcanti, 2010: Evaluation of the 1520-0434(2003)18,530:RSATFP.2.0.CO;2. ETA regional model high resolution forecasts for two cases of ——, and D. O. Blanchard, 1998: A baseline climatology of sounding- intense rainfall over the Serra do Mar region (in Portuguese). derived supercell and tornado forecast parameters. Wea. Fore- Braz. Meteor. Mag., 25, 501–512. [Available online at http:// casting, 13, 1148–1164, doi:10.1175/1520-0434(1998)013,1148: www.scielo.br/pdf/rbmet/v25n4/09.pdf.] ABCOSD.2.0.CO;2. Velasco, I., and J. M. Fritsch, 1987: Mesoscale convective com- J. Geophys. Res., ——, S. Richardson, J. M. Straka, P. M. Markowski, and D. O. plexes in the Americas. 92, 9561–9613. Vera, C., and Coauthors, 2006: The South American Low-Level Jet Blanchard, 2000: The association of significant tornadoes with a Experiment. Bull. Amer. Meteor. Soc., 87, 63–77, doi:10.1175/ baroclinic boundary on 2 June 1995. Mon. Wea. Rev., 128, 174– BAMS-87-1-63. 191, doi:10.1175/1520-0493(2000)128,0174:TAOSTW.2.0.CO;2. Wakimoto, R. M., and H. Cai, 2000: Analysis of a nontornadic storm Saha, S., and Coauthors, 2010: The NCEP Climate Forecast during VORTEX 95. Mon. Wea. Rev., 128, 565–592, doi:10.1175/ System Reanalysis. Bull. Amer. Meteor. Soc., 91, 1015–1057, 1520-0493(2000)128,0565:AOANSD.2.0.CO;2. doi:10.1175/2010BAMS3001.1. Wood, V. T., and R. A. Brown, 1997: Effects of radar sampling Salio, P., M. Nicolini, and A. C. Saulo, 2002: Chaco low-level jet on single-Doppler velocity signatures of mesocyclones events characterization during the austral season. and tornadoes. Wea. Forecasting, 12, 928–938, doi:10.1175/ J. Geophys. Res., 107, 4816, doi:10.1029/2001JD001315. 1520-0434(1997)012,0928:EORSOS.2.0.CO;2. Satyamurty, P., and L. F. Mattos, 1989: Climatological lower tro- Zipser, E. J., D. J. Cecil, C. Liu, S. W. Nesbitt, and D. P. Yorty, pospheric frontogenesis in the midlatitudes due to horizontal 2006: Where are the most intense thunderstorms on Earth? deformation and divergence. Mon. Wea. Rev., 117, 1355–1364, Bull. Amer. Meteor. Soc., 87, 1057–1071, doi:10.1175/ doi:10.1175/1520-0493(1989)117,1355:CLTFIT.2.0.CO;2. BAMS-87-8-1057.

Unauthenticated | Downloaded 09/28/21 08:21 PM UTC