Until recently, it was believed, that tornadoes were rather rare events in Brazil, and very few of those reported had been observed within radar range. However, two tornadoes occurred during the afternoon of 25 May 2004 in the west of the State of São Paulo, and their tornado-vortex signatures had been observed by the S-band Doppler radars located in Bauru and Presidente Prudente, respectively. Another tornadic storm was recorded by the Bauru radar on 24 May 2005, which will be the main subject of this paper in comparison to the previous events. These well-documented occurrences prompted a detailed investigation, in an attempt to find relevant signatures in radar and lightning observations, which could be used for nowcasting and an early alert system. This is vitally important for the SIHESP (Portuguese acronym for “Hydrometeorological System of the State of São Paulo”) Project, which is currently being implemented, to provide early warnings of severe storms to local authorities, Civil Defense Organizations, industry and the general public.

The synoptic situation on 25 May 2005 showed a true severe weather outbreak, dominated by a cold front, which moved rapidly in a north-easterly direction (35-40 km.h^-1) from southern Paraná (12UT=09LT) to the central State of São Paulo (21LT). This intensified the already strong divergence at 200 hPa over the State, and together with the embedded jet stream, created areas of extreme instability, resulting in widespread pre-frontal rainfalls over the southern parts of the State of São Paulo. Embedded nuclei of extremely intense precipitation were accompanied by strong winds, causing severe damage in several towns of the central interior and a major flood in the City of São Paulo. At least one of the cells spawned a tornado, while another one created an exceptionally strong windstorm with cyclonic convergence. This is probably the first time that a multiple-vortex tornado had been recorded on a video in the southern hemisphere. The short video (17:30:57-17:33:20) clearly shows small tornado satellites rotating around the main tornadic axis. This phenomenon is only observed with very intense tornadoes generated by supercells.

At 14:31 (all times in LT=UT-3), an isolated cell was detected about 50km east-south-east of the Bauru radar, which developed into a small supercell within 30 min, with Z ³ 50 dBZ and echo tops (10 dBZ) around 13 km, lasting for more than 3 hours, while it moved towards east-south-east at 64 km.h^-1. A hook echo, characteristic of tornadic storms, was already observed at 16:08 at a range of 130km, together with a strong cyclonic circulation visible in the radial velocity field (Vr ranging from –22 to +12 m.s^-1. A hook echo, characteristic of tornadic storms, was already observed at 16:08 at a range of 130km, together with a strong cyclonic circulation visible in the radial velocity field (Vr ranging from –22 to +12 m.s^-1 indicative of rotation). The first touch-down probably occurred around 17:00, followed by another one shortly before 17:30, confirmed by a sudden absence of cloud-to-ground strokes from this cell, as observed by the Brazilian Lightning Detection Network (RINDAT). No variation of lightning parameters (peak current, multiplicity, polarity) was found for the tornadic cell. It is noteworthy, that during a 24-hour period about 20 thousand lightning strokes were recorded in the State of São Paulo by RINDAT, which is an absolute record for May during seven years of observations. The tornado reached F3 intensity in the Fujita scale during its mature stage, characterizing a significant event, with an estimated damage of US $ 42 million. Preliminary analysis indicates, that the destruction path of the tornado extended for approximately 15 km, with a width of up to 200 m. The trajectory of the tornado followed more or less the motion of the parent thunderstorm.

Output from the Regional and Meso-Eta models (40x40 km and 10x10 km resolution, respectively), such as model-generated profiles of wind and thermodynamic variables, as well as severe weather parameters (CAPE, bulk Richardson number shear, etc) replacing predicted by observed surface winds, is used to verify the predictability of such small-scale severe events. The predicted severe weather parameters did display skill in highlighting the existence of a pre-storm environment that was favorable for significant severe weather development in the region of the tornado that afternoon. Details of this analysis, as well as results from high-resolution simulations with the ARPS model, will be discussed as part of the presentation.

This study highlights the importance of Doppler radar and real-time lightning observations, especially as a useful tool for nowcasting techniques, to predict the development of extremely severe rain, hail or wind storms in Southeast Brazil. However, the signatures obtained so far still need to be fine-tuned with more cases, in order to derive more precise algorithms to issue automatic warnings for the forecaster.