32nd Conference on Broadcast Meteorology/31st Conference on Radar Meteorology/Fifth Conference on Coastal Atmospheric and Oceanic Prediction and Processes

Sunday, 10 August 2003
The Bauru Flash Flood of 8 February 2001 and Options for Improved Warnings
Gerhard Held, Universidade Estadual Paulista, Bauru, Brazil; and J. L. Gomes
Poster PDF (441.0 kB)
The beginning of February 2001 was characterized by weather conditions typical for summer in the central State of São Paulo. On three days, relatively isolated, almost stationary storms developed into intense cells, accumulating vast amounts of precipitation above cloud base by means of a cyclonic shear, which created strong updrafts, as observed in the radial velocity field of the Bauru S-band Doppler radar, resulting in parts of Bauru being flooded.

The most severe flood was caused by a storm occurring on 8 February 2001, between 18:16 and 19:31, over the southern and western catchment of the Rio Bauru. Storms already began to develop within the radar range around noon (all times in LT=UT-3), growing in size and intensity, and merging into large complexes, especially in the north-north-east to east-south-east sector between 60-200 km range. At around 18:00, due to new development rapidly progressing from north and north-west within the 50 km range, the town of Bauru appeared to be threatened and a general warning for extremely heavy rainfall was issued to the Civil Defense Authority, solely based on radar observations.

The first indication of an extremely severe cell, just north-west of the catchment, was observed at 18:16 and manifested in a dramatic tilt of the echo core, with a significant vertical shear of radial velocities from +13 m.s-1 at cloud base to –11m.s-1 near the echo top (18:31), accumulating precipitation aloft. At 18:46, a well-pronounced shear in radial velocities (up to –1.8x10-3 s-1) could be observed near the ground along a 50 km long, radially-oriented line along the 325° azimuth, persisting for more than 30 min. Heavy rain in the catchment only began to fall at 19:16, with rainfall rates of up to 200 mm.h-1, causing flash floods in tributaries and the main river, which resulted in the loss of five lives through drowning and three due to collapsing structures, with an estimated material damage of about US $ 1.5 million.

The operational Regional ETA model (grid resolution of 40 km), initiated on 6 February 2001 at 21:00, predicted 48 hours ahead virtually no rain within the 240 km range of the Bauru radar for the 6-hour period ending at 21:00. However, post-analysis runs of the non-hydrostatic ETA model, with a grid resolution of 10 km for 6-hour time steps, centered over Bauru, and initiated on 6 February 2001, 21:00, using boundary conditions from the ETA Regional model, predicted the rainfall area in the north-west sector very well 48 hours ahead, but underestimated the amount by about a factor of 5 (£10 mm accumulated in 6 hours, compared to 50 mm measured in Bauru).

Further post analysis of this case revealed, that if a microburst (MB) warning module would have been available to the radar meteorologist in real-time, a more specific warning for the north-west sector could have already been issued to the Civil Defense Authorities at 17:01, when the first MB was flagged at <20 km range, followed by frequent warnings in the same region until the flooding event.

Supplementary URL: http://www.ipmet.unesp.br/publications/p4c.1_figures.htm