U.S. light sweet crude for October delivery closed at $67.25 a barrel, after spiking above $68.

This week, the price fell when the report came out, then started rising.

U.S. data released Wednesday morning showed gasoline inventories in the world's biggest energy consumer tumbled 3.2 million barrels last week, or nearly 2 percent, with two weeks of the peak U.S. driving season to go.

"The gasoline draw is a bullish number, without doubt. The decline was much bigger than the 1.1 million barrels draw the Street was anticipating," said Jim Ritterbusch, analyst at Ritterbusch and Associates.

The market was also watching Tropical Storm Katrina gather strength over the Bahamas and take aim at Florida's southern tip and the oil producing Gulf of Mexico, which accounts for up to a quarter of U.S. oil output.

U.S. light sweet crude for October delivery closed at $67.35 a barrel, up $1.64, after spiking as high as $67.40.

Natural gas will be more expensive but heating oil will be non-existant.

atellite imagery...Doppler radar data from the Bahamas and Miami...
and reconnaissance wind data indicate TD-12 has become much better
organized this morning and has strengthened into Tropical Storm
Katrina. The initial intensity of 35 kt is based on an 1153z recon
925 mb flight-level wind report of 48 kt in the northeast quadrant
...Which equals about 36 kt at the surface using a Standard 75
percent reduction factor for that level. This intensity is also
supported by a consensus intensity estimate of t2.5/35 kt from all
three satellite agencies. The next aircraft is expected to
investigate Katrina this afternoon.


The initial motion estimate is 330/07. Katrina's center may be
developing a little more to the north of the previous fixes in
response to the bursts of deep convection that have been developing
in the northeast quadrant of the larger circulation envelope.
However...radar data suggests that small vortices or mesocyclones
are being generated within the convective bursts...and then
propagating westward along the north side of larger circulation.
The initial position remains roughly in the geometric center of all
the small vortices noted in radar data. Overall...recon data
indicate the wind field continues to consolidate. The forecast
track remains basically unchanged form the previous advisories. The
subtropical ridge to the north of Katrina that currently extends
east-west along 30-31n latitude is expected to slowly build
eastward...causing the cyclone to turn more westward after 24 hours
and cross the southern Florida Peninsula. After emerging over the
eastern Gulf of Mexico in 72 hours...the western portion of the
ridge is forecast to weaken and allow Katrina to move
northwestward.

Katrina has developed a symmetrical upper-level outflow pattern
within a relatively weak shear environment. Although the wind field
is currently elongated east-west...steady intensification seems
reasonable at this time...especially as an upper-level low just
southwest of the system moves away to the west. Katrina is expected
to become a hurricane prior to landfall...which is similar to the
SHIPS and GFDL models.


Forecaster Stewart




forecast positions and Max winds


initial 24/1500z 24.7n 76.7w 35 kt
12hr VT 25/0000z 25.4n 77.4w 40 kt
24hr VT 25/1200z 25.9n 78.4w 45 kt
36hr VT 26/0000z 26.0n 79.2w 55 kt
48hr VT 26/1200z 26.1n 80.1w 65 kt
72hr VT 27/1200z 26.3n 82.5w 40 kt
96hr VT 28/1200z 27.0n 84.5w 55 kt
120hr VT 29/1200z 29.0n 86.0w 65 kt