Visual observations were made from the 14th floor of a building at the University of New South Wales (UNSW).
The plume from the fires at Cordeaux Reservoir was noted from UNSW by 1430. Cumulus was observed to the south-south-west, and had clearly changed into cumulonimbus by 1550.
Figure 2 is a radar image taken at 1540. It shows a
precipitation zone over the ocean about 40 km south-south-west of UNSW,
co-incident with the thunderstorm observed from UNSW. The radar reflectivity
peak corresponds to a zone of intense precipitation, with a rainfall rate
greater than 100 
. This was probably the source of cold
downdrafts and outflows. The centre of this zone is also marked on
figure 1.
Figure 2: Weather radar image taken at 1540.
At 1556 an unusual, low cloud was observed from UNSW, moving in an east or north-easterly direction. It was clearly made up of bushfire smoke, and although it was not a cloud in the standard sense, the word `cloud' will be used here to distinguish it from the initial smoke plume. The officer on duty at Sydney Airport's meteorological station estimated its height at about 1000~m and described it as ``stratus'', although he later told one of the authors that he had been unable to decide on its nature.
This stage is shown in figure 3. The dark cloud in the top right-hand quadrant of figure 3 is the base of the cumulonimbus. The smoke cloud does not extend all the way to the cumulonimbus base at the right-hand side. The mean height of the smoke plume is about half the height of the head of the smoke cloud; the smoke plume can just be made out in figure 3 as a sinuous band in the middle of the bright zone. An aircraft is descending to land on the east-west runway of Sydney Airport (runway 25).
Figure 3: Photograph taken from UNSW at 1555:55, in the direction of Cronulla. To see a
full-resolution image, click
here. All the
photographs have been digitized and standard histogram-equalization algorithms
have been applied to the images. The histogram equalization operates on the
overall intensity field, and as a result differences in colour tint between the
images may be created.
By 1601 the head of the smoke cloud was off Kurnell. This stage is shown in figure 4, where the head appears in sharp relief against the backdrop of cumulonimbus. An anemometer at Kurnell, from which data was logged every two seconds, showed no wind change at the time, confirming that the phenomenon was offshore.
Figure 4: Photograph taken from UNSW at 1601:22, in the direction of Kurnell. To see a full-resolution
image, click here.
Image details as figure 3.
The smoke cloud's head crossed a line through Long Bay at 1609, as shown in figure 5, at this stage appearing more diffuse. The sunlight coming from the west continues to highlight the smoke cloud against the dark background.
Figure 5: Photograph taken from UNSW at 1608:50, in the direction of Long Bay. To see a full-resolution
image, click here.
Image details as figure 3.
The phenomenon crossed a line through Coogee at 16:24-25, as shown in
figure 6. By this stage the rays of sunlight are
nearly normal to the smoke cloud, hence it appears black. Its structure appears
less well-defined than at the stage of figure 4.
The dark cloud at mid-levels is in the background. An aircraft can be seen
heading approximately west-south-west prior to turning to land on the east-west
runway. An anemometer mounted on the Ocean Reference Station (ORS1), situated
about 4 km east-north-east of Coogee and mounted 4.5 m above sea level, shows a
change from 5-6 
south-westerly to 10-11
west-south-westerly at 1625.
Figure 6: Photograph taken from UNSW at 1624:33, in the direction of Coogee. To see a full-resolution
image, click here.
Image details as figure 3.
It is likely that the phenomenon was a thunderstorm-generated gravity current, since it travelled away from a probable source of dense, cold outflow air (the thunderstorm reflectivity peak) and advected smoke normal to the prevailing wind. In estimating the distance the current travelled, it will be assumed that it propagated parallel to the coastline. The actual plan-form of the current front might have been linear or, as is more likely, a segment of an arc proceeding from a point-source disturbance. Hence the following inferences apply to the observation if it were a segment of an arc-like front or a portion of a linear front. It cannot have had a significant component of propagation velocity towards the coastline because it was never observed to cross the coast despite its probable origin only 8 km from shore, compared with a long-shore component of travel of 45 km from origin to Coogee. Furthermore, the prevailing wind would have tended to advect it offshore. The wind change recorded at ORS1 was west-south-westerly. This is consistent with shore-parallel and offshore propagation; however, caution should be exercised in relating the wind direction behind a gravity-current head to the direction of propagation, because gravity currents are expected to have three-dimensional internal flows (Simpson 1987).
Hence, given the angles of the lines
along which the photographs were taken, an assumption of shore-parallel
propagation can only result in an underestimation of the distance travelled and
hence the speed of propagation will be a lower-bound estimate. The distance
travelled from 1556 to 1624 is therefore at least 45 km. Using the segment of
travel from Kurnell to Coogee for best accuracy, the average speed of
propagation is at least 20 (40 knots). This is consistent
with the speeds typically observed for atmospheric gravity currents (Simpson
1987).
The feeder current within the head is faster than the propagation speed of the
gravity current. Thus, wind speeds within the head were a minimum of
20 (40 knots) and likely to be significantly higher, both
because of the underestimate of the propagation speed and because the feeder
current is faster than this. The speed of 10-11 recorded over
the ocean off Coogee may be because
ORS1 was on the lateral
periphery of the current. The gravity-current flow may also have begun to slow
down by that stage.
A minimum bound on the height of the current can be estimated from a blow-up of the photograph in figure 4, using the height of the tallest chimney at the Kurnell refinery. This sort of estimate would only be centred on the true height if the current were directly above Kurnell; since it is an unknown distance behind Kurnell, the true height of the current will be greater, owing to perspective. The chimney is 71 m above sea level, implying that the top of the nose of the current was at least 800 m above sea level and its body was at least 1000 m high, the latter estimate being consistent with the observation from Sydney Airport's meteorological station. Such heights are consistent with other atmospheric gravity currents (Simpson 1987).
