25 November 2007

Australian Bushfire Activity: 7-16 Nov 2007

Below is the latest 10-day hotspot composite for Australia from the NASA MODIS Rapid Response Global Fire Mapper website. The hotspots were compiled over the period from 7-16 November 2007.

Over the period, the major focus of fire activity in Australia remains in the northern portion of the continent, primarily NT and QLD. Comparing to previous maps, a distinct reduction in fire activity over the Top End is noted.

In QLD, many of these fires were reported in the last national update, particularly those in the Gulf and central portions of the state. Most of the fire noted in that region went out shortly after the date of the last report (which was 2 days into the current period). The fire activity on the western side of Cape York also continued, although the level of activity is gradually decreasing. Southeastern portions of the Peninsula, near Cooktown, continued to burn.

As before, the central coastal and southeastern regions of the states continued to see isolated hotspots, but no major fires have been reported or identified.

The Gulf country regions of both QLD and NT remained quite active, particularly on the NT side of the border. There are two main areas in NT near the coast -- one about 40 km NW of Booraloola, the other about 100 km to east of the town. These fires were roughly 60 000 to 100 000 ha in size. Fires also continued in eastern Arnhem Land, near the coast.

Across a broad swath of the western Top End, extending inland into central NT, some significant rain was observed early in the month. This has helped damp down the fire activity in those regions. Still, lightning did likely initiate some storms in Central NT, near Yuendumu, as discussed in this post.

Over in WA, significant fires continued to burn in west Kimberley, 100 km (or so) to the east of Derby. In the northern reaches of the Great Sandy Desert, a large fire burnt for most of the period, encompassing on the order of 500 000 ha. Several other large fires were noted to the east of this regions, out towards the general vicinity of Halls Creek and towards the NT border. The largest of these fires were on the order of 50 000 to 80 000 ha of area burnt.

The Pilbara region of WA also saw some bushfire activity during this period. This was widely scattered throughout the region. While widely scattered, some of these fires did affects areas on the order of perhaps 20 000 to 40 000 ha. Similarly in the southern deseret regions of WA, north of the Nullarbor and east of Laverton.

In the southwestern part of the state, multiple hotspots are seen in the forests of the region. One fire, in the Jarrah forest near Collie, was believed to have been deliberately lit. Others may be controlled burns, although in some cases these are wildfires likely wildfires.

Tasmania also saw some significant wildfire activity during this period. Several widely spaced hotspots are visible in the composite. Several difficult to control fires were reported on the Tasmanian West Coast, near Strahan and Zeehan. Arson is suspected in these fires.

A few scattered hotspots are noted in NSW, especially the NE corner and VIC. Interestingly, the greatest density of hotspots in both states are over Sydney and Melbourne. These are obviously spurious hotspots, die to some other cause. There have been no bushfires within the city limits of either town during this period. I actually did notices the Sydney example when scanning MODIS imagery...They all occurred in one overpass, suggesting they are not individual house fires or some such occurring over the entire period. There are many known errors associated with the hotspot detection. See the FAQ on the NAFI page for a start...

A Word on Size Estimates

Earlier, while looking at at some of the size calculations, I noticed a few discrepancies between areas I computed earlier using the spot information in NAFI and the later fire scar information. Namely, the fire scar estimates seemed to be lower, by perhaps as much as 25% lower. I don't fully understand the source of this difference-- for example,it could be the large symbols covering too much area or parts of the automatic fire scar mapper missing some burnt areas or just general sloppiness/imprecision on my part. There are differences in the resolution of the data used to compute the two different things, with higher resolution data in the fire scars. Whatever the cause, it provides some sort of useful bound on the accuracy of the estimates (say +/- 20% or so in the worse case). I always use agency estimates when available, and compare with other published sources when I can. This is probably not a major issue for most of you, but it should be kept in mind

No comments: