A LANDSAT MSS-DERIVED FIRE HISTORY OF KAKADU NATIONAL-PARK, MONSOONALNORTHERN AUSTRALIA, 1980-94 - SEASONAL EXTENT, FREQUENCY AND PATCHINESS

Citation
J. Russellsmith et al., A LANDSAT MSS-DERIVED FIRE HISTORY OF KAKADU NATIONAL-PARK, MONSOONALNORTHERN AUSTRALIA, 1980-94 - SEASONAL EXTENT, FREQUENCY AND PATCHINESS, Journal of Applied Ecology, 34(3), 1997, pp. 748-766
Citations number
61
Categorie Soggetti
Ecology
Journal title
ISSN journal
00218901
Volume
34
Issue
3
Year of publication
1997
Pages
748 - 766
Database
ISI
SICI code
0021-8901(1997)34:3<748:ALMFHO>2.0.ZU;2-T
Abstract
1. A 15-year fire history (1980-94) was assembled for Kakadu National Park, a 20 000 km(2) World Heritage property in monsoonal northern Aus tralia, based on interpretation of LANDSAT MSS imagery sampled at leas t three times over the 7-month dry season. 2. Detailed ground-truthing was undertaken at the end of the early dry season period (May-July) f or both 1993 and 1994; ground-truth data were not available for previo us years. Overall agreement was greater than 80% in both years. In sum , these data inspire a relatively high degree of confidence in the int erpreted fire history of the Park for any one year, at least at the la ndscape and habitat scales examined here. 3. An average of 46% of the Park was found to be burnt each year over the 15 years of records, wit h 25% burnt in the early, and 21% burnt in the late, dry season. The d ata indicate a pronounced shift from a fire regime dominated by late ( typically more intense and potentially extensive) dry season fires up until the mid-1980s, to one dominated by early (typically of low inten sity and patchy) dry season fire subsequently. 4. Whereas an average o f 55% of lowland savanna habitats has been burnt annually, 28% of habi tats occupying both sandstone plateau and riverine landforms have been burnt each year. The great majority of burning in relatively fire-sen sitive sandstone habitats continues to be in the late dry season. Data indicate a marked increase in the extent of burning on floodplains an d in associated fire-sensitive Melaleuca forests and lowland rainfores ts, from 1990; this increase is attributed to increased herbaceous fue l loads associated with the removal of feral Asian water buffalo. 5. P roximity analyses indicate that slightly more early dry season burning has been undertaken close to roads, and at greater distances from set tlements; no proximity differences were discernible for fires late in the dry season. Burning has been concentrated close to lowland drainag e lines, both in the early and late dry season. 6. Lowland savanna sit es are burnt on average 3 out of 5 years. In contrast, the majority of sandstone plateau and riverine floodplain sites have burnt on average 0-4 times and 0-3 times, respectively, over the 15 years of records. 7. The median size of contiguously burnt areas (patches) has been decl ining steadily over the 15 years of records, from upwards of 300 ha in itially to c. 60 ha in 1994. 8. It is concluded that although the asse mbled data are imperfect in that they underrepresent wet season burns and very small fires, are prone to positional errors of up to c. 300 m , and because little confidence can be placed in the sequential fire h istories for any 1-ha site, they afford an effective means by which th e Park's fire management programme can be monitored, its problems iden tified, and its needs assessed. The demonstrable utility of this progr amme has led to its recent extension on other significant conservation properties in monsoonal northern Australia; the procedure offers pote ntial for wider adoption in savanna regions generally.