Australian Fire Regimes: Contemporary Patterns (April 1998 - March 2000) and Changes Since European Settlement

Australia: State of the Environment Second Technical Paper Series (Biodiversity), Series 2
J Russell-Smith, R Craig, AM Gill, R Smith & J Williams
Department of the Environment and Heritage, 2002
ISBN 0 642 54801 3

Appendix 1: The Relationship Between Hot Spots from NOAA AVHRR Imagery and Fires in Southwestern Australian Forests

PDF file

Abstract

The task of relating 'fire hot spots' (FHS) detected daily by satellite to actual fires on the ground is a complex one. There are difficulties of satellite detection including obfuscation by cloud and by sunglint due to low-angle views. On the ground, fire statistics may be collated by a number of agencies in different ways and with different degrees of precision. In this study we have used daily maps of FHS from NOAA AVHRR imagery for the period September through May, 1998-99, and for the same period in 1999-2000 and compared them with locations of prescribed and unplanned fires on the same days. FHS were detected on a daily basis and grouped into 'events' (clusters). By comparing events from day to day for temporal and geographic overlap, multiple counting was avoided. Being explicit and quantitative about the possible errors due to poor imagery or cloud obfuscation in detecting FHS was important to the evaluation of the results. Close co-operation with the agency determining the presence of real fires whether unplanned or prescribed (Western Australian Department of Conservation and Land Management) was found to be essential. We determined the numbers of fires not matched by 'events', numbers of fires that were matched with 'events', and numbers of 'events' evident apparently without fire occurrence. In the forested southwest many fires were undetected by satellite and there were many 'events' unmatched by real fires. Inadequate imagery (especially due to cloud obfuscation) was a major reason for the non detection of fires. Low intensity fires under tree canopies may be a reason for poor detection even when appropriate imagery was available. A further problem with poor detection, and the occurrences of many unmatched 'events', could be due to large errors in matching locations (up to 3km error for automatic processing of images for FHS).

Authors

A Malcolm Gill
Honorary Research Fellow, Centre for Plant Biodiversity Research, CSIRO Plant Industry.

Gerard van Didden
Consultant, Sawyers Valley, Western Australia, formerly with Western Australian Department of Conservation and Land Management, Fire Section.

Peter HR Moore
Senior Technical Officer, Centre for Plant Biodiversity Research, CSIRO Plant Industry.

Citation

Gill, A. M., van Didden, G. & Moore, H. R. (2002) The Relationship Between Hot Spots from NOAA AVHRR Imagery and Fires in Southwestern Australian Forests. In Australian Fire Regimes: Contemporary Patterns (April 1998 - March 2000) and Changes Since European Settlement, Russell-Smith, J., Craig, R., Gill, A.M., Smith, R. and Williams, J. 2002. Australia State of the Environment Second Technical Paper Series (Biodiversity), Department of the Environment and Heritage, Canberra. http://www.ea.gov.au/soe/techpapers/index.html