Historical Fire Detection of Tropical Forest from NDVI Time-series Data: Case Study on Jambi, Indonesia

Dyah R. Panuju (1Department of Soil Science and Land Resources, Bogor Agricultural University. Jalan Meranti. Bogor 16680. Indonesia.)
Bambang H. Trisasongko (1Department of Soil Science and Land Resources, Bogor Agricultural University. Jalan Meranti. Bogor 16680. Indonesia. Email: [e-mail])
Budi Susetyo (2Department of Statistics, Bogor Agricultural University. Bogor 16680. Indonesia.)
Mahmud A. Raimadoya (3Department of Civil and Environmental Engineering, Bogor Agricultural University. Bogor 16680. Indonesia.)
Brian G. Lees (4School of Physical, Environmental and Mathematical Sciences, University of New South Wales at Australian Defence Force Academy. Northcott Drive, Canberra ACT 2600. Australia)

Journal of Mathematical and Fundamental Sciences
Vol. 42 No. 1 (2010)

In addition to forest encroachment, forest fire is a serious problem in Indonesia. Attempts at managing its widespread and frequent occurrence has led to intensive use of remote sensing data. Coarse resolution images have been employed to derive hot spots as an indicator of forest fire. However, most efforts to verify the hot spot data and to verify fire accidents have been restricted to the use of medium or high resolution data. At present, it is difficult to verify solely upon those data due to severe cloud cover and low revisit time. In this paper, we present a method to validate forest fire using NDVI time series data. With the freely available NDVI data from SPOT VEGETATION, we successfully detected changes in time series data which were associated with fire accidents.