Fire risk assessment in savanna ecosystems with multi-temporal satellite data
Verbesselt, J.; Somers, B.; Lhermitte, S.; van Aardt, J.; Jonckheere, I.; Coppin, P.
Communications in Agricultural and Applied Biological Sciences 70(2): 23-26
2005
ISSN/ISBN: 1379-1176 PMID: 16366268 Document Number: 594407
A study was conducted at the Kruger National Park (KNP) in northeastern South Africa to improve the fire risk assessment capacity of the SPOT VEGETATION satellite by combining the vegetation amount (iRVI) and vegetation state related indices (ARND). In situ fire activity data was used to evaluate the combination of satellite derived iRVI and ARND indices with binary logistic regression. The S10 products of SPOT VEGETATION data for fire risk assessment were acquired from April 1998 to September 2003. A comprehensive fire history database of the KNP was used to evaluate the SPOT VEGETATION capacity for fire risk assessment, and the date, cause and position of each fire were extracted from the database for the study period 1998-2003. Analysis of the interaction effects between iRVI and ARND using the Wald test statistic did not show significant relation. The partial deviance chi-square test was used to verify wether iRVI improved the binary logistic model with ARND as explanatory variable, and it showed that iRVI significantly improved the model containing ARND at a 95% confidence level. Results demonstrated that the combination of a fire risk index related to vegetation state and an end-of-wet-season herbaceous biomass related index improved the fire risk assessment capacity of the SPOT VEGETATION satellite for the low lying savannas of KNP.