An optimized burned area detection method based on the GESAVI

摘要

This paper presents a methodology for burned area detection based on the application of a vegetation index, the Generalized Soil-Adjusted Vegetation Index (GESAVI), to a multitemporal IRS-WiFS image series. A combination of two existing radiometric normalization methods is applied to the multitemporal series in order to perform an accurate change detection analysis. The GESAVI belongs to the soil-adjusted vegetation indices (SAVI) family and has been developed to normalize soil effects in the canopy response. It is defined in terms of the soil line parameters (A and B) and a soil adjustment coefficient (Z) as: GESAVI=(NIR-BR-A)/(R+Z) where Z is related to the red reflectance at the cross point of the soil line and the vegetation isolines. The results show the suitability of the GESAVI for the discrimination of burnt areas and the potentiality of this index for the evaluation of vegetation regeneration in those areas. When compared with traditional indices (SAVI and NDVI), the GESAVI shows improved temporal discrimination of vegetation changes, which leads consequently to the optimization of burnt area detection.