Assessing Vegetation Regrowth after Fire Based on Time Series of SPOT-VEGETATION Data

Time series analysis of the regeneration index (RI) derived from Normalized Difference Vegetation Index (NDVI) satellite imagery offers the potential to quantify the post-fire vegetation growth with spatial and temporal accuracy. Up to now, application of the RI on time series of coarse to moderate spatial resolution imagery remains however limited, due to the lack of adequate alternatives to delineate control plots on large scale. The primary goal of this study was therefore to develop a robust methodology for the selection of control plots and for the application of the RI on time series of coarse spatial resolution imagery. In the first section, two Time Series Similarity (TSS) approaches are introduced as alternative for the classical control plot selection approaches based on reference layers. The proposed approaches are compared with the classical approach using unburnt focal pixels. The results of this comparison revealed the usefulness of the Root Mean Square Distance TSS_RMSD approach to delineate control pixels. In the second section, quality indicators for the proposed approaches were derived based on pre-fire similarity, since the pre-and post-fire similarity showed a linear relationship. Consequently, pre-fire similarity permits to assess the quality of control pixel before using them in RI calculations. In the third section, the proposed TSS_RMSD approach was implemented for every bum! pixel in Ihe study area and the RI time series were calculated. Based on the RI profiles, recovery indicators were calculated for all burnt pixels that relied the spatio-temporal vegetation regrowth.