Assessment of forest fire seasonality using MODIS fire potential: A time series approach

We utilized Fire Potential Index (FPI) to assess quantitatively fire season in a Mediterranean and a Temperate ecoregion of Iberian Peninsula by means of Time Series Analysis. The FPI was derived using MODIS NDWI and NDVI time series as indicators of vegetation status. Temperature and relative humidity time series from meteorological stations were interpolated using Universal Kriging. The average FPI time series were computed for each ecoregion for the period running from January 2001 to December 2005 and Time Series Analysis was applied to these series. Results show that FPINDWI and FPINDVI were able to describe fire season in two different ecoregions. Using Time Series Analysis was possible to evaluate and model fire dynamics. By means of spectral analysis and autoregressive models it was possible to detect hidden periodic components and model FPI dynamics with previous FPI information. Large differences between the Mediterranean and the Temperate ecoregions were demonstrated and subtle variability in the Temperate region could be explained. This quantitative approach represents an operational new tool for assessing subtle variations in fire season due to climate change.