Burned area extraction based on time-series Modis 250m NDVI of Guangxi province

A new method based on geography knowledge and growth information of vegetation was described in this paper to extract burned areas from Guangxi province between year 2006 to 2010 using time-series MOderate Resolution Imaging Spectroradiometer (MODIS) Terra 16-day composite 250 m Normalized Difference Vegetation Index (NDVI) products (MOD13Q1). The aim of this work was to explore a method which can be used in the production of burned area mapping in large scale. The algorithm results were compared with those produced by Maximum Likelihood (ML) method extracting from Landsat Thematic Mapper (TM) imagery for a site of Guangxi province. The performance of each of these methods was evaluated by comparison with the sub-class, coded 24, of forest in the National Land Use Data (NLUD) of year 2010. Analyzing the area of burned fields extracted by both methods, it was discovered that ML method with a severe over extraction and the new method more reasonable. It could be concluded that it was feasible to extract burned areas from time-series MODIS NDVI based on knowledge and this method was potential to be applied in the production of burned area mapping for large scale.