Tratamento de ruídos e caracterização de fisionomias do cerrado utilizando séries temporais do sensor modis

Abstract

Cerrado is formed by a mosaic of grassland, shrubland and woodland physiognomies with a typical phenological cycle. Thus, MODIS data provide daily measurements which allows to monitor the seasonal phenology of the vegetation. The objective of this work was to characterize savanna formations, forest formations and cerrado areas converted by anthropic actions, by using temporal series of MODIS NDVI and EVI after noise reduction. The adopted methodology should be divided into the following steps: (a) elaboration of the temporal cube with NDVI and EVI, in which the z profile corresponds to temporal signature, (b) noise elimination, (c) detection of temporal signature. The Minimum Noise Fraction Transformation (MNF) method was applied to reduce noise in temporal signature. The results showed that the NDVI values were higher than the EVI; and there was a relationship with the seasons of the year. The forest formations presented the highest values of NDVI and EVI, showing the lowest variations among the seasons. The converted areas of Cerrado presented the lowest values in both indices, and their values decreased in the beginning of the dry season, probably because it was the harvesting season. The study concluded that the NDVI and EVI temporal series are useful in differentiation among vegetation types.