The Positional and Thematic Accuracy for Analysis of Multi-Temporal Satellite Images on Mangrove Areas

The Positional and Thematic Accuracy for Analysis of Multi-Temporal Satellite Images on Mangrove Areas
Paulo Rodrigo Zanin¹, Carlos Antonio O. Vieira²

1.Universidade Federal de Santa Catarina, Campus Universitário Reitor João David Ferreira Lima, Trindade-Florianópolis-SC, CEP 88040-970 (paulorzgeo@gmail.com)
2.Universidade Federal de Santa Catarina, Campus Universitário Reitor João David Ferreira Lima, Trindade-Florianópolis-SC, CEP 88040-970 (carlos.vieira@ufsc.br)

Abstract: This article presents standard procedures on the positional and thematic accuracy for analysis of multi-temporal satellite images on mangrove areas. The Landsat TM 5 imagery for the years: 1986, 1997, and 2010 were used to identify and quantify the anthropogenic pressure, through the process of urbanization on coastal marine ecosystems surrounding neighborhoods at the city of Florianópolis – SC, Brazil. Geometric correction of these images reached a positional accuracy ¼ pixels. On the thematic accuracy, a supervised classification, using a maximum likelihood classifier, was validated using Kappa Index statistics (Conglaton and Green, 1999). The results show that high classification accuracies were reached. However, marine processes, such as coastal erosion on the shoreline and dynamics of ecosystems (e.g., natural succession of species of different vegetation types, as well as the seasonal variation of reflectance of plants) are factors to consider when assessing the reliability of image classifications. 

 Keywords: Mangrove Areas, Multi-Temporal Analysis, Land Cover Classification.

AttachmentSize
ZaninAccuracy2012.pdf148.47 KB