The semi-arid region of Brazil has great potential for storing water in alluvial areas, and may give support to small-scale family farming. There is a need to establish a methodology for mapping these areas with the objective of future studies to implement underground dams and suitable management of the soils that occur on alluvial terraces. In this way, the aim of this study was to apply Landsat-Thematic Mapper images together with collateral data such as the drainage network, map of soil classes and elevation data to assist in the classification of alluvial terraces. The study area was Pajeú River basin located in the 'Caatinga' ecosystem of the Pernambuco state. Topographic assessment of the alluvial areas was done by using SRTM (Shuttle Radar Topography Mission) data. It was found that the single use of orbital data brings incongruous classification, but with the inclusion of collateral data it was possible to obtain better results in classification. The drainage network was also important in masking the spurious classification. The analysis of the topography of the classified areas was inadequate using SRTM data.
alluvial soils; supervised classification; remote sensing; relief
