Images processing such as linear contrast stretch, ration band, multiplication band and principal components transformation were used to enhance in TM - Landsat data features related to landslides occurrence as landslide scar, antropic features and relief features. The study was carried out among a sector of Serra do Mar in the city of Caraguatatuba (São Paulo, State, Brazil) using images obtained at January 1996.

The composition of band TM 473 RGB contributed most effectively to discriminate landslide scar differences. Antropic features and landslide scar features were highlighted by both ration band composition (5/7, 4/3, 4/1 RGB) and principal components composition (321 RGB), but only principal components composition could distinguished then. The multiplication band composition (5x4, 2x4, 7x4 RGB) contributed better to highlight relief features related to landslide as valley configuration and slope configuration.

The above techniques were supported by aerial photographs, topographic maps, thematic maps and field observation. The information obtained by these processing were integrated with MNT data and thematic data, generating the landslide hazard map.