AbstractsGeography &GIS

Semi-automatic identification of neogenic deposits by using high resolution digital surface models in Southeastern Brazil

by Sonia Maria Lima Silva

Institution: Universität Tübingen
Year: 2015
Record ID: 1118137
Full text PDF: http://hdl.handle.net/10900/59310


The Cenozoic deposits occupy large areas of the Brazilian territory with parts subject to intense human occupation. Nevertheless they are still little known and researched. One of the most important Cenozoic sedimentary units in Brazil corresponds to Barreiras Formation (Neogene), which is located along almost all of the Brazilian coast. Geomorphologically the deposits of the Barreiras Formation are associated with a relief of tabular forms, dissected into deep valleys with steep slopes. This research aims to develop a semi-automatic method of detailed scale mapping Barreiras Formation deposits by employing Digital Surface Models (DSMs) obtained from high resolution satellite images. The justification of this research is based on both: the lack of cartographic databases on the scale of 1:25.000, as well as the lack of maps representing these deposits in more detailed scale necessary for their proper description and interpretation. The selected area is located in the northern region of the State of Rio de Janeiro, between the city of Campos dos Goytacazes and the border to the State of Espírito Santo. In this area the most significant occurrence of Barreiras Formation deposits in the State of Rio de Janeiro is found. The main goal to be achieved in this research is the identification of such deposits by evaluating their geomorphological characteristics which are modeled by specific variables that represent these aspects. The study on the potential of the use of stereoscopic capabilities of high resolution images coming from ALOS/PRISM sensor, which made the generation of DSMs possible, together with the availability of Rational Polynomial Coefficients (RPCs), contributed to minimize the costs of acquiring field data, since most of the information needed to identify these deposits were obtained directly from these DSMs. The method used in this research consisted of: (1) the automatic generation and vertical accuracy control of DSMs; (2) the orthorectification of images and horizontal accuracy control of orthoimages; (3) the preparation of DSMs for the terrain analyses; (4) the terrain analyses; and (5) the generation and qualitative and quantitative evaluation of the final map. The generation of DSMs and the orthorectification of the images were carried out by using the empirical model of rational functions, using the RPCs available in the images. The terrain analyses were based on the interpretation and combination of specific morphometric variables using two different methods: (a) combinatorial OR; and (b) maximum entropy. The best result was obtained with the first method, combining the four variables: altitude, slope, curvature and terrain roughness index. Beforehand, these variables were classified according to intervals representing the Barreiras Formation. The assessment of the final map was made by comparing such map to another one, generated by using visual interpretation method, in a scale comparable to the scale of the map generated in this research. The level of detail in drawing the boundaries of the…