Geoestatística na otimização de modelos digitais de elevação e no planejamento da mecanização florestal a partir de imagens SRTM e ASTER

Abstract

Digital elevation models (DEMs) are critical in determining the planning of forest mechanization. The objective was to evaluate the geostatistics in optimizing digital elevation models and planning of forest mechanization from SRTM and ASTER images. In land use mapping was determined the areas of permanent preservation, legal reserve and logging. The derivatives were evaluated DEMs SRTM images 30 and 90 m spatial resolution (ER) and Aster RE 30 m by means of geostatistical analysis gave the krigados models, namely: krigado SRTM 30m, 90m SRTM krigado and Aster krigado. The reference MDE was extracted from the interpolation equidistant contour lines 10 meters in 1: 25.000 obtained by planialtimetric survey with geodetic GPS. Maps were generated with three slope classes used in forest mechanization of assessed MOUs. The original MOUs and krigados were evaluated by descriptive analysis, square root of the mean squared elevation error (RMSE), visual and numerical evaluation. The results showed that the RMSE was not sufficient to assess the models' accuracy to determine the planning of forest mechanization. The analyzes showed that the use of geostatistics allowed optimize free MOUs, highlighting the SRTM 30 m, with potential to be used in the planning of forest mechanization.