Avaliação da qualidade de dados altimétricos “alos world 3D – 30 m”

Abstract

Digital Elevation Models (DEM) at a global level are sources of data for several disciplines of knowledge, implying a growing search for the best global models in terms of accuracy. Thus, the present study had as its main objective to evaluate the altimetric quality and propose corrections at a national level for the global MDE Alos World 3D – 30m (AW3D30), also evaluating correlations of the vertical errors of the AW3D30 with land use and land cover classes, derived from of the MapBiomas project, and applying corrections based on these correlations. Finally, it validated the correction method by framing the PEC-PCD in the original and corrected MDE. The materials of this study consisted of altimetric data that represent the field truth, the MDE AW3D30 and Raster of land use and land cover from the MapBiomas Platform. The methodology was structured starting with the acquisition, treatment and normalization of the aforementioned data, then errors were estimated in the MDE AW3D30 against field truth data, then statistical studies evaluated the consistency of the data and the altimetric quality of the AW3D30 by MapBiomas class . In the sequence, the work established parameters of altimetric correction using an empirical modeling, this model was applied and validated in different regions. Given the exposed methodology, this study was able to build corrected AW3D30 models for three validation areas, in the states of Pernambuco, São Paulo and Bahia. The results showed a significant improvement in altimetric accuracy, measured by the RMSE, whose values increased from 8.22m, 4.06m and 1.87m before correction to 4.55m, 2.48m and 1.26m after correction, respectively. The results encourage the application of the corrective methodology proposed in this study throughout the national territory, since it reduced the RMSE of the AW3D30 by 38.70%, on average, for the validation areas.