Avaliação de Modelos Digitais de Elevação provenientes de dados de sensoriamento remoto de distribuição gratuita

Abstract

The Digital Elevation Models (DEM) are an important data model to representation of the altimetric variations of the Earth surface. Free DEMs from remote sensing have gained prominence, and most applications are concentrated in the area of Geosciences, mainly geomorphological studies. In choosing the DEM to be used, the accuracy of the product, the mapping scale, the characteristics of the area of application, as well as the errors arising from the data acquisition process, must be taken into account. Given that DEM users are available to different resolutions with different acquisition and processing methods, it is important to develop studies that evaluate the altimetric accuracy of the DEMs and investigate variables that influence the occurrence of errors in the representation of the registered elevation. This work aimed to evaluate altimetric accuracy of the following DEMs available free of charge and from remote sensing: SRTM C-band (SRTM v3), ASTER GDEM v2, TOPODATA, both with spatial resolution of 1 arc-second that equivalent to approximately 30 meters and SRTM C-band with spatial resolution of 3 arc-seconds that is approximately 90 meters (SRTM v2). In addition to investigating the relationship between the observed errors and the influence variables: land slope classes (flat, smooth corrugated, corrugated, mountainous and steep) and type of ground cover (water, urban - average of two storey, urban - high buildings, vegetation). LIDAR data with 25 cm vertical accuracy were used as reference for assessment validation. The study area is a densely populated urban area (Belo Horizonte - MG). The evaluation took into account remote sensing principles and used the methods of analysis of histograms, topographic profiles, correlation coefficient, regression analysis, RMSE, analysis of the statistical and spatial distribution of errors and Cartographic Accuracy Standard of Digital Cartographic Products (PEC-PCD) of Brazil. The results indicated strong positive linear correlation and the existence of a functional relationship between the evaluated models and the reference model. The model that presented the smallest altimetric errors to represent the study area was SRTM v3. The analysis of the statistical frequency distribution of the errors determined to the SRTM v3 gain in relation to the other models. Considering 1: 50,000 mapping scale the PEC-PCD classification indicated class B standard for SRTM, class C for ASTER GDEM, class D for TOPODATA and SRTM v2 did not score on this mapping scale. It was verified that, in both models evaluated, the altimetric errors increase as the slope increases. The SRTM v3 presented the smallest altimetric errors to represent all classes of geomorphological units, with exception of steep geomorphological units (slopes above 75%), for which ASTER GDEM proved to be more advantageous. The stereoscopy and interferometry methods presented limitations to represent the altimetry in water bodies. The increase in altimetric errors in relation to the types of land cover occurred in the following order: urban (high buildings), water, vegetation, open land and urban (average of two storey). With the representation of the spatial distribution of altimetric errors of the models it was possible to visualize the dispersion or concentration of the errors and where they are located.