Análise comparativa dos relevos obtidos por modelos digitais de elevação em uma zona urbana

Abstract

The increasing use of Digital Elevation Models (DEMs) in urban applications requires careful assessments of their vertical accuracy, particularly in contexts where terrain characteristics affect infrastructure planning. Therefore, this study aimed to evaluate and compare the vertical accuracy of different DEMs applied to the urban area of Crateús, Ceará State, Brazil, using topographic data from an urban water supply network as reference. The analyzed models were TOPODATA, FABDEM, Copernicus DEM (GLO-30), ANADEM, and ALOS PALSAR 12.5 m. All DEMs were clipped to the urban area, reprojected to the SIRGAS 2000 / UTM Zone 24S coordinate reference system, and, when necessary, vertically corrected to orthometric heights based on the mean geoid undulation from the MAPGEO2015 model. Vertical accuracy was assessed through point-by-point comparison between DEM-derived elevations and reference values, using elevation error (ΔZ). The analysis included statistical metrics such as mean error (bias), mean squared error (MSE), and root mean square error (RMSE), as well as graphical and spatial representations. The results showed a positive linear relationship between all DEMs and the reference elevations, with differences in error magnitude and dispersion. FABDEM, ANADEM, and Copernicus DEM presented superior vertical performance, characterized by lower error variability and greater concentration of ΔZ values near zero. In contrast, TOPODATA and ALOS PALSAR 12.5 m exhibited higher dispersion and a systematic tendency toward elevation overestimation, mainly related to their surface-based nature as Digital Surface Models. These findings demonstrate that the DEM type (DTM or DSM) directly influences vertical accuracy in urban environments.