Avaliação da acurácia vertical e calibração de modelos digitais de elevação na bacia hidrográfica do arroio Cambaí, Itaqui-RS

Abstract

Digital Elevation Models (DEMs) are digital data designed to represent the altitudes of the physical surface using data from remote sensing, photogrammetry, GNSS positioning and topographic surveys. DEMs are useful in various areas, such as urban planning, water resource management, hydrological modeling and environmental studies. However, these models are subject to various types of error that can affect their accuracy and reliability, especially vertical error, which refers to the difference between the actual elevation of the terrain and the elevation represented by the DEM. This study sought to analyze the altimetric quality of four quasi-global DEMs, using descriptive statistics to verify the absolute vertical accuracy of the data in the Arroio Cambaí watershed, located in the municipality of Itaqui - RS. The MDEs analyzed were: ALOS World/AW3D30, ASTER GDEM/ASTWBD, Copernicus DEM (COP-30) and SRTM NASADEM. To carry out statistical analysis, data with three-dimensional coordinates derived from GNSS positioning and available on the SIGEF/INCRA platform were used as control points. The evaluation methodology involved applying a statistical analysis to the altitude of the GNSS data and the altitude of the DEMs, applying descriptive statistics such as Root Mean Square Error (RMSE), linear regression and coefficient of determination, which allowed the accuracy of each DEM to be analyzed. The results showed that the Copernicus DEM had the best accuracy (RMSE = 2.44 m) for the region. And that after applying a linear adjustment, a 6.13% improvement in altimetric quality was obtained. The NASADEM model (RMSE = 2.87 m) improved by 9.30%, ALOS (RMSE = 3.02 m) by 23.95% and the ASTER GDEM model (RMSE = 6.71 m) by 22.39%. In summary, the best results were obtained with the Copernicus DEM model, followed by ALOS, SRTM NASADEM and, lastly, the ASTER GDEM model. After complete analysis of the results, the most suitable model for the region studied was the Copernicus DEM, which showed the best correlation and the lowest RMSE of 2.29 m after calibration. The study revealed that the calibration of DEMs can have a direct impact on the quality of geospatial information used in environmental studies and management. The need for a prior assessment to choose the most appropriate model to represent a specific region is evident, contributing to the advancement of knowledge in digital representations.