A Comparative Study about Vertical Accuracy of Four Freely Available Digital Elevation Models: A Case Study in the Balsas River Watershed, Brazil

Zuleide Alves Ferreira,Pedro Cabral

doi:10.3390/ijgi11020106

Abstract

Digital elevation models (DEMs) provide important support to research since these data are freely available for almost all areas of the terrestrial surface. Thus, it is important to assess their accuracy for correct applicability regarding the correct use scale. Therefore, this paper aims to assess the vertical accuracy of ALOS PALSAR, GMTED2010, SRTM, and Topodata DEMs according to the Brazilian Cartographic Accuracy Standard through the official high accuracy network data of the Brazilian Geodetic System. This study also seeks to investigate whether the altimetric error is correlated with altitude and slope in the study area. Our results showed that the four assessed DEMs in this study demonstrated satisfactory accuracy to provide mappings in scales up to 1:100,000 because more than 90% of the extracted points presented altimetric errors of less than 25 m when compared with the reference points from the high accuracy network of the Brazilian Geodetic System. Regarding the altimetric error, we could not find a significant correlation with altitude or slope in the study area. In this sense, future DEMs assessments should be based on the investigation of other factors that may influence altimetric error.

Highlights

Digital elevation models (DEMs) provide an important topographic product that is fundamental for many scientific and commercial applications [1,2]
Results show that regarding the mean error and mean absolute error, the values of the statistical analysis are similar for the four DEMs (Table 3)
ALOS PALSAR, Shuttle Radar Topography Mission (SRTM), and Topodata DEMs show similarity in all statistical indicators, and it is possible to notice that GMTED2010 shows the worst performance mainly when we consider the root mean square (RMSE) (7.48 m) and the error range (54.00 m), i.e., the difference between the minimum and maximum altimetric errors

Summary

Introduction

Digital elevation models (DEMs) provide an important topographic product that is fundamental for many scientific and commercial applications [1,2]. Several DEM products from many sources have been made freely available to geoinformation users in the last decade, so it is important to investigate their possible applications by assessing their accuracy [3]. DEM products accuracy has been regularly investigated to evaluate their applicative potentialities, improving mapping methods [4]. Most of these experiments are performed by comparing the extracted data from DEMs to a set of reference data, i.e., control points, through accuracy statistical indicators, such as mean difference, standard deviation, or root mean square error [4]. The Decree n◦ 89,817/1984 determines criteria for cartographic products classification regarding their accuracy and the distribution of errors using a statistical indicator of positional quality named the “Cartographic

Objectives

Methods

Results

Discussion

Conclusion