On the optimization of digital elevation models to obtain geomorphological and hydrological variables: Evaluation of the bidimensional convolution on alos, aster-gdem and srtm products

Abstract

The geomorphological and hydrological variables reflect the characteristics of a watershed and constitute essential data in spatial analysis of the terrain. With the dissemination of digital data, freely available digital elevation models (DEM) based on satellite data are being increasingly used. However, these models have known limitations inherent to errors resulting from the data acquisition process, compromising the extraction of spatial information derived from them. The present work aim to evaluate the application of a two-dimensional convolution technique in three DEM: ALOS (Advanced Land Observing Satellite), ASTER-GDEM (Advanced Spaceborne Thermal Emission and Reflection Radiometer-Global Digital Elevation Model) and SRTM (Shuttle Radar Topographic Mission) as well as to verify the influence of the tool in the optimization of these products in geomorphological and hydrological variables. The DEM were compared to conventional topographic data and further evaluated based on root of the mean square error (RMSE) other statistical tests. The results showed that the elevation models can be considerably optimized with the use of the convolution technique, but for this it is essential to adopt an adequate window size​ on the neighboring pixels. The technique was able to reduce the irregularities on the surface, showing an improved representation of the slope and accumulated flow maps. The analyzes shows that those geoprocessing tools available in GIS packages can promote a gain in the quality of free DEM, favoring the acquisition of morphological variables with greater accuracy.