Analysis and Correction of Digital Elevation Models for Plain Areas

• Autores: Guevara Ochoa, Cristian; Vives, Luis; Zimmermann, Erik; Masson, Ignacio; Fajardo González, Luisa Fernanda; Scioli, Carlos
• Año de publicación: 2019
• Idioma: inglés
• Tipo de recurso: artículo
• Materia: Oceanografía, Hidrología, Recursos Hídricos; DEM; Modelos digitales de elevación del terreno; Cuenca del arroyo del Azul
• Institución: Comisión de Investigaciones Científicas de la Provincia de Buenos Aires

Water movement modeling in plain areas requires digital elevation models (DEMs) adequately representing the morphological and geomorphological land patterns including the presence of civil structures that could affect water flow patterns. This has a direct effect on water accumulation and water flow direction. The objectives of this work were to analyze, compare and improve DEMs so surface water movement in plain areas could be predicted. In order to do that, we evaluated the accuracy of a digital elevation data set consisting in 4064 points measured with a differential global positioning system (GPS) in a plain area of central Buenos Aires province. Three DEMs were analyzed: (1) the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER), (2) the Shuttle Radar Topography Mission (SRTM) and (3) the Advanced Land Observing Satellite with the Phased Array Type L-Band Synthetic Aperture Radar (ALOS PALSAR). Several topographic attributes (i.e., height, surface area, land slope, delimitation of geomorphological units, civil structures, basin boundaries and streams network) and different interpolation methods were analyzed. The results showed that both the SRTM and the ALOS PALSAR DEMs had a ± 4.4 m root mean square error (RMSE) in contrast to the ASTER DEM which had a ± 9 m RMSE. Our analysis proved that the best DEM representing the study area is the SRTM. The most suitable interpolation methods applied to the SRTM were the inverse distance weighting and the ANUDEM, whereas the spline method displayed the lowest vertical accuracy. With the proposed method we obtained a DEM for the study area with a ± 3.2 m RMSE, a 33% error reduction compared to the raw DEM