Evaluation of Open Source Dems for Regional Hydrology Analysis in a Medium-Large Basin

Abstract

Nowadays exist a variety of topographic source with different spatialresolution obtained by using distinct methodologies. Choosing a data sourcethat is the more accurately possible in area relief, has become fundamental inhydrological studies. The aim of this study is to compare the accuracy of fouropen source DEMs (TOPO_IGN, SRTM_30m, SRTM_90m and ASTER-GDEM)in the Quequen Grande River Catchment (QGRC), in the south-west of BuenosAires Province in Argentina, in two different ways. Determining the verticalerror of four source DEMs using geodesic points established by the NationalGeographic Institute of Argentina (GCP_IGN); and examining the associationof DEMs error with hydrological parameters associated to drainage networkobtained in each case. The results confirmed differences between GCP_IGNand open source DEMs elevation values. The Root-Mean-Square Error (RMSE)value shows that the uncertainty of elevation measurement is bigger in highaltitudinal zones. This is due to two factors (i) the complex landscape pattern inareas characterized by high elevation values within flat study area; and (ii) thelevel of precision and spatial resolution of each open source DEM. The calculatedslope is overestimated in all DEMs, indicating a positive bias and due to this, thatslope accuracy is influenced by terrain roughness and pixel size of DEM, mainly.The extract catchments and drainage information from all DEMs area also affectthe outcome. The terrain morphology strongly influences the DEM accuracy.So, coming to a conclusion, as in a QGRC, that include zones with high slopesand flat areas, SRTM_30m can be used to obtain height data. SRTM_90m canbe used to obtain automatic drainage and catchment delineation. Both DEMsare similar to the obtained by using IGN topographic data. Lastly, the ASTERGDEMmodel although possessing a high spatial resolution, turned out givingthe lesser appropriate topographic information. These results can be useful forfurther hydrological, agronomical and environmental studies that employ theseDEMs in modeling exercises.