1 Introduction Digital elevation model (DEM) is useful and important. It is used in several applications such as: slope calculation for landslide occurrences, flood alerts, drainage network and watershed delimitation, hydrological models, agricultural studies, protected geographical areas, among others. High resolution altimetry data sets are expensive. This data is obtained only for small regions. In countries like Brazil, with continental dimensions, it is time consuming to analyse large areas using high resolution data sets. The common solution is to use the datasets available as SRTM [1] with horizontal resolution of 90 meters (SRTM-90) and 30 meters (SRTM-30), and Aster GDEM [2] with 30 meter resolution. These data sets are freely available for the entire Earth surface for latitudes lower than 60 o . The main limitation of using these data is that in a dense forest they represent the tree canopy surface, and not the terrain surface. If in a satellite collected image a forest parcel was cut, the altimetry values on the parcel are different from those in its neighbourhood, leading to some errors in the extracted drainage. Another problem related to Aster GDEM data set images of different dates used for composing the stereoscopy mosaic, is that they don’t usually have the same spectral response. SRTM data set presents another problem: large aquatic areas, such as large rivers and lakes are represented by flat levels. Flat areas must be properly treated to assure a correct water flow path determination. The SRTM-30 data was recently made available, but it does not have valid values at all grid positions. These positions, called voids, are marked with the value -32767, and it must be substituted by some estimated altimetry data. TerraHidro system [3], which is a software system for hydrological studies, was employed to perform this task. A linear interpolator was implemented in TerraHidro: it uses the four nearest values from the void position taken from a lower resolution grid of the same area, usually an SRTM-90 data grid. This work shows qualitatively the precision of drainages extracted by the TerraHidro system, using SRTM-90 and SRTM-30 data sets, in comparison with the pan-European Catchment Characterization and Modelling version 2.1 (CCM2), river and catchment database [4] developed and generated by the JRC [5]. The paper is organized as follows: Section 2 briefly presents the TerraHidro system and the geographic region used in this work, and Section 3 shows the results and some comparison. 2 Materials and methods The United Kingdom was the geographic region used in the development of this work. Although it is not a significantly large area, it is an isolated area, bordered only by the ocean. The SRTM-90 and SRTM-30 data of this region were used. Figure 1 shows in grey levels the SRTM-30 data set. Assessment of the drainage network extracted by the TerraHidro system using the CCM2 drainage as reference data João Ricardo de Freitas Oliveira, Sergio Rosim, Alexandre Copertino Jardim National Institute for Space Research - INPE Av. Dos Astronautas, 1758 São José dos Campos, Brazil {joao, sergio, alexandre}@dpi.inpe.br Abstract The objective of this study is to compare the drainage networks extracted by the TerraHidro system, developed in the Image Processing Division (DPI) of the National Institute for Space Research (INPE), using SRTM data with resolutions of 30 and 90 meters, with the existing drainages in the pan-European drainage network database, called Catchment Characterization and Modelling version 2.1 (CCM2), river and catchment database developed at the Institute for Environment and Sustainability (IES) of the Joint Research Centre (JRC). In other words, CCM2 data set was used as reference data for qualitative analysis of the extracted drainages by TerraHidro. The SRTM 30m data contains altimetry points with value -32767, called void points, that must be substituted by some estimated altimetry data. TerraHidro automatically performs these corrections using any available altimetry data set grid as an alternative value grid. In this work we used the SRTM 90m as the alternative grid. To do so, TerraHidro uses a bilinear interpolator, which performs a linear interpolation weighting by the inverse of the distance using the four nearest values. A conversion process of these drainages, called upscaling, was executed in order to adapt them to lower resolutions, in this case 900m. Again, this new set of drainage was compared with the reference data. Finally, a procedure called HAND was executed and the result is displayed indicating areas with varying levels of flooding potential. The data used in this work are the SRTM 90m and SRTM 30m from the UK. The basic TerraHidro features have also been described. Keywords: TerraHidro, drainage network, SRTM, HAND, upscaling.