Cell Size Dependence of Threshold Conditions for the Delineation of Drainage Networks from Gridded Elevation Data

P. Tarolli1, G. Dalla Fontana1, G. Moretti2, S. Orlandini2
1 Department of Land and Agroforest Environments, University of Padova, Agripolis,
Viale dell’Università, 16 – 35020 Legnaro (Padova), Italy
Telephone: +39 049 827 2695
Fax: +39 049 827 2686
2 Department of Mechanical & Civil Engineering, University of Modena & Reggio Emilia,
Via Vignolese, 905 – 41100 Modena, Italy
Telephone: +39 059 205 6105
Fax: +39 059 205 6126

The drainage network is the pattern of tributaries and master streams in a drainage basin as delineated on a planimetric map. In theory, the network includes all the minor rills which are definite watercourses, even including all the ephemeral channels in the furthermost headwaters. In practice, the detail of the drainage network is dependent on the scale of the map used to trace the channels (Leopold et al. 1964, p. 131). When preparing a topographic map the headward limits of the blue lines do not reflect any statistical characteristic of streamflow occurrence, nor differences in the hydrologic response of the headwater due to the various combinations of climate, topography and geology. In actual fact, they are drawn to fit a rather personalized aesthetic (Leopold 1994, p. 228). However, in light of recent field studies on the channel head (e.g., Dietrich and Dunne 1993), of increasing availability of accurate digital elevation data due to the LiDAR (Light Detection And Ranging) technology (e.g., Tarolli and Tarboton, 2006; Carter et al. 2007; Cavalli et al. 2008; Vianello et al. 2009; Tarolli and Dalla Fontana 2009), and of recent advances in terrain analysis (e.g., Gallant and Wilson 2000; Moretti and Orlandini 2008; Orlandini and Moretti 2009), a rationale for the delineation of drainage networks can be sought.

A field definition of the channel head is provided by Dietrich and Dunne (1993) as the upstream boundary of concentrated water flow and sediment transport between definable banks. Although it is not easy to provide a globally useful criterion for a well-defined bank, it is commonly accepted that the bank is recognizable as a morphological feature independent of the flow. In this perspective, a detailed description of hydrologic flows may not be required in order to predict channel heads, and meaningful predictive models can be formulated by combining terrain analysis and generalizations from field facts. Field and theoretical studies addressing the problem of defining where channels begin have led to the definition of different threshold conditions for channel initiation. O’Callaghan and Mark (1984) and Tarboton et al. (1988) defined channel networks on a digital elevation model as those pixels that have an accumulated drainage area greater than some “threshold support area.” Montgomery and Dietrich (1988) proposed to use a threshold on a power function of both drainage area and the local slope. Howard (1994) considered a threshold on the gradient divergence normalized by mean gradient. Peckham (1995) investigated a method based on Strahler’s (1957) classification of drainage networks extracted from digital elevation data, and iterated pruning of exterior links.

In the present study, these criteria are evaluated by using accurate field observations of channel heads and channel network in the Rio Cordon catchment (Eastern Italian Alps, Fig. 1), gridded elevation data obtained from high-precision LiDAR surveys (Cavalli et al. 2008), and advanced methods for the delineation of drainage basins and surface flow paths from these data (Orlandini et al. 2003; Orlandini and Moretti 2009). The dependence of threshold conditions for the delineation of drainage networks on the size of grid cells involved is investigated.



  • "International Society for Geomorphometry (ISG) is a non-commercial, nongovernmental association of researchers and experts that are open for free exchange of knowledge and opinions about various aspects of DEM processing and Digital Terrain Modeling."

We use cookies to give you the best experience. Disclaimer