Estimating Cell-To-Cell Land Surface Drainage Paths from Digital Channel Networks, with an Application to the Amazon Basin

Estimating Cell-To-Cell Land Surface Drainage Paths from Digital Channel Networks, with an Application to the Amazon Basin

Journal of Hydrology 315 (2005) 167–182 www.elsevier.com/locate/jhydrol Estimating cell-to-cell land surface drainage paths from digital channel networks, with an application to the Amazon basin Emilio Mayorgaa,*, Miles G. Logsdona,1, Maria Victoria R. Ballesterb,2, Jeffrey E. Richeya,1 aSchool of Oceanography, University of Washington, Box 355351, Seattle, WA 98195-5351, USA bCentro de Energia Nuclear na Agricultura, Universidade de Sa˜o Paulo, Piracicaba, SP 13416-000, Brazil Received 18 May 2004; revised 28 February 2005; accepted 18 March 2005 Abstract Cell-to-cell surface flow paths are commonly derived from gridded digital elevation models (DEM) by choosing the direction of steepest descent to one of the eight surrounding cells. However, adequate DEMs often are not available. We developed a topography-independent method for creating gridded, land and stream drainage direction maps based on corrected vector river networks. We applied it to the Digital Chart of the World river network in the Amazon basin gridded at 0.0058 resolution; in this basin, low relief and poor topographic data have prevented the effective use of DEM-based methods. We geo-registered 224 hydrographic gages against the processed network and compared extracted vs. published drainage areas. Drainage areas ranged from 227 to 4,620,000 km2. Median relative error was 4.5%, increasing in smaller basins to 94% in basins %2000 km2. The effective limit of reliability may differ from 2000 km2 across the basin. The drainage direction map and derivative datasets represent an improvement over existing datasets for regional research in the Amazon basin. Methods exploiting vector networks complement terrain approaches, and combined they may yield advances in the automated extraction of drainage maps and handling of topologically realistic river systems. q 2005 Elsevier B.V. All rights reserved. Keywords: River networks; Drainage basins; Flow direction; Watershed extraction; Geographic information systems; Amazonia 1. Introduction regions through the transport of water, sediments, and chemical constituents. The dynamics at a given River channel networks are fundamental land channel reach can be understood only in the context surface features that integrate diverse and distant of the characteristics of its drainage area and the upstream river network. A systematic linkage of the * Corresponding author. Fax: C1 206 685 3351. land to its drainage channels involves the delineation E-mail addresses: [email protected] (E. Mayorga), of flow paths from each element on the land surface to [email protected] (M.G. Logsdon), [email protected] a channel reach and down the river network to the (M.V.R. Ballester), [email protected] (J.E. Richey). mouth. Such integration is required in many signifi- 1 C Fax: 1 206 685 3351. cant earth science and resource management 2 Fax: C55 19 3429 4610. 0022-1694/$ - see front matter q 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.jhydrol.2005.03.023 168 E. Mayorga et al. / Journal of Hydrology 315 (2005) 167–182 applications, including the validation of land surface Vector maps of channel networks are widely hydrology models using observed river discharge available, detailed, and verifiable sources of relevant (Lohmann et al., 1998) and studies of the effects of information. Compared to elevation, channels are basin characteristics and land use on river nutrient and more easily and accurately extracted from aerial and carbon fluxes (Krysanova et al., 1998; Ludwig et al., satellite remote sensing, and errors more easily 1996; Smith et al., 1997). identified and corrected. However, available digital Surface flow paths are commonly derived from a networks are often used only for visual display or gridded digital elevation model (DEM), where the proximity analysis (river buffers); datasets supporting elevation values are stored in regular grid cells. primarily such applications may include a number of Several algorithms exist to determine the direction of topological network complexities, including water flow from cell to cell based on terrain analysis disconnected reaches, two-bank channel represen- and DEMs (Martz and Garbrecht, 1992; Moore et al., tation, and flow bifurcation. For small watersheds, 1991). The most convenient and widely used method topologically corrected networks are often used in is the D8 algorithm (Hogg et al., 1997; Maidment, network analysis, and in hydraulic channel routing 1993), in which the direction of flow out of each cell models based on lumped sub-watersheds or hydro- corresponds to the direction of steepest descent to one logical representative units, where watershed bound- of the eight surrounding cells. A consequence of this aries may be extracted manually (Fread, 1992; method is that water and materials in a cell can flow to Maidment, 1993). only one of its neighbors, a simplification that may The use of digitized channel networks to assist in result in drainage artifacts (Costa-Cabral and Burges, delineating basin-wide land and river surface flow paths has been limited. In the ‘stream burning’ 1994). Nevertheless, other important parameters can method, a river network from an external source is be derived easily and unambiguously once unique incised into a DEM by artificially lowering the flow direction paths are delineated; examples include elevation of cells that are part of this network, forcing flow accumulation grids (the number of cells draining flow paths to drain into and follow the imposed to any given cell), distance along the network, river network (Graham et al., 1999; Liang and Mackay, networks at a desired scale, and drainage basins 2000; Renssen and Knoop, 2000). This scheme (e.g. Vo¨ro¨smarty et al., 2000a,b). Utilities to calculate reduces the impact of inaccuracies in DEMs, but these parameters from flow direction grids have still requires an adequate DEM while ensuring been incorporated in most Geographical Information realistic flow paths only at the larger scales for Systems (GIS) software, such as ArcInfo (ESRI, which a river network is available. Sekulin et al. 1997) and GRASS (Neteler and Mitasova, 2004). (1992) used a channel network and a nearest-neighbor The use of DEMs and unique cell-to-cell flow paths method only to extract gridded drainage areas without greatly facilitates the extraction of surface flow the use of a DEM. More recently, Olivera and Raina characteristics, river networks, and drainage (2003) developed the Network Tracing Method basins, and the development of hydrological and (NTM) to generate coarse gridded river networks biogeochemical applications. Nevertheless, DEMs of from topologically corrected and simplified vector sufficient resolution and accuracy do not exist in large networks, independently of DEMs. This novel scheme areas of the world, especially in developing countries combines network gridding and upscaling while and in low-relief forested regions such as the Amazon preserving original channel lengths more consistently lowlands; low-relief areas are specially challenging than previous upscaling methods (e.g. Fekete et al., even when high-resolution DEMs are available. In 2001; Olivera et al., 2002). However, it neglects cells such cases, alternative methods are needed to derive not containing channels in the vector dataset; such cell-to-cell surface flow paths covering the entire land cells may represent streams missing from the vector surface, at a resolution high enough to support a network or areas dominated by subsurface flow. variety of applications. Do¨ll and Lehner (2002) We present an approach related to but more discuss additional limitations of deriving drainage exhaustive than the one used by Sekulin et al. (1992). directions from DEMs. Like NTM, it relies on topologically corrected and E. Mayorga et al. / Journal of Hydrology 315 (2005) 167–182 169 simplified channel networks to derive gridded cell-to- complexities (Fig. 2). Correction or simplification of cell surface flow paths. However, the resulting flow such features may be facilitated through automated paths encompass both true rivers and the entire land methods in a GIS, but ultimately must rely on visual surface. Rather than incorporating a network upscaling inspection and manual corrections. Errors in drainage procedure, this approach creates a base gridded flow- network configuration must be checked and corrected direction dataset using the finest cell resolution that is against ancillary data such as atlases and air photogra- computationally practical. It is intended for use in phy. Commonly available network datasets will regions where existing DEMs are inadequate, but typically require extensive pre-processing. detailed channel network maps are available. With a Two other datasets are required. First, unique cell- complete land and river flow direction grid created, to-cell flow paths require a single outlet. Rivers in surface flow path analysis functions can be applied in estuaries and deltas are often split into multiple the same manner as with flow direction grids derived channels and mouths (see Fig. 2), and this complexity from DEMs. After describing the method, we illustrate is depicted faithfully in vector channel networks. and evaluate it with a large-scale application to the A single mouth point must be chosen manually; all entire Amazon basin using a corrected version of the other draining channels may either be forced to Digital Chart of the World (Danko, 1992)river reverse their direction of flow or may be cut-off at network; the derived datasets are freely available and arbitrary points. Reliance on DEMs and D8 to extract may be used in hydrological and biogeochemical flow directions and river networks sidesteps the issue modeling and in studies linking river observations to because the use of elevation gradients and unique cell- their corresponding drained areas, particularly in to-cell paths enforces an arbitrary unique mouth point, connection with the Large-scale Biosphere- even when additional processing is involved. Second, atmosphere experiment in Amazonia (LBA). the basin boundary corresponding to the chosen mouth point is needed. This polygon may be created manually by visual inspection or may originate in 2.

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    16 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us