NetMap's Technical Help Guide

1.1 Building Synthetic Stream Networks

User version not yet available, under development (2015 - 2016)
The "BuildGrids" (bldgs) fortran program (Miller 2002, 2013) is used to generate the stream layer from a DEM and other supporting information such as a channel mask. There are a number of sub routines, including one for merging DEMs, including of different resolution. Users can customize the upward channel extent and thus channel density, the size of drainage wings, and the length of channel segments (from a minimum node spacing equal to the DEM resolution). Information on the derivation of the analytic river is provided below.
Building the node based analytic stream layer is based on several user defined attributes including critical drainage area, planform curvature, hillslope gradient and critical channel length.
Critical drainage area is drainage area per unit contour length and various thresholds can be used to identify provisional locations of channel heads or channel extent (Figure 1).  Critical area is equal to area multiplied by slope gradient-squared (AS2). Threshold values vary with steep and less steep topography, for example, less than 25%, greater than 25% etc.  Plan curvature is the curvature of the surface on a cell-by-cell basis, as fitted through that cell and its eight surrounding neighbors. Curvature is the second derivative of the surface, or the slope-of-the-slope. The plan curvature is perpendicular to the direction of the maximum slope (Figure 2). Algorithms for flow direction and channel delineation are described by Clarke et al. 2008 but see latest Technical Help covering the River Builder.
Figure 1. NetMap’s channel delineation method requires calculating the drainage area per unit contour length, distinct for both steep and low gradient areas. Example is from northern California.
Figure 2. The plan curvature is shown for a basin defined as the second derivative of the surface, or the slope-of-the-slope. The plan curvature is perpendicular to the direction of the maximum slope. The plan curvature is combined with the specific drainage area (Figure 1) and a minimum channel length threshold to define channels and the upward extent of channels.  Algorithms for flow direction and channel delineation are described by Clarke et al. 2008 but see latest Technical Help covering the River Builder.
Using both specific drainage area per unit contour length and pan curvature thresholds, a user can manually calibrate the upward extent of channel heads, that is, define the channel density. Too few channels may be created or too many channels. When threshold values are incorrect, there will be “channel feathering” which is manifest as numerous, closely spaced channels extending up on planar hillslopes (Figure 3). The analyst must “calibrate” the thresholds to eliminate or reduce channel feathering (using visual clues of channel locations in landscapes) while providing for a high density of channels. This is done by defining the inflection point between specific drainage area and channel density, given that the threshold for plan curvature is met (Figure 4). The pixel cells that are classified as being a “channel” must extend contiguously for a certain minimum distance, often ranging between 30 to 50 meters, set by the analyst.
Figure 3. Analytic stream networks derived from DEMs can have too high a drainage density. Here is an example of a Analytic channel network with “channel feathering”, defined as closely spaced channels on planar hillslopes. See Figure 4.
Figure 4. To determine the headward extent of the channel network in NetMap requires calculating the relationship between specific drainage area and channel density. There is an inflection point that forms that indicates the transition to an environment of channel feathering (e.g., Figure 3)
Once the appropriate thresholds are set for plan curvature, specific drainage area (Figure 4) and critical channel length, the channel network is defined within NetMap (Figure 5).
Figure 5. An appropriate channel network for a basin in northern California is created by adjusting thresholds for plan curvature and specific drainage area (Figure 4) so that no channel feathering has occurred. Contrast this network with the inaccurate network shown in Figure 3. The NHD is also shown for comparison, and the NetMap network has a higher network density, and thus more headwater channels.
The overall strategy in NetMap is to create high density drainage networks so as not to eliminate headwater streams, since information about network extent (or channel heads) is generally absent for most landscapes. With a liberal network within NetMap tools, a user can remove headwater channels that do not exist, and the network is renumbered. The ‘Channel Head’ tool is located in the Fluvial Morphology Module. In this way, a user can customize NetMap’s stream layer to better match specific landscape locations. Another way to tailor NetMap’s stream layer, in terms of drainage density (channel heads) and other features such as segment breaks and segment lengths scales is to use the River Builder tool. These tools allow for more flexibility and tailoring of a stream layer to specific landscapes than non-Analytic networks, like the NHD (described in more detail below).
DEM derived stream networks depend on topographic enforcement of the derived channel network, and thus Analytic networks are more accurate in higher relief terrain with topographic controls on channel position.  In areas of low relief such as in wide low gradient meandering or braided rivers, DEM derived networks may be inaccurate, since it is difficult for a computer model to determine exact locations of channels in wide valleys of little relief (e.g., Figure 6). Thus, NetMap’s stream layer can be mirrored onto other stream layers, such as the cartographic NHD (derived from imagery by mapping personnel).  The analyst must define what gradient threshold to use to “enforce” NetMap’s stream layer using another stream layer, e.g., the NHD. However, because of the dynamic shifting behavior of river channels across wide floodplains, even the NHD become spatially inaccurate over time (Figure 6).

Made with help of Dr.Explain

Copyright TerrainWorks 2014