Understanding the Many Flavors of Channel Area - and getting to deciding how to denote these in Channel Area projects.

[joewheaton]

In #486 we distilled the different types of Channel Areas you can use down to:

• I. NHD derived channel area
• II. NHD network line
• III. Manually digitized network line (including manually edited II)
• IV. Manually digitized channel area (including manually edited I)
• V. TauDEM derived network line (discussed here)
• VI. TauDEM derived channel area (i.e. buffer from V)

However, I want to highlight this for people that was buried back in an old discussion https://github.com/Riverscapes/rs-web-monorepo/discussions/207, which lays out why Channel Area is so important!

Why Channel Area Matters so much to so many of our models

There are tons of reasons, but the most obvious is that we care about the riverscapes, and the channel area represents the "river" part of that. Topography matters too, but the way almost all of our models work, is really sensitive to channel position:

• TauDEM - HAND outputs are based explicitly on what is flagged as a drainage cell and calculating height above those. The resulting relative elevation model that HAND produces is explicitly normalized relative to channel area, so it matters.
• VBET - VBET runs off of HAND and channel area. The specification of the drainage network (as an area) explicilty controls what part of the drainage network we are calculating valley bottoms for, and distance from channel area is also a line of evidence in the model. Moreover, channel areas themselves, by defintion MUST be part of the valley bottom.
• BRAT - For beaver, aquatic habitat is protection from predation. Where the water is, is where you are safe. The capacity model logic of BRAT is based on looking at suitable building materials within a distance from this protection (logic being it is safer the closer you are to the water and they only take risks further away when it is really tasty... e.g. Aspen).
• RCAT - RCAT currently cares more about mapping vegetation where VBET says is valley bottom (assuming that's area that could potentially support riparian). So to the extent VBET cares about channel area, RCAT does. Beyond that, RCAT's accuracy could be improved by removing the channel area from the riparian calculations and veg sampling.
• Confinement - An accurate estimate of confinement is highly sensitive to the mapping of the channel position on the valley bottom and where that channel abuts a valley bottom margin.
• RIM - RIM is the map of inundated area within the valley bottom. Riverscapes lacking structural forcing, tend to be entirely free-flowing at low-flows and will have inundated areas contained within the channel area. However, riverscapes with structural forcing by things like beaver dams, will often have inundated areas that extend well beyond the channel area. We tend to report inundated areas as a proportion of valley bottom area, but reporting them as a proportion of channel area is helpful to. Moreover, we derive cenerlines and/or thalwegs from these innundated areas as well as complexity measuers from the skeletized flow area which show difluences, confluences, channel heads and can be used to get at relative flow length (total length of CL of flow paths divided by riverscape length)

• GNAT - GNAT calculates a load of metrics that are based on the geometry of the channel area. For example:

  • Channel Area
  • Channel CL Length
  • Thalweg Length
  • Many flavors of Sinuosity
  • Braided index
  • Channel Width (integrated)
  • Confluence/Diffluence/Channel Head counts, mapping and density

• River Styler - River styling is based most directly on Confinement and GNAT ouputs. The accuracy of these measures is improved with better channel area mapping.

I know most of you know all of this, but it is important to remind ourselves why something so basic is so important to all our models. If there is ONE place to direct people to work to improve the quality of any local assessment (i.e. in QRiS) or to improve the quality of a network model, it is improving the ChannelArea mapping.