This forum is shutting down! Please post new discussions at

Burning streams into DEM for nutrient retention model

SPardoSPardo Member
edited October 2013 in Terrestrial Models
Hello all,

My colleague was using the "DEM Reconditioning" tool in the Arc Hydro toolbar to burn NHD flowlines into our DEM and it more or less just flattens the entire DEM into a gentle 200 km coastal slope, which we figure can't be the intended effect of burning the hydrology into the DEM.

We noticed that there is a "Burn Stream Slope" tool but the NHD flowlines is not the correct input as we are prompted for "FromElev" and "ToElev" fields that are not part of the accompanying attribute table.

So 2 questions linger:

1. Should we definitely be using the Burn Stream Slope tool?
2. What NHD data should be used for the "Stream" input if not the flowlines?

Thanks for the feedback!

Arc Hydro paths for your reference:
Terrain Preprocessing > DEM Manipulation > DEM Reconditioning
Terrain Preprocessing > DEM Manipulation > Burn Stream Slope


  • Update: "burning" the streams using DEM Reconditioning actually works fine, but simply looks odd because the stream network pixels are so much lower than the rest of the DEM thereby impacting the stretching of the elevation values and the symbolization.
  • swolnyswolny Member, NatCap Staff
    Hi Sam -

    I'm just wondering if you played with the inputs for Stream buffer, and Smooth/Sharp drop/raise in the DEM Reconditioning tool. By default, for example, the "Sharp drop/raise" value is set to 1000, which causes the final stream network to be 1000 units below what it was before and is quite extreme. You can probably get output that looks better to you by changing some of these values.

    ~ Stacie
  • Stacie,

    You're correct in your assumption that we did not change the default. Great suggestion! Given that the max elevation in our watershed is only about 350 meters, -1000 is terribly extreme and -100 would have probably done the trick as well.

    Thanks for the feedback.
  • Hi, am I right in assuming then, from this discussion, that carving a 'realistic' value for streams into a DEM is not the way to go. Rather it is better to use a highly exaggerated number ? Any rule of thumb ? How does this choice relate to the resolution of the DEM (i.e. Aster vs SRTM, or resampled for National Level studies) ?
    thanks, Ben
  • swolnyswolny Member, NatCap Staff
    It's really hard to answer that question. Many different methods of stream burning are out there, some change the elevation by a lot (one rule of thumb, for example, is to raise all non-stream cells by the value of the highest elevation in the DEM), others by a little. See this paper for an old, but interesting, overview of techniques.

    It is, however, important to consider resolution, as the DEM should not be of coarser resolution than the stream layer. Also in that paper are examples of what happens when you burn streams into DEMs that have a coarser resolution. To really do it right, there are a lot of factors to consider...

    ~ Stacie
  • Many thanks Stacie ! Very useful input.
This discussion has been closed.