2

I'm quite new to ArcMap and I have a question regarding the flow accumulation tool. What I did so far:

  1. converted TIN into raster (sample_distance/sampling_method distance: CELLSIZE 1 m)
  2. calculated flow direction
  3. filled sinks
  4. calculated flow accumulation
  5. calculated two new rasters with a threshold of 100 and 500 cells, respectively
  6. converted each stream raster into a polyline feature.

What bothers/surprises me is that many of these polylines are unnaturally straight. I expected a much more irregular pattern. I have actually been in the field and seen the area. There are no artificial channels that could explain why the lines are so straight.

Is there any mistake in my calculation process?

Any step I missed?

Have you ever had a similar output when using the flow accumulation tool?

Or would you say that the result makes sense?

enter image description here

  • What is the source of your TIN? As in how and from what the TIN created? Have you tried using ArcHydro? I find it a lot easier to work with when it comes to working out flow paths and subsequent catchments. – Keagan Allan Apr 16 '18 at 13:08
  • 1
    You should fill sinks BEFORE computing flow direction. Flat areas will create these straight channels, just an artefact of processing, I guess it is a scale related issue. – Hornbydd Apr 16 '18 at 13:21
  • I haven't tried ArcHydro yet. Thanks for your recommendation! – Caschmi Apr 20 '18 at 14:08
  • Yes, you are right, I should have filled sinks before computing flow direction. So I did the whole procedure again. The patterns remains exactly the same though. – Caschmi Apr 20 '18 at 14:20
  • @Keagan Allan, I don't have much information regarding the TIN. I only know that the elevation model was ordered from a state office and that the data stem from an airborne laserscanning in 2014. I will check with my colleague who provided the TIN. Is there any caution to be taken when converting TIN to raster and "forcing" the cell size? I set the cell size to 1 m because the default size was an odd number of 2.xxxx that I didn't want to use for further calculations. – Caschmi Apr 20 '18 at 14:58
1

Note that it's important to fill sinks before computing flow direction. Sometimes this issue can occur after filling the sinks, but as there are visibly man-made features in this raster (e.g, the road), this might not be so unrealistic. Other times, this occurs from over interpolation of a raster (creating highly refined flowpaths for the courseness of your grid). One way to check is by using the "Flow Path Tracing" tool in ArcHydro. If you identify that it's not following the true flow path, the quickest way to resolve this is by first burning the streams, if you have a known location in the form of some shapefile. Check Dean Djokic's report, "Comprehensive Terrain Preproccessing Using ArcHydro Tools"

  • 1
    Thanks for pointing out the wrong order. As I just commented above, I went through all steps again (in the right order) but the pattern of the flow accumulation didn't change. I agree in what you say about straight lines along man-made features. The pattern seems very realistic along the road, for instance. However, in the green area W/SW and E/SE of the intersection, there are no such features, so I struggle to find an explanation for the pattern. I will try the Flow Path Tracer tool in order to check for over-interpolation. – Caschmi Apr 20 '18 at 14:42
1

This kind of flat flow accumulation is very common where the DEM is flat (e.g., water bodies, LiDAR doesn't penetrate water!) or there are downstream structures that convey water through the ground surface (e.g., culverts, again LiDAR doesn't penetrate the ground, but most of the time, pier bridges are OK). You can try hydro enforcing (or DEM burning) through those artificial structures. In your screenshot, I can see a couple streams flowing under the road (culverts maybe).

  1. Burn the DEM though those structures
  2. Fill the burned DEM
  3. Calculate flow direction & accumulation using the burned & filled DEM

Judging from the terrain only (satellite imagery would help a lot, but...), I tried to add some "burn lines" (red lines) where there may be streams blocked by the road (high ground obstructing the flow). Hydrologically, streams are not discontinued at these points; rather they flow "under" the road (culverts). Usually, this under-road stream cannot be well represented by DEM, so we manually "connect" those disconnected streams by lowering (burning) the road so upstream cells can drain to the downstream more "naturally."

In a similar way, you can force flow accumulation by manually burning streams where the terrain is flat, but there is actually a defined stream from the satellite imagery.

enter image description here

This presentation may be useful.

  • Thanks for your help! What does "DEM burning" and "burn the DEM through those structures" mean? Can you specify on that step? Thank you! – Caschmi Jul 6 '18 at 7:31
  • No problem. I added some "burn lines" and more details above. There is an ArcHydro tool for burning streams (hydro enforcing or DEM reconditioning). Please refer to the presentation in the link. – Huidae Cho Jul 7 '18 at 9:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.