Flow Direction tool not calculating approperiate values from a burned DEM

[JontyKnox]

I am conducting some watershed work in conjunction with land use etc and I recently had to burn my streams into my DEM so a proper flow accumulation of the area could be calculated.
The burn of the streams went well, but now when I try and calculate the flow direction instead of getting the standard 1,2,4,8,16,32,64 and 128 I have values ranging from 1 to 248. Most of the pixels are the standard values but there are in the region of 10,000 pixels that aren't.
Needless to say this completely throws off my flow accumulation grid and I can no longer calculate my watersheds.

[jamiefinney]

I had this exact same problem when i was running this stage and i could never get it to work. I'm fairly new to arc hydro and have been playing around with if for the last couple of weeks i only started to get it working properly when i installed a version from here: http://www.hec.usace.army.mil/software/hec-geohms/download.html or maybe here: http://www.hec.usace.army.mil/software/hec-geodozer/index.html now it seems to work fine and a few extra tools to boot!

These downloads contain both arc hydro and hec-geohms / dozer i've got them both installed at the moment hope this helps

[JimMonchak]

Have you tried the "Fill Sinks" operation again?  I've run into the same problem after building walls or burning in the streams.  So, you may need to re-run the fill sinks after either of those operations-- then re-run the flow direction tool.

Jim

[JacquesTardie]

I'm having the same problem here, except that my values are stopping at 196.

Going to try and refill the DEM I'm using.

Anyone found the reason this happens and a solution to fix it?

[AniaWajda]

Hello! I have run into this problem today (my DEM values were topped at 196 and there were much more categories than a standard D8 analysis should allow for) - I know that my response almost 9 years later might not be helpful to you, but maybe it would help someone else in the future!

Here are the steps I followed - my results were what I expected:

1. Obtain DTMs
2. Mosaic the DTMs into a single image (Mosaic to New Raster tool)
3. Filter tool to fix major issues with inconsistency in the DTM data (Filter tool)
4. Fill the inconsistencies (aka sinks) in the source layer (Fill tool) using the filtered mosaic
5. Calculate flow direction (Flow Direction tool) - should have following numbers: 1 2 4 8 16 32 64 128 only

Best of luck everyone!

Ania

[deleted-user-yIALGElyA_K9]

Ania,

Would you mind sharing more details about Step 3, your filtering process. We are looking into doing this with our high resolution LiDAR DEMS (sub-meter). Would love to know more.

Thanks,

Jacob

Indiana DNR Division of Water

[AniaWajda]

Hello Jacob,

I used the default that ArcGIS provides in their Filter tool, using Low pass 3-by-3 filter. Our DTMs were at 2 m resolution (provincial acquisition) so I am not sure how to help with your higher resolution data. Full disclosure, I am an undergraduate student working on her final year project, so I am by no means an authority on the subject! I hope someone else would be able to provide you with some direction!

Best regards,

Ania

[ChristineDartiguenave]

For clarification, the values 1,2,4,8,16,32,64 and 128 in the Flow Direction raster indicates that all cells flow into one and only one other cell. When one cell may flow into multiple other cells, the value stored in the flow direction cell is the sum of all potential directions. This usually means that you will end up with sinks - a sink is an area where the flow direction is not precisely determined as going into one and only one direction. When you fill the input dem, you are eliminating the sinks and forcing the water to find a way to flow out. This is the dendritic scenario. Sinks may be real and we have ways to handle them - this is the deranged scenario.

Christine Dartiguenave

Esri Water Resources Team

[AniaWajda]

Thank you for that clarification, Christine! That's very interesting - I am just beginning my journey with topography analysis (I am in my final year of undergrad) so learning new things is always appreciated. I think for my purposes and the type of terrain I deal with, filling the sinks was an appropriate decision (supported by the supervising professor) but it is very interesting to learn that there are other ways of dealing with actual sinks.

Best regards,

Ania

[AdamDabrowski]

As far as I understand, FlowDirection doesn't work properly when there are sinks. Sometimes after burning rivers they become new sinks therefore the algorithm cannot work. The solution is first to fill the DEM, then to check the network so it doesn't have any gaps between river segments, burning it and then running FlowDirection.