Strange banding in SDR outputs

Hello all,

I’m seeing some strange straight line bands in the SDR model outputs. It only seems to crop up for certain regions, but does spread over quite sizable areas where it does appear. The image below shows an example for the USLE map output.


I thought this might be an issue with the DEM, but it has occurred with 2 different DEMs. There are also no weird straight lines in the land cover data either. Is this something anyone here has encountered before, and any idea how much it is affecting the results?

Many thanks,

Hi @lukezw -

I’ve seen that happen when I’ve resampled the DEM (usually during reprojection) using Nearest Neighbor interpolation. If this is the case for you, try starting from the original DEM and reproject with Bilinear or Cubic and see if it changes anything.

You can also look at files in the Intermediate output folder to see where this pattern starts, to narrow down which layer or calculation is causing it. It’s likely to be when flow direction is calculated.

~ Stacie


Hi Stacie,

Thanks for your response. I couldn’t remember what resampling technique I used, so I tried again with bilinear and the straight lines disapperead.

However, I discovered to my surprise that the flow direction output produced by the SDR model has values all over the place. When I use the same DEM in RouteDEM, it produces a flow direction map which has 8 integer flow directions only (ranging from 0-7), which is the sort of flow direction output I would expect. Is there something wrong with the SDR flow direction output, or is it expected behavior that it doesn’t stick to the 8 integer values? This occurs with both the old and corrected (bilinear sampled) DEM.


1 Like

Hi Luke -

Glad to hear that bilinear helped.

RouteDEM provides two flow direction algorithms - D8 and Multiple Flow Direction (MFD). D8 forces all water to go in one direction, and will produce only the 8 integer values, while MFD works very differently, it allows water to flow in multiple directions from each pixel, and produces intermediate values. SDR uses MFD, and there is not currently a way to choose D8 as an option within the freshwater models. So if you use RouteDEM with the MFD option, theoretically you should get the same result as you get in SDR.

Of note, it looks like the SDR User Guide still says that it uses the D-infinity algorithm for flow direction. This is old/incorrect and (I’m pretty sure - verifying with @jdouglass) needs to be updated to say MFD. I’ll make sure to do that.

~ Stacie

1 Like

Great thanks for explaining that and reassuring me that the flow direction algorithm is working as expected.

@swolny, yes, I can confirm that SWY uses MFD, though @rich once told me that MFD could be classified as a D-inifinity implementation which is why I had left the UG chapter as-is. @rich, could you confirm this?

I think I might have said that Dinf is a MFD? I’ll change that in that chapter when I get to adding D8 as an option too.

Thanks @jdouglass and @rich. Whatever term is used, let’s just keep it consistent within the User Guide, and between the User Guide and tools, otherwise it can be confusing (at least it is for me.)

~ Stacie

Okay! I’ll change it right now.

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.