Lack of concordance between the stream calculated by Delineatelt and that of the NDR result

Hello community.
As noted in the guide, I used the Delineatelt model to generate the watershed file before compiling the NDR model. However, the output files do not have the same finesse as that of the stream file of the Delineatelt model. My question, how am I going to proceed to have the same fineness of the hydrographic network of the output files of the NDR model?

Hi salaha, thanks for reaching out to us. I wonder if there is some confusion here.

From what I can see, the top map is total nitrogen export (from the guide: “A pixel level map showing how much nitrogen from each pixel eventually reaches the stream” ([units: kg/pixel/year).

The bottom map is the stream network (with values of 0 - black, and 1 - white, where there is channel). If you want to see the output streams, look in the “intermediate_outputs” folder and it will be named “stream…”

So, they are completely different things. If you add a range of colors in your top map, you’ll see that there are incremental values, rather than 0 & 1. Does that make sense to you?

See how the guide describes n_total_export here… NDR: Nutrient Delivery Ratio — InVEST® documentation

I hope this helps.

"Thank you dear Nadine for the response elements.

Yes, indeed, there is no confusion, but I would like to know how I will proceed to avoid the large thickness of the stream of output files?"

Hi Nadine.
Sorry for not articulating my issue well earlier. Actually, the thickness I was referring to was about the width of the rivers. The only thing bothering me about the output files from the NDR is that the rivers are wider. I’ll send you a photo of the N_total from another result I saw, although it’s not the same area, the rivers have a smaller width.

Hi salaha,Can you post the legend and tell me what color you think represents rivers?


Hi dear Nadine, yes indeed, this is an output from NDR.

But for my output as well (second image), what’s in white resembles the rivers, and in the legend, there’s no white color. What do the white lines represent? And how can I make them disappear from the file?

I’m sorry, but now I see your issue. Here is another post with the same questions, with an answer from my co-worker, Stacie. This is a current model limitation, but I think we are working on providing an option to generate streams using D8 routing.

I’m sorry, it took me a while to understand your question. Let me know if the option suggested by Stacie works for you.


Thank you very much dear Nadine. Also sorry for not being too clear in explaining my problem.

No, you were actually clear, I just forgot about this issue and I am new at responding to community support requests, so I’m very sorry. I will bookmark this question and follow up with our software team to find out if/when they will allow the D8 option.

Thanks so much for your patience.

Hi @salaha -

Our software team is in the process of adding the D8 flow direction option to all of our models that include flow routing, which will generally produce streams that are 1 pixel wide, instead of the wider ones often produced by MFD. This will take some work, and I don’t know what the timeline is for it to be released, but don’t expect it soon.

~ Stacie

Hi dear Swolny,

Thank you very much for your response. Could you please provide me with any other options you might have?

Honestly, I don’t have any other resources. At the moment, I am trying to burn streams into a DEM to use in NDR, so far unsuccessfully. This is definitely an issue that I don’t have a good workaround for.

~ Stacie

Just a quick update. @salaha, if you use ArcGIS, I just had success using ArcHydro’s DEM Reconditioning tool to burn a stream network into the DEM, which caused NDR to create results where the stream network is not wide (it follows the raster version of the stream network that I burned).

This is worth trying (despite ArcHydro’s poor documentation, even after all these years), but only works if you have a good stream layer to use for burning. If you don’t have a separate stream network to use, you could try using the InVEST RouteDEM tool with the D8 flow direction option, and use the resulting stream layer for burning (assuming that it looks good to you).

In case anyone in the future wants to try this, here are some quick notes:

  • Download for ArcHydro
  • Install it using the installation instructions. The next time you open an Arc document, ArcHydro will appear in the Toolbox list.
  • Use the tool Terrain Preprocessing > DEM Manipulation > DEM Reconditioning:
    • The Input Stream Raster should have a value of 1 where there are streams, and a value of 0 for the background (non-stream) pixels, covering the rest of the study area. (The ArcHydro documentation does not tell you this, but the process did not work for me when I had a background value of NoData).
    • Number of Cells for Stream Buffer: 3
    • Smooth Drop in Z units: 10
    • Sharp Drop in Z units: 100
      • You can play with these 3 values, but this worked fine for creating the correct stream network in NDR in my case.

If anyone else tries this, please let us know how it worked for you.

~ Stacie

Thank you very much, I will give it a try and keep you posted on the outcome that will be obtained.