Find Saddle Points empty output

[Pedro Venâncio]

Dear Whitebox Geospatial team,

I'm trying to find the saddle points from a DTM of a vast region in Portugal.

I've tried with an DTM of 25x25 meters, and then another of 80x80 meters. I also have tried with a smaller region, but in all attempts, I always get empty shapefiles or all pixels of zero from the Find Saddle Points algorithm.

The steps I'm doing are:

1. Import the GeoTIFF DTM to .dep format;

2. Fill sinks with Breach depressions (fast);

3. Create the flow pointer grid with D8 Flow Pointer (direction);

4. Create the valley network with D8 and Rho8 Flow Accumulation (with total catchment area option);

5. Create the ridge network with Branch Length;

6. Run the Find Saddle Points with:

    - Ridge Network file: from step 5;

    - Valley Network file: from step 4;

    - D8 Flow Pointer file: from step 3;

    - DEM file: from step 2.

But the output is always empty.

Am I missing something? Should I configure some thresholds somewhere?

Thank you very much!

Best regards,

Pedro Venâncio

[dr.john.lindsay]

Hello Pedro and thank you for asking your question. First, I just want to point out that Whitebox GAT, the software that you are using, is now legacy software and is no longer supported. We have moved development of the Whitebox project onto the WhiteboxTools platform. But as you are likely aware, this particular tool has not been ported over to the new platform (more on this in a bit).

Okay, why might this be happening for you? Well, in your workflow, you do you not indicate that you are thresholding your flow accumulation nor branch length rasters to create Boolean valley and ridge network rasters. I suspect that that is the cause of your issue. Are you specifying the raw flow accumulation and branch length rasters as the inputs to the Find Saddle Points tool? If so, then you must first threshold them to create Boolean rasters, i.e. a 'stream channel network' is consistent of areas of high flow accumulation and a 'ridge network' is consistent of areas of high branch length.

In fact, I have very recently re-visited this tool, called LowPointsOnHeadwaterDivides, and have created an equivalent for the WhiteboxTools platform. It is actually a greatly simplified workflow since the user must specify an input DEM and a stream network only. Unfortunately however, this tool will not be part of the 'open-core' of WBT, but will rather become part of the new set of plugins that we plan to offer for sale starting next month.

Regards,

John