Has anyone else noticed this in ArcMap 10.7?
I have a custom .pyt toolbox that uses snap pour point and watershed tools to create a local watershed for user input sites. For the last 3 years, it has consistently run, taking about 90 seconds each time - I've used it thousands of times. Since my upgrade to ArcMap 10.7, the custom tool now takes 50 minutes to run. I included reporting to see what was going on and I discovered it was taking nearly 15 minutes to snap pour points and 33 minutes to delineate the watershed consistently for every input site I tried.
I tried running the snap pour point tool manually from the toolbox and it also took much longer than expected, around 9 minutes.
Is anyone else having trouble with the speed of the watershed tools in ArcMap 10.7? Did something change?
Hi Molly,
Sorry to hear that you are having performance issues with the Snap Pour Point and Watershed tools. Could you please share the sample datasets that are causing this issue and I can investigate further.
For Snap Pour Point tool,
1) Input pour point (Raster or feature)
2) Input accumulation raster (Raster)
3) Snap distance
4) Pour point field
For Watershed tool,
1) Input flow direction raster
2) Input pour point data (Is this the output from the Snap Pour Point tool? If so, you don't need to provide this)
3) Pour point field
Are you using any environment settings? (Such as Parallel Processing Factor)
Lastly, could you also provide any information on the machine configuration you are using?
OS Name:
Processor:
Memory (RAM):
Disk space and type (SSD or HDD):
Best Regards,
Neeraj Rajasekar | Spatial Analyst Product Engineer
Hi Neeraj,
The sample data can be downloaded here: watershed_data.gdb.zip - Google Drive - the GDB includes the input raster for the pour point tool, the flow accumulation raster and the flow direction raster. To answer your questions,
For snap pour point tool:
1) Input pour point (Raster or feature) - included in GDB
2) Input accumulation raster (Raster) - included in GDB
3) Snap distance - 0
4) Pour point field - Value
For Watershed tool,
1) Input flow direction raster - included in GDB
2) Input pour point data (Is this the output from the Snap Pour Point tool? If so, you don't need to provide this) - output from snap pour point tool
3) Pour point field - Value
Are you using any environment settings? (Such as Parallel Processing Factor)
I originally had no environment settings other than default. But, in testing, I tried increasing the parallel processing factor and it had no impact on the speed of the tool.
Lastly, could you also provide any information on the machine configuration you are using?
OS Name: Windows 10 Pro
Processor: Intel(R) Xeon(R) CPU E3-1245 v5 @3.50GHz
Memory (RAM): 16.0 GB
Disk space and type (SSD or HDD): HDD 224 GB total, 45.2 GB free space
I also made the switch to Windows 10 around the same time I upgraded to ArcMap 10.7, but I'm not sure if I ran these tools between the switch or not.
Thank you for checking this out!
Hi Molly,
I tried the Snap Pour Point and Watershed tools with your data. I'm unable to reproduce the performance issues you are encountering. For me these tools complete processing in 3.6 seconds for Snap Pour Point and 9.5 seconds for Watershed.
The machine I tested on isn't that different to what you have (I have 32 GB RAM and a bit more SSD disk space), but that shouldn't result in such a vast difference, especially since the dataset isn't very big.
Is there any other information you could share about your run?
- What output format was used? (I tried TIFF and FGDB and both result in similar time).
- Are you running through a python script? Does it take the same time through the application?
- Could you pleas confirm the software version used. Are you using ArcMAP 10.7.1 or 10.7?
Thanks,
Neeraj