Count Overlapping Features taking too long to run

515
2
03-30-2022 04:15 AM
MatthewLehane
New Contributor

I have an existing line feature class that I am attempting to run the "Count Overlapping Features" tool on. However, this tool is taking painstakingly long to run, over a day so far. The current feature class has about 4,500 existing features. I assume it will dissolve into many more features than that. Is it supposed to take this long? Is there anything I can do to speed it up? I have an i9 CPU, 64GB of ram, and a GTX 3070 graphics card.

Furthermore, I ran the "Dissolve" tool on this same feature class and it completed with a warning in less than a minute, the warning stated that not all features could be split because I hit maximum memory usage.. which I did not even know was possible in ArcPro. It split into over 11,000 features before it must of hit the maximum memory usage and did not complete the dissolve on all features. 

0 Kudos
2 Replies
DanPatterson
MVP Esteemed Contributor

Locally stored data?

In the same coordinate system? (otherwise you have added a projection and temporary data into the mix)

Pairwise Dissolve (Analysis)—ArcGIS Pro | Documentation might be better than Dissolve

count overlaps doesn't use tiling/subdivisions

Tiled processing of large datasets—ArcGIS Pro | Documentation

which means you might have examine the options.

Count Overlapping Features (Analysis)—ArcGIS Pro | Documentation

would be impacted by not only the number of features but the total number of vertices as well


... sort of retired...
DuncanHornby
MVP Notable Contributor

You give no indication of the topological nature of the overlaps. Do you have sparse overlaps or dense overlaps (think of hundreds of lines stack on top of each other). Are the lines long and overlapping other lines  in multiple places?

It's always best practise to include images of your data because just describing a problem is usually hopelessly insufficient in describing the data. So edit your question to include that information.

I suspect without seeing/knowing anything about your data is that it's densely stacked. If you read the Usage section of the help file on Count Overlapping Features it clearly states output is a planarized union of the inputs. A few thousand stacked lines of varying length and overlap will quickly generate millions if not into the 100's of millions of segments... And then you wonder why it takes so long....

0 Kudos