At ArcGIS Desktop 10.5, I have a polygon file geodatabase feature class. Many of its polygons are spatially coincident; that is, if I click on a single polygon in Arcmap to select it, the resulting selection will have multiple records, all sharing the same exact polygon outline. Here's a screenshot example where one polygon is selected, resulting in 74 selected records. In my mind, they represent 74 polygons stacked upon each other.
The attribute table contains a long integer field, noise_value. I need to sum this field for each input polygon, which, as I mentioned above, may be comprised of multiple spatially coincident polygons, and output the result to a new, flattened, polygon feature class containing the summed value.
In my example, the 74 overlapping polygons would be reduced to a single output polygon with a summed noise_value. Of course, this summation would have to be repeated for all input polygons.
I've tried these approaches, and all have failed:
1. Flatten the input feature class while applying the field map merge rule "SUM" for the noise_value attribute.
I tried this approach with the Feature Class to Feature Class tool as well as the Merge tool. In each case the output was identical to the input; no flattening or summing occurred.
2. Create a flattened output feature class and then spatially join it to the original feature class, summing by noise_value with field map merge rule SUM.
I tried this approach with the Feature to Polygon tool, and also with the Union tool to create the flattened output feature class. Both cases failed, with a "Topoengine error". Research indicates that this type of error can be traced to inadequate hardware resources, but I'm using a computer with an i7 processor, 16GB ram and a 1 TB SSD.
3. Use the Dissolve tool (@BERA's suggestion) with noise_value as the statistics field and Statistics Type = SUM.
Since what I want to do is flatten (dissolve) the geometry, I left the Dissolve Field blank. The result did not work for me because it created a single output polygon with one overall summed noise_value.
I've run out of ideas, and am looking for advice.