I'm stuck and its a critical juncture on some data I'm prepping for a project. My initial gut feeling is that I have a spatial index issue on my data that I cannot appear to resolve by dropping and recreating the indexes!. Would you have any idea of how to resolve my issue ?
The example I have is buildings locations (points) and I need to select them via select by location where the points are completely within a polygon which has an attribute, based on this selection I update attributes and go onto the next process - so all very simple. I have done this process and whilst QC'ing the results I have spotted some anomalies - some points are not being selected by the polygon although they are definitely inside the polygon. (see index_issue1.png)
The polygons have has been created by a union on cover and a buffer of buildings points a 2.9m, and then the building buffer removed. (see index_issue2.png), the over all polygon is as per example attached (see index_issue3.png)
I have tried this process in a file .gdb and Oracle both with the same results. If I create another polygon without all the little holes I can select everything I need as per expected , if I edit the geometry of the point and update it by a fraction of a millimetre then the object is selected in the select by location process using the original and cookie cut polygons, what I'd like to see if there is another process I should be using? a better way to ensure I'm not loosing any data? is this a common occurrence ..
My SDE is 10.3.1 and my client is 10.3.1 also. (Using ST_Geometry types)
Any help would be hugely appreciated, I'd like to keep the little hair I have left.
you mention little holes, but they aren't shown in the first two images. Is there a chance that the errant point is on the border, or what appears to be close to a border of one of those. And are your coordinates projected or geographic?
Thanks for taking the time out to look at the issue,
In index_issue1.png in the original post - the pink is the polygon I'm using as a selector, the red dots ( and the cyan dots - selected) are the buildings and the white circles with the thin black line are the 'holes' in my polygon. This example shows that all the points in the pink polygon are selected except for the one with the arrow pointing at it, it clearly should be and there is no geometry issue at this location.. I curiously have another example where a point in the circles gets selected - but that's another issue
thanks for the clarification Paul... It seems that you have ruled out exporting this to a new file and the projection issue, that leaves the source of the data, but I see no reason why this could happen since you indicate editing it by a millimeter fixes things, which should also be fixed when things are exported
Yep tried everything , file, personal and sde but none offered me a fix. The shifting the geometry is not viable - as I have to identify these as the fail to be selected first ,
I guess I'll have to leave it well enough alone ....... unless somebody come up with a alternative or spots something I'm doing incorrectly.
Cheers and thanks for taking the time out
I assume you have already checked the geometry of both layers. As a workaround, if you set a very small search distance does that somehow work better? Assuming that doesn't work if you run an overlay can you then use that to make queries and updates, and then join it back to your original data?