I have two tables, 'protein' and 'data_aggreg', in a PostGIS database. I am simply trying to query the points that intersect (all of the points in 'data_aggreg' share exactly the same geometry as a portion of the points in 'protein' table).
This is typically a simple operation. However, this time PostGIS and QGIS return nothing and ArcGIS returns the correct selection of points. The data in both QGIS and ArcGIS are sourced directly from the same PostGIS database via their respective methods of connecting to external DBs (ie, no shapefiles, csv's, etc are being used for this analysis). Both tables are in projected coordinate system EPSG 26912 and both tables are single point geometry.
When I run ST_Intersects in PostGIS on the two geom columns nothing is returned. When I run the 'select by location' tool in QGIS nothing is returned (I've also tried spatial join with same effect).
Syntax for ST_Intersects, which returns nothing:
SELECT protein.protein, data_aggreg.pro_2016 from production.protein, scratch.data_aggreg where ST_Intersects(geom26912, geom);
However when I run 'select by location' in ArcGIS I get the expected result. That is, the subset of points in the 'protein' table that share the same geometry with 'data_aggreg' are correctly selected. Also, when I change my PostGIS 'where' query from ST_Intersects to geom1 = geom2 I also get the expected result.
Syntax when using '=' in PostGIS, which returns the expected result:
SELECT protein.protein, data_aggreg.pro_2016 from production.protein, scratch.data_aggreg where geom26912 = geom;
It is my understanding that if any part of features share any space then ST_Intersects will return 'true'. So it follows that if geom1 = geom2 then ST_Intersects should return 'true'. But this is not the case.
So why the difference in results within PostGIS when using ST_Intersects vs '=', and why the difference among PostGIS when using ST_Intersects, QGIS 'select by location', and ArcGIS 'select by location'?