I need to convert raster data to point data so I can use the points for interpolation via kriging. My problem is that my resolution is 5 ft and I'm working on a whole city. The data I am converting to points is grouped by zone with No Data values between; most/all of the points in each zone have the same value. I need the high resolution because some zones are linear, just 2-3 cells wide and about 20 cells long, while others are huge behemoths. I need to capture the smaller zones but I don't need 2000 points in the larger zones; because the values are all the same, I might need 10 or 20. I experimented with lowering the resolution to 50 and 100 ft but I lose much of the smaller zones. My raster is floating data so converting to polygons to do any other vector stuff isn't possible, I don't think.
Am I doomed to have a billion points and manually remove ones from the big data chunks?
What kind of processing time am I looking at if I do 5 ft Raster to Point and have one point per cell and I interpolate those? It already took several hours to interpolate between 10 points via kriging over the whole area.
I would not like to sacrifice my data resolution for the small zones if I can help it. They are very important and represent the variability I need to capture.