I have a data set that is classified into 7 landcover classes. I've been asked to expand this data set to a new boundary.

How can I interpolate my existing data out to meet the new boundary?

In the example below:

  • A is the current data set with two landcover types (1 & 2)
  • B shows the new boundary that I need to expand and fill
  • C shows the desired interpolated outcome where 1 & 2 have been interpolated to the new boundary

Schematic example

This needs to be done for several hundred locations - some of which are slivers where the differences are minimal so I would like to avoid manually making the changes.

I suspect this may not be possible in the current form as I can't figure out how to tell the GIS which boundaries it needs to follow and the image above is a simplified example of the sort of boundaries involved.

Buffering is out as this would result in overlaps. A couple of solutions I've thought about include creating a point grid over the existing data and create thiessen polygons. Alternatively the data could be rasterised and then carry out some form of interpolation for the missing areas.

I'm using ArcGIS with a standard licence.

  • So your current data is vector? What about Snap? Eliminate? I'm wondering if the cases might be so different that a single tool wouldn't really work well for all cases. You could use buffering, union the buffer result, and use any areas that didn't overlap to incorporate into the existing polys. Areas that did might need a smaller buffer or something. If you went to raster Euclidean Allocation might work. Without actually seeing some different areas that need to be addressed, it's hard to suggest solutions that might apply globally. Is interpolation feasible vs a source classification? – Chris W May 14 '15 at 22:52
  • With no answers offered in 18 months I think it is time to edit to let us know what you have tried and where you are stuck in more detail. – PolyGeo Oct 2 '16 at 8:20