Team,
The default or first option for any raster re-sampling / re-projection geoprocessing tools is "nearest neighbor". This re-sampling algorithm is explicitly for integer or categorical data. The other two options are for continuous data and represent interpolation schemes.
Because "nearest neighbor" is the default re-sampling algorithm, when re-sampling or re-projecting a continuous data raster, numerical artifacts are introduced.
I work with continuous data almost explicitly (digital elevation models, hydrologic/snow data, etc). I also teach undergrads GIS. I have been teaching them about the difference in these re-sampling algorithms. However, having to choose an appropriate interpolation scheme for re-sampling continuous data (bilinear or cubic), but it is often a step that is neglected.
Can ESRI change the default re-sampling algorithm to either bilinear or cubic? OR can they add some sort of warning / prompt / flag on the geoprocessing tools that advises the user as to the appropriate sampling scheme?
Let me know what you think!
Joel Sholtes
Colorado Mesa University
