Quality estimation of a spline interpolation with barriers

Question

I interpolated the age of minerals for the Inner Tien Shan mountains (Kyrgyzstan) using the ArcGIS tool "spline with barriers". I use ~ 200 points for my interpolation. The barrier that I use is a fault map (polylines, green in example). For my further interpretations, I would like to estimate the quality of my interpolation. At the moment I am using the point density for this purpose, but this is only correct for regions where no barriers are (see example below).

Is there a solution, where I can calculate the point density with taking my barriers into account or are there maybe different approaches to estimate the quality of my interpolation?

Example:

Here D1,D2 and D3 have all the same point density of 36.5%, whereas my interpolation quality is different in each region.

Answer 1

What you need to do is to separate a part of your input points (at least 10% of your data), then with the 90% data interpolate your variable (in your case spline with barriers).

After that you must validate your results using RMSE for example, save the 10% of your data in a new shapefile and use extract values to points in the toolbox (spatial analyst->Extraction). You can find the RMSE equation in this site.

Answer 2

I don't think that interpolation between points, unless you have really good coverage will work for minerals age. It seems that proximity polygons inside individual areas will do better job. With spare coverage it is worth to consider methods developed by Hutchinson (Australian university), where cotinious raster of something helps interpolate values at XY points. Climatologist often use elevation model as such raster to help with temperature mapping. The trick is to find raster, e.g. depth to the layer in your case? Privet Tian-Shanyu...