# What could be wrong with interpolating values extracted from an interpolated surface (grid)

275
2
11-26-2021 09:52 AM
by
Occasional Contributor II

I want to assess the impact of sampling density on the accuracy of interpolation using a particular interpolation method. The idea is to sample at different resolutions (grid-based), for example, 1 sample per 2kmx2km grid, 3kmx3km grid , 1 sample per 5kmx5km grid, etc. These sample points would be extracted from existing interpolated grid at the various sampling densities. The question is this: Given the fact that the supposed observed data (samples extracted from existing interpolated grid) are indeed interpolated, what possible impact may it have on the result of this study. Note that, the intent is to assess the interpolation accuracy using Root Mean Square Error (RMSE), Mean Error (ME), etc.

Thanks,

Tags (2)
1 Solution

Accepted Solutions
Esri Regular Contributor

This seems like a pretty reasonable study outline to me.

If I had to nitpick, I would suggest simulating data from the interpolation model rather than extracting the results of a single interpolation.  This allows you to do multiple repetitions for each gridding, where the data for each repetition comes from the same model and differs only in their random components.

That being said, in Geostatistical Analyst, you can only simulate from univariate Simple Kriging models, so that might not be an option.

I'm definitely interested in the results, I've done some ad hoc experimentation along these lines but not enough to draw any solid conclusions.

2 Replies
MVP Esteemed Contributor

good exam question!  What have you come up with so far?

... sort of retired...
Esri Regular Contributor

This seems like a pretty reasonable study outline to me.

If I had to nitpick, I would suggest simulating data from the interpolation model rather than extracting the results of a single interpolation.  This allows you to do multiple repetitions for each gridding, where the data for each repetition comes from the same model and differs only in their random components.

That being said, in Geostatistical Analyst, you can only simulate from univariate Simple Kriging models, so that might not be an option.

I'm definitely interested in the results, I've done some ad hoc experimentation along these lines but not enough to draw any solid conclusions.