I am a bit confused about the resulting semivariogram I've got after detrending. The value of the sill is zero and it seems that the theoretical variogram is non existent. What does this result indicate? I have attached the two semivariogram (detrended and with trend).
It's hard to say for sure from the pictures, but it looks like your data (before detrending) does not have a lot of spatial autocorrelation. Then when you do detrending, you're removing even more spatial autocorrelation, and there's nothing left to perform kriging on.
Then when you do detrending, you're removing even more spatial autocorrelation, and there's nothing left to perform kriging on.
Are you sure your data needed detrending?
This is my interpretation too. Looks like the data doesn't need detrending. Try and read some of the documentation to understand what detrending is about and when there is, or isn't, a need to apply it. You shouldn't just use every tool or option available in Geostatistical Analyst just because it is there, you need to apply them based on knowledge of your dataset and statistics provided by Geostatistical Analyst.
Simply viewing the dataset's values symbolized with an appropriate legend in ArcMap, should tell you a lot about a possible need to detrend the data.
