The objective of my study is to compare the accuracy of the model SK (simple kriging) and OK (ordinary kriging). When I use OK to apply on my data, there is a little bit variation on the prediction map but when I use the SK to apply on my data, there is no variation on the prediction map.

My data is the water quality that I collected from the lake:

enter image description here

My problem is there is no variation on my prediction map when I use SK to apply on my data. What should I do?

This is the image of it semivariogram:

enter image description here

enter image description here enter image description here

  • Data provided as an image of a table isn't very useful - I've scanned that and created a text file which I've included in my answer but you should try and include data than can be read as text in your questions. – Spacedman Jun 20 at 11:26

You have 11 points and a very flat variogram. This means there appears to be no spatial structure and your data is random noise. Any best prediction at a non-sample location is going to be the sample average.

You could modify the variogram binning or variogram parameters and get something with a bit of spatial structure, but that might be more luck than actual ground-truth spatial correlation.

If anyone else wants to try Kriging on your data, here it is in text format (I ran the image through gocr and then fixed the mistakes, so there may still be errors).


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.