Kriging giving variable results?

359
1
05-15-2014 03:10 AM
ErnestDunwoody
New Contributor III
Hello Expert Krigeres,

I have  wheat yield data from a Header for a paddock. After cleaning up the messy data (around the edges of the paddock etc) and
then interpolating it (1.66m resolution= default) a number of times by using the Kriging  tool set in Spatial Analyst (using default Ordinary and Spherical settings) I get slightly different results, including -ive values. All settings such as resolution the same. Why?

If I do the same thing and change the output resolution (2 m) I get different results from the first resolution at which I kriged it. I understand that. But when I do it again and again at the second resolution I sometime get the same results and sometimes get different results.

The results are different enough to throw the yield calibrations out of whack.

Details are:
Yield points are about 5 m apart in the rows and the rows are 16 m apart.

I have attached the point file. I'm interpolating on the Yield field in that file.

Sure would appreciate any suggestions for why this might be happening and how I can get stable kriging values.

Thank you,
Ernest Dunwoody
0 Kudos
1 Reply
SteveLynch
Esri Regular Contributor
Ernest

Kriging can produce negative values from an all positive input dataset.

I used your data and repeated the kriging 100 times without changing any of the parameters. All the outputs are identical, which is what we would expect.

By the way, you have coincident data. The kriging method in Spatial Analyst will only use one of the coincident values, whereas, kriging in Geostatistical Analyst can use all the coincident data.

Steve
0 Kudos