Removing Trends: Clarification

[KirstenFalzarano]

Hi There,
  I thought I understood what a trend was, but now I am a bit confused. In my dataset, I am using kriging to interpolate a soil surface based on elevation measurements. The surface covers an area of 9x8m, with a slope of 30cm across approximately 9m (nw to sw direction). When I look at the data in trend analysis, there is a first order trend, which corresponds to the slope. Is the sloping trend something that I want to remove, or is this the source of the autocorrelation in my data? According to one of the geostatistical analyst tutorials, if your data is represented by a sloping hill (plane) you can keep the trend in the data, unless it is too simple to represent the true surface. Is the easiest solution to try both and pick the best cross-validation results? Finally, I am creating interpolated surfaces for several different soil layers and comparing them. Is it best to use the same method/parameters for all? What if one method (ie ordinary kriging with trend removed) works better for one layer and not another? Should the method that works the best overall be selected?

Thank you for any advice - this forum has been VERY helpful to me!

[EricKrause]

The problem is that trend, autocorrelation, and anisotropy can all look the same.  Even if you see something that looks like a trend, you may be able to account for it through autocorrelation or anisotropy.  The path you choose should be based on your knowledge of the data and model diagnostics like cross-validation.

As for whether to use the same methodology for the different soil layers, that's a tough question to answer.  If you use a fundamentally different model (for example, remove trend on one layer and do anisotropic corrections on another), you're implying that the underlying physics of the soil surfaces are different.  The physics of the soil surfaces may well be different, but that's a decision that you need to make as an expert in your domain.

[KirstenFalzarano]

Thank you, Eric!
   That makes sense. I guess the problem is that none of the kriging methods I've tried produces drastic differences, so choosing the right one is a bit difficult! Thank you for taking the time to answer my questions!