CoKriging Order of Trend Removal

[MelissaSlater]

I am trying to find a good interpolation model for some precipitation data and would like to remove the elevational trend but I am unsure on the order of trend removal. I have my precip data as Dataset#1 and my DEM as Dataset#2. I was planing to use the Simple Kriging with Normal Score transformation, but don't know which trend removal order to chose for each dataset. Any info on what the orders mean (i.e. first vs constant) or best practices for elevation trend removal would be greatly appreciated.
Thanks!
Mel

[EricKrause]

Trend removal is done by fitting a polynomial equation of the x- and y-coordinates.  So, a first-order trend removal will remove linear trends.  Similarly, a second-order polynomial will remove quadratic trends, and higher-order trends will use higher-order polynomials.

The best mechanism for deciding which order is to use the Trend Analysis ESDA tool.  Look at the graphic in that topic.  When you select "Trend on Projection" you can change the order of polynomial.  Try different orders to see what fits your data best, and use that same order as order for trend removal.

[mariaaglq]

Hi Eric, I hope you don't mind my little follow-up question 🙏

In the example that Melissa gave, having precipitation data with an elevation trend, please imagine that we have decided to use cokriging to interpolate precipitation -with elevation as auxiliary variable. Aren't we using that elevation component of the precipitation data as a guide for our predictions, therefore shouldn't we keep that elevation trend instead of removing it (because the cokriging model will make good use of it)?

Thank you very much for your time