Re: negative universal kriging values

[Edzer J. Pebesma]

femke wrote: Dear gstat list,   I am getting negative universal kriging values when the minimum should be zero.  Following some of the discussions on the list, I tried adding a slight nugget which didn't help.  Does anyone have any suggestions as to how this might be avoided?   Femke, think of universal kriging as linear regression (for the trend) + simple kriging of residuals. Linear regression planes are linear, and therefore always become zero at some stage; it is fairly easy to calculate where, it depends on your predictor variable values. To avoid this, you could work on the log-scale, and back-transform predictions. However, there may be a number of reasons why you would prefer not to do this. -- Edzer