It sounds like it's a matter of how much of your XYZ data covers your domain and what interpolation method you are using, assuming you're going with the 2D interpolator, nearest neighbour would -in essence- give different results than linear interpolation for example. (given that your XYZ data is healthy)
Hence my questions to the original poster (OP) to find out which version of Blue Kenue is being used and what, exactly, is the problem experienced. Both of those things are still unclear for me, so I cannot provide any help or guidance at this time. The more recent versions of Blue Kenue will likely behave differently than the older versions of Blue Kenue due to the large internal changes that are ongoing.
Kindly,
Alan - Current Blue Kenue Lead Developer