Kriging Around an Island?

Hello Forum,

I have been working with several different interpolation tools to create the most accurate surface from sample point data of a waterbody surrounding an island. The only problem is that the types of interpolators I have found that can apply a barrier are IDW, Spline with Barrier, and Kernel Interpolation with Barriers. These techniques have given varying results with Spline and Kernel being the only ones that allowed data to travel around barriers (data points are not read by "line of sight"). The real question is how to apply statistics to the process to better understand the accuracy of the resulting surface. For that reason and a push from my committee, I would like to apply a kriging interpolation, but continue to fall short trying to have the data work around the shoreline rather than through it. Is it possible to apply a barrier in this function on a square processing extent that contains a horseshoe shaped island, keeping sample points inside and outside the horseshoe from interacting?

Thank you,
Justin