Interpolation of agronomic data from plot to field scale: using a clustered versus a spatially randomized block design