We prepared punctual and block-kriged maps of sodium adsoption ratio (SAR) of an experimental plot measuring 162 x 96 m for four depths. Positive definite, transitive, and isotropic spherical models were fitted based on grid sampling at 12-m intervals at depths of 0 to 30 (d4), 30 to 60 (d2), 60 to 90 (d3), and 90 to 120 (d4) cm. Small patches in point-kriged maps disappeared in block-kriged ones. The within-block (144 m2) variance—1.328, 1.619, 3.704, and 7.731 in the four depths, respectively—was directly proportional to the nugget variance. The standard error (SE) of block-kriging was less than half that of point-kriging. The SE of the interpolator increased as a function of depth in direct proportion to nugget variance. We also validated the kriging model for delineating regionalized variation. The kriged map revealed that sodicity was comparatively lower and more homogeneous in the area represented by the lower half of the map, compared with the upper half. The lower half of the field was, therefore, found to be suitable for planting such relatively sensitive species as kinnow (citrus) and peaches. The upper part (around P1 in the map) could be utilized for such relatively tolerant species as pear and pomegranate. To arrive at an independent random error for making valid treatment comparisons, particularly in the upper portions, the model must also account for inherent soil variability.
© Williams & Wilkins 1988. All Rights Reserved.