A response-surface calibration model for rapid and versatile site-specific lime-requirement predictions in south-eastern Australia

Abstract

The development of response-surface calibration models that may be used in conjunction with the lime requirement buffer methods is described. The buffer methods tested were the Woodruff, New Wooruff, Mehlich, and Shoemaker, McLean and Pratt lime-requirement buffers. Model predictions were compared with those obtained from multivariate models and buffer methods calibrated using conventional linear regressions. The multivariate models described lime requirement as a function of a number of soil variables. All of the models were validated against soil : CaCO ₃ incubations using a statistical jackknifing procedure for error and bias estimations. The advantages of the derived response-surface models were their improved prediction accuracy and flexibility, with a choice of any target pH _{CaCl 2} value between 5.5 and 7 without need for individual calibrations. The response-surface model for the Woodruff buffer method produced the most accurate predictions of lime requirement. The uncertainty of its lime requirement predictions for acid soil in an agricultural field at Kelso, New South Wales, Australia, measured by 95% confidence intervals, was 0.22 Mg/ha. A spatial analysis of lime requirement for the field showed a range of 4–11 Mg/ha. This range provides a reason for site-specific lime applications. Under- and over-applications resulting from a ‘blanket’ 7.13 Mg/ha single-rate application of lime over the field were estimated to range from –4 to 2.9 Mg/ha.