Spatial Dependence and Interpolation of Soil Properties in West Sumatra, Indonesia: II. Co‐regionalization and Co‐kriging

Abstract

Co‐kriging was used to interpolate values of topsoil 0.5 M NaHCO3‐extractable P at 234 locations in West Sumatra, Indonesia, by exploiting its spatial covariance with a more densely sampled property, 25% HCl‐extractable P. NaHCO3‐P and HCl‐P had been sampled at 52 and 107 locations, respectively, in a nongeometric pattern across the 106 650 has region. Isotropic semivariograms of NaHCO3‐P and HCl‐P and their cross semivariogram showed spatial dependence over 6.3, 4.2, and 10.4 km, respectively, at the sampling scale used. The map of co‐kriged values for NaHCO3‐P showed a similar regional pattern but more local detail than that achieved by kriging from NaHCO3‐P samples alone. Co‐kriging reduced estimation variances (relative to kriging) by up to 40% in areas where sampling density of NaHCO3‐P was lowest. Co‐kriging variances exceeded those of kriging by up to 10% in areas where sampling density of NaHCO3‐P was high. In such areas, the covariate HCl‐P had little effect on the interpolated value but still added a component to the estimation variance. Co‐kriging could not be used to interpolate values for extreme locations where there were no NaHCO3‐P samples within the radius of the kriging neighborhood. Despite local improvement in estimation precision the full benefits of co‐kriging interpolation are not universally obtained from nongeometric variate‐covariate sampling patterns.

Improving Map Accuracy of Soil Variables Using Soil Electrical Conductivity as a Covariate

“…However, both methods generated similar trends in the variability of the soil parameters. Comparable results have been found for topsoil silt cokriged with subsoil silt and sand as covariates (McBratney and Webster, 1983a), NaHCO 3 -extractable P cokriged with 25% HCl-extractable P (Trangmar et al, 1986), NO 3 cokriged with soil EC, (Zhang et al, 1992) and soil EC of soil paste extract cokriged with apparent soil EC (Vaughan et al, 1995). Improved local detail of the cokriging maps was due to the finer sampling grid of the covariate, soil EC (McBratney and Webster, 1983a).…”

“…The pasture of study was oriented mostly in one dimension, and there were insufficient sample pairs of the soil variables for the n = 30 and n = 15 sampling schemes to obtain well-structured directional semivariograms (Trangmar et al, 1986). Therefore, it was assumed that all semivariograms were isotropic.…”

“…McBratney and Webster (1983a) estimated topsoil silt content utilizing its spatial interrelationship with subsoil sand and subsoil silt as secondary variables. Trangmar et al (1986) explored the use of densely sampled 25% HCl-extractable P as a covariate for estimating the variable of interest, NaHCO 3 -extractable P. Noninvasively collected measures have also served as useful secondary variables. Leenaers et al (1990) improved the quality of interpolation maps of zinc concentration in the Geul floodplain in the southern Netherlands by incorporating elevation data.…”

Modeling spatial variability and transport processes in a glacial till soil

“…However, when not enough sample pairs are obtained in the major or diagonal directions to obtain wellstructured directional semivariograms, isotropicity is assumed and an isotropic semivariogram is estimated based on all the samples. This phenomenon has been encountered in earlier studies Trangmar et al,, 1986b;Mohanty and Kanwar, 1992] and also in this study. Moreover, the assumption is well supported inasmuch as the well-structured auto semivariograms are the foundation for wellstructured cross semivariograms.…”

“…demonstrated the cross-correlation between K"t and macrovoids. Trangmar et al [1986b] used cokriging to interpolate values of 0.5 M NaHCOs-extractable P in the top soil by exploiting its spatial covariance with a more densely sampled property. McBratney and Webster [1983] and Samra et al…”