Comparison of Ordinary and Lognormal Kriging on Skewed Data of Total Cadmium in Forest Soils of Sweden

Spatial statistical analysis of georeferenced data of total cadmium (TCd) in forest soils of Sweden was assumed to providemore advantageous maps than traditional interpolated maps. However, 264 measurements of TCd in O-horizon of forest soils displayed skewed frequency distribution. Since atypicalobservations affect badly the variogram, outliers wereidentified, different data transformations were tested andordinary (OK) and lognormal kriging (LK) scenarios werecompared based on cross-validation. Results were comparedusing overall measures of predictors, e.g. traditionalmean squared prediction error (MSPE), mean of krigingvariances, variance ratio, median of internallystandardised residuals, and assessments of classificationaccuracy, such as percentage of correctly predictedsamples and within-class MSPE.One outlier was identified based on the absolute value of skewness of value differences less or equal to one in data pairs separated at certain lag classes. Mapping categories characterised by percentage of correct classification and within-class MSPE were found to be essential in comparison of kriging results additionally to the overall measures. In comparison of kriging methods, OK predicted high values moreaccurately and LK was more effective to predict low and mediumvalues. Thus, OK was suggested for mapping high concentration of TCd and other pollutants. Percentage of correctly predictedsamples and within-class MSPE were found to be dependent on kriging method, as well as on the number and limits of categories.

[1]  Noel A Cressie,et al.  Statistics for Spatial Data. , 1992 .

[2]  R. Webster,et al.  Sampling To Monitor Soil In England And Wales , 1999 .

[3]  G. W. Snedecor.  Statistical Methods , 1964 .

[4]  J. N. R. Jeffers.  Analysis of spatial data , 1983 .

[5]  W. W. Muir,et al.  Regression Diagnostics: Identifying Influential Data and Sources of Collinearity , 1980 .

[6]  R. Webster,et al.  Statistical Methods in Soil and Land Resource Survey. , 1990 .

[7]  S. Verma,et al.  Importance of Skewness and Kurtosis Statistical Tests for Outlier Detection and Elimination in Evaluation of Geochemical Reference Materials , 1998 .

[8]  K. Juang,et al.  A Comparison of Three Kriging Methods Using Auxiliary Variables in Heavy-Metal Contaminated Soils , 1998 .

[9]  B. M. Davis.  Uses and abuses of cross-validation in geostatistics , 1987 .

[10]  R. Webster,et al.  Sample adequately to estimate variograms of soil properties , 1992 .

[11]  B. Kedem,et al.  Bayesian Prediction of Transformed Gaussian Random Fields , 1997 .

[12]  D. Helsel,et al.  Statistical methods in water resources , 2020, Techniques and Methods.

[13]  Mike Rees,et al.  5. Statistics for Spatial Data , 1993 .

[14]  C. Roth.  Is Lognormal Kriging Suitable for Local Estimation? , 1998 .

[15]  C. T. Haan,et al.  Statistical Methods In Hydrology , 1977 .

[16]  R. M. Lark,et al.  A comparison of some robust estimators of the variogram for use in soil survey , 2000 .

[17]  R. Reese.  Geostatistics for Environmental Scientists , 2001 .

[18]  R. Webster,et al.  Coregionalization of trace metals in the soil in the Swiss Jura , 1994 .

[19]  A. Alriksson.  Regional Variability of Cd, Hg, Pb and C Concentrations in Different Horizons of Swedish Forest Soils , 2001 .