Effects of sampling scale on statistical parameters estimation and spatial variability analysis of soil nutrition

Abstract

Based on nine soil nutrient datasets from different sampling scale, the effects of sampling scale on spatial variability analysis were analyzed. Results indicated that (i) with the sampling scale decreasing, the estimated mean of regional soil nutrient decreased, the coefficient variation increased, and the global trend of soil nutrient spatial distribution weakened. However, sampling scale did not change the fitting model of spatial variability; (ii) when sampling in a large scale and the weak self-correlation among soil samples, the relatively small number of samples will be enough for the regional statistical analysis to estimate parameters, but insufficient for spatial variability analysis and interpolation; (iii) when the sampling number is larger than the optimum (In the paper, the optimum sampling number is 112), the accuracy of regional statistical parameters, spatial variability characters and interpolation will increase with soil sampling scale decreasing. When sampling scale is lager than 0.2, it will fulfill the demand for the interpolation of soil nutrition; (iv) Spatial pattern of samples have a more significant effect on spatial range and accuracy of spatial interpolation than sampling scale. Copyright © 2010, TSI ® Press Printed in the USA. All rights reserved.