LAUSR

Abstract Variations in soil properties are both scale- and location- specific. Empirical mode decomposition (EMD) has been used to separate variations in soil properties at different scales. However, the technique has not been fully employed to characterize the 2-dimensional scale-specific variations in soil properties. Across a 21 million ha region in Eastern China, 2863 topsoil samples were collected on a semi-regular grid. Two-dimensional EMD algorithm applied to soil observations was used to extract scale- and location- specific variations in elevation data, soil clay content, soil organic matter (SOM) and pH. It was found that the variations in elevation and pH were mostly within large scales while the variations in clay content and SOM were mostly within intermediate to large scales. The variations were also location-specific due to differences in soil type as a function of elevation, parent materials, and land use. Correlations between the controlling factors (elevation and clay) and soil properties (SOM and pH) were also found to be scale- and location- specific, indicating a complex interrelationship between elevation, parent material, vegetation zonation, microbiological communities, leaching and transport of CaCO 3 , runoff-erosion and fertilization. The results provided guidance for scale- and location- specific soil and land management practices. It was concluded that this method can be applied elsewhere to improve digital soil mapping of soil properties and better understand the variations and interrelationships in soil properties as a function of changing environmental factors and anthropogenic activities. To fully evaluate the performance of the 2-d EMD algorithm, soil samples that were collected on an irregular grid across large spatial extents should be used in the future.