Curvelet transform to study scale-dependent anisotropic soil spatial variation

Abstract

Information on soil spatial variability is important for optimal management of agricultural and natural resources. Systematic studies to characterize and quantify soil spatial variability have identified various issues including sale dependence and anisotropy. In this research, we have introduced curvelet transform to characterize scale-dependent anisotropic soil spatial variation. The new curvelet transform is a multi-scale transform with strong directional sensitivity. It separates overall variations in soil properties in to a number of spatial scales and directions. It combines multiple methods including wavelet and ridgelet transforms. The curvelet transform is ideally suited for the presentation of soil variability information containing abrupt values or displaying discontinuity in its spatial distribution. Spatial variability in soil potassium (K) measured using airborne radiometric survey was characterized using the curvelet transform and is presented as a case study. Soil K data from radiometric survey is often used to characterize soil and its properties. Overall variation in soil K was separated and quantified at different scales and directions, which were indicative of the scales of different landscape modification processes and their directions. Percent contribution towards the total variance at different scales and directions indicated the importance of those processes that modified the landscape. The curvelet transform provided explicit information at different scales and directions to understand the variability in landscape processes in the study area. The spatial variability information at a wide range of scales, locations, and directions can also be used in multi-scale directional soil mapping, scale specific prediction of soil properties, and filtering, smoothing and denoising of satellite derived data. © 2013 Elsevier B.V.