Separating scale-specific soil spatial variability: A comparison of multi-resolution analysis and empirical mode decomposition

Abstract Soil spatial variability is scale dependent. In separating soil spatial variability at multiple scales, wavelet based multi-resolution analysis (MRA) is an established method, whereas empirical mode decomposition (EMD) has just been introduced in soil science. A careful comparison between these methods is necessary and is the goal of this research. Here a brief description of the methods is provided and they are compared using soil water storage (SWS) data observed along a 576 m transect. The MRA separated the variations of a spatial series into predefined scale intervals, each of which contributed differently to the overall variance of the series. EMD separated the overall variation into different mode functions (known as Intrinsic Mode Functions; IMFs) representing different scales as they are present in the series. The EMD did not use any predefined basis (such as mother wavelet in wavelet transform) for scale separation. The proportion of overall variance contributed at each scale was used to identify the most dominant scale. Correlation between the scale components (MRA products and IMFs) and different factors controlling SWS along the transect enabled identification of the dominant controls of SWS and the scales at which they occur.