Scale-Specific Variability to Understand Complex Soil Processes.

Soil varies from location to location and the information on soil spatial variability is important for sustainable resource management. However, the variability as controlled by different factors and processes operating at different scales and in different intensities makes it challenging to understand the underlying soil processes. Quantification of these variability and their dominant controls at multiple scales in space and time can only lead to a better understanding on the underlying soil processes. The variability in soil water storage (SWS) measured down to 1.4 m (0.2 m depth interval) at 128 regularly spaced locations along a transect of 576 m over a five-year period from the Hummocky landscape of central Canada was quantified in this study to understand the underlying hydrological dynamics. The similarity between spatial patterns of SWS over the entire study period was stronger within a season (intra-season) than the same season from different years (inter-annual) and between seasons (inter-season). The variability at multiple scales and locations was quantified using the wavelet transform. The strongest large scale (>72 m) variability contributed from the macro-topography and a moderate medium scale (18-72 m) variability contributed from the landform elements were persistent over the entire measurement period (time stability). The locations and the scales of the most persistent spatial patterns over time and depth were quantified using the wavelet coherency. The changes in the persistent patterns indicated the changes in the scales and locations of underlying hydrological processes, which can be used to identify change in sampling domain. The similarities/dissimilarities in the spatial pattern between the surface and sub-surface measurements at different scales and locations were used to infer the whole profile hydrological dynamics (depth persistence). Scale-specific dominant controls were identified after separating the variance contribution of each scale towards the overall variance using the Hilbert-Huang transform. The large scale macro-topographical control and medium scale landform control were much stronger than very large scale soil textural control on SWS. The scale-specific relationship with controlling factors improved the prediction of SWS.