Spatial patterns of saturated hydraulic conductivity and its controlling factors for forested soilscapes

Accurate estimation of saturated hydraulic conductivity (Ks) in soils is essential for various hydrological applications. Ks is difficult to characterize because of its high variability even over short distances, and measurement methods typically require considerable time and resources. Consequently, researchers often use a limited number of measurements for characterizing Ks or use various soil properties for indirect estimation via pedotransfer functions. This dissertation investigates spatial patterns of Ks at different sampling scales along a complex forested soilscape ranging from relatively clay-rich to sand-rich soils. Although there is a tight link in the spatial patterns of soil texture, color and mineralogy as a function of topography along this tropical rainforest catena (i.e. toposequence), there is no similar dependency with Ks, and univariate statistics show no significant (alpha = 0.05) difference in Ks between soil types. Spatial patterns of Ks are linked to topography and texture only where soils are comprised of ≥80% sand. Based on surface and subsurface analyses of physical and biological processes, it appears that Ks is largely controlled by macropores from roots and animals (i.e. biopores), but textural porosity largely controls Ks for coarse textures with ≥80% sand. At all sampling scales (lags of 25, 10, 1 and 0.25 m), there is little to no autocorrelation in Ks and no apparent link with topography or soil properties, and 'structure' does not emerge from 'noise,' except for transects extending over soil boundaries separating coarse (≥80% sand) and less coarse textures. Because pedotransfer functions using readily available soil information do not properly account for macropores generated by bioturbation, these functions do not provide reasonable point estimates of Ks or accurate descriptions of its spatial patterns. This data-intensive study demonstrates a general lack of spatial structure and predictability in Ks for forest soils at all scales that modelers must consider in hydrologic investigations. This lack of spatial structure in Ks can be expected in any forest setting with considerable small-scale biological activity. An important implication of this study is that a single Ks value can be assigned for multiple soil types in a forest setting.