Direct and Indirect Controls of Soil Water Storage in a Hummocky Landscape. II.

Soil water is the principle limiting factor in semi arid agricultural production and a key element in environmental health. The storage of soil water is not controlled by a single factor, rather a complex suite of environmental factors and processes work independently or in combination to determine soil water storage. The effect of some factors can be measured directly, whereas, others that are supposed to exist, cannot be measured directly. This study examines the complex relationship between soil water storage and its direct and indirect controls in a Hummocky landscape using Structural Equation (SE) Modeling. Soil water storage was measured up to 1.4 m depth along a 128 point transect established at St. Denis National Wildlife Area, Saskatchewan, Canada. Time Domian Reflectometry and Neutron Moisture Meter were used to measure the soil water over three years period.  Controlling factors including elevation, sand, silt, clay and organic carbon were measured. Different terrain indices were calculated from the Digital Elevation Model of the study area. The initial Structural Equation Model (SEM) was developed including two indirect or latent controlling factors (topography and texture) and 15 observed factors as the indicators of latent factors. Almost all variations in soil water storage were explained by topography as indicated by large standardized path coefficient in the SEM. Among the different terrain indices, relative elevation, wetness index, aspect and solar radiation effect represented the topography best. The relationship between soil water storage and texture was insignificant. This relationship clearly indicated topography to be major control of soil water storage. The consistency of the relationship over different seasons indicated season independent control of soil water storage. The total effect of each factor on soil water storage was calculated from the direct and indirect effects. SE Modeling is a better tool in explaining multivariate relationship between controlling factors.