3.2 Controls of primary soil moisture patterns
The spatial patterns of soil moisture show the higher Pearson correlation coefficients with soil-topography properties at each measurement depth (Table 1). The results generally indicated that terrain features were larger contributors to the variance in soil moisture than the soil properties. While most of topographical attributes (e.g., topography wetness index and slope) had strong correlations with the derived EOFs, only soil texture among the soil parameter showed significant correlations. Soil organic matter displayed lower correlations at the surface soil (0-20 cm). Depth to bedrock, which is related to both soil thickness and topography, seems to have had a large influence on soil moisture variability at all depths. This result may be confirmed by examination of the soil moisture values within the wet locations which are characterized by the soils with >1 m depth to bedrock. Elevation, slope, and curvature were negatively related to soil moisture contents, while upslope contributing area, depth to bedrock, topographic wetness index, and percent silt and clay were positively correlated to the soil moisture contents. This result is likely due to the fact that most soils with the deep soil profiles are generally limited to lower elevations and concave slope areas (i.e., valley floor and swales) where soil moisture is normally the highest. Henninger et al. (1976) reported that soil moisture increased toward the near-stream zone within a predominantly agricultural watershed, which was a result of both topographic convergence and moderately to poorly drained soils within the near-stream zone. Our regression analysis indicated that soil texture did not exert a strong influence on the soil moisture spatial distributions at the catchment scale, particularly at soil depth intervals between 0.4 and 0.6 m. This finding is due to the relatively small variations in soil textural properties throughout the measured locations for the different soil-landform units (Takagi and Lin, 2011). Famiglietti et al. (1998) reported that under wet conditions, the best correlations existed with porosity and hydraulic conductivity along a profile of a 200-m length; whereas under dry conditions, the relative elevation, aspect and clay content provided the best correlations.
The EOF analysis was repeated for the spatial anomaly data in two categories (depth/wetness), and the degree to which these factors affected the soil moisture distribution was calculated (Table 1). At the Shale Hills, the correlation coefficient values generally increased with soil depth for elevation, slope, and percent silt and clay content, whereas the highest values were observed at intermediate depths (0.4 m) for curvature, depth to bedrock, and topographic wetness index. These results indicate that soil moisture becomes strongly aligned with convergent topography and suggests that lateral flow processes may be important driver of soil moisture redistribution at these depths. The increased influence of the parameters with depth may relate to the seasonal changes of soil moisture which undergoes more dramatic changes near the soil surface. As indicated in Takagi and Lin (2012), the subsurface soil moisture exhibited weak temporal variability in the correlation coefficient values that suggested the dampened effects of climate and hydrological fluxes. Thus, the subsurface soil moisture distribution in this catchment is a function of both topographic parameters and soil depth. An observation that was reinforced by the transient hydrological fluxes such as the presence of the ephemeral shallow water table that seasonally exist within the valley.