Characteristics and controls of variability in soil moisture and groundwater in a headwater catchment

Abstract

Abstract. Hydrological processes, including runoff generation, depend on the distribution of water in a catchment, which varies in space and time. This paper presents experimental results from a headwater research catchment in New Zealand, where we made distributed measurements of streamflow, soil moisture and groundwater levels, sampling across a range of aspects, hillslope positions, distances from stream and depths. Our aim was to assess the controls, types and implications of spatial and temporal variability in soil moisture and groundwater tables.We found that temporal variability in soil moisture and water table is strongly controlled by the seasonal cycle in potential evapotranspiration, for both the mean and extremes of their distributions. Groundwater is a larger water storage component than soil moisture, and this general difference increases even more with increasing catchment wetness. The spatial standard deviation of both soil moisture and groundwater is larger in winter than in summer. It peaks during rainfall events due to partial saturation of the catchment, and also rises in spring as different locations dry out at different rates. The most important controls on spatial variability in storage are aspect and distance from the stream. South-facing and near-stream locations have higher water tables and showed soil moisture responses for more events. Typical hydrological models do not explicitly account for aspect, but our results suggest that it is an important factor in hillslope runoff generation.Co-measurement of soil moisture and water table level allowed us to identify relationships between the two. Locations where water tables peaked closer to the surface had consistently wetter soils and higher water tables. These wetter sites were the same across seasons. However, patterns of strong soil moisture responses to summer storms did not correspond to the wetter sites.Total catchment spatial variability is composed of multiple variability sources, and the dominant type is sensitive to those stores that are close to a threshold such as field capacity or saturation. Therefore, we classified spatial variability as "summer mode" or "winter mode". In "summer mode", variability is controlled by shallow processes, e.g. interaction of water with soils and vegetation. In "winter mode", variability is controlled by deeper processes, e.g. groundwater movement and bypass flow. Double streamflow peaks observed during some events show the direct impact of groundwater variability on runoff generation. Our results suggest that emergent catchment behaviour depends on the combination of these multiple, time varying components of storage variability.

Linking hydrologic signatures to hydrologic processes: A review

“…Experimentation with flexible model structures in addition to experimental knowledge attributes a short time‐to‐peak in runoff and a strong threshold in the storage‐discharge relationship to saturation excess overland flow (Fenicia et al, ). The watershed storage can be approximated using several methods, some using only rainfall and streamflow (such as an antecedent precipitation index or storm rainfall total, optionally adjusted by streamflow as a proxy for storage) but others requiring additional data, such as antecedent soil moisture (Detty & McGuire, ; McMillan & Srinivasan, ).…”

“…This signature indicates riparian runoff generation (Peak 1) followed by subsurface runoff generation (Peak 2). Discussed further in McMillan and Srinivasan ()…”

Deriving hydrological signatures from soil moisture data

“…As described in the previous paragraph, soil moisture distributions commonly display bimodal behaviour with distinct, stable states for summer and winter. This behaviour has been shown to be indicative of a whole‐catchment transition between a summer mode and winter mode, each displaying characteristic sources of hydrologic variability and dominant flow paths (McMillan and Srinivasan, ). It is essential for hydrologic models to correctly simulate these seasonal transitions in order to predict catchment wetness states and responses to precipitation, and errors in transition predictions are often cited as a cause for poor model performance during spring (drying) and autumn (wetting) seasons (Pinol et al, ).…”

“…Others have used remotely sensed soil moisture data for calibration and data assimilation (Wanders et al, ; Brocca et al, ). Some previous work has investigated soil moisture signatures for process understanding, as a complement to flow data to identify response thresholds (McMillan et al, ; Scaife and Band, ), and many others have used soil moisture data as a qualitative aid to interpreting processes in experimental catchments (Western et al, ; Penna et al, ; McMillan and Srinivasan, ; Geris et al, ). There remains significant potential to use soil moisture data to design quantitative signatures that characterise soil water dynamics, a core component of process‐oriented hydrological models.…”

Factors affecting soil moisture spatial variability for a humid forest hillslope

“…Pan and Peters‐Lidard () explained that the relationship can be changed if the mean soil moisture was greater or less than a threshold level for the soil moisture. Several studies have evaluated the impact of soil properties, topography, meteorological forcing, and groundwater on the variability of soil moisture (Famiglietti et al, ; McMillan & Srinivasan, ; Rosenbaum et al, ; Vereecken et al, ). Particularly, soil properties (soil structure and soil texture) were frequently linked to soil moisture variability (Famiglietti et al, ; Martinez et al, ; Vereecken et al, ).…”