Patterns of groundwater and soil moisture variability: hard data, soft data and dominant controls

Abstract

Soil moisture and groundwater storage are important for understanding and predicting rainfall-runoff processes in watersheds. Research over the last 100 years has revealed detailed process understanding of infiltration and subsurface flow processes, but most studies have been restricted to small plots or hillslopes. However, catchment-scale hydrological functioning is not necessarily dominated by the same factors that control the response on the plot- and hillslope scale. Spatial patterns, such as groundwater and soil moisture patterns in the case of this thesis, reflect the spatial organization of natural hydrological systems. These patterns can be analyzed in terms of connectivity between runoff generation areas in different parts of the catchments and the stream network to investigate functional relations between pattern connectivity and runoff response. Deciphering the dominant factors and small scale processes that control these patterns, helps to better understand how surface and/or subsurface flowpaths are established that efficiently contribute to runoff and dominate the runoff response at the catchment-scale. As time, effort and expenses for obtaining direct measurements of catchment-scale spatial patterns is high, qualitative or so called “soft” data can be a useful complement to quantitative or “hard” data. In this thesis a new qualitative method called the “Boots & Trousers” method for mapping spatial patterns of soil moisture in humid environments and an adapted version for semi-arid conditions is proposed and systematically tested. Both methods are based on qualitative wetness indicators that one can see, feel or hear on the soil surface and are intuitive to local people from their every-day experience in outdoor activities (Switzerland) or crop growing and brick making (Tanzania). Both schemes were systematically tested to determine the correlation between qualitative wetness classes and quantitative differences in soil water content and for the agreement among classifications by different raters. It could be shown that the qualitative wetness classes reflect actual differences in volumetric water content. Neither experience, nor a certain level of education were a prerequisite for robust wetness classifications but a detailed introduction and training resulted in higher agreement among individual raters. The classifications for wet sampling points showed the highest agreement with the hard soil moisture data, while intermediate wetness classes seemed to be more difficult to assign. Some raters had a tendency to systematically rate specific wetness classes as too wet or too dry but when the raters were familiar with the application of the scheme, the mean offset was small and typically within the range of one wetness class. In addition, the dominant topographic controls of median groundwater levels and groundwater response timing were investigated in a 20 ha pre-alpine catchment in the Alptal, Switzerland, with low permeability soils in order to predict spatial patterns of groundwater response for non-monitored sites. From the analysis of 51 groundwater monitoring sites and 133 rainfall events between 2010 and 2012 in the study catchment, it was shown that median groundwater levels were correlated to topographic indices including slope, curvature, Topographic Wetness Index (TWI) and upslope contributing area. The strength of correlation between groundwater levels and TWI decreased at the beginning of rainfall events, indicating large spatial differences in groundwater responses, and increased after peak flow, when groundwater levels could be considered as being spatially close to a steady state. Median groundwater response times were also correlated to topographic indices and decreased with increasing TWI for sites with TWI < 6, while wetter sites responded almost immediately to rainfall. Rainfall intensity was more important than antecedent moisture conditions for the slope of this functional relation. The results of this thesis show that qualitative methods like the proposed “Boots & Trousers” method are reliable supplements of quantitative methods to capture the spatial variability in shallow soil moisture in different environments. They are fast to apply, require no experience, no measuring device and still provide robust and reliable results. They are therefore particularly suitable for mapping spatial soil moisture patterns in developing countries or remote areas. The findings on the variability in groundwater highlight the importance of topography on median groundwater levels and groundwater response timing in mountain catchments with a low permeability soil suggesting differences in the dominant controls and runoff processes, compared to flatter watersheds with transmissive soils. The findings of this thesis are expected to be transferable to other catchments with similar character and therefore further our understanding to make predictions of soil moisture and groundwater storage and the runoff response of ungauged catchments.