文摘
River management and restoration measures are of increasing importance for integrated water resources management (IWRM) as well as for ecosystem services. However, often river management mainly considers engineering and construction aspects only and the hydrogeological settings as the properties and functions of ancient fluvial systems are neglected which often do not lead to the desired outcome. Knowledge of the distribution of sediment units could contribute to a more efficient restoration. In this study, we present two noninvasive approaches for delineation of fluvial sediment architecture that can form a basis for the restoration, particularly in areas where site disturbance is not permitted. We investigate the floodplain of a heavily modified low-mountain river in Switzerland using different hydrogeophysical methods. In the first approach, we use data from electromagnetic induction (EMI) with four different integral depths (0.75-?m) and gamma-spectrometry as well as the elevation data as input for a K-means cluster algorithm. The generated cluster map of the surface combines the main characteristics from multilayered input data and delineates areas of varying soil properties. The resulting map provides an indication of areas with different sedimentary units. In the second approach, we develop a new iterative method for the generation of a geological structure model (GSM) by means of various EMI forward models. We vary the geological input parameters based on the measured data until the predicted EMI maps match the measured EMI values. Subsequently, we use the best matched input data for the GSM generation. The derived GSM provides a 3D delineation of possible ancient stream courses. A comparison with an independent ground penetrating radar (GPR) profile confirmed the delineations on the cluster map as well as the vertical changes of the GSM qualitatively. Thus, each of the approaches had the capacity for detecting sedimentary units with distinct hydraulic properties as an indication of former stream courses. The developed methodology presents a promising tool for the characterization of test sites with no additional subsurface information.