Subglacial landforms reveal basal ice flow patterns of the Last Glacial Maximum Rhine glacier

We present new insights into the ice flow dynamics of the Last Glacial Maximum (LGM) Rhine glacier based on a comprehensive inventory of glacially streamlined bedforms. High-resolution LiDAR data was used to map ice-marginal moraines and more than 2500 subglacial landforms located in the ~6000 km²-sized footprint of the former piedmont lobe. Orientation and morphometry of mapped bedforms were subsequently used to deduce paleo ice flow lines. Most of the subglacial landforms in the dataset are drumlins, but glacial lineations and subglacial ribs (Rogen/ribbed moraines) are also present in the study area. Streamlined bedforms predominantly occur in fields internal to the frontal moraine set of the inner (Stein am Rhein ice margin) of two LGM ice marginal complexes (Kamleitner et al., 2023). We interpret these landforms to have been shaped isochronously during the late LGM readvance (Kamleitner et al., 2023; Schreiner, 1992) to and the active stabilization at the Stein am Rhein ice marginal position. Deviating drumlin orientations (e.g. cross-cutting relationships) are rare within the Stein am Rhein flow set, supporting the hypothesis of contemporaneous formation. Bedform orientations of this flow set are the basis for inferring the ice flow patterns during the Stein am Rhein stadial. Continuous fields of flow are interpolated by applying the recently presented kriging approach of Ng and Hughes (2019). The reconstructed directions show radial ice flow emanating from the mouth of the confined Alpenrhein Valley that fans out towards the Stein am Rhein frontal moraines. Flow lines converge due to compression in narrow valley sections and diverge around topographic highs. Basal ice flow during the late LGM Stein am Rhein readvance was strongly controlled by topography. Derived paleo flow lines are combined with information from bedform elongation that allows to confine potential areas of relatively fast flowing ice. We find these to largely overlap with known overdeepenings, in line with predictions from numerical simulations (Cohen et al., 2018).