BOTTOM UP OR TOP DOWN? USING TIME-SERIES ANALYSIS TO TRACK EROSIONAL SIGNALS IN AN ANALOG LANDSCAPE (Invited Presentation)

Inferring geologic history and predicting future conditions depends partly on our understanding of how landscapes respond to external forcing. Changes in base level are thought to propagate through landscapes from the bottom up, with rapidly-responding channel networks sending erosional signals to hillslopes. However, paleo-erosion records in fluvial terraces and lake cores suggest that changes in hillslope sediment transport (i.e., diffusive disturbances) could drive erosion from the top down. Identifying these mechanisms in landscapes requires (1) known external forcing and (2) erosion rates with high spatiotemporal resolution and extent. In natural settings, reconstructing climate and tectonics is complicated by several factors, including an incomplete sedimentary record and unknown initial conditions. Similarly, erosion rate measurements are generally limited to single points on hillslopes (i.e., soil pits) or basin-averaged rates derived from stream sediments. Here, we overcome these difficulties by using an analog eroding landscape to examine the transmission of base level. We ran five experiments at the St. Anthony Falls eXperimental Landscape Model under steady uplift and rainfall, systematically varying the relative strength of hillslope and fluvial erosion and running each experiment to flux steady state. To quantify erosional signal propagation, we calculated autocorrelation and covariance in space and time series of elevation. If base level is communicated from the bottom up through the channel network, we expect the elevation on a given hillslope to be correlated with the elevation in the channel to which it drains, with a lag that reflects the distance between the two points. Alternatively, if most signals come from the top down, we expect the elevation in a given channel to be correlated with the hillslopes in its watershed, with a lag that reflects the transport time of hillslope material to the channels. While we observe both types of signal transmission in our analog landscapes, our preliminary results suggest that top-down erosional signals dominate in landscapes with more efficient hillslope processes. We also find a clear signature of drainage basin reorganization in individual elevation time series, emphasizing the need to incorporate topologic change in our models of landscape evolution.