Title: Predicting Landscape Form and Dynamics and Various Modeling Approaches: A Case Study Implementing Landscape Connectivity with Topographic Filtering Model

Abstract: A wide range of modeling approaches exists for simulating landscape processes, each with a specific formulation of geomorphic transport laws and predictive capability combining human, physical, and social forces that drive landscape changes. How we build a landscape simulation model depends on a suite of factors, including what we model, the scale of prediction, why we model, and the current extent of knowledge. The widespread availability of high-fidelity topography combined with advances in geospatial analysis offers the opportunity to reimagine approaches to the difficult problem of predicting sediment delivery from watersheds. Here we present a model that uses high-resolution topography to filter sediment sources to quantify sediment delivery to the watershed outlet. It is a reduced-complexity, top-down model that defines transfer functions--topographic filters--between spatially distributed sediment sources and spatially integrated sediment delivery. The goal of the model is to forecast changes in watershed suspended sediment delivery in response to spatially distributed changes in sediment source magnitude or delivery, whether a result of watershed drivers or intentional management actions. Such an application requires the context of a watershed model that accounts for all sediment sources, enforces sediment mass balance throughout the spatial domain, and accommodates sediment storage and delivery over time. The model is developed for a HUC-8 watershed with a flat upland dominated by corn-soybean agriculture and deeply incised valleys near the watershed outlet with large sediment contributions from near-channel sources. Topofilter computes the delivery and storage of field-derived sediment according to its spatial and structural connectivity to the stream channel network; subsequently, delivery of both field- and near-channel-derived sediment along with floodplain storage are computed in the stream channel network to the watershed outlet. The model outputs provide a spatially rich representation of sediment delivery and storage on the field and along the stream that is consistent with available independent information on sediment accumulations and fluxes. Rather than a single best-calibrated solution, Topofilter uses the Generalized Likelihood Uncertainty Estimate (GLUE) approach to develop many possible solutions with sediment delivery rates expressed as probability distribution functions across the watershed. The ensemble of simulation outputs provides a useful basis for estimating uncertainty in sediment delivery and the effectiveness of different landscape management allocations across a watershed.