Sediment Fingerprinting in Intensively Managed Landscapes: Application of an Enhanced Bayesian Un-mixing Framework that accounts for Spatiotemporal Heterogeneity to Study Intra-Seasonal Trends in Source Contributions

Abstract:

We apply an enhanced revision of a Bayesian un-mixing framework for estimating sediment source contributions to an intensively managed watershed in the US Midwest that is characterized by spatiotemporal heterogeneity. The framework has two key parameters, namely a and b, that account for spatial origin attributes and the time history of sediments delivered to the watershed outlet, respectively. The probabilistic treatment of these parameters incorporates variability in source erosion processes, as well as the delivery times and storage of eroded material within the watershed. Uncertainty in source contribution estimates in our approach is quantified naturally through the use of Markov Chain Monte Carlo simulations for estimating the posterior probability density functions. The studied watershed is the 26 km² South Amana Sub-Watershed located within the Clear Creek Watershed (CCW), IA, which is part of the Critical Zone Observatory for Intensively Managed Landscapes (IML-CZO). Utilizing stable isotopes of C and N, the Bayesian framework predicted trends in mean source contributions and uncertainty that were in agreement with observations from other studies. Terrestrial sources were found to dominate sediment contributions in the earlier stages of the growing season when land cover was small and the hydrologic forcing was large. Instream sources became more dominate during the latter stages when land cover was well-established and more extensive. Also, the effects of spatial heterogeneity and sediment travel time and delivery on uncertainty in sources contribution estimates were adequately captured. The results clearly showed a reduction in uncertainty when watershed connectivity was greatest and considerable amounts of material were delivered to the collection point at the outlet. On-going application of the framework to the Upper Sangamon River Basin (USRB), which also part of the IML-CZO and has distinct features from CCW, is expected to shed more light on the effects of connectivity and other landscape attributes on sediment source dynamics over the growing season.