An enhanced Bayesian fingerprinting framework for studying sediment source dynamics in intensively managed landscapes

Abstract

An enhanced revision of the Fox and Papanicolaou (hereafter referred to as “F‐P”) (2008a) Bayesian, Markov Chain Monte Carlo fingerprinting framework for estimating sediment source contributions and their associated uncertainties is presented. The F‐P framework included two key deterministic parameters, α and β, that, respectively, reflected the spatial origin attributes of sources and the time history of eroded material delivered to and collected at the watershed outlet. However, the deterministic treatment of α and β is limited to cases with well‐defined spatial partitioning of sources, high sediment delivery, and relatively short travel times with little variability in transport within the watershed. For event‐based studies in intensively managed landscapes, this may be inadequate since landscape heterogeneity results in variabilities in source contributions, their pathways, delivery times, and storage within the watershed. Thus, probabilistic treatments of α and β are implemented in the enhanced framework to account for these variabilities. To evaluate the effects of the treatments of α and β on source partitioning, both frameworks are applied to the South Amana Subwatershed (SASW) in the U.S. midwest. The enhanced framework is found to estimate mean source contributions that are in good agreement with estimates from other studies in SASW. The enhanced framework is also able to produce expected trends in uncertainty during the study period, unlike the F‐P framework, which does not perform as expected. Overall, the enhanced framework is found to be less sensitive to changes in α and β than the F‐P framework, and, therefore, is more robust and desirable from a management standpoint.

The Intensively Managed Landscape Critical Zone Observatory: A Scientific Testbed for Understanding Critical Zone Processes in Agroecosystems

“…A unique, event‐based sampling framework has been developed in IML‐CZO to capture the high spatial and temporal variability of water–sediment–nutrient transport processes that have been accelerated by intensive land management. The event‐based monitoring includes measurements on the landscape of water table fluctuations, enrichment ratios, and roughness as well as instream measurements of bank erosion, hysteresis, sediment sources, and sedimentation (e.g., Abban et al, 2016; Blair et al, 2018; Lee et al, 2017; Neal and Anders, 2015; Schilling et al, 2018; Wilson et al, 2012; Yu and Rhoads, 2018).…”

An extended Bayesian sediment fingerprinting mixing model for the full Bayes treatment of geochemical uncertainties

“…A key advancement in quantifying the uncertainties associated with the sediment fingerprinting procedure has been the emergence of Bayesian mixing models (Abban et al, ; D'Haen et al, ; Massoudieh, Gellis, Banks, & Wieczorek, ; Nosrati, Govers, Semmens, & Ward, ; Stewart, Massoudieh, & Gellis, ) as an alternative to the more commonly applied least‐squares “frequentist” mixing model approaches (Collins et al, ; Martínez‐Carreras et al, ; Walling, Collins, & McMellin, ). Uncertainties include spatial and temporal variability in source and riverine sediment geochemistry, analytical instrument error, mixing model error, and unknown residual error such as nonconservative sediment transport (Sherriff, Franks, Rowan, Fenton, & Ó'hUallacháin, ; Small, Rowan, & Franks, ).…”

Monte Carlo fingerprinting of the terrestrial sources of different particle size fractions of coastal sediment deposits using geochemical tracers: some lessons for the user community

“…Many studies (e.g., Franks and Rowan 2000;Collins et al 2013aCollins et al , b, 2014Stone et al 2014;Liu et al 2016b;Pulley and Collins 2018;Habibi et al 2019) have used a frequentist approach incorporating a Monte Carlo framework for uncertainty analysis. Alternatively, Bayesian approaches have also been used to evaluate uncertainties associated with the results generated by sediment fingerprinting (Nosrati et al 2018;Gholami et al 2017b;Cooper et al 2014Cooper et al , 2015Abban et al 2016;Cooper and Krueger 2017;Habibi et al 2019). Here, a critical factor influencing the choice of data processing framework concerns whether the source and sediment tracer data satisfy the requirements of Bayesian methods, including exhibiting normal distributions.…”