Sediment transport models, utilized to guide land management in the Great Barrier Reef (GBR), assume settling velocities for individual silt and clay particles on the order of 0.01 mm/s; however, silts and clays once flocculated exhibit settling velocities on the order of 1 mm/s. In this study, in‐situ (n = 144,912) and laboratory‐dispersed (n = 64) particle size measurements collected using laser diffractometry were compared from nine rivers discharging along 800 km of GBR coastline to determine the extent of in‐situ flocculation. Environmental controls on in‐situ particle size were investigated using decision tree algorithms trained on coeval measurements of salinity, shear rate, and turbidity. Comparison of in‐situ and dispersed particle size measurements demonstrate that suspended‐sediment across all catchments flocculated into larger aggregates with an order‐of‐magnitude difference in median particle size between in‐situ (D_50v = 132 μm, σ = 60 μm for all data) and dispersed (D_50v = 15 μm, σ = 11 μm for all data) particles. Machine learning algorithms showed excellent promise predicting various measures of in‐situ particle size. Model validation R² ranged from 0.72 to 0.99, inclusion of catchment as a categorical variable only marginally (<1%) increased R². These results demonstrate that flocculation is prevalent across all rivers surveyed and that hydrodynamics are more important than inter‐catchment differences (e.g., differences in climate, geology, or land‐use). Implications of widespread flocculation on the determination of end‐of‐catchment sediments loads and subsequent dispersal patterns across continental shelves are discussed to inform the refinement and monitoring of GBR sediment targets.

The settling velocity (w_s) in estuarine environments can impact whether a region is eroding or accreting sediment on the bed, yet determining this rate can be an indirect process requiring a number of assumptions. Accurate determination of w_s is especially needed for numerical models to reproduce observed sediment concentrations at the appropriate timescale. We collected information on suspended sediment flocculation at a channel site (13 m deep) and a shallows site (4 m deep) within South San Francisco Estuary, alongside timeseries of flow, wave statistics, turbulent shear, and bottle samples analyzed for both w_s and particle size. Using the measurements of floc size and settling velocity, we performed a sensitivity analysis on the unknown parameters in the general explicit formula for settling velocity. The collected particle size distribution data show that multiple classes of flocs are present; these are characterized as flocculi, microflocs, and macroflocs. We show that w_s of flocculi is closest to w_s for the full distribution. The determined parameter values lead to near‐bed mass‐weighted settling velocities (standard deviation) of 1.18 (0.55) and 0.22 (0.15) mm/s at the channel and shallows sites, respectively. Modeling efforts can use this work to help select an appropriate sediment model and parameter values.

Optical and acoustic backscatter measurements are routinely utilized to monitor suspended‐sediment concentration (M); however, both measurements are affected by changes in particle size and density. In this study, optical and acoustic backscatter measurements are combined to a single parameter, the sediment composition index (SCI), to quantify M, mean particle radius by number (a_o), the fraction of sediment <20 μm by diameter (), and particle bulk apparent density (ρ_bulk). Data are analyzed from Chesapeake Bay and five rivers of Queensland, Australia. SCI is utilized to predict the ratio of M to acoustic backscatter under changes in a_o and ρ_bulk (R² ranged from 0.6 to 0.98 across all data sets) and combined with acoustic backscatter to predict estimates of M that are independent of changes in a_o and ρ_bulk. SCI is proportional to log₁₀(a_o) and for SCI from acoustic backscatter measured at 6 MHz (R² = 0.8 and 0.74, respectively, p‐value < 0.001, n = 133), while SCI(log₁₀(a_o)) and SCI() from acoustic backscatter measured at 2 MHz or lower are sensitive to changes in floc fractal dimension. Estimates of ρ_bulk from SCI are biased by changes in particle size (R² is 0.1–0.5 across all datasets). This study builds upon recent work that derived SCI to quantify composition of sand and mud in suspension and demonstrates the utility of the approach in systems transporting flocculated silt and clay. Future research directions are discussed.