We quantified spatial and temporal patterns of suspended sediment transport over two growing seasons (May-November, 2016; June-September, 2017) using an array of optical-backscatter turbidity sensors) equipped with automatic anti-fouling wipers. We were not able to collect measurements or draw conclusions during the winter months when sediment transport is influenced by ice rafting, rather than simple advection of SSC. We deployed the sensors across the marsh platform and adjacent tidal channel in a shore-normal transect. Based on direct observations of flooding, we concluded the sensor transect was parallel to the direction of initial marsh flooding. The sensor in the channel (RBR Duo equipped with Seapoint turbidity probe) was located 35cm above the bed and 3m from the edge of the marsh. Sensors were fastened to rigid grates for deployment on the marsh, with the grid flush with the marsh soil surface and the actual sensor window located 7cm above the surface. The sampling point is 5cm in front of the sensor window. To minimize interference with plants, we removed vegetation from within this distance and placed a ceramic tile on the ground to prevent regrowth directly in front of the sensor window. In 2016, sensors were placed at 2.7m and 17m from the marsh edge (RBR Solo and RBR Duo, respectively). In 2017, five sensors were installed on the marsh surface at 1.25 m (RBR Solo), 2.7 m (RBR Duo), 9.3 m (RBR Duo), 17 m (YSI 6600), and 24 m (YSI Exo) from the marsh edge .The sensor located 17 m into the marsh in 2017 malfunctioned and did not record data. All other sensors measured turbidity every 15 minutes for the entire deployment, totaling nearly 100,000 measurements, with select sensors also measuring pressure.The sensor error in turbidity measurements is <2% and significant sensor drift is uncommon (http://seapoint.com/pdf/stm_ds.pdf).

Suspect data points were removed from the time series record following Ganju et al., 2005. We used a recursive filter to remove points which were greater than 10 NTU higher than adjacent time steps. The record was visually analyzed to ensure points being removed represented values anomalously higher than surrounding values. These removed data points represent times when a sensor was obstructed, fouled, or not submerged.

Turbidity data was converted to SSC via laboratory calibrations using sediment collected from the site and in situ field sampling. In the lab, we created sediment-water slurries with a range of SSC and measured them with an additional turbidity sensor. In the field, we measured turbidity with an additional sensor at various locations around the site and at different tidal stages. We collected a water sample in conjunction with each turbidity measurement. This ensures that the calibration is accurate to the average sediment characteristics from across the site. We compared sensor turbidity measurements to total suspended solid measurements obtained via vacuum filtration of water samples and sediment-water slurries. The y-intercept value was set to zero, resulting in the equation SSC (mg/L) =2.26*Sensor Turbidity (NTU) (R2=0.98, n=32, p<<0.001). The average error is 19% over the entire data range and 21% for SSC less than 100 mg/L . There were no outliers which suggests the various locations sampled had sediment of similar optic properties. All field sensors were in turn individually calibrated to the sensor used in the lab via turbidity standards. In other words, all turbidity sensors were calibrated to one sensor which was then used to calibrate to SSC.

Pressure was measured by the channel sensor and converted to depth by removing atmospheric pressure . Values equal to or below the sensor elevation (0.35m) are listed as 0.35.

Ganju, N.K., Schoellhamer, D.H. & Bergamaschi, B.A., (2005). Suspended sediment fluxes in a tidal wetland: Measurement, controlling factors, and error analysis. Estuaries 28(6), 812-822.