Optically based quantification of fluxes of mercury, methyl mercury, and polychlorinated biphenyls (PCBs) at Berry's Creek tidal estuary, New Jersey

Abstract

Berry's Creek is an urban tidal estuary of the Hackensack River (New Jersey, U.S.A.), with over 10 km of waterways and over 3 km2 of tidal marsh. More than a century of industrial activity has resulted in elevated contaminant concentrations in sediment, water, and biota. Mercury, methyl mercury, and polychlorinated biphenyls (PCBs) have been identified as the primary chemicals of concern (COCs) due to their persistence and bioaccumulation potential in aquatic and nearshore wildlife. As part of a multi‐phased remedial investigation program, OPTically‐based In‐situ Characterization System (OPTICS) field studies were conducted in Berry's Creek to characterize water column contaminant dynamics, specifically to (1) improve understanding of the interrelationship(s) between surface sediment and water column contaminants and (2) quantify the transport of water column COCs between waterways and marshes. Results from OPTICS monitoring indicated that particulate resuspension from the upper 5 mm of the sediment bed during flood and ebb tides and storm flows is the primary process by which COCs are exchanged from bedded sediment to the water column. Analysis of mass exchange of COCs indicated that COCs associated with resuspended particulates are transported from the waterways to the marshes, where they are retained. Dissolved COC exchange from marshes to the waterways was minor (less than 8% of total COC mass in water). Overall, marshes were observed to be net sinks for COCs.

OPTically‐based In situ Characterization System (OPTICS) to quantify concentrations and mass fluxes of mercury and methylmercury in South River, Virginia, USA

“…The median ratio was near 1.0, and mean absolute model bias was within 15% of mean concentrations. Similar to the findings by Bergamaschi et al () and Chang et al (), modelled versus measured particulate mercury and methylmercury concentrations exhibited a similar or lower degree of variance compared with uncertainties determined for discrete water sample field duplicates (Table ). This lends confidence to the optically derived time series of chemical concentrations and TSS.…”

“…Similar to the results presented by Bergamaschi et al () and Chang et al (), comparisons between laboratory‐measured and OPTICS‐derived (“measured” and “modelled,” respectively) concentrations of particulate mercury and methylmercury in the South River indicated excellent agreement (Figure ; Table ). TSS, which was determined through linear regression between laboratory‐based TSS and optical backscattering measurements, also exhibited strong statistical similarity between modelled and measured.…”

“…Note that for previous OPTICS‐related studies (Bergamaschi et al, ; Bergamaschi, Fleck, et al, ; Bergamaschi, Krabbenhoft, et al, ; Chang et al, ), statistical evaluation of the accuracy and precision of the OPTICS‐derived chemical concentrations were performed using model cross‐validation or validation with fully independent datasets. Here, discrete (measured) data were relatively unvarying over the majority of the study period (baseflow), leading to biases when performing cross‐validation or bootstrapping methods; hence, all discrete data were used to calibrate the model.…”

“…Therefore, because optical parameters can characterize the biogeochemistry of natural waters and its substances (particle concentration, composition, and size distribution; Babin, Morel, Fournier‐Sicre, Fell, & Stramski, ; Twardowski et al, ; Boss, Twardowski, & Herring, ; Boss et al, ), optical measurements have the potential to quantify concentrations of chemical contaminants. Optically based chemical characterization methods have been shown to be effective for determining concentrations of unfiltered, filtered, and particulate mercury and methylmercury (Bergamaschi et al, , , ; Chang, Martin, Whitehead, Jones, & Spada, ), and unfiltered polychlorinated biphenyls (PCBs) (Chang et al, ).…”

“…OPTICS modelling involved partial least squares (PLS) statistical regression analysis of time series data derived from in situ physical, optical, and water quality measurements in conjunction with periodic discrete manually sampled water column chemical concentration data to derive hourly time series records of chemical contaminant variability following methods described by Bergamaschi et al () and Chang et al (; Figure ). Briefly, PLS analysis combines multiple linear regression and principal component analysis.…”