Capturing the Dynamics of Dissolved Organic Carbon (DOC) in Tidal Saltmarsh Estuaries Using Remote-Sensing-Informed Models

The fluxes of dissolved organic carbon (DOC) through tidal marsh-influenced estuaries remain poorly quantified and have been identified as a missing component in carbon-cycle models. The extreme variability inherent to these ecosystems of the land-ocean interface challenge our ability to capture DOC-concentration dynamics and to calculate accurate DOC fluxes. In situ discrete and continuous measurements provide high-quality estimates of DOC concentration, but these strategies are constrained spatially and temporally and can be costly to operate. Here, field measurements and high-spatial-resolution remote sensing were used to train and validate a predictive model of DOC-concentration distributions in the Plum Island Estuary (PIE), a mesotidal saltmarsh-influenced estuary in Massachusetts. A large set of field measurements collected between 2017 and 2023 was used to develop and validate an empirical algorithm to retrieve DOC concentration with a ±15% uncertainty from Sentinel-2 imagery. Implementation on 141 useable images produced a 6-year time series (2017–2023) of DOC distributions along the thalweg. Analysis of the time series helped identify river discharge, tidal water level (WL), and a marsh enhanced vegetation index 2 as predictors of DOC distribution in the estuary, and facilitated the training and validation of a simple model estimating the distribution. This simple model was able to predict DOC along the PIE thalweg within ±16% of the in situ measurements. Implementation for three years (2020–2022) illustrated how this type of remote-sensing-informed models can be coupled with the outputs hydrodynamic models to calculate DOC fluxes in tidal marsh-influenced estuaries and estimate DOC export to the coastal ocean.